TLA Line data Source code
1 : /*-------------------------------------------------------------------------
2 : *
3 : * xml.c
4 : * XML data type support.
5 : *
6 : *
7 : * Portions Copyright (c) 1996-2023, PostgreSQL Global Development Group
8 : * Portions Copyright (c) 1994, Regents of the University of California
9 : *
10 : * src/backend/utils/adt/xml.c
11 : *
12 : *-------------------------------------------------------------------------
13 : */
14 :
15 : /*
16 : * Generally, XML type support is only available when libxml use was
17 : * configured during the build. But even if that is not done, the
18 : * type and all the functions are available, but most of them will
19 : * fail. For one thing, this avoids having to manage variant catalog
20 : * installations. But it also has nice effects such as that you can
21 : * dump a database containing XML type data even if the server is not
22 : * linked with libxml. Thus, make sure xml_out() works even if nothing
23 : * else does.
24 : */
25 :
26 : /*
27 : * Notes on memory management:
28 : *
29 : * Sometimes libxml allocates global structures in the hope that it can reuse
30 : * them later on. This makes it impractical to change the xmlMemSetup
31 : * functions on-the-fly; that is likely to lead to trying to pfree() chunks
32 : * allocated with malloc() or vice versa. Since libxml might be used by
33 : * loadable modules, eg libperl, our only safe choices are to change the
34 : * functions at postmaster/backend launch or not at all. Since we'd rather
35 : * not activate libxml in sessions that might never use it, the latter choice
36 : * is the preferred one. However, for debugging purposes it can be awfully
37 : * handy to constrain libxml's allocations to be done in a specific palloc
38 : * context, where they're easy to track. Therefore there is code here that
39 : * can be enabled in debug builds to redirect libxml's allocations into a
40 : * special context LibxmlContext. It's not recommended to turn this on in
41 : * a production build because of the possibility of bad interactions with
42 : * external modules.
43 : */
44 : /* #define USE_LIBXMLCONTEXT */
45 :
46 : #include "postgres.h"
47 :
48 : #ifdef USE_LIBXML
49 : #include <libxml/chvalid.h>
50 : #include <libxml/parser.h>
51 : #include <libxml/parserInternals.h>
52 : #include <libxml/tree.h>
53 : #include <libxml/uri.h>
54 : #include <libxml/xmlerror.h>
55 : #include <libxml/xmlsave.h>
56 : #include <libxml/xmlversion.h>
57 : #include <libxml/xmlwriter.h>
58 : #include <libxml/xpath.h>
59 : #include <libxml/xpathInternals.h>
60 :
61 : /*
62 : * We used to check for xmlStructuredErrorContext via a configure test; but
63 : * that doesn't work on Windows, so instead use this grottier method of
64 : * testing the library version number.
65 : */
66 : #if LIBXML_VERSION >= 20704
67 : #define HAVE_XMLSTRUCTUREDERRORCONTEXT 1
68 : #endif
69 : #endif /* USE_LIBXML */
70 :
71 : #include "access/htup_details.h"
72 : #include "access/table.h"
73 : #include "catalog/namespace.h"
74 : #include "catalog/pg_class.h"
75 : #include "catalog/pg_type.h"
76 : #include "commands/dbcommands.h"
77 : #include "executor/spi.h"
78 : #include "executor/tablefunc.h"
79 : #include "fmgr.h"
80 : #include "lib/stringinfo.h"
81 : #include "libpq/pqformat.h"
82 : #include "mb/pg_wchar.h"
83 : #include "miscadmin.h"
84 : #include "nodes/execnodes.h"
85 : #include "nodes/miscnodes.h"
86 : #include "nodes/nodeFuncs.h"
87 : #include "utils/array.h"
88 : #include "utils/builtins.h"
89 : #include "utils/date.h"
90 : #include "utils/datetime.h"
91 : #include "utils/lsyscache.h"
92 : #include "utils/memutils.h"
93 : #include "utils/rel.h"
94 : #include "utils/syscache.h"
95 : #include "utils/xml.h"
96 :
97 :
98 : /* GUC variables */
99 : int xmlbinary = XMLBINARY_BASE64;
100 : int xmloption = XMLOPTION_CONTENT;
101 :
102 : #ifdef USE_LIBXML
103 :
104 : /* random number to identify PgXmlErrorContext */
105 : #define ERRCXT_MAGIC 68275028
106 :
107 : struct PgXmlErrorContext
108 : {
109 : int magic;
110 : /* strictness argument passed to pg_xml_init */
111 : PgXmlStrictness strictness;
112 : /* current error status and accumulated message, if any */
113 : bool err_occurred;
114 : StringInfoData err_buf;
115 : /* previous libxml error handling state (saved by pg_xml_init) */
116 : xmlStructuredErrorFunc saved_errfunc;
117 : void *saved_errcxt;
118 : /* previous libxml entity handler (saved by pg_xml_init) */
119 : xmlExternalEntityLoader saved_entityfunc;
120 : };
121 :
122 : static xmlParserInputPtr xmlPgEntityLoader(const char *URL, const char *ID,
123 : xmlParserCtxtPtr ctxt);
124 : static void xml_errsave(Node *escontext, PgXmlErrorContext *errcxt,
125 : int sqlcode, const char *msg);
126 : static void xml_errorHandler(void *data, xmlErrorPtr error);
127 : static int errdetail_for_xml_code(int code);
128 : static void chopStringInfoNewlines(StringInfo str);
129 : static void appendStringInfoLineSeparator(StringInfo str);
130 :
131 : #ifdef USE_LIBXMLCONTEXT
132 :
133 : static MemoryContext LibxmlContext = NULL;
134 :
135 : static void xml_memory_init(void);
136 : static void *xml_palloc(size_t size);
137 : static void *xml_repalloc(void *ptr, size_t size);
138 : static void xml_pfree(void *ptr);
139 : static char *xml_pstrdup(const char *string);
140 : #endif /* USE_LIBXMLCONTEXT */
141 :
142 : static xmlChar *xml_text2xmlChar(text *in);
143 : static int parse_xml_decl(const xmlChar *str, size_t *lenp,
144 : xmlChar **version, xmlChar **encoding, int *standalone);
145 : static bool print_xml_decl(StringInfo buf, const xmlChar *version,
146 : pg_enc encoding, int standalone);
147 : static bool xml_doctype_in_content(const xmlChar *str);
148 : static xmlDocPtr xml_parse(text *data, XmlOptionType xmloption_arg,
149 : bool preserve_whitespace, int encoding,
150 : XmlOptionType *parsed_xmloptiontype,
151 : xmlNodePtr *parsed_nodes,
152 : Node *escontext);
153 : static text *xml_xmlnodetoxmltype(xmlNodePtr cur, PgXmlErrorContext *xmlerrcxt);
154 : static int xml_xpathobjtoxmlarray(xmlXPathObjectPtr xpathobj,
155 : ArrayBuildState *astate,
156 : PgXmlErrorContext *xmlerrcxt);
157 : static xmlChar *pg_xmlCharStrndup(const char *str, size_t len);
158 : #endif /* USE_LIBXML */
159 :
160 : static void xmldata_root_element_start(StringInfo result, const char *eltname,
161 : const char *xmlschema, const char *targetns,
162 : bool top_level);
163 : static void xmldata_root_element_end(StringInfo result, const char *eltname);
164 : static StringInfo query_to_xml_internal(const char *query, char *tablename,
165 : const char *xmlschema, bool nulls, bool tableforest,
166 : const char *targetns, bool top_level);
167 : static const char *map_sql_table_to_xmlschema(TupleDesc tupdesc, Oid relid,
168 : bool nulls, bool tableforest, const char *targetns);
169 : static const char *map_sql_schema_to_xmlschema_types(Oid nspid,
170 : List *relid_list, bool nulls,
171 : bool tableforest, const char *targetns);
172 : static const char *map_sql_catalog_to_xmlschema_types(List *nspid_list,
173 : bool nulls, bool tableforest,
174 : const char *targetns);
175 : static const char *map_sql_type_to_xml_name(Oid typeoid, int typmod);
176 : static const char *map_sql_typecoll_to_xmlschema_types(List *tupdesc_list);
177 : static const char *map_sql_type_to_xmlschema_type(Oid typeoid, int typmod);
178 : static void SPI_sql_row_to_xmlelement(uint64 rownum, StringInfo result,
179 : char *tablename, bool nulls, bool tableforest,
180 : const char *targetns, bool top_level);
181 :
182 : /* XMLTABLE support */
183 : #ifdef USE_LIBXML
184 : /* random number to identify XmlTableContext */
185 : #define XMLTABLE_CONTEXT_MAGIC 46922182
186 : typedef struct XmlTableBuilderData
187 : {
188 : int magic;
189 : int natts;
190 : long int row_count;
191 : PgXmlErrorContext *xmlerrcxt;
192 : xmlParserCtxtPtr ctxt;
193 : xmlDocPtr doc;
194 : xmlXPathContextPtr xpathcxt;
195 : xmlXPathCompExprPtr xpathcomp;
196 : xmlXPathObjectPtr xpathobj;
197 : xmlXPathCompExprPtr *xpathscomp;
198 : } XmlTableBuilderData;
199 : #endif
200 :
201 : static void XmlTableInitOpaque(struct TableFuncScanState *state, int natts);
202 : static void XmlTableSetDocument(struct TableFuncScanState *state, Datum value);
203 : static void XmlTableSetNamespace(struct TableFuncScanState *state, const char *name,
204 : const char *uri);
205 : static void XmlTableSetRowFilter(struct TableFuncScanState *state, const char *path);
206 : static void XmlTableSetColumnFilter(struct TableFuncScanState *state,
207 : const char *path, int colnum);
208 : static bool XmlTableFetchRow(struct TableFuncScanState *state);
209 : static Datum XmlTableGetValue(struct TableFuncScanState *state, int colnum,
210 : Oid typid, int32 typmod, bool *isnull);
211 : static void XmlTableDestroyOpaque(struct TableFuncScanState *state);
212 :
213 : const TableFuncRoutine XmlTableRoutine =
214 : {
215 : XmlTableInitOpaque,
216 : XmlTableSetDocument,
217 : XmlTableSetNamespace,
218 : XmlTableSetRowFilter,
219 : XmlTableSetColumnFilter,
220 : XmlTableFetchRow,
221 : XmlTableGetValue,
222 : XmlTableDestroyOpaque
223 : };
224 :
225 : #define NO_XML_SUPPORT() \
226 : ereport(ERROR, \
227 : (errcode(ERRCODE_FEATURE_NOT_SUPPORTED), \
228 : errmsg("unsupported XML feature"), \
229 : errdetail("This functionality requires the server to be built with libxml support.")))
230 :
231 :
232 : /* from SQL/XML:2008 section 4.9 */
233 : #define NAMESPACE_XSD "http://www.w3.org/2001/XMLSchema"
234 : #define NAMESPACE_XSI "http://www.w3.org/2001/XMLSchema-instance"
235 : #define NAMESPACE_SQLXML "http://standards.iso.org/iso/9075/2003/sqlxml"
236 :
237 :
238 : #ifdef USE_LIBXML
239 :
240 : static int
241 UIC 0 : xmlChar_to_encoding(const xmlChar *encoding_name)
242 : {
243 0 : int encoding = pg_char_to_encoding((const char *) encoding_name);
244 :
245 0 : if (encoding < 0)
246 0 : ereport(ERROR,
247 EUB : (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
248 : errmsg("invalid encoding name \"%s\"",
249 : (const char *) encoding_name)));
250 UIC 0 : return encoding;
251 EUB : }
252 : #endif
253 :
254 :
255 : /*
256 : * xml_in uses a plain C string to VARDATA conversion, so for the time being
257 : * we use the conversion function for the text datatype.
258 : *
259 : * This is only acceptable so long as xmltype and text use the same
260 : * representation.
261 : */
262 : Datum
263 GIC 411 : xml_in(PG_FUNCTION_ARGS)
264 : {
265 : #ifdef USE_LIBXML
266 411 : char *s = PG_GETARG_CSTRING(0);
267 : xmltype *vardata;
268 : xmlDocPtr doc;
269 ECB :
270 : /* Build the result object. */
271 GIC 411 : vardata = (xmltype *) cstring_to_text(s);
272 :
273 ECB : /*
274 : * Parse the data to check if it is well-formed XML data.
275 : *
276 : * Note: we don't need to worry about whether a soft error is detected.
277 : */
278 GNC 411 : doc = xml_parse(vardata, xmloption, true, GetDatabaseEncoding(),
279 411 : NULL, NULL, fcinfo->context);
280 387 : if (doc != NULL)
281 381 : xmlFreeDoc(doc);
282 :
283 GIC 387 : PG_RETURN_XML_P(vardata);
284 : #else
285 : NO_XML_SUPPORT();
286 : return 0;
287 : #endif
288 ECB : }
289 :
290 :
291 : #define PG_XML_DEFAULT_VERSION "1.0"
292 :
293 :
294 : /*
295 : * xml_out_internal uses a plain VARDATA to C string conversion, so for the
296 : * time being we use the conversion function for the text datatype.
297 : *
298 : * This is only acceptable so long as xmltype and text use the same
299 : * representation.
300 : */
301 : static char *
302 GIC 11386 : xml_out_internal(xmltype *x, pg_enc target_encoding)
303 : {
304 11386 : char *str = text_to_cstring((text *) x);
305 :
306 : #ifdef USE_LIBXML
307 11386 : size_t len = strlen(str);
308 : xmlChar *version;
309 : int standalone;
310 : int res_code;
311 :
312 CBC 11386 : if ((res_code = parse_xml_decl((xmlChar *) str,
313 : &len, &version, NULL, &standalone)) == 0)
314 ECB : {
315 : StringInfoData buf;
316 :
317 CBC 11386 : initStringInfo(&buf);
318 :
319 GIC 11386 : if (!print_xml_decl(&buf, version, target_encoding, standalone))
320 : {
321 : /*
322 ECB : * If we are not going to produce an XML declaration, eat a single
323 : * newline in the original string to prevent empty first lines in
324 : * the output.
325 : */
326 GIC 11362 : if (*(str + len) == '\n')
327 CBC 5 : len += 1;
328 : }
329 11386 : appendStringInfoString(&buf, str + len);
330 :
331 GIC 11386 : pfree(str);
332 :
333 11386 : return buf.data;
334 : }
335 :
336 UNC 0 : ereport(WARNING,
337 : errcode(ERRCODE_INTERNAL_ERROR),
338 : errmsg_internal("could not parse XML declaration in stored value"),
339 : errdetail_for_xml_code(res_code));
340 ECB : #endif
341 UIC 0 : return str;
342 ECB : }
343 :
344 :
345 : Datum
346 GIC 11254 : xml_out(PG_FUNCTION_ARGS)
347 EUB : {
348 GIC 11254 : xmltype *x = PG_GETARG_XML_P(0);
349 :
350 : /*
351 : * xml_out removes the encoding property in all cases. This is because we
352 EUB : * cannot control from here whether the datum will be converted to a
353 : * different client encoding, so we'd do more harm than good by including
354 : * it.
355 : */
356 GIC 11254 : PG_RETURN_CSTRING(xml_out_internal(x, 0));
357 ECB : }
358 :
359 :
360 : Datum
361 UIC 0 : xml_recv(PG_FUNCTION_ARGS)
362 : {
363 : #ifdef USE_LIBXML
364 0 : StringInfo buf = (StringInfo) PG_GETARG_POINTER(0);
365 : xmltype *result;
366 : char *str;
367 ECB : char *newstr;
368 : int nbytes;
369 : xmlDocPtr doc;
370 UIC 0 : xmlChar *encodingStr = NULL;
371 : int encoding;
372 EUB :
373 : /*
374 : * Read the data in raw format. We don't know yet what the encoding is, as
375 : * that information is embedded in the xml declaration; so we have to
376 : * parse that before converting to server encoding.
377 : */
378 UIC 0 : nbytes = buf->len - buf->cursor;
379 0 : str = (char *) pq_getmsgbytes(buf, nbytes);
380 :
381 EUB : /*
382 : * We need a null-terminated string to pass to parse_xml_decl(). Rather
383 : * than make a separate copy, make the temporary result one byte bigger
384 : * than it needs to be.
385 : */
386 UIC 0 : result = palloc(nbytes + 1 + VARHDRSZ);
387 0 : SET_VARSIZE(result, nbytes + VARHDRSZ);
388 0 : memcpy(VARDATA(result), str, nbytes);
389 UBC 0 : str = VARDATA(result);
390 0 : str[nbytes] = '\0';
391 :
392 UIC 0 : parse_xml_decl((const xmlChar *) str, NULL, NULL, &encodingStr, NULL);
393 :
394 : /*
395 : * If encoding wasn't explicitly specified in the XML header, treat it as
396 : * UTF-8, as that's the default in XML. This is different from xml_in(),
397 EUB : * where the input has to go through the normal client to server encoding
398 : * conversion.
399 : */
400 UBC 0 : encoding = encodingStr ? xmlChar_to_encoding(encodingStr) : PG_UTF8;
401 EUB :
402 : /*
403 : * Parse the data to check if it is well-formed XML data. Assume that
404 : * xml_parse will throw ERROR if not.
405 : */
406 UNC 0 : doc = xml_parse(result, xmloption, true, encoding, NULL, NULL, NULL);
407 UIC 0 : xmlFreeDoc(doc);
408 :
409 : /* Now that we know what we're dealing with, convert to server encoding */
410 0 : newstr = pg_any_to_server(str, nbytes, encoding);
411 EUB :
412 UIC 0 : if (newstr != str)
413 : {
414 0 : pfree(result);
415 0 : result = (xmltype *) cstring_to_text(newstr);
416 0 : pfree(newstr);
417 EUB : }
418 :
419 UIC 0 : PG_RETURN_XML_P(result);
420 : #else
421 EUB : NO_XML_SUPPORT();
422 : return 0;
423 : #endif
424 : }
425 :
426 :
427 : Datum
428 UIC 0 : xml_send(PG_FUNCTION_ARGS)
429 : {
430 UBC 0 : xmltype *x = PG_GETARG_XML_P(0);
431 : char *outval;
432 : StringInfoData buf;
433 :
434 : /*
435 : * xml_out_internal doesn't convert the encoding, it just prints the right
436 : * declaration. pq_sendtext will do the conversion.
437 : */
438 UIC 0 : outval = xml_out_internal(x, pg_get_client_encoding());
439 EUB :
440 UIC 0 : pq_begintypsend(&buf);
441 UBC 0 : pq_sendtext(&buf, outval, strlen(outval));
442 UIC 0 : pfree(outval);
443 0 : PG_RETURN_BYTEA_P(pq_endtypsend(&buf));
444 : }
445 :
446 :
447 : #ifdef USE_LIBXML
448 : static void
449 GBC 66 : appendStringInfoText(StringInfo str, const text *t)
450 : {
451 66 : appendBinaryStringInfo(str, VARDATA_ANY(t), VARSIZE_ANY_EXHDR(t));
452 66 : }
453 EUB : #endif
454 :
455 :
456 : static xmltype *
457 GIC 10889 : stringinfo_to_xmltype(StringInfo buf)
458 : {
459 10889 : return (xmltype *) cstring_to_text_with_len(buf->data, buf->len);
460 ECB : }
461 :
462 :
463 : static xmltype *
464 GIC 39 : cstring_to_xmltype(const char *string)
465 : {
466 39 : return (xmltype *) cstring_to_text(string);
467 : }
468 ECB :
469 :
470 : #ifdef USE_LIBXML
471 : static xmltype *
472 GIC 10950 : xmlBuffer_to_xmltype(xmlBufferPtr buf)
473 : {
474 10950 : return (xmltype *) cstring_to_text_with_len((const char *) xmlBufferContent(buf),
475 ECB : xmlBufferLength(buf));
476 : }
477 : #endif
478 :
479 :
480 : Datum
481 GIC 21 : xmlcomment(PG_FUNCTION_ARGS)
482 : {
483 ECB : #ifdef USE_LIBXML
484 GIC 21 : text *arg = PG_GETARG_TEXT_PP(0);
485 CBC 21 : char *argdata = VARDATA_ANY(arg);
486 GIC 21 : int len = VARSIZE_ANY_EXHDR(arg);
487 : StringInfoData buf;
488 : int i;
489 :
490 : /* check for "--" in string or "-" at the end */
491 90 : for (i = 1; i < len; i++)
492 ECB : {
493 GIC 72 : if (argdata[i] == '-' && argdata[i - 1] == '-')
494 3 : ereport(ERROR,
495 ECB : (errcode(ERRCODE_INVALID_XML_COMMENT),
496 : errmsg("invalid XML comment")));
497 : }
498 GIC 18 : if (len > 0 && argdata[len - 1] == '-')
499 3 : ereport(ERROR,
500 : (errcode(ERRCODE_INVALID_XML_COMMENT),
501 : errmsg("invalid XML comment")));
502 ECB :
503 GIC 15 : initStringInfo(&buf);
504 CBC 15 : appendStringInfoString(&buf, "<!--");
505 15 : appendStringInfoText(&buf, arg);
506 GIC 15 : appendStringInfoString(&buf, "-->");
507 :
508 15 : PG_RETURN_XML_P(stringinfo_to_xmltype(&buf));
509 ECB : #else
510 : NO_XML_SUPPORT();
511 : return 0;
512 : #endif
513 : }
514 :
515 :
516 :
517 : /*
518 : * TODO: xmlconcat needs to merge the notations and unparsed entities
519 : * of the argument values. Not very important in practice, though.
520 : */
521 : xmltype *
522 GIC 10764 : xmlconcat(List *args)
523 : {
524 : #ifdef USE_LIBXML
525 10764 : int global_standalone = 1;
526 10764 : xmlChar *global_version = NULL;
527 10764 : bool global_version_no_value = false;
528 : StringInfoData buf;
529 : ListCell *v;
530 :
531 10764 : initStringInfo(&buf);
532 32295 : foreach(v, args)
533 ECB : {
534 GIC 21531 : xmltype *x = DatumGetXmlP(PointerGetDatum(lfirst(v)));
535 : size_t len;
536 ECB : xmlChar *version;
537 : int standalone;
538 : char *str;
539 :
540 GIC 21531 : len = VARSIZE(x) - VARHDRSZ;
541 21531 : str = text_to_cstring((text *) x);
542 ECB :
543 CBC 21531 : parse_xml_decl((xmlChar *) str, &len, &version, NULL, &standalone);
544 :
545 21531 : if (standalone == 0 && global_standalone == 1)
546 UIC 0 : global_standalone = 0;
547 GIC 21531 : if (standalone < 0)
548 21525 : global_standalone = -1;
549 :
550 21531 : if (!version)
551 CBC 21522 : global_version_no_value = true;
552 9 : else if (!global_version)
553 GIC 6 : global_version = version;
554 CBC 3 : else if (xmlStrcmp(version, global_version) != 0)
555 UIC 0 : global_version_no_value = true;
556 ECB :
557 GBC 21531 : appendStringInfoString(&buf, str + len);
558 CBC 21531 : pfree(str);
559 ECB : }
560 :
561 CBC 10764 : if (!global_version_no_value || global_standalone >= 0)
562 ECB : {
563 : StringInfoData buf2;
564 :
565 CBC 3 : initStringInfo(&buf2);
566 EUB :
567 GIC 3 : print_xml_decl(&buf2,
568 CBC 3 : (!global_version_no_value) ? global_version : NULL,
569 ECB : 0,
570 : global_standalone);
571 :
572 CBC 3 : appendBinaryStringInfo(&buf2, buf.data, buf.len);
573 GIC 3 : buf = buf2;
574 : }
575 :
576 CBC 10764 : return stringinfo_to_xmltype(&buf);
577 : #else
578 ECB : NO_XML_SUPPORT();
579 : return NULL;
580 : #endif
581 : }
582 :
583 :
584 : /*
585 : * XMLAGG support
586 : */
587 : Datum
588 GIC 10752 : xmlconcat2(PG_FUNCTION_ARGS)
589 : {
590 10752 : if (PG_ARGISNULL(0))
591 : {
592 9 : if (PG_ARGISNULL(1))
593 UIC 0 : PG_RETURN_NULL();
594 : else
595 GIC 9 : PG_RETURN_XML_P(PG_GETARG_XML_P(1));
596 : }
597 10743 : else if (PG_ARGISNULL(1))
598 UIC 0 : PG_RETURN_XML_P(PG_GETARG_XML_P(0));
599 ECB : else
600 GIC 10743 : PG_RETURN_XML_P(xmlconcat(list_make2(PG_GETARG_XML_P(0),
601 ECB : PG_GETARG_XML_P(1))));
602 : }
603 :
604 EUB :
605 : Datum
606 CBC 5 : texttoxml(PG_FUNCTION_ARGS)
607 : {
608 5 : text *data = PG_GETARG_TEXT_PP(0);
609 EUB :
610 GIC 5 : PG_RETURN_XML_P(xmlparse(data, xmloption, true));
611 ECB : }
612 :
613 :
614 : Datum
615 UIC 0 : xmltotext(PG_FUNCTION_ARGS)
616 : {
617 LBC 0 : xmltype *data = PG_GETARG_XML_P(0);
618 :
619 ECB : /* It's actually binary compatible. */
620 UIC 0 : PG_RETURN_TEXT_P((text *) data);
621 ECB : }
622 :
623 :
624 : text *
625 GNC 84 : xmltotext_with_options(xmltype *data, XmlOptionType xmloption_arg, bool indent)
626 EUB : {
627 : #ifdef USE_LIBXML
628 : text *volatile result;
629 : xmlDocPtr doc;
630 : XmlOptionType parsed_xmloptiontype;
631 : xmlNodePtr content_nodes;
632 GNC 84 : volatile xmlBufferPtr buf = NULL;
633 84 : volatile xmlSaveCtxtPtr ctxt = NULL;
634 84 : ErrorSaveContext escontext = {T_ErrorSaveContext};
635 : PgXmlErrorContext *xmlerrcxt;
636 : #endif
637 :
638 84 : if (xmloption_arg != XMLOPTION_DOCUMENT && !indent)
639 : {
640 : /*
641 : * We don't actually need to do anything, so just return the
642 : * binary-compatible input. For backwards-compatibility reasons,
643 : * allow such cases to succeed even without USE_LIBXML.
644 : */
645 18 : return (text *) data;
646 : }
647 :
648 : #ifdef USE_LIBXML
649 : /* Parse the input according to the xmloption */
650 66 : doc = xml_parse(data, xmloption_arg, true, GetDatabaseEncoding(),
651 : &parsed_xmloptiontype, &content_nodes,
652 : (Node *) &escontext);
653 66 : if (doc == NULL || escontext.error_occurred)
654 : {
655 15 : if (doc)
656 UNC 0 : xmlFreeDoc(doc);
657 : /* A soft error must be failure to conform to XMLOPTION_DOCUMENT */
658 GBC 15 : ereport(ERROR,
659 : (errcode(ERRCODE_NOT_AN_XML_DOCUMENT),
660 : errmsg("not an XML document")));
661 : }
662 EUB :
663 : /* If we weren't asked to indent, we're done. */
664 GNC 51 : if (!indent)
665 : {
666 9 : xmlFreeDoc(doc);
667 9 : return (text *) data;
668 : }
669 :
670 : /* Otherwise, we gotta spin up some error handling. */
671 42 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
672 :
673 42 : PG_TRY();
674 : {
675 42 : size_t decl_len = 0;
676 :
677 : /* The serialized data will go into this buffer. */
678 42 : buf = xmlBufferCreate();
679 :
680 42 : if (buf == NULL || xmlerrcxt->err_occurred)
681 UNC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
682 : "could not allocate xmlBuffer");
683 :
684 : /* Detect whether there's an XML declaration */
685 GNC 42 : parse_xml_decl(xml_text2xmlChar(data), &decl_len, NULL, NULL, NULL);
686 :
687 : /*
688 : * Emit declaration only if the input had one. Note: some versions of
689 : * xmlSaveToBuffer leak memory if a non-null encoding argument is
690 : * passed, so don't do that. We don't want any encoding conversion
691 : * anyway.
692 : */
693 42 : if (decl_len == 0)
694 36 : ctxt = xmlSaveToBuffer(buf, NULL,
695 : XML_SAVE_NO_DECL | XML_SAVE_FORMAT);
696 : else
697 6 : ctxt = xmlSaveToBuffer(buf, NULL,
698 : XML_SAVE_FORMAT);
699 :
700 42 : if (ctxt == NULL || xmlerrcxt->err_occurred)
701 UNC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
702 : "could not allocate xmlSaveCtxt");
703 :
704 GNC 42 : if (parsed_xmloptiontype == XMLOPTION_DOCUMENT)
705 : {
706 : /* If it's a document, saving is easy. */
707 18 : if (xmlSaveDoc(ctxt, doc) == -1 || xmlerrcxt->err_occurred)
708 UNC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
709 : "could not save document to xmlBuffer");
710 : }
711 GNC 24 : else if (content_nodes != NULL)
712 : {
713 : /*
714 : * Deal with the case where we have non-singly-rooted XML.
715 : * libxml's dump functions don't work well for that without help.
716 : * We build a fake root node that serves as a container for the
717 : * content nodes, and then iterate over the nodes.
718 : */
719 : xmlNodePtr root;
720 : xmlNodePtr newline;
721 :
722 21 : root = xmlNewNode(NULL, (const xmlChar *) "content-root");
723 21 : if (root == NULL || xmlerrcxt->err_occurred)
724 UNC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
725 : "could not allocate xml node");
726 :
727 : /* This attaches root to doc, so we need not free it separately. */
728 GNC 21 : xmlDocSetRootElement(doc, root);
729 21 : xmlAddChild(root, content_nodes);
730 :
731 : /*
732 : * We use this node to insert newlines in the dump. Note: in at
733 : * least some libxml versions, xmlNewDocText would not attach the
734 : * node to the document even if we passed it. Therefore, manage
735 : * freeing of this node manually, and pass NULL here to make sure
736 : * there's not a dangling link.
737 : */
738 21 : newline = xmlNewDocText(NULL, (const xmlChar *) "\n");
739 21 : if (newline == NULL || xmlerrcxt->err_occurred)
740 UNC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
741 : "could not allocate xml node");
742 :
743 GNC 54 : for (xmlNodePtr node = root->children; node; node = node->next)
744 : {
745 : /* insert newlines between nodes */
746 33 : if (node->type != XML_TEXT_NODE && node->prev != NULL)
747 : {
748 9 : if (xmlSaveTree(ctxt, newline) == -1 || xmlerrcxt->err_occurred)
749 : {
750 UNC 0 : xmlFreeNode(newline);
751 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
752 : "could not save newline to xmlBuffer");
753 : }
754 : }
755 :
756 GNC 33 : if (xmlSaveTree(ctxt, node) == -1 || xmlerrcxt->err_occurred)
757 : {
758 UNC 0 : xmlFreeNode(newline);
759 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
760 : "could not save content to xmlBuffer");
761 : }
762 : }
763 :
764 GNC 21 : xmlFreeNode(newline);
765 : }
766 :
767 42 : if (xmlSaveClose(ctxt) == -1 || xmlerrcxt->err_occurred)
768 : {
769 UNC 0 : ctxt = NULL; /* don't try to close it again */
770 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
771 : "could not close xmlSaveCtxtPtr");
772 : }
773 :
774 GNC 42 : result = (text *) xmlBuffer_to_xmltype(buf);
775 : }
776 UNC 0 : PG_CATCH();
777 : {
778 0 : if (ctxt)
779 0 : xmlSaveClose(ctxt);
780 0 : if (buf)
781 0 : xmlBufferFree(buf);
782 0 : if (doc)
783 0 : xmlFreeDoc(doc);
784 :
785 0 : pg_xml_done(xmlerrcxt, true);
786 :
787 0 : PG_RE_THROW();
788 : }
789 GNC 42 : PG_END_TRY();
790 :
791 42 : xmlBufferFree(buf);
792 42 : xmlFreeDoc(doc);
793 :
794 42 : pg_xml_done(xmlerrcxt, false);
795 :
796 42 : return result;
797 : #else
798 : NO_XML_SUPPORT();
799 : return NULL;
800 : #endif
801 : }
802 :
803 ECB :
804 : xmltype *
805 GIC 10830 : xmlelement(XmlExpr *xexpr,
806 : Datum *named_argvalue, bool *named_argnull,
807 : Datum *argvalue, bool *argnull)
808 : {
809 : #ifdef USE_LIBXML
810 ECB : xmltype *result;
811 : List *named_arg_strings;
812 : List *arg_strings;
813 : int i;
814 : ListCell *arg;
815 : ListCell *narg;
816 : PgXmlErrorContext *xmlerrcxt;
817 GIC 10830 : volatile xmlBufferPtr buf = NULL;
818 10830 : volatile xmlTextWriterPtr writer = NULL;
819 :
820 : /*
821 : * All arguments are already evaluated, and their values are passed in the
822 : * named_argvalue/named_argnull or argvalue/argnull arrays. This avoids
823 ECB : * issues if one of the arguments involves a call to some other function
824 : * or subsystem that wants to use libxml on its own terms. We examine the
825 : * original XmlExpr to identify the numbers and types of the arguments.
826 : */
827 GIC 10830 : named_arg_strings = NIL;
828 CBC 10830 : i = 0;
829 GIC 10854 : foreach(arg, xexpr->named_args)
830 : {
831 CBC 27 : Expr *e = (Expr *) lfirst(arg);
832 : char *str;
833 ECB :
834 GBC 27 : if (named_argnull[i])
835 UIC 0 : str = NULL;
836 ECB : else
837 GIC 27 : str = map_sql_value_to_xml_value(named_argvalue[i],
838 : exprType((Node *) e),
839 : false);
840 24 : named_arg_strings = lappend(named_arg_strings, str);
841 24 : i++;
842 ECB : }
843 :
844 CBC 10827 : arg_strings = NIL;
845 10827 : i = 0;
846 GIC 21642 : foreach(arg, xexpr->args)
847 : {
848 10815 : Expr *e = (Expr *) lfirst(arg);
849 ECB : char *str;
850 :
851 : /* here we can just forget NULL elements immediately */
852 GIC 10815 : if (!argnull[i])
853 ECB : {
854 GIC 10815 : str = map_sql_value_to_xml_value(argvalue[i],
855 : exprType((Node *) e),
856 ECB : true);
857 GIC 10815 : arg_strings = lappend(arg_strings, str);
858 ECB : }
859 GBC 10815 : i++;
860 : }
861 :
862 GIC 10827 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
863 ECB :
864 GIC 10827 : PG_TRY();
865 : {
866 10827 : buf = xmlBufferCreate();
867 10827 : if (buf == NULL || xmlerrcxt->err_occurred)
868 UIC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
869 : "could not allocate xmlBuffer");
870 GIC 10827 : writer = xmlNewTextWriterMemory(buf, 0);
871 CBC 10827 : if (writer == NULL || xmlerrcxt->err_occurred)
872 LBC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
873 : "could not allocate xmlTextWriter");
874 :
875 CBC 10827 : xmlTextWriterStartElement(writer, (xmlChar *) xexpr->name);
876 :
877 GIC 10851 : forboth(arg, named_arg_strings, narg, xexpr->arg_names)
878 ECB : {
879 GBC 24 : char *str = (char *) lfirst(arg);
880 GIC 24 : char *argname = strVal(lfirst(narg));
881 :
882 CBC 24 : if (str)
883 GIC 24 : xmlTextWriterWriteAttribute(writer,
884 : (xmlChar *) argname,
885 ECB : (xmlChar *) str);
886 EUB : }
887 :
888 GIC 21642 : foreach(arg, arg_strings)
889 ECB : {
890 GIC 10815 : char *str = (char *) lfirst(arg);
891 :
892 10815 : xmlTextWriterWriteRaw(writer, (xmlChar *) str);
893 : }
894 :
895 10827 : xmlTextWriterEndElement(writer);
896 :
897 : /* we MUST do this now to flush data out to the buffer ... */
898 10827 : xmlFreeTextWriter(writer);
899 10827 : writer = NULL;
900 ECB :
901 CBC 10827 : result = xmlBuffer_to_xmltype(buf);
902 EUB : }
903 UIC 0 : PG_CATCH();
904 : {
905 0 : if (writer)
906 LBC 0 : xmlFreeTextWriter(writer);
907 0 : if (buf)
908 UIC 0 : xmlBufferFree(buf);
909 :
910 0 : pg_xml_done(xmlerrcxt, true);
911 :
912 0 : PG_RE_THROW();
913 : }
914 GIC 10827 : PG_END_TRY();
915 :
916 CBC 10827 : xmlBufferFree(buf);
917 ECB :
918 GBC 10827 : pg_xml_done(xmlerrcxt, false);
919 :
920 GIC 10827 : return result;
921 ECB : #else
922 : NO_XML_SUPPORT();
923 : return NULL;
924 : #endif
925 : }
926 :
927 :
928 EUB : xmltype *
929 GBC 71 : xmlparse(text *data, XmlOptionType xmloption_arg, bool preserve_whitespace)
930 : {
931 : #ifdef USE_LIBXML
932 : xmlDocPtr doc;
933 :
934 CBC 71 : doc = xml_parse(data, xmloption_arg, preserve_whitespace,
935 : GetDatabaseEncoding(), NULL, NULL, NULL);
936 GBC 47 : xmlFreeDoc(doc);
937 EUB :
938 GIC 47 : return (xmltype *) data;
939 : #else
940 : NO_XML_SUPPORT();
941 : return NULL;
942 ECB : #endif
943 : }
944 :
945 :
946 : xmltype *
947 GBC 36 : xmlpi(const char *target, text *arg, bool arg_is_null, bool *result_is_null)
948 EUB : {
949 : #ifdef USE_LIBXML
950 : xmltype *result;
951 : StringInfoData buf;
952 ECB :
953 GIC 36 : if (pg_strcasecmp(target, "xml") == 0)
954 GBC 6 : ereport(ERROR,
955 : (errcode(ERRCODE_SYNTAX_ERROR), /* really */
956 EUB : errmsg("invalid XML processing instruction"),
957 : errdetail("XML processing instruction target name cannot be \"%s\".", target)));
958 :
959 : /*
960 : * Following the SQL standard, the null check comes after the syntax check
961 : * above.
962 : */
963 GBC 30 : *result_is_null = arg_is_null;
964 GIC 30 : if (*result_is_null)
965 GBC 6 : return NULL;
966 :
967 CBC 24 : initStringInfo(&buf);
968 :
969 24 : appendStringInfo(&buf, "<?%s", target);
970 ECB :
971 GIC 24 : if (arg != NULL)
972 ECB : {
973 : char *string;
974 :
975 GIC 12 : string = text_to_cstring(arg);
976 12 : if (strstr(string, "?>") != NULL)
977 3 : ereport(ERROR,
978 : (errcode(ERRCODE_INVALID_XML_PROCESSING_INSTRUCTION),
979 : errmsg("invalid XML processing instruction"),
980 : errdetail("XML processing instruction cannot contain \"?>\".")));
981 :
982 9 : appendStringInfoChar(&buf, ' ');
983 CBC 9 : appendStringInfoString(&buf, string + strspn(string, " "));
984 GIC 9 : pfree(string);
985 : }
986 21 : appendStringInfoString(&buf, "?>");
987 :
988 21 : result = stringinfo_to_xmltype(&buf);
989 21 : pfree(buf.data);
990 21 : return result;
991 : #else
992 : NO_XML_SUPPORT();
993 : return NULL;
994 : #endif
995 ECB : }
996 :
997 :
998 : xmltype *
999 GIC 30 : xmlroot(xmltype *data, text *version, int standalone)
1000 : {
1001 : #ifdef USE_LIBXML
1002 : char *str;
1003 : size_t len;
1004 : xmlChar *orig_version;
1005 ECB : int orig_standalone;
1006 : StringInfoData buf;
1007 :
1008 GIC 30 : len = VARSIZE(data) - VARHDRSZ;
1009 CBC 30 : str = text_to_cstring((text *) data);
1010 :
1011 GIC 30 : parse_xml_decl((xmlChar *) str, &len, &orig_version, NULL, &orig_standalone);
1012 ECB :
1013 GBC 30 : if (version)
1014 GIC 12 : orig_version = xml_text2xmlChar(version);
1015 ECB : else
1016 GIC 18 : orig_version = NULL;
1017 :
1018 CBC 30 : switch (standalone)
1019 ECB : {
1020 GIC 9 : case XML_STANDALONE_YES:
1021 9 : orig_standalone = 1;
1022 CBC 9 : break;
1023 6 : case XML_STANDALONE_NO:
1024 6 : orig_standalone = 0;
1025 GIC 6 : break;
1026 CBC 6 : case XML_STANDALONE_NO_VALUE:
1027 GIC 6 : orig_standalone = -1;
1028 6 : break;
1029 9 : case XML_STANDALONE_OMITTED:
1030 ECB : /* leave original value */
1031 GIC 9 : break;
1032 ECB : }
1033 :
1034 GIC 30 : initStringInfo(&buf);
1035 CBC 30 : print_xml_decl(&buf, orig_version, 0, orig_standalone);
1036 GIC 30 : appendStringInfoString(&buf, str + len);
1037 ECB :
1038 GIC 30 : return stringinfo_to_xmltype(&buf);
1039 : #else
1040 ECB : NO_XML_SUPPORT();
1041 : return NULL;
1042 : #endif
1043 : }
1044 :
1045 :
1046 EUB : /*
1047 : * Validate document (given as string) against DTD (given as external link)
1048 ECB : *
1049 : * This has been removed because it is a security hole: unprivileged users
1050 EUB : * should not be able to use Postgres to fetch arbitrary external files,
1051 : * which unfortunately is exactly what libxml is willing to do with the DTD
1052 : * parameter.
1053 ECB : */
1054 : Datum
1055 LBC 0 : xmlvalidate(PG_FUNCTION_ARGS)
1056 : {
1057 0 : ereport(ERROR,
1058 ECB : (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
1059 : errmsg("xmlvalidate is not implemented")));
1060 : return 0;
1061 : }
1062 :
1063 :
1064 : bool
1065 GIC 12 : xml_is_document(xmltype *arg)
1066 ECB : {
1067 : #ifdef USE_LIBXML
1068 : xmlDocPtr doc;
1069 GNC 12 : ErrorSaveContext escontext = {T_ErrorSaveContext};
1070 ECB :
1071 : /*
1072 : * We'll report "true" if no soft error is reported by xml_parse().
1073 : */
1074 GNC 12 : doc = xml_parse((text *) arg, XMLOPTION_DOCUMENT, true,
1075 : GetDatabaseEncoding(), NULL, NULL, (Node *) &escontext);
1076 GIC 12 : if (doc)
1077 6 : xmlFreeDoc(doc);
1078 :
1079 GNC 12 : return !escontext.error_occurred;
1080 : #else /* not USE_LIBXML */
1081 : NO_XML_SUPPORT();
1082 : return false;
1083 : #endif /* not USE_LIBXML */
1084 ECB : }
1085 :
1086 :
1087 : #ifdef USE_LIBXML
1088 :
1089 : /*
1090 : * pg_xml_init_library --- set up for use of libxml
1091 : *
1092 : * This should be called by each function that is about to use libxml
1093 : * facilities but doesn't require error handling. It initializes libxml
1094 : * and verifies compatibility with the loaded libxml version. These are
1095 : * once-per-session activities.
1096 : *
1097 : * TODO: xmlChar is utf8-char, make proper tuning (initdb with enc!=utf8 and
1098 : * check)
1099 : */
1100 : void
1101 GIC 45661 : pg_xml_init_library(void)
1102 ECB : {
1103 : static bool first_time = true;
1104 :
1105 GIC 45661 : if (first_time)
1106 : {
1107 : /* Stuff we need do only once per session */
1108 ECB :
1109 : /*
1110 : * Currently, we have no pure UTF-8 support for internals -- check if
1111 : * we can work.
1112 : */
1113 : if (sizeof(char) != sizeof(xmlChar))
1114 : ereport(ERROR,
1115 : (errmsg("could not initialize XML library"),
1116 : errdetail("libxml2 has incompatible char type: sizeof(char)=%zu, sizeof(xmlChar)=%zu.",
1117 : sizeof(char), sizeof(xmlChar))));
1118 :
1119 : #ifdef USE_LIBXMLCONTEXT
1120 : /* Set up libxml's memory allocation our way */
1121 : xml_memory_init();
1122 : #endif
1123 :
1124 : /* Check library compatibility */
1125 GIC 13 : LIBXML_TEST_VERSION;
1126 ECB :
1127 GIC 13 : first_time = false;
1128 : }
1129 45661 : }
1130 ECB :
1131 : /*
1132 : * pg_xml_init --- set up for use of libxml and register an error handler
1133 : *
1134 : * This should be called by each function that is about to use libxml
1135 : * facilities and requires error handling. It initializes libxml with
1136 : * pg_xml_init_library() and establishes our libxml error handler.
1137 : *
1138 : * strictness determines which errors are reported and which are ignored.
1139 : *
1140 : * Calls to this function MUST be followed by a PG_TRY block that guarantees
1141 : * that pg_xml_done() is called during either normal or error exit.
1142 : *
1143 : * This is exported for use by contrib/xml2, as well as other code that might
1144 : * wish to share use of this module's libxml error handler.
1145 : */
1146 : PgXmlErrorContext *
1147 GIC 11923 : pg_xml_init(PgXmlStrictness strictness)
1148 : {
1149 : PgXmlErrorContext *errcxt;
1150 : void *new_errcxt;
1151 :
1152 : /* Do one-time setup if needed */
1153 11923 : pg_xml_init_library();
1154 ECB :
1155 : /* Create error handling context structure */
1156 GIC 11923 : errcxt = (PgXmlErrorContext *) palloc(sizeof(PgXmlErrorContext));
1157 11923 : errcxt->magic = ERRCXT_MAGIC;
1158 11923 : errcxt->strictness = strictness;
1159 11923 : errcxt->err_occurred = false;
1160 11923 : initStringInfo(&errcxt->err_buf);
1161 :
1162 : /*
1163 ECB : * Save original error handler and install ours. libxml originally didn't
1164 : * distinguish between the contexts for generic and for structured error
1165 : * handlers. If we're using an old libxml version, we must thus save the
1166 : * generic error context, even though we're using a structured error
1167 : * handler.
1168 : */
1169 CBC 11923 : errcxt->saved_errfunc = xmlStructuredError;
1170 :
1171 ECB : #ifdef HAVE_XMLSTRUCTUREDERRORCONTEXT
1172 GIC 11923 : errcxt->saved_errcxt = xmlStructuredErrorContext;
1173 ECB : #else
1174 : errcxt->saved_errcxt = xmlGenericErrorContext;
1175 : #endif
1176 :
1177 CBC 11923 : xmlSetStructuredErrorFunc((void *) errcxt, xml_errorHandler);
1178 ECB :
1179 : /*
1180 : * Verify that xmlSetStructuredErrorFunc set the context variable we
1181 : * expected it to. If not, the error context pointer we just saved is not
1182 : * the correct thing to restore, and since that leaves us without a way to
1183 : * restore the context in pg_xml_done, we must fail.
1184 : *
1185 : * The only known situation in which this test fails is if we compile with
1186 : * headers from a libxml2 that doesn't track the structured error context
1187 : * separately (< 2.7.4), but at runtime use a version that does, or vice
1188 : * versa. The libxml2 authors did not treat that change as constituting
1189 : * an ABI break, so the LIBXML_TEST_VERSION test in pg_xml_init_library
1190 : * fails to protect us from this.
1191 : */
1192 :
1193 : #ifdef HAVE_XMLSTRUCTUREDERRORCONTEXT
1194 GIC 11923 : new_errcxt = xmlStructuredErrorContext;
1195 : #else
1196 : new_errcxt = xmlGenericErrorContext;
1197 : #endif
1198 :
1199 11923 : if (new_errcxt != (void *) errcxt)
1200 UIC 0 : ereport(ERROR,
1201 : (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
1202 : errmsg("could not set up XML error handler"),
1203 : errhint("This probably indicates that the version of libxml2"
1204 : " being used is not compatible with the libxml2"
1205 : " header files that PostgreSQL was built with.")));
1206 :
1207 : /*
1208 : * Also, install an entity loader to prevent unwanted fetches of external
1209 : * files and URLs.
1210 EUB : */
1211 GIC 11923 : errcxt->saved_entityfunc = xmlGetExternalEntityLoader();
1212 GBC 11923 : xmlSetExternalEntityLoader(xmlPgEntityLoader);
1213 :
1214 GIC 11923 : return errcxt;
1215 : }
1216 :
1217 :
1218 : /*
1219 : * pg_xml_done --- restore previous libxml error handling
1220 ECB : *
1221 : * Resets libxml's global error-handling state to what it was before
1222 : * pg_xml_init() was called.
1223 : *
1224 : * This routine verifies that all pending errors have been dealt with
1225 : * (in assert-enabled builds, anyway).
1226 : */
1227 : void
1228 GIC 11923 : pg_xml_done(PgXmlErrorContext *errcxt, bool isError)
1229 ECB : {
1230 : void *cur_errcxt;
1231 :
1232 : /* An assert seems like enough protection here */
1233 GIC 11923 : Assert(errcxt->magic == ERRCXT_MAGIC);
1234 ECB :
1235 : /*
1236 : * In a normal exit, there should be no un-handled libxml errors. But we
1237 : * shouldn't try to enforce this during error recovery, since the longjmp
1238 : * could have been thrown before xml_ereport had a chance to run.
1239 : */
1240 GIC 11923 : Assert(!errcxt->err_occurred || isError);
1241 :
1242 : /*
1243 : * Check that libxml's global state is correct, warn if not. This is a
1244 : * real test and not an Assert because it has a higher probability of
1245 : * happening.
1246 : */
1247 : #ifdef HAVE_XMLSTRUCTUREDERRORCONTEXT
1248 11923 : cur_errcxt = xmlStructuredErrorContext;
1249 : #else
1250 : cur_errcxt = xmlGenericErrorContext;
1251 : #endif
1252 :
1253 11923 : if (cur_errcxt != (void *) errcxt)
1254 UIC 0 : elog(WARNING, "libxml error handling state is out of sync with xml.c");
1255 :
1256 ECB : /* Restore the saved handlers */
1257 GIC 11923 : xmlSetStructuredErrorFunc(errcxt->saved_errcxt, errcxt->saved_errfunc);
1258 11923 : xmlSetExternalEntityLoader(errcxt->saved_entityfunc);
1259 :
1260 ECB : /*
1261 : * Mark the struct as invalid, just in case somebody somehow manages to
1262 : * call xml_errorHandler or xml_ereport with it.
1263 : */
1264 GIC 11923 : errcxt->magic = 0;
1265 :
1266 : /* Release memory */
1267 11923 : pfree(errcxt->err_buf.data);
1268 11923 : pfree(errcxt);
1269 11923 : }
1270 :
1271 :
1272 : /*
1273 : * pg_xml_error_occurred() --- test the error flag
1274 : */
1275 : bool
1276 UIC 0 : pg_xml_error_occurred(PgXmlErrorContext *errcxt)
1277 : {
1278 0 : return errcxt->err_occurred;
1279 : }
1280 ECB :
1281 :
1282 : /*
1283 : * SQL/XML allows storing "XML documents" or "XML content". "XML
1284 : * documents" are specified by the XML specification and are parsed
1285 : * easily by libxml. "XML content" is specified by SQL/XML as the
1286 : * production "XMLDecl? content". But libxml can only parse the
1287 : * "content" part, so we have to parse the XML declaration ourselves
1288 : * to complete this.
1289 : */
1290 :
1291 : #define CHECK_XML_SPACE(p) \
1292 : do { \
1293 : if (!xmlIsBlank_ch(*(p))) \
1294 : return XML_ERR_SPACE_REQUIRED; \
1295 : } while (0)
1296 :
1297 : #define SKIP_XML_SPACE(p) \
1298 : while (xmlIsBlank_ch(*(p))) (p)++
1299 :
1300 : /* Letter | Digit | '.' | '-' | '_' | ':' | CombiningChar | Extender */
1301 : /* Beware of multiple evaluations of argument! */
1302 : #define PG_XMLISNAMECHAR(c) \
1303 : (xmlIsBaseChar_ch(c) || xmlIsIdeographicQ(c) \
1304 : || xmlIsDigit_ch(c) \
1305 : || c == '.' || c == '-' || c == '_' || c == ':' \
1306 : || xmlIsCombiningQ(c) \
1307 : || xmlIsExtender_ch(c))
1308 :
1309 : /* pnstrdup, but deal with xmlChar not char; len is measured in xmlChars */
1310 : static xmlChar *
1311 CBC 100 : xml_pnstrdup(const xmlChar *str, size_t len)
1312 ECB : {
1313 : xmlChar *result;
1314 :
1315 CBC 100 : result = (xmlChar *) palloc((len + 1) * sizeof(xmlChar));
1316 GIC 100 : memcpy(result, str, len * sizeof(xmlChar));
1317 100 : result[len] = 0;
1318 100 : return result;
1319 : }
1320 :
1321 : /* Ditto, except input is char* */
1322 : static xmlChar *
1323 1212 : pg_xmlCharStrndup(const char *str, size_t len)
1324 ECB : {
1325 : xmlChar *result;
1326 :
1327 CBC 1212 : result = (xmlChar *) palloc((len + 1) * sizeof(xmlChar));
1328 GIC 1212 : memcpy(result, str, len);
1329 1212 : result[len] = '\0';
1330 :
1331 1212 : return result;
1332 ECB : }
1333 :
1334 : /*
1335 : * Copy xmlChar string to PostgreSQL-owned memory, freeing the input.
1336 : *
1337 : * The input xmlChar is freed regardless of success of the copy.
1338 : */
1339 : static char *
1340 GIC 54293 : xml_pstrdup_and_free(xmlChar *str)
1341 : {
1342 : char *result;
1343 :
1344 54293 : if (str)
1345 : {
1346 54293 : PG_TRY();
1347 : {
1348 54293 : result = pstrdup((char *) str);
1349 ECB : }
1350 UIC 0 : PG_FINALLY();
1351 : {
1352 GIC 54293 : xmlFree(str);
1353 : }
1354 CBC 54293 : PG_END_TRY();
1355 EUB : }
1356 : else
1357 UIC 0 : result = NULL;
1358 :
1359 GIC 54293 : return result;
1360 : }
1361 :
1362 : /*
1363 : * str is the null-terminated input string. Remaining arguments are
1364 : * output arguments; each can be NULL if value is not wanted.
1365 : * version and encoding are returned as locally-palloc'd strings.
1366 ECB : * Result is 0 if OK, an error code if not.
1367 : */
1368 : static int
1369 CBC 33738 : parse_xml_decl(const xmlChar *str, size_t *lenp,
1370 : xmlChar **version, xmlChar **encoding, int *standalone)
1371 : {
1372 : const xmlChar *p;
1373 : const xmlChar *save_p;
1374 : size_t len;
1375 : int utf8char;
1376 : int utf8len;
1377 :
1378 : /*
1379 : * Only initialize libxml. We don't need error handling here, but we do
1380 : * need to make sure libxml is initialized before calling any of its
1381 : * functions. Note that this is safe (and a no-op) if caller has already
1382 : * done pg_xml_init().
1383 ECB : */
1384 GIC 33738 : pg_xml_init_library();
1385 :
1386 : /* Initialize output arguments to "not present" */
1387 33738 : if (version)
1388 CBC 33420 : *version = NULL;
1389 GIC 33738 : if (encoding)
1390 UIC 0 : *encoding = NULL;
1391 GIC 33738 : if (standalone)
1392 33420 : *standalone = -1;
1393 :
1394 33738 : p = str;
1395 ECB :
1396 GIC 33738 : if (xmlStrncmp(p, (xmlChar *) "<?xml", 5) != 0)
1397 33623 : goto finished;
1398 :
1399 : /*
1400 : * If next char is a name char, it's a PI like <?xml-stylesheet ...?>
1401 : * rather than an XMLDecl, so we have done what we came to do and found no
1402 : * XMLDecl.
1403 ECB : *
1404 : * We need an input length value for xmlGetUTF8Char, but there's no need
1405 : * to count the whole document size, so use strnlen not strlen.
1406 : */
1407 GIC 115 : utf8len = strnlen((const char *) (p + 5), MAX_MULTIBYTE_CHAR_LEN);
1408 CBC 115 : utf8char = xmlGetUTF8Char(p + 5, &utf8len);
1409 GBC 115 : if (PG_XMLISNAMECHAR(utf8char))
1410 GIC 6 : goto finished;
1411 :
1412 CBC 109 : p += 5;
1413 ECB :
1414 : /* version */
1415 GIC 109 : CHECK_XML_SPACE(p);
1416 218 : SKIP_XML_SPACE(p);
1417 109 : if (xmlStrncmp(p, (xmlChar *) "version", 7) != 0)
1418 UIC 0 : return XML_ERR_VERSION_MISSING;
1419 CBC 109 : p += 7;
1420 GIC 109 : SKIP_XML_SPACE(p);
1421 109 : if (*p != '=')
1422 LBC 0 : return XML_ERR_VERSION_MISSING;
1423 CBC 109 : p += 1;
1424 109 : SKIP_XML_SPACE(p);
1425 :
1426 GIC 109 : if (*p == '\'' || *p == '"')
1427 109 : {
1428 : const xmlChar *q;
1429 :
1430 109 : q = xmlStrchr(p + 1, *p);
1431 GBC 109 : if (!q)
1432 UIC 0 : return XML_ERR_VERSION_MISSING;
1433 EUB :
1434 GIC 109 : if (version)
1435 100 : *version = xml_pnstrdup(p + 1, q - p - 1);
1436 109 : p = q + 1;
1437 : }
1438 : else
1439 UIC 0 : return XML_ERR_VERSION_MISSING;
1440 :
1441 : /* encoding */
1442 GIC 109 : save_p = p;
1443 190 : SKIP_XML_SPACE(p);
1444 109 : if (xmlStrncmp(p, (xmlChar *) "encoding", 8) == 0)
1445 : {
1446 27 : CHECK_XML_SPACE(save_p);
1447 27 : p += 8;
1448 27 : SKIP_XML_SPACE(p);
1449 27 : if (*p != '=')
1450 UIC 0 : return XML_ERR_MISSING_ENCODING;
1451 GIC 27 : p += 1;
1452 27 : SKIP_XML_SPACE(p);
1453 :
1454 27 : if (*p == '\'' || *p == '"')
1455 27 : {
1456 : const xmlChar *q;
1457 :
1458 27 : q = xmlStrchr(p + 1, *p);
1459 27 : if (!q)
1460 UIC 0 : return XML_ERR_MISSING_ENCODING;
1461 :
1462 GIC 27 : if (encoding)
1463 UIC 0 : *encoding = xml_pnstrdup(p + 1, q - p - 1);
1464 GIC 27 : p = q + 1;
1465 : }
1466 ECB : else
1467 UIC 0 : return XML_ERR_MISSING_ENCODING;
1468 : }
1469 : else
1470 ECB : {
1471 CBC 82 : p = save_p;
1472 ECB : }
1473 :
1474 : /* standalone */
1475 GIC 109 : save_p = p;
1476 163 : SKIP_XML_SPACE(p);
1477 109 : if (xmlStrncmp(p, (xmlChar *) "standalone", 10) == 0)
1478 ECB : {
1479 GIC 54 : CHECK_XML_SPACE(save_p);
1480 54 : p += 10;
1481 54 : SKIP_XML_SPACE(p);
1482 CBC 54 : if (*p != '=')
1483 LBC 0 : return XML_ERR_STANDALONE_VALUE;
1484 CBC 54 : p += 1;
1485 GIC 54 : SKIP_XML_SPACE(p);
1486 CBC 108 : if (xmlStrncmp(p, (xmlChar *) "'yes'", 5) == 0 ||
1487 GIC 54 : xmlStrncmp(p, (xmlChar *) "\"yes\"", 5) == 0)
1488 : {
1489 30 : if (standalone)
1490 30 : *standalone = 1;
1491 30 : p += 5;
1492 : }
1493 48 : else if (xmlStrncmp(p, (xmlChar *) "'no'", 4) == 0 ||
1494 24 : xmlStrncmp(p, (xmlChar *) "\"no\"", 4) == 0)
1495 ECB : {
1496 GIC 18 : if (standalone)
1497 18 : *standalone = 0;
1498 18 : p += 4;
1499 ECB : }
1500 : else
1501 CBC 6 : return XML_ERR_STANDALONE_VALUE;
1502 : }
1503 ECB : else
1504 : {
1505 GBC 55 : p = save_p;
1506 : }
1507 ECB :
1508 GIC 103 : SKIP_XML_SPACE(p);
1509 CBC 103 : if (xmlStrncmp(p, (xmlChar *) "?>", 2) != 0)
1510 UIC 0 : return XML_ERR_XMLDECL_NOT_FINISHED;
1511 GIC 103 : p += 2;
1512 EUB :
1513 GIC 33732 : finished:
1514 CBC 33732 : len = p - str;
1515 :
1516 GIC 37182 : for (p = str; p < str + len; p++)
1517 3450 : if (*p > 127)
1518 UIC 0 : return XML_ERR_INVALID_CHAR;
1519 :
1520 GIC 33732 : if (lenp)
1521 33732 : *lenp = len;
1522 :
1523 33732 : return XML_ERR_OK;
1524 ECB : }
1525 :
1526 :
1527 : /*
1528 : * Write an XML declaration. On output, we adjust the XML declaration
1529 : * as follows. (These rules are the moral equivalent of the clause
1530 : * "Serialization of an XML value" in the SQL standard.)
1531 : *
1532 : * We try to avoid generating an XML declaration if possible. This is
1533 : * so that you don't get trivial things like xml '<foo/>' resulting in
1534 : * '<?xml version="1.0"?><foo/>', which would surely be annoying. We
1535 : * must provide a declaration if the standalone property is specified
1536 : * or if we include an encoding declaration. If we have a
1537 : * declaration, we must specify a version (XML requires this).
1538 : * Otherwise we only make a declaration if the version is not "1.0",
1539 : * which is the default version specified in SQL:2003.
1540 : */
1541 : static bool
1542 CBC 11419 : print_xml_decl(StringInfo buf, const xmlChar *version,
1543 ECB : pg_enc encoding, int standalone)
1544 : {
1545 GBC 11419 : if ((version && strcmp((const char *) version, PG_XML_DEFAULT_VERSION) != 0)
1546 CBC 11401 : || (encoding && encoding != PG_UTF8)
1547 11401 : || standalone != -1)
1548 : {
1549 48 : appendStringInfoString(buf, "<?xml");
1550 :
1551 48 : if (version)
1552 36 : appendStringInfo(buf, " version=\"%s\"", version);
1553 : else
1554 GIC 12 : appendStringInfo(buf, " version=\"%s\"", PG_XML_DEFAULT_VERSION);
1555 :
1556 48 : if (encoding && encoding != PG_UTF8)
1557 : {
1558 : /*
1559 : * XXX might be useful to convert this to IANA names (ISO-8859-1
1560 : * instead of LATIN1 etc.); needs field experience
1561 : */
1562 LBC 0 : appendStringInfo(buf, " encoding=\"%s\"",
1563 ECB : pg_encoding_to_char(encoding));
1564 : }
1565 :
1566 GIC 48 : if (standalone == 1)
1567 CBC 24 : appendStringInfoString(buf, " standalone=\"yes\"");
1568 GIC 24 : else if (standalone == 0)
1569 12 : appendStringInfoString(buf, " standalone=\"no\"");
1570 CBC 48 : appendStringInfoString(buf, "?>");
1571 ECB :
1572 CBC 48 : return true;
1573 EUB : }
1574 ECB : else
1575 CBC 11371 : return false;
1576 ECB : }
1577 EUB :
1578 ECB : /*
1579 : * Test whether an input that is to be parsed as CONTENT contains a DTD.
1580 : *
1581 : * The SQL/XML:2003 definition of CONTENT ("XMLDecl? content") is not
1582 : * satisfied by a document with a DTD, which is a bit of a wart, as it means
1583 : * the CONTENT type is not a proper superset of DOCUMENT. SQL/XML:2006 and
1584 : * later fix that, by redefining content with reference to the "more
1585 : * permissive" Document Node of the XQuery/XPath Data Model, such that any
1586 : * DOCUMENT value is indeed also a CONTENT value. That definition is more
1587 EUB : * useful, as CONTENT becomes usable for parsing input of unknown form (think
1588 : * pg_restore).
1589 ECB : *
1590 : * As used below in parse_xml when parsing for CONTENT, libxml does not give
1591 : * us the 2006+ behavior, but only the 2003; it will choke if the input has
1592 : * a DTD. But we can provide the 2006+ definition of CONTENT easily enough,
1593 : * by detecting this case first and simply doing the parse as DOCUMENT.
1594 EUB : *
1595 : * A DTD can be found arbitrarily far in, but that would be a contrived case;
1596 : * it will ordinarily start within a few dozen characters. The only things
1597 ECB : * that can precede it are an XMLDecl (here, the caller will have called
1598 : * parse_xml_decl already), whitespace, comments, and processing instructions.
1599 : * This function need only return true if it sees a valid sequence of such
1600 : * things leading to <!DOCTYPE. It can simply return false in any other
1601 : * cases, including malformed input; that will mean the input gets parsed as
1602 : * CONTENT as originally planned, with libxml reporting any errors.
1603 : *
1604 : * This is only to be called from xml_parse, when pg_xml_init has already
1605 EUB : * been called. The input is already in UTF8 encoding.
1606 ECB : */
1607 : static bool
1608 GIC 467 : xml_doctype_in_content(const xmlChar *str)
1609 ECB : {
1610 CBC 467 : const xmlChar *p = str;
1611 :
1612 : for (;;)
1613 18 : {
1614 ECB : const xmlChar *e;
1615 EUB :
1616 GIC 532 : SKIP_XML_SPACE(p);
1617 CBC 485 : if (*p != '<')
1618 GBC 97 : return false;
1619 CBC 388 : p++;
1620 :
1621 GIC 388 : if (*p == '!')
1622 EUB : {
1623 GIC 36 : p++;
1624 :
1625 : /* if we see <!DOCTYPE, we can return true */
1626 CBC 36 : if (xmlStrncmp(p, (xmlChar *) "DOCTYPE", 7) == 0)
1627 GIC 21 : return true;
1628 :
1629 : /* otherwise, if it's not a comment, fail */
1630 CBC 15 : if (xmlStrncmp(p, (xmlChar *) "--", 2) != 0)
1631 LBC 0 : return false;
1632 ECB : /* find end of comment: find -- and a > must follow */
1633 GIC 15 : p = xmlStrstr(p + 2, (xmlChar *) "--");
1634 CBC 15 : if (!p || p[2] != '>')
1635 LBC 0 : return false;
1636 ECB : /* advance over comment, and keep scanning */
1637 CBC 15 : p += 3;
1638 GBC 15 : continue;
1639 ECB : }
1640 :
1641 : /* otherwise, if it's not a PI <?target something?>, fail */
1642 CBC 352 : if (*p != '?')
1643 GIC 349 : return false;
1644 CBC 3 : p++;
1645 ECB :
1646 : /* find end of PI (the string ?> is forbidden within a PI) */
1647 GIC 3 : e = xmlStrstr(p, (xmlChar *) "?>");
1648 CBC 3 : if (!e)
1649 LBC 0 : return false;
1650 :
1651 ECB : /* advance over PI, keep scanning */
1652 CBC 3 : p = e + 2;
1653 ECB : }
1654 : }
1655 :
1656 :
1657 : /*
1658 : * Convert a text object to XML internal representation
1659 : *
1660 : * data is the source data (must not be toasted!), encoding is its encoding,
1661 : * and xmloption_arg and preserve_whitespace are options for the
1662 : * transformation.
1663 : *
1664 : * If parsed_xmloptiontype isn't NULL, *parsed_xmloptiontype is set to the
1665 : * XmlOptionType actually used to parse the input (typically the same as
1666 : * xmloption_arg, but a DOCTYPE node in the input can force DOCUMENT mode).
1667 : *
1668 : * If parsed_nodes isn't NULL and the input is not an XML document, the list
1669 : * of parsed nodes from the xmlParseBalancedChunkMemory call will be returned
1670 : * to *parsed_nodes.
1671 : *
1672 : * Errors normally result in ereport(ERROR), but if escontext is an
1673 : * ErrorSaveContext, then "safe" errors are reported there instead, and the
1674 : * caller must check SOFT_ERROR_OCCURRED() to see whether that happened.
1675 : *
1676 : * Note: it is caller's responsibility to xmlFreeDoc() the result,
1677 : * else a permanent memory leak will ensue! But note the result could
1678 : * be NULL after a soft error.
1679 : *
1680 : * TODO maybe libxml2's xmlreader is better? (do not construct DOM,
1681 : * yet do not use SAX - see xmlreader.c)
1682 EUB : */
1683 ECB : static xmlDocPtr
1684 GNC 617 : xml_parse(text *data, XmlOptionType xmloption_arg,
1685 : bool preserve_whitespace, int encoding,
1686 : XmlOptionType *parsed_xmloptiontype, xmlNodePtr *parsed_nodes,
1687 : Node *escontext)
1688 ECB : {
1689 : int32 len;
1690 : xmlChar *string;
1691 : xmlChar *utf8string;
1692 EUB : PgXmlErrorContext *xmlerrcxt;
1693 GIC 617 : volatile xmlParserCtxtPtr ctxt = NULL;
1694 CBC 617 : volatile xmlDocPtr doc = NULL;
1695 ECB :
1696 : /*
1697 : * This step looks annoyingly redundant, but we must do it to have a
1698 : * null-terminated string in case encoding conversion isn't required.
1699 : */
1700 GIC 617 : len = VARSIZE_ANY_EXHDR(data); /* will be useful later */
1701 CBC 617 : string = xml_text2xmlChar(data);
1702 :
1703 : /*
1704 : * If the data isn't UTF8, we must translate before giving it to libxml.
1705 : *
1706 : * XXX ideally, we'd catch any encoding conversion failure and return a
1707 : * soft error. However, failure to convert to UTF8 should be pretty darn
1708 : * rare, so for now this is left undone.
1709 : */
1710 GIC 617 : utf8string = pg_do_encoding_conversion(string,
1711 : len,
1712 : encoding,
1713 : PG_UTF8);
1714 :
1715 : /* Start up libxml and its parser */
1716 617 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_WELLFORMED);
1717 :
1718 : /* Use a TRY block to ensure we clean up correctly */
1719 617 : PG_TRY();
1720 : {
1721 617 : bool parse_as_document = false;
1722 : int res_code;
1723 617 : size_t count = 0;
1724 617 : xmlChar *version = NULL;
1725 617 : int standalone = 0;
1726 :
1727 : /* Any errors here are reported as hard ereport's */
1728 CBC 617 : xmlInitParser();
1729 :
1730 GIC 617 : ctxt = xmlNewParserCtxt();
1731 CBC 617 : if (ctxt == NULL || xmlerrcxt->err_occurred)
1732 LBC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
1733 ECB : "could not allocate parser context");
1734 :
1735 : /* Decide whether to parse as document or content */
1736 GIC 617 : if (xmloption_arg == XMLOPTION_DOCUMENT)
1737 CBC 144 : parse_as_document = true;
1738 ECB : else
1739 : {
1740 : /* Parse and skip over the XML declaration, if any */
1741 GIC 473 : res_code = parse_xml_decl(utf8string,
1742 ECB : &count, &version, NULL, &standalone);
1743 GIC 473 : if (res_code != 0)
1744 : {
1745 GNC 6 : errsave(escontext,
1746 : errcode(ERRCODE_INVALID_XML_CONTENT),
1747 : errmsg_internal("invalid XML content: invalid XML declaration"),
1748 : errdetail_for_xml_code(res_code));
1749 6 : goto fail;
1750 : }
1751 :
1752 EUB : /* Is there a DOCTYPE element? */
1753 GIC 467 : if (xml_doctype_in_content(utf8string + count))
1754 21 : parse_as_document = true;
1755 : }
1756 ECB :
1757 : /* initialize output parameters */
1758 GNC 611 : if (parsed_xmloptiontype != NULL)
1759 66 : *parsed_xmloptiontype = parse_as_document ? XMLOPTION_DOCUMENT :
1760 : XMLOPTION_CONTENT;
1761 611 : if (parsed_nodes != NULL)
1762 66 : *parsed_nodes = NULL;
1763 :
1764 CBC 611 : if (parse_as_document)
1765 ECB : {
1766 : /*
1767 : * Note, that here we try to apply DTD defaults
1768 : * (XML_PARSE_DTDATTR) according to SQL/XML:2008 GR 10.16.7.d:
1769 : * 'Default values defined by internal DTD are applied'. As for
1770 : * external DTDs, we try to support them too, (see SQL/XML:2008 GR
1771 : * 10.16.7.e)
1772 : */
1773 GIC 165 : doc = xmlCtxtReadDoc(ctxt, utf8string,
1774 : NULL,
1775 : "UTF-8",
1776 : XML_PARSE_NOENT | XML_PARSE_DTDATTR
1777 : | (preserve_whitespace ? 0 : XML_PARSE_NOBLANKS));
1778 165 : if (doc == NULL || xmlerrcxt->err_occurred)
1779 : {
1780 : /* Use original option to decide which error code to report */
1781 72 : if (xmloption_arg == XMLOPTION_DOCUMENT)
1782 GNC 69 : xml_errsave(escontext, xmlerrcxt,
1783 : ERRCODE_INVALID_XML_DOCUMENT,
1784 : "invalid XML document");
1785 : else
1786 3 : xml_errsave(escontext, xmlerrcxt,
1787 : ERRCODE_INVALID_XML_CONTENT,
1788 : "invalid XML content");
1789 48 : goto fail;
1790 : }
1791 : }
1792 : else
1793 : {
1794 GIC 446 : doc = xmlNewDoc(version);
1795 GNC 446 : if (doc == NULL || xmlerrcxt->err_occurred)
1796 UNC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
1797 : "could not allocate XML document");
1798 :
1799 GIC 446 : Assert(doc->encoding == NULL);
1800 446 : doc->encoding = xmlStrdup((const xmlChar *) "UTF-8");
1801 GNC 446 : if (doc->encoding == NULL || xmlerrcxt->err_occurred)
1802 UNC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
1803 : "could not allocate XML document");
1804 GIC 446 : doc->standalone = standalone;
1805 :
1806 : /* allow empty content */
1807 446 : if (*(utf8string + count))
1808 : {
1809 868 : res_code = xmlParseBalancedChunkMemory(doc, NULL, NULL, 0,
1810 GNC 434 : utf8string + count,
1811 : parsed_nodes);
1812 GIC 434 : if (res_code != 0 || xmlerrcxt->err_occurred)
1813 : {
1814 GNC 30 : xml_errsave(escontext, xmlerrcxt,
1815 : ERRCODE_INVALID_XML_CONTENT,
1816 : "invalid XML content");
1817 6 : goto fail;
1818 : }
1819 : }
1820 ECB : }
1821 :
1822 GNC 569 : fail:
1823 : ;
1824 : }
1825 CBC 48 : PG_CATCH();
1826 : {
1827 GIC 48 : if (doc != NULL)
1828 CBC 24 : xmlFreeDoc(doc);
1829 GIC 48 : if (ctxt != NULL)
1830 48 : xmlFreeParserCtxt(ctxt);
1831 ECB :
1832 CBC 48 : pg_xml_done(xmlerrcxt, true);
1833 ECB :
1834 CBC 48 : PG_RE_THROW();
1835 : }
1836 569 : PG_END_TRY();
1837 :
1838 569 : xmlFreeParserCtxt(ctxt);
1839 :
1840 GIC 569 : pg_xml_done(xmlerrcxt, false);
1841 ECB :
1842 CBC 569 : return doc;
1843 : }
1844 :
1845 ECB :
1846 EUB : /*
1847 : * xmlChar<->text conversions
1848 ECB : */
1849 : static xmlChar *
1850 GBC 671 : xml_text2xmlChar(text *in)
1851 : {
1852 CBC 671 : return (xmlChar *) text_to_cstring(in);
1853 ECB : }
1854 :
1855 :
1856 : #ifdef USE_LIBXMLCONTEXT
1857 :
1858 : /*
1859 : * Manage the special context used for all libxml allocations (but only
1860 : * in special debug builds; see notes at top of file)
1861 : */
1862 : static void
1863 : xml_memory_init(void)
1864 EUB : {
1865 : /* Create memory context if not there already */
1866 : if (LibxmlContext == NULL)
1867 ECB : LibxmlContext = AllocSetContextCreate(TopMemoryContext,
1868 : "Libxml context",
1869 : ALLOCSET_DEFAULT_SIZES);
1870 :
1871 : /* Re-establish the callbacks even if already set */
1872 : xmlMemSetup(xml_pfree, xml_palloc, xml_repalloc, xml_pstrdup);
1873 : }
1874 :
1875 : /*
1876 : * Wrappers for memory management functions
1877 : */
1878 : static void *
1879 : xml_palloc(size_t size)
1880 : {
1881 : return MemoryContextAlloc(LibxmlContext, size);
1882 : }
1883 :
1884 :
1885 : static void *
1886 : xml_repalloc(void *ptr, size_t size)
1887 : {
1888 : return repalloc(ptr, size);
1889 : }
1890 :
1891 :
1892 : static void
1893 : xml_pfree(void *ptr)
1894 : {
1895 : /* At least some parts of libxml assume xmlFree(NULL) is allowed */
1896 : if (ptr)
1897 : pfree(ptr);
1898 : }
1899 :
1900 :
1901 : static char *
1902 : xml_pstrdup(const char *string)
1903 : {
1904 : return MemoryContextStrdup(LibxmlContext, string);
1905 : }
1906 : #endif /* USE_LIBXMLCONTEXT */
1907 :
1908 :
1909 : /*
1910 : * xmlPgEntityLoader --- entity loader callback function
1911 : *
1912 : * Silently prevent any external entity URL from being loaded. We don't want
1913 : * to throw an error, so instead make the entity appear to expand to an empty
1914 : * string.
1915 : *
1916 : * We would prefer to allow loading entities that exist in the system's
1917 : * global XML catalog; but the available libxml2 APIs make that a complex
1918 : * and fragile task. For now, just shut down all external access.
1919 : */
1920 : static xmlParserInputPtr
1921 GIC 9 : xmlPgEntityLoader(const char *URL, const char *ID,
1922 : xmlParserCtxtPtr ctxt)
1923 : {
1924 9 : return xmlNewStringInputStream(ctxt, (const xmlChar *) "");
1925 ECB : }
1926 :
1927 :
1928 : /*
1929 : * xml_ereport --- report an XML-related error
1930 : *
1931 : * The "msg" is the SQL-level message; some can be adopted from the SQL/XML
1932 : * standard. This function adds libxml's native error message, if any, as
1933 : * detail.
1934 : *
1935 : * This is exported for modules that want to share the core libxml error
1936 : * handler. Note that pg_xml_init() *must* have been called previously.
1937 : */
1938 : void
1939 CBC 7 : xml_ereport(PgXmlErrorContext *errcxt, int level, int sqlcode, const char *msg)
1940 ECB : {
1941 : char *detail;
1942 :
1943 : /* Defend against someone passing us a bogus context struct */
1944 GIC 7 : if (errcxt->magic != ERRCXT_MAGIC)
1945 LBC 0 : elog(ERROR, "xml_ereport called with invalid PgXmlErrorContext");
1946 ECB :
1947 EUB : /* Flag that the current libxml error has been reported */
1948 GIC 7 : errcxt->err_occurred = false;
1949 :
1950 : /* Include detail only if we have some text from libxml */
1951 CBC 7 : if (errcxt->err_buf.len > 0)
1952 6 : detail = errcxt->err_buf.data;
1953 : else
1954 GIC 1 : detail = NULL;
1955 :
1956 CBC 7 : ereport(level,
1957 : (errcode(sqlcode),
1958 ECB : errmsg_internal("%s", msg),
1959 : detail ? errdetail_internal("%s", detail) : 0));
1960 LBC 0 : }
1961 :
1962 :
1963 : /*
1964 : * xml_errsave --- save an XML-related error
1965 : *
1966 : * If escontext is an ErrorSaveContext, error details are saved into it,
1967 : * and control returns normally.
1968 : *
1969 : * Otherwise, the error is thrown, so that this is equivalent to
1970 : * xml_ereport() with level == ERROR.
1971 : *
1972 : * This should be used only for errors that we're sure we do not need
1973 : * a transaction abort to clean up after.
1974 : */
1975 : static void
1976 GNC 102 : xml_errsave(Node *escontext, PgXmlErrorContext *errcxt,
1977 : int sqlcode, const char *msg)
1978 : {
1979 : char *detail;
1980 :
1981 : /* Defend against someone passing us a bogus context struct */
1982 102 : if (errcxt->magic != ERRCXT_MAGIC)
1983 UNC 0 : elog(ERROR, "xml_errsave called with invalid PgXmlErrorContext");
1984 :
1985 : /* Flag that the current libxml error has been reported */
1986 GNC 102 : errcxt->err_occurred = false;
1987 :
1988 : /* Include detail only if we have some text from libxml */
1989 102 : if (errcxt->err_buf.len > 0)
1990 102 : detail = errcxt->err_buf.data;
1991 : else
1992 UNC 0 : detail = NULL;
1993 :
1994 GNC 102 : errsave(escontext,
1995 : (errcode(sqlcode),
1996 : errmsg_internal("%s", msg),
1997 : detail ? errdetail_internal("%s", detail) : 0));
1998 54 : }
1999 :
2000 :
2001 : /*
2002 ECB : * Error handler for libxml errors and warnings
2003 : */
2004 : static void
2005 GIC 199 : xml_errorHandler(void *data, xmlErrorPtr error)
2006 ECB : {
2007 CBC 199 : PgXmlErrorContext *xmlerrcxt = (PgXmlErrorContext *) data;
2008 GIC 199 : xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) error->ctxt;
2009 199 : xmlParserInputPtr input = (ctxt != NULL) ? ctxt->input : NULL;
2010 199 : xmlNodePtr node = error->node;
2011 CBC 199 : const xmlChar *name = (node != NULL &&
2012 199 : node->type == XML_ELEMENT_NODE) ? node->name : NULL;
2013 GIC 199 : int domain = error->domain;
2014 CBC 199 : int level = error->level;
2015 ECB : StringInfo errorBuf;
2016 :
2017 : /*
2018 : * Defend against someone passing us a bogus context struct.
2019 : *
2020 : * We force a backend exit if this check fails because longjmp'ing out of
2021 : * libxml would likely render it unsafe to use further.
2022 : */
2023 GIC 199 : if (xmlerrcxt->magic != ERRCXT_MAGIC)
2024 UIC 0 : elog(FATAL, "xml_errorHandler called with invalid PgXmlErrorContext");
2025 :
2026 ECB : /*----------
2027 : * Older libxml versions report some errors differently.
2028 : * First, some errors were previously reported as coming from the parser
2029 : * domain but are now reported as coming from the namespace domain.
2030 : * Second, some warnings were upgraded to errors.
2031 : * We attempt to compensate for that here.
2032 : *----------
2033 : */
2034 CBC 199 : switch (error->code)
2035 ECB : {
2036 GIC 15 : case XML_WAR_NS_URI:
2037 15 : level = XML_ERR_ERROR;
2038 15 : domain = XML_FROM_NAMESPACE;
2039 CBC 15 : break;
2040 :
2041 GIC 27 : case XML_ERR_NS_DECL_ERROR:
2042 ECB : case XML_WAR_NS_URI_RELATIVE:
2043 : case XML_WAR_NS_COLUMN:
2044 : case XML_NS_ERR_XML_NAMESPACE:
2045 : case XML_NS_ERR_UNDEFINED_NAMESPACE:
2046 : case XML_NS_ERR_QNAME:
2047 : case XML_NS_ERR_ATTRIBUTE_REDEFINED:
2048 : case XML_NS_ERR_EMPTY:
2049 GBC 27 : domain = XML_FROM_NAMESPACE;
2050 GIC 27 : break;
2051 : }
2052 ECB :
2053 : /* Decide whether to act on the error or not */
2054 CBC 199 : switch (domain)
2055 EUB : {
2056 GIC 157 : case XML_FROM_PARSER:
2057 ECB : case XML_FROM_NONE:
2058 : case XML_FROM_MEMORY:
2059 : case XML_FROM_IO:
2060 :
2061 : /*
2062 : * Suppress warnings about undeclared entities. We need to do
2063 : * this to avoid problems due to not loading DTD definitions.
2064 : */
2065 CBC 157 : if (error->code == XML_WAR_UNDECLARED_ENTITY)
2066 GIC 3 : return;
2067 ECB :
2068 : /* Otherwise, accept error regardless of the parsing purpose */
2069 GIC 154 : break;
2070 ECB :
2071 GIC 42 : default:
2072 : /* Ignore error if only doing well-formedness check */
2073 42 : if (xmlerrcxt->strictness == PG_XML_STRICTNESS_WELLFORMED)
2074 33 : return;
2075 CBC 9 : break;
2076 : }
2077 :
2078 ECB : /* Prepare error message in errorBuf */
2079 GIC 163 : errorBuf = makeStringInfo();
2080 ECB :
2081 CBC 163 : if (error->line > 0)
2082 163 : appendStringInfo(errorBuf, "line %d: ", error->line);
2083 163 : if (name != NULL)
2084 UIC 0 : appendStringInfo(errorBuf, "element %s: ", name);
2085 CBC 163 : if (error->message != NULL)
2086 GIC 163 : appendStringInfoString(errorBuf, error->message);
2087 ECB : else
2088 UIC 0 : appendStringInfoString(errorBuf, "(no message provided)");
2089 ECB :
2090 : /*
2091 : * Append context information to errorBuf.
2092 : *
2093 : * xmlParserPrintFileContext() uses libxml's "generic" error handler to
2094 : * write the context. Since we don't want to duplicate libxml
2095 : * functionality here, we set up a generic error handler temporarily.
2096 : *
2097 : * We use appendStringInfo() directly as libxml's generic error handler.
2098 : * This should work because it has essentially the same signature as
2099 : * libxml expects, namely (void *ptr, const char *msg, ...).
2100 : */
2101 GIC 163 : if (input != NULL)
2102 : {
2103 CBC 163 : xmlGenericErrorFunc errFuncSaved = xmlGenericError;
2104 GIC 163 : void *errCtxSaved = xmlGenericErrorContext;
2105 ECB :
2106 GIC 163 : xmlSetGenericErrorFunc((void *) errorBuf,
2107 : (xmlGenericErrorFunc) appendStringInfo);
2108 :
2109 : /* Add context information to errorBuf */
2110 163 : appendStringInfoLineSeparator(errorBuf);
2111 :
2112 163 : xmlParserPrintFileContext(input);
2113 :
2114 : /* Restore generic error func */
2115 163 : xmlSetGenericErrorFunc(errCtxSaved, errFuncSaved);
2116 : }
2117 :
2118 : /* Get rid of any trailing newlines in errorBuf */
2119 163 : chopStringInfoNewlines(errorBuf);
2120 :
2121 : /*
2122 : * Legacy error handling mode. err_occurred is never set, we just add the
2123 : * message to err_buf. This mode exists because the xml2 contrib module
2124 : * uses our error-handling infrastructure, but we don't want to change its
2125 : * behaviour since it's deprecated anyway. This is also why we don't
2126 : * distinguish between notices, warnings and errors here --- the old-style
2127 : * generic error handler wouldn't have done that either.
2128 : */
2129 163 : if (xmlerrcxt->strictness == PG_XML_STRICTNESS_LEGACY)
2130 : {
2131 1 : appendStringInfoLineSeparator(&xmlerrcxt->err_buf);
2132 1 : appendBinaryStringInfo(&xmlerrcxt->err_buf, errorBuf->data,
2133 : errorBuf->len);
2134 :
2135 1 : pfree(errorBuf->data);
2136 1 : pfree(errorBuf);
2137 1 : return;
2138 : }
2139 :
2140 : /*
2141 : * We don't want to ereport() here because that'd probably leave libxml in
2142 : * an inconsistent state. Instead, we remember the error and ereport()
2143 : * from xml_ereport().
2144 : *
2145 : * Warnings and notices can be reported immediately since they won't cause
2146 : * a longjmp() out of libxml.
2147 : */
2148 162 : if (level >= XML_ERR_ERROR)
2149 : {
2150 159 : appendStringInfoLineSeparator(&xmlerrcxt->err_buf);
2151 159 : appendBinaryStringInfo(&xmlerrcxt->err_buf, errorBuf->data,
2152 : errorBuf->len);
2153 :
2154 159 : xmlerrcxt->err_occurred = true;
2155 : }
2156 3 : else if (level >= XML_ERR_WARNING)
2157 : {
2158 3 : ereport(WARNING,
2159 : (errmsg_internal("%s", errorBuf->data)));
2160 : }
2161 : else
2162 : {
2163 UIC 0 : ereport(NOTICE,
2164 : (errmsg_internal("%s", errorBuf->data)));
2165 : }
2166 :
2167 GIC 162 : pfree(errorBuf->data);
2168 162 : pfree(errorBuf);
2169 : }
2170 :
2171 :
2172 : /*
2173 : * Convert libxml error codes into textual errdetail messages.
2174 : *
2175 : * This should be called within an ereport or errsave invocation,
2176 : * just as errdetail would be.
2177 : *
2178 : * At the moment, we only need to cover those codes that we
2179 ECB : * may raise in this file.
2180 : */
2181 : static int
2182 GNC 3 : errdetail_for_xml_code(int code)
2183 : {
2184 : const char *det;
2185 :
2186 GIC 3 : switch (code)
2187 : {
2188 UIC 0 : case XML_ERR_INVALID_CHAR:
2189 0 : det = gettext_noop("Invalid character value.");
2190 0 : break;
2191 0 : case XML_ERR_SPACE_REQUIRED:
2192 0 : det = gettext_noop("Space required.");
2193 LBC 0 : break;
2194 GIC 3 : case XML_ERR_STANDALONE_VALUE:
2195 3 : det = gettext_noop("standalone accepts only 'yes' or 'no'.");
2196 3 : break;
2197 UIC 0 : case XML_ERR_VERSION_MISSING:
2198 LBC 0 : det = gettext_noop("Malformed declaration: missing version.");
2199 UBC 0 : break;
2200 UIC 0 : case XML_ERR_MISSING_ENCODING:
2201 0 : det = gettext_noop("Missing encoding in text declaration.");
2202 LBC 0 : break;
2203 UIC 0 : case XML_ERR_XMLDECL_NOT_FINISHED:
2204 0 : det = gettext_noop("Parsing XML declaration: '?>' expected.");
2205 LBC 0 : break;
2206 0 : default:
2207 UIC 0 : det = gettext_noop("Unrecognized libxml error code: %d.");
2208 LBC 0 : break;
2209 : }
2210 ECB :
2211 GNC 3 : return errdetail(det, code);
2212 : }
2213 :
2214 :
2215 : /*
2216 : * Remove all trailing newlines from a StringInfo string
2217 : */
2218 : static void
2219 GIC 486 : chopStringInfoNewlines(StringInfo str)
2220 : {
2221 812 : while (str->len > 0 && str->data[str->len - 1] == '\n')
2222 326 : str->data[--str->len] = '\0';
2223 486 : }
2224 :
2225 :
2226 : /*
2227 ECB : * Append a newline after removing any existing trailing newlines
2228 : */
2229 : static void
2230 GIC 323 : appendStringInfoLineSeparator(StringInfo str)
2231 : {
2232 323 : chopStringInfoNewlines(str);
2233 CBC 323 : if (str->len > 0)
2234 GBC 214 : appendStringInfoChar(str, '\n');
2235 GIC 323 : }
2236 :
2237 ECB :
2238 : /*
2239 : * Convert one char in the current server encoding to a Unicode codepoint.
2240 : */
2241 : static pg_wchar
2242 GIC 9140 : sqlchar_to_unicode(const char *s)
2243 EUB : {
2244 : char *utf8string;
2245 ECB : pg_wchar ret[2]; /* need space for trailing zero */
2246 :
2247 : /* note we're not assuming s is null-terminated */
2248 GIC 9140 : utf8string = pg_server_to_any(s, pg_mblen(s), PG_UTF8);
2249 ECB :
2250 GIC 9140 : pg_encoding_mb2wchar_with_len(PG_UTF8, utf8string, ret,
2251 : pg_encoding_mblen(PG_UTF8, utf8string));
2252 :
2253 9140 : if (utf8string != s)
2254 UIC 0 : pfree(utf8string);
2255 :
2256 CBC 9140 : return ret[0];
2257 : }
2258 ECB :
2259 :
2260 : static bool
2261 CBC 1819 : is_valid_xml_namefirst(pg_wchar c)
2262 ECB : {
2263 : /* (Letter | '_' | ':') */
2264 CBC 1822 : return (xmlIsBaseCharQ(c) || xmlIsIdeographicQ(c)
2265 3641 : || c == '_' || c == ':');
2266 : }
2267 :
2268 :
2269 : static bool
2270 GIC 7321 : is_valid_xml_namechar(pg_wchar c)
2271 : {
2272 : /* Letter | Digit | '.' | '-' | '_' | ':' | CombiningChar | Extender */
2273 7766 : return (xmlIsBaseCharQ(c) || xmlIsIdeographicQ(c)
2274 CBC 445 : || xmlIsDigitQ(c)
2275 GBC 127 : || c == '.' || c == '-' || c == '_' || c == ':'
2276 GIC 6 : || xmlIsCombiningQ(c)
2277 15532 : || xmlIsExtenderQ(c));
2278 : }
2279 : #endif /* USE_LIBXML */
2280 :
2281 :
2282 : /*
2283 : * Map SQL identifier to XML name; see SQL/XML:2008 section 9.1.
2284 : */
2285 ECB : char *
2286 GIC 1826 : map_sql_identifier_to_xml_name(const char *ident, bool fully_escaped,
2287 ECB : bool escape_period)
2288 : {
2289 : #ifdef USE_LIBXML
2290 : StringInfoData buf;
2291 : const char *p;
2292 :
2293 : /*
2294 : * SQL/XML doesn't make use of this case anywhere, so it's probably a
2295 : * mistake.
2296 : */
2297 GIC 1826 : Assert(fully_escaped || !escape_period);
2298 :
2299 1826 : initStringInfo(&buf);
2300 ECB :
2301 CBC 10976 : for (p = ident; *p; p += pg_mblen(p))
2302 : {
2303 GIC 9150 : if (*p == ':' && (p == ident || fully_escaped))
2304 7 : appendStringInfoString(&buf, "_x003A_");
2305 CBC 9143 : else if (*p == '_' && *(p + 1) == 'x')
2306 GIC 3 : appendStringInfoString(&buf, "_x005F_");
2307 CBC 10784 : else if (fully_escaped && p == ident &&
2308 GIC 1644 : pg_strncasecmp(p, "xml", 3) == 0)
2309 : {
2310 UIC 0 : if (*p == 'x')
2311 0 : appendStringInfoString(&buf, "_x0078_");
2312 : else
2313 0 : appendStringInfoString(&buf, "_x0058_");
2314 : }
2315 GIC 9140 : else if (escape_period && *p == '.')
2316 LBC 0 : appendStringInfoString(&buf, "_x002E_");
2317 ECB : else
2318 : {
2319 GIC 9140 : pg_wchar u = sqlchar_to_unicode(p);
2320 ECB :
2321 GIC 18280 : if ((p == ident)
2322 CBC 1819 : ? !is_valid_xml_namefirst(u)
2323 GIC 7321 : : !is_valid_xml_namechar(u))
2324 CBC 9 : appendStringInfo(&buf, "_x%04X_", (unsigned int) u);
2325 ECB : else
2326 CBC 9131 : appendBinaryStringInfo(&buf, p, pg_mblen(p));
2327 : }
2328 : }
2329 :
2330 1826 : return buf.data;
2331 : #else /* not USE_LIBXML */
2332 ECB : NO_XML_SUPPORT();
2333 : return NULL;
2334 : #endif /* not USE_LIBXML */
2335 EUB : }
2336 ECB :
2337 :
2338 : /*
2339 EUB : * Map XML name to SQL identifier; see SQL/XML:2008 section 9.3.
2340 : */
2341 : char *
2342 GIC 64 : map_xml_name_to_sql_identifier(const char *name)
2343 : {
2344 : StringInfoData buf;
2345 : const char *p;
2346 :
2347 64 : initStringInfo(&buf);
2348 :
2349 352 : for (p = name; *p; p += pg_mblen(p))
2350 : {
2351 288 : if (*p == '_' && *(p + 1) == 'x'
2352 CBC 8 : && isxdigit((unsigned char) *(p + 2))
2353 GIC 8 : && isxdigit((unsigned char) *(p + 3))
2354 CBC 8 : && isxdigit((unsigned char) *(p + 4))
2355 8 : && isxdigit((unsigned char) *(p + 5))
2356 GIC 8 : && *(p + 6) == '_')
2357 CBC 8 : {
2358 : char cbuf[MAX_UNICODE_EQUIVALENT_STRING + 1];
2359 : unsigned int u;
2360 :
2361 8 : sscanf(p + 2, "%X", &u);
2362 GIC 8 : pg_unicode_to_server(u, (unsigned char *) cbuf);
2363 CBC 8 : appendStringInfoString(&buf, cbuf);
2364 GIC 8 : p += 6;
2365 : }
2366 ECB : else
2367 GIC 280 : appendBinaryStringInfo(&buf, p, pg_mblen(p));
2368 : }
2369 :
2370 CBC 64 : return buf.data;
2371 : }
2372 :
2373 : /*
2374 : * Map SQL value to XML value; see SQL/XML:2008 section 9.8.
2375 : *
2376 : * When xml_escape_strings is true, then certain characters in string
2377 : * values are replaced by entity references (< etc.), as specified
2378 : * in SQL/XML:2008 section 9.8 GR 9) a) iii). This is normally what is
2379 : * wanted. The false case is mainly useful when the resulting value
2380 ECB : * is used with xmlTextWriterWriteAttribute() to write out an
2381 : * attribute, because that function does the escaping itself.
2382 : */
2383 : char *
2384 GIC 64943 : map_sql_value_to_xml_value(Datum value, Oid type, bool xml_escape_strings)
2385 : {
2386 CBC 64943 : if (type_is_array_domain(type))
2387 ECB : {
2388 : ArrayType *array;
2389 : Oid elmtype;
2390 : int16 elmlen;
2391 : bool elmbyval;
2392 : char elmalign;
2393 : int num_elems;
2394 : Datum *elem_values;
2395 : bool *elem_nulls;
2396 : StringInfoData buf;
2397 : int i;
2398 :
2399 CBC 3 : array = DatumGetArrayTypeP(value);
2400 GIC 3 : elmtype = ARR_ELEMTYPE(array);
2401 CBC 3 : get_typlenbyvalalign(elmtype, &elmlen, &elmbyval, &elmalign);
2402 ECB :
2403 GIC 3 : deconstruct_array(array, elmtype,
2404 : elmlen, elmbyval, elmalign,
2405 ECB : &elem_values, &elem_nulls,
2406 : &num_elems);
2407 :
2408 GIC 3 : initStringInfo(&buf);
2409 ECB :
2410 GIC 12 : for (i = 0; i < num_elems; i++)
2411 : {
2412 9 : if (elem_nulls[i])
2413 UIC 0 : continue;
2414 GBC 9 : appendStringInfoString(&buf, "<element>");
2415 GIC 9 : appendStringInfoString(&buf,
2416 9 : map_sql_value_to_xml_value(elem_values[i],
2417 : elmtype, true));
2418 CBC 9 : appendStringInfoString(&buf, "</element>");
2419 ECB : }
2420 :
2421 GIC 3 : pfree(elem_values);
2422 3 : pfree(elem_nulls);
2423 :
2424 3 : return buf.data;
2425 : }
2426 : else
2427 : {
2428 : Oid typeOut;
2429 : bool isvarlena;
2430 : char *str;
2431 :
2432 : /*
2433 ECB : * Flatten domains; the special-case treatments below should apply to,
2434 : * eg, domains over boolean not just boolean.
2435 : */
2436 GIC 64940 : type = getBaseType(type);
2437 ECB :
2438 : /*
2439 EUB : * Special XSD formatting for some data types
2440 : */
2441 GBC 64940 : switch (type)
2442 EUB : {
2443 GBC 33 : case BOOLOID:
2444 33 : if (DatumGetBool(value))
2445 CBC 30 : return "true";
2446 ECB : else
2447 CBC 3 : return "false";
2448 EUB :
2449 GBC 24 : case DATEOID:
2450 EUB : {
2451 : DateADT date;
2452 : struct pg_tm tm;
2453 : char buf[MAXDATELEN + 1];
2454 :
2455 GBC 24 : date = DatumGetDateADT(value);
2456 EUB : /* XSD doesn't support infinite values */
2457 GBC 24 : if (DATE_NOT_FINITE(date))
2458 UBC 0 : ereport(ERROR,
2459 EUB : (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
2460 : errmsg("date out of range"),
2461 : errdetail("XML does not support infinite date values.")));
2462 CBC 24 : j2date(date + POSTGRES_EPOCH_JDATE,
2463 : &(tm.tm_year), &(tm.tm_mon), &(tm.tm_mday));
2464 GIC 24 : EncodeDateOnly(&tm, USE_XSD_DATES, buf);
2465 :
2466 24 : return pstrdup(buf);
2467 : }
2468 :
2469 18 : case TIMESTAMPOID:
2470 ECB : {
2471 : Timestamp timestamp;
2472 : struct pg_tm tm;
2473 : fsec_t fsec;
2474 : char buf[MAXDATELEN + 1];
2475 :
2476 GIC 18 : timestamp = DatumGetTimestamp(value);
2477 :
2478 : /* XSD doesn't support infinite values */
2479 18 : if (TIMESTAMP_NOT_FINITE(timestamp))
2480 3 : ereport(ERROR,
2481 ECB : (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
2482 : errmsg("timestamp out of range"),
2483 : errdetail("XML does not support infinite timestamp values.")));
2484 CBC 15 : else if (timestamp2tm(timestamp, NULL, &tm, &fsec, NULL, NULL) == 0)
2485 15 : EncodeDateTime(&tm, fsec, false, 0, NULL, USE_XSD_DATES, buf);
2486 ECB : else
2487 UIC 0 : ereport(ERROR,
2488 : (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
2489 : errmsg("timestamp out of range")));
2490 :
2491 GIC 15 : return pstrdup(buf);
2492 : }
2493 ECB :
2494 GIC 12 : case TIMESTAMPTZOID:
2495 : {
2496 : TimestampTz timestamp;
2497 : struct pg_tm tm;
2498 : int tz;
2499 ECB : fsec_t fsec;
2500 GIC 12 : const char *tzn = NULL;
2501 ECB : char buf[MAXDATELEN + 1];
2502 :
2503 GIC 12 : timestamp = DatumGetTimestamp(value);
2504 ECB :
2505 EUB : /* XSD doesn't support infinite values */
2506 GIC 12 : if (TIMESTAMP_NOT_FINITE(timestamp))
2507 LBC 0 : ereport(ERROR,
2508 : (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
2509 : errmsg("timestamp out of range"),
2510 : errdetail("XML does not support infinite timestamp values.")));
2511 GIC 12 : else if (timestamp2tm(timestamp, &tz, &tm, &fsec, &tzn, NULL) == 0)
2512 CBC 12 : EncodeDateTime(&tm, fsec, true, tz, tzn, USE_XSD_DATES, buf);
2513 : else
2514 UIC 0 : ereport(ERROR,
2515 ECB : (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
2516 : errmsg("timestamp out of range")));
2517 :
2518 GIC 12 : return pstrdup(buf);
2519 : }
2520 :
2521 ECB : #ifdef USE_LIBXML
2522 GIC 18 : case BYTEAOID:
2523 : {
2524 CBC 18 : bytea *bstr = DatumGetByteaPP(value);
2525 ECB : PgXmlErrorContext *xmlerrcxt;
2526 CBC 18 : volatile xmlBufferPtr buf = NULL;
2527 18 : volatile xmlTextWriterPtr writer = NULL;
2528 ECB : char *result;
2529 :
2530 GIC 18 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
2531 :
2532 18 : PG_TRY();
2533 : {
2534 18 : buf = xmlBufferCreate();
2535 18 : if (buf == NULL || xmlerrcxt->err_occurred)
2536 UIC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
2537 ECB : "could not allocate xmlBuffer");
2538 GIC 18 : writer = xmlNewTextWriterMemory(buf, 0);
2539 18 : if (writer == NULL || xmlerrcxt->err_occurred)
2540 UIC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
2541 : "could not allocate xmlTextWriter");
2542 :
2543 GIC 18 : if (xmlbinary == XMLBINARY_BASE64)
2544 15 : xmlTextWriterWriteBase64(writer, VARDATA_ANY(bstr),
2545 15 : 0, VARSIZE_ANY_EXHDR(bstr));
2546 : else
2547 3 : xmlTextWriterWriteBinHex(writer, VARDATA_ANY(bstr),
2548 CBC 3 : 0, VARSIZE_ANY_EXHDR(bstr));
2549 :
2550 ECB : /* we MUST do this now to flush data out to the buffer */
2551 GIC 18 : xmlFreeTextWriter(writer);
2552 CBC 18 : writer = NULL;
2553 :
2554 18 : result = pstrdup((const char *) xmlBufferContent(buf));
2555 ECB : }
2556 LBC 0 : PG_CATCH();
2557 ECB : {
2558 LBC 0 : if (writer)
2559 0 : xmlFreeTextWriter(writer);
2560 UIC 0 : if (buf)
2561 UBC 0 : xmlBufferFree(buf);
2562 EUB :
2563 UIC 0 : pg_xml_done(xmlerrcxt, true);
2564 EUB :
2565 UIC 0 : PG_RE_THROW();
2566 ECB : }
2567 GBC 18 : PG_END_TRY();
2568 :
2569 GIC 18 : xmlBufferFree(buf);
2570 ECB :
2571 GIC 18 : pg_xml_done(xmlerrcxt, false);
2572 ECB :
2573 CBC 18 : return result;
2574 ECB : }
2575 : #endif /* USE_LIBXML */
2576 :
2577 : }
2578 :
2579 : /*
2580 : * otherwise, just use the type's native text representation
2581 : */
2582 GIC 64835 : getTypeOutputInfo(type, &typeOut, &isvarlena);
2583 64835 : str = OidOutputFunctionCall(typeOut, value);
2584 :
2585 : /* ... exactly as-is for XML, and when escaping is not wanted */
2586 64835 : if (type == XMLOID || !xml_escape_strings)
2587 10785 : return str;
2588 :
2589 : /* otherwise, translate special characters as needed */
2590 54050 : return escape_xml(str);
2591 : }
2592 : }
2593 ECB :
2594 :
2595 : /*
2596 : * Escape characters in text that have special meanings in XML.
2597 : *
2598 : * Returns a palloc'd string.
2599 : *
2600 : * NB: this is intentionally not dependent on libxml.
2601 : */
2602 : char *
2603 CBC 54173 : escape_xml(const char *str)
2604 ECB : {
2605 : StringInfoData buf;
2606 : const char *p;
2607 :
2608 CBC 54173 : initStringInfo(&buf);
2609 GIC 334366 : for (p = str; *p; p++)
2610 : {
2611 280193 : switch (*p)
2612 ECB : {
2613 LBC 0 : case '&':
2614 0 : appendStringInfoString(&buf, "&");
2615 0 : break;
2616 GIC 18 : case '<':
2617 18 : appendStringInfoString(&buf, "<");
2618 CBC 18 : break;
2619 GIC 12 : case '>':
2620 12 : appendStringInfoString(&buf, ">");
2621 CBC 12 : break;
2622 UIC 0 : case '\r':
2623 0 : appendStringInfoString(&buf, "
");
2624 0 : break;
2625 GIC 280163 : default:
2626 280163 : appendStringInfoCharMacro(&buf, *p);
2627 280163 : break;
2628 : }
2629 : }
2630 54173 : return buf.data;
2631 : }
2632 :
2633 :
2634 : static char *
2635 CBC 12 : _SPI_strdup(const char *s)
2636 : {
2637 12 : size_t len = strlen(s) + 1;
2638 GIC 12 : char *ret = SPI_palloc(len);
2639 :
2640 12 : memcpy(ret, s, len);
2641 12 : return ret;
2642 : }
2643 :
2644 :
2645 : /*
2646 : * SQL to XML mapping functions
2647 : *
2648 : * What follows below was at one point intentionally organized so that
2649 : * you can read along in the SQL/XML standard. The functions are
2650 ECB : * mostly split up the way the clauses lay out in the standards
2651 : * document, and the identifiers are also aligned with the standard
2652 : * text. Unfortunately, SQL/XML:2006 reordered the clauses
2653 : * differently than SQL/XML:2003, so the order below doesn't make much
2654 : * sense anymore.
2655 : *
2656 : * There are many things going on there:
2657 : *
2658 : * There are two kinds of mappings: Mapping SQL data (table contents)
2659 : * to XML documents, and mapping SQL structure (the "schema") to XML
2660 : * Schema. And there are functions that do both at the same time.
2661 : *
2662 : * Then you can map a database, a schema, or a table, each in both
2663 : * ways. This breaks down recursively: Mapping a database invokes
2664 EUB : * mapping schemas, which invokes mapping tables, which invokes
2665 ECB : * mapping rows, which invokes mapping columns, although you can't
2666 : * call the last two from the outside. Because of this, there are a
2667 : * number of xyz_internal() functions which are to be called both from
2668 : * the function manager wrapper and from some upper layer in a
2669 : * recursive call.
2670 : *
2671 : * See the documentation about what the common function arguments
2672 : * nulls, tableforest, and targetns mean.
2673 : *
2674 : * Some style guidelines for XML output: Use double quotes for quoting
2675 : * XML attributes. Indent XML elements by two spaces, but remember
2676 : * that a lot of code is called recursively at different levels, so
2677 : * it's better not to indent rather than create output that indents
2678 : * and outdents weirdly. Add newlines to make the output look nice.
2679 : */
2680 :
2681 :
2682 : /*
2683 : * Visibility of objects for XML mappings; see SQL/XML:2008 section
2684 : * 4.10.8.
2685 : */
2686 :
2687 : /*
2688 : * Given a query, which must return type oid as first column, produce
2689 : * a list of Oids with the query results.
2690 : */
2691 : static List *
2692 CBC 18 : query_to_oid_list(const char *query)
2693 : {
2694 ECB : uint64 i;
2695 CBC 18 : List *list = NIL;
2696 ECB : int spi_result;
2697 :
2698 CBC 18 : spi_result = SPI_execute(query, true, 0);
2699 GIC 18 : if (spi_result != SPI_OK_SELECT)
2700 LBC 0 : elog(ERROR, "SPI_execute returned %s for %s",
2701 : SPI_result_code_string(spi_result), query);
2702 :
2703 GIC 54 : for (i = 0; i < SPI_processed; i++)
2704 : {
2705 : Datum oid;
2706 ECB : bool isnull;
2707 :
2708 CBC 36 : oid = SPI_getbinval(SPI_tuptable->vals[i],
2709 GBC 36 : SPI_tuptable->tupdesc,
2710 : 1,
2711 : &isnull);
2712 GIC 36 : if (!isnull)
2713 CBC 36 : list = lappend_oid(list, DatumGetObjectId(oid));
2714 : }
2715 ECB :
2716 GIC 18 : return list;
2717 ECB : }
2718 :
2719 :
2720 : static List *
2721 GIC 18 : schema_get_xml_visible_tables(Oid nspid)
2722 : {
2723 : StringInfoData query;
2724 :
2725 18 : initStringInfo(&query);
2726 18 : appendStringInfo(&query, "SELECT oid FROM pg_catalog.pg_class"
2727 ECB : " WHERE relnamespace = %u AND relkind IN ("
2728 : CppAsString2(RELKIND_RELATION) ","
2729 : CppAsString2(RELKIND_MATVIEW) ","
2730 : CppAsString2(RELKIND_VIEW) ")"
2731 : " AND pg_catalog.has_table_privilege (oid, 'SELECT')"
2732 : " ORDER BY relname;", nspid);
2733 :
2734 GIC 18 : return query_to_oid_list(query.data);
2735 ECB : }
2736 :
2737 :
2738 EUB : /*
2739 : * Including the system schemas is probably not useful for a database
2740 : * mapping.
2741 : */
2742 ECB : #define XML_VISIBLE_SCHEMAS_EXCLUDE "(nspname ~ '^pg_' OR nspname = 'information_schema')"
2743 :
2744 : #define XML_VISIBLE_SCHEMAS "SELECT oid FROM pg_catalog.pg_namespace WHERE pg_catalog.has_schema_privilege (oid, 'USAGE') AND NOT " XML_VISIBLE_SCHEMAS_EXCLUDE
2745 :
2746 :
2747 : static List *
2748 UIC 0 : database_get_xml_visible_schemas(void)
2749 : {
2750 0 : return query_to_oid_list(XML_VISIBLE_SCHEMAS " ORDER BY nspname;");
2751 ECB : }
2752 :
2753 :
2754 : static List *
2755 UIC 0 : database_get_xml_visible_tables(void)
2756 : {
2757 ECB : /* At the moment there is no order required here. */
2758 UBC 0 : return query_to_oid_list("SELECT oid FROM pg_catalog.pg_class"
2759 : " WHERE relkind IN ("
2760 : CppAsString2(RELKIND_RELATION) ","
2761 : CppAsString2(RELKIND_MATVIEW) ","
2762 ECB : CppAsString2(RELKIND_VIEW) ")"
2763 : " AND pg_catalog.has_table_privilege(pg_class.oid, 'SELECT')"
2764 : " AND relnamespace IN (" XML_VISIBLE_SCHEMAS ");");
2765 EUB : }
2766 :
2767 :
2768 : /*
2769 ECB : * Map SQL table to XML and/or XML Schema document; see SQL/XML:2008
2770 : * section 9.11.
2771 : */
2772 :
2773 : static StringInfo
2774 GIC 48 : table_to_xml_internal(Oid relid,
2775 ECB : const char *xmlschema, bool nulls, bool tableforest,
2776 : const char *targetns, bool top_level)
2777 : {
2778 : StringInfoData query;
2779 :
2780 GIC 48 : initStringInfo(&query);
2781 CBC 48 : appendStringInfo(&query, "SELECT * FROM %s",
2782 : DatumGetCString(DirectFunctionCall1(regclassout,
2783 ECB : ObjectIdGetDatum(relid))));
2784 GIC 48 : return query_to_xml_internal(query.data, get_rel_name(relid),
2785 ECB : xmlschema, nulls, tableforest,
2786 : targetns, top_level);
2787 EUB : }
2788 :
2789 ECB :
2790 : Datum
2791 GBC 18 : table_to_xml(PG_FUNCTION_ARGS)
2792 : {
2793 GIC 18 : Oid relid = PG_GETARG_OID(0);
2794 CBC 18 : bool nulls = PG_GETARG_BOOL(1);
2795 18 : bool tableforest = PG_GETARG_BOOL(2);
2796 18 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
2797 :
2798 18 : PG_RETURN_XML_P(stringinfo_to_xmltype(table_to_xml_internal(relid, NULL,
2799 ECB : nulls, tableforest,
2800 : targetns, true)));
2801 : }
2802 :
2803 :
2804 : Datum
2805 CBC 5 : query_to_xml(PG_FUNCTION_ARGS)
2806 : {
2807 GBC 5 : char *query = text_to_cstring(PG_GETARG_TEXT_PP(0));
2808 GIC 5 : bool nulls = PG_GETARG_BOOL(1);
2809 GBC 5 : bool tableforest = PG_GETARG_BOOL(2);
2810 5 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
2811 EUB :
2812 GBC 5 : PG_RETURN_XML_P(stringinfo_to_xmltype(query_to_xml_internal(query, NULL,
2813 : NULL, nulls, tableforest,
2814 EUB : targetns, true)));
2815 : }
2816 :
2817 :
2818 ECB : Datum
2819 GIC 6 : cursor_to_xml(PG_FUNCTION_ARGS)
2820 ECB : {
2821 GIC 6 : char *name = text_to_cstring(PG_GETARG_TEXT_PP(0));
2822 CBC 6 : int32 count = PG_GETARG_INT32(1);
2823 GIC 6 : bool nulls = PG_GETARG_BOOL(2);
2824 CBC 6 : bool tableforest = PG_GETARG_BOOL(3);
2825 GIC 6 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(4));
2826 :
2827 : StringInfoData result;
2828 : Portal portal;
2829 : uint64 i;
2830 :
2831 6 : initStringInfo(&result);
2832 :
2833 CBC 6 : if (!tableforest)
2834 ECB : {
2835 GIC 3 : xmldata_root_element_start(&result, "table", NULL, targetns, true);
2836 3 : appendStringInfoChar(&result, '\n');
2837 ECB : }
2838 :
2839 GIC 6 : SPI_connect();
2840 6 : portal = SPI_cursor_find(name);
2841 CBC 6 : if (portal == NULL)
2842 UIC 0 : ereport(ERROR,
2843 : (errcode(ERRCODE_UNDEFINED_CURSOR),
2844 : errmsg("cursor \"%s\" does not exist", name)));
2845 :
2846 GIC 6 : SPI_cursor_fetch(portal, true, count);
2847 24 : for (i = 0; i < SPI_processed; i++)
2848 18 : SPI_sql_row_to_xmlelement(i, &result, NULL, nulls,
2849 : tableforest, targetns, true);
2850 :
2851 6 : SPI_finish();
2852 :
2853 6 : if (!tableforest)
2854 CBC 3 : xmldata_root_element_end(&result, "table");
2855 :
2856 GIC 6 : PG_RETURN_XML_P(stringinfo_to_xmltype(&result));
2857 : }
2858 :
2859 ECB :
2860 : /*
2861 : * Write the start tag of the root element of a data mapping.
2862 : *
2863 : * top_level means that this is the very top level of the eventual
2864 EUB : * output. For example, when the user calls table_to_xml, then a call
2865 : * with a table name to this function is the top level. When the user
2866 : * calls database_to_xml, then a call with a schema name to this
2867 ECB : * function is not the top level. If top_level is false, then the XML
2868 : * namespace declarations are omitted, because they supposedly already
2869 : * appeared earlier in the output. Repeating them is not wrong, but
2870 : * it looks ugly.
2871 : */
2872 : static void
2873 GBC 119 : xmldata_root_element_start(StringInfo result, const char *eltname,
2874 EUB : const char *xmlschema, const char *targetns,
2875 : bool top_level)
2876 ECB : {
2877 : /* This isn't really wrong but currently makes no sense. */
2878 CBC 119 : Assert(top_level || !xmlschema);
2879 :
2880 GIC 119 : appendStringInfo(result, "<%s", eltname);
2881 CBC 119 : if (top_level)
2882 : {
2883 GIC 89 : appendStringInfoString(result, " xmlns:xsi=\"" NAMESPACE_XSI "\"");
2884 89 : if (strlen(targetns) > 0)
2885 15 : appendStringInfo(result, " xmlns=\"%s\"", targetns);
2886 ECB : }
2887 GIC 119 : if (xmlschema)
2888 ECB : {
2889 : /* FIXME: better targets */
2890 GIC 9 : if (strlen(targetns) > 0)
2891 CBC 3 : appendStringInfo(result, " xsi:schemaLocation=\"%s #\"", targetns);
2892 ECB : else
2893 GIC 6 : appendStringInfoString(result, " xsi:noNamespaceSchemaLocation=\"#\"");
2894 : }
2895 119 : appendStringInfoString(result, ">\n");
2896 119 : }
2897 :
2898 :
2899 : static void
2900 119 : xmldata_root_element_end(StringInfo result, const char *eltname)
2901 : {
2902 119 : appendStringInfo(result, "</%s>\n", eltname);
2903 119 : }
2904 :
2905 :
2906 : static StringInfo
2907 56 : query_to_xml_internal(const char *query, char *tablename,
2908 : const char *xmlschema, bool nulls, bool tableforest,
2909 : const char *targetns, bool top_level)
2910 : {
2911 : StringInfo result;
2912 : char *xmltn;
2913 : uint64 i;
2914 :
2915 56 : if (tablename)
2916 48 : xmltn = map_sql_identifier_to_xml_name(tablename, true, false);
2917 : else
2918 8 : xmltn = "table";
2919 :
2920 56 : result = makeStringInfo();
2921 :
2922 56 : SPI_connect();
2923 56 : if (SPI_execute(query, true, 0) != SPI_OK_SELECT)
2924 UIC 0 : ereport(ERROR,
2925 : (errcode(ERRCODE_DATA_EXCEPTION),
2926 : errmsg("invalid query")));
2927 :
2928 GIC 56 : if (!tableforest)
2929 : {
2930 26 : xmldata_root_element_start(result, xmltn, xmlschema,
2931 : targetns, top_level);
2932 26 : appendStringInfoChar(result, '\n');
2933 : }
2934 :
2935 56 : if (xmlschema)
2936 15 : appendStringInfo(result, "%s\n\n", xmlschema);
2937 :
2938 194 : for (i = 0; i < SPI_processed; i++)
2939 138 : SPI_sql_row_to_xmlelement(i, result, tablename, nulls,
2940 : tableforest, targetns, top_level);
2941 :
2942 56 : if (!tableforest)
2943 CBC 26 : xmldata_root_element_end(result, xmltn);
2944 :
2945 GIC 56 : SPI_finish();
2946 ECB :
2947 GIC 56 : return result;
2948 : }
2949 ECB :
2950 :
2951 EUB : Datum
2952 GIC 15 : table_to_xmlschema(PG_FUNCTION_ARGS)
2953 : {
2954 CBC 15 : Oid relid = PG_GETARG_OID(0);
2955 GIC 15 : bool nulls = PG_GETARG_BOOL(1);
2956 15 : bool tableforest = PG_GETARG_BOOL(2);
2957 15 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
2958 : const char *result;
2959 ECB : Relation rel;
2960 :
2961 GIC 15 : rel = table_open(relid, AccessShareLock);
2962 15 : result = map_sql_table_to_xmlschema(rel->rd_att, relid, nulls,
2963 ECB : tableforest, targetns);
2964 CBC 15 : table_close(rel, NoLock);
2965 :
2966 GIC 15 : PG_RETURN_XML_P(cstring_to_xmltype(result));
2967 ECB : }
2968 :
2969 :
2970 : Datum
2971 GIC 3 : query_to_xmlschema(PG_FUNCTION_ARGS)
2972 ECB : {
2973 GIC 3 : char *query = text_to_cstring(PG_GETARG_TEXT_PP(0));
2974 3 : bool nulls = PG_GETARG_BOOL(1);
2975 3 : bool tableforest = PG_GETARG_BOOL(2);
2976 CBC 3 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
2977 ECB : const char *result;
2978 : SPIPlanPtr plan;
2979 : Portal portal;
2980 :
2981 GIC 3 : SPI_connect();
2982 :
2983 3 : if ((plan = SPI_prepare(query, 0, NULL)) == NULL)
2984 UIC 0 : elog(ERROR, "SPI_prepare(\"%s\") failed", query);
2985 ECB :
2986 GIC 3 : if ((portal = SPI_cursor_open(NULL, plan, NULL, NULL, true)) == NULL)
2987 UIC 0 : elog(ERROR, "SPI_cursor_open(\"%s\") failed", query);
2988 :
2989 GIC 3 : result = _SPI_strdup(map_sql_table_to_xmlschema(portal->tupDesc,
2990 : InvalidOid, nulls,
2991 : tableforest, targetns));
2992 3 : SPI_cursor_close(portal);
2993 3 : SPI_finish();
2994 :
2995 3 : PG_RETURN_XML_P(cstring_to_xmltype(result));
2996 : }
2997 :
2998 :
2999 EUB : Datum
3000 GIC 6 : cursor_to_xmlschema(PG_FUNCTION_ARGS)
3001 EUB : {
3002 GIC 6 : char *name = text_to_cstring(PG_GETARG_TEXT_PP(0));
3003 6 : bool nulls = PG_GETARG_BOOL(1);
3004 6 : bool tableforest = PG_GETARG_BOOL(2);
3005 6 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3006 EUB : const char *xmlschema;
3007 : Portal portal;
3008 :
3009 GBC 6 : SPI_connect();
3010 GIC 6 : portal = SPI_cursor_find(name);
3011 6 : if (portal == NULL)
3012 UIC 0 : ereport(ERROR,
3013 : (errcode(ERRCODE_UNDEFINED_CURSOR),
3014 : errmsg("cursor \"%s\" does not exist", name)));
3015 :
3016 GIC 6 : xmlschema = _SPI_strdup(map_sql_table_to_xmlschema(portal->tupDesc,
3017 : InvalidOid, nulls,
3018 : tableforest, targetns));
3019 6 : SPI_finish();
3020 :
3021 6 : PG_RETURN_XML_P(cstring_to_xmltype(xmlschema));
3022 : }
3023 :
3024 :
3025 ECB : Datum
3026 GIC 12 : table_to_xml_and_xmlschema(PG_FUNCTION_ARGS)
3027 : {
3028 12 : Oid relid = PG_GETARG_OID(0);
3029 12 : bool nulls = PG_GETARG_BOOL(1);
3030 12 : bool tableforest = PG_GETARG_BOOL(2);
3031 CBC 12 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3032 ECB : Relation rel;
3033 : const char *xmlschema;
3034 :
3035 CBC 12 : rel = table_open(relid, AccessShareLock);
3036 GIC 12 : xmlschema = map_sql_table_to_xmlschema(rel->rd_att, relid, nulls,
3037 : tableforest, targetns);
3038 12 : table_close(rel, NoLock);
3039 :
3040 12 : PG_RETURN_XML_P(stringinfo_to_xmltype(table_to_xml_internal(relid,
3041 : xmlschema, nulls, tableforest,
3042 ECB : targetns, true)));
3043 : }
3044 :
3045 :
3046 : Datum
3047 CBC 3 : query_to_xml_and_xmlschema(PG_FUNCTION_ARGS)
3048 : {
3049 3 : char *query = text_to_cstring(PG_GETARG_TEXT_PP(0));
3050 GIC 3 : bool nulls = PG_GETARG_BOOL(1);
3051 3 : bool tableforest = PG_GETARG_BOOL(2);
3052 3 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3053 :
3054 : const char *xmlschema;
3055 : SPIPlanPtr plan;
3056 ECB : Portal portal;
3057 :
3058 CBC 3 : SPI_connect();
3059 ECB :
3060 CBC 3 : if ((plan = SPI_prepare(query, 0, NULL)) == NULL)
3061 LBC 0 : elog(ERROR, "SPI_prepare(\"%s\") failed", query);
3062 :
3063 CBC 3 : if ((portal = SPI_cursor_open(NULL, plan, NULL, NULL, true)) == NULL)
3064 UIC 0 : elog(ERROR, "SPI_cursor_open(\"%s\") failed", query);
3065 :
3066 GIC 3 : xmlschema = _SPI_strdup(map_sql_table_to_xmlschema(portal->tupDesc,
3067 : InvalidOid, nulls, tableforest, targetns));
3068 3 : SPI_cursor_close(portal);
3069 3 : SPI_finish();
3070 ECB :
3071 GIC 3 : PG_RETURN_XML_P(stringinfo_to_xmltype(query_to_xml_internal(query, NULL,
3072 ECB : xmlschema, nulls, tableforest,
3073 : targetns, true)));
3074 : }
3075 :
3076 :
3077 : /*
3078 : * Map SQL schema to XML and/or XML Schema document; see SQL/XML:2008
3079 : * sections 9.13, 9.14.
3080 : */
3081 :
3082 : static StringInfo
3083 GIC 9 : schema_to_xml_internal(Oid nspid, const char *xmlschema, bool nulls,
3084 ECB : bool tableforest, const char *targetns, bool top_level)
3085 : {
3086 : StringInfo result;
3087 : char *xmlsn;
3088 : List *relid_list;
3089 : ListCell *cell;
3090 :
3091 CBC 9 : xmlsn = map_sql_identifier_to_xml_name(get_namespace_name(nspid),
3092 ECB : true, false);
3093 GBC 9 : result = makeStringInfo();
3094 :
3095 GIC 9 : xmldata_root_element_start(result, xmlsn, xmlschema, targetns, top_level);
3096 9 : appendStringInfoChar(result, '\n');
3097 ECB :
3098 CBC 9 : if (xmlschema)
3099 3 : appendStringInfo(result, "%s\n\n", xmlschema);
3100 :
3101 GIC 9 : SPI_connect();
3102 ECB :
3103 GIC 9 : relid_list = schema_get_xml_visible_tables(nspid);
3104 ECB :
3105 CBC 27 : foreach(cell, relid_list)
3106 : {
3107 18 : Oid relid = lfirst_oid(cell);
3108 : StringInfo subres;
3109 :
3110 GIC 18 : subres = table_to_xml_internal(relid, NULL, nulls, tableforest,
3111 : targetns, false);
3112 :
3113 18 : appendBinaryStringInfo(result, subres->data, subres->len);
3114 18 : appendStringInfoChar(result, '\n');
3115 : }
3116 :
3117 9 : SPI_finish();
3118 :
3119 9 : xmldata_root_element_end(result, xmlsn);
3120 :
3121 9 : return result;
3122 : }
3123 :
3124 ECB :
3125 : Datum
3126 GIC 6 : schema_to_xml(PG_FUNCTION_ARGS)
3127 : {
3128 6 : Name name = PG_GETARG_NAME(0);
3129 CBC 6 : bool nulls = PG_GETARG_BOOL(1);
3130 GIC 6 : bool tableforest = PG_GETARG_BOOL(2);
3131 CBC 6 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3132 ECB :
3133 : char *schemaname;
3134 : Oid nspid;
3135 :
3136 CBC 6 : schemaname = NameStr(*name);
3137 GIC 6 : nspid = LookupExplicitNamespace(schemaname, false);
3138 ECB :
3139 GIC 6 : PG_RETURN_XML_P(stringinfo_to_xmltype(schema_to_xml_internal(nspid, NULL,
3140 : nulls, tableforest, targetns, true)));
3141 ECB : }
3142 :
3143 :
3144 : /*
3145 : * Write the start element of the root element of an XML Schema mapping.
3146 : */
3147 : static void
3148 GIC 48 : xsd_schema_element_start(StringInfo result, const char *targetns)
3149 : {
3150 48 : appendStringInfoString(result,
3151 ECB : "<xsd:schema\n"
3152 : " xmlns:xsd=\"" NAMESPACE_XSD "\"");
3153 CBC 48 : if (strlen(targetns) > 0)
3154 9 : appendStringInfo(result,
3155 : "\n"
3156 : " targetNamespace=\"%s\"\n"
3157 : " elementFormDefault=\"qualified\"",
3158 ECB : targetns);
3159 GIC 48 : appendStringInfoString(result,
3160 : ">\n\n");
3161 48 : }
3162 :
3163 :
3164 : static void
3165 48 : xsd_schema_element_end(StringInfo result)
3166 ECB : {
3167 CBC 48 : appendStringInfoString(result, "</xsd:schema>");
3168 GIC 48 : }
3169 ECB :
3170 :
3171 : static StringInfo
3172 GIC 9 : schema_to_xmlschema_internal(const char *schemaname, bool nulls,
3173 ECB : bool tableforest, const char *targetns)
3174 : {
3175 EUB : Oid nspid;
3176 : List *relid_list;
3177 : List *tupdesc_list;
3178 : ListCell *cell;
3179 ECB : StringInfo result;
3180 :
3181 CBC 9 : result = makeStringInfo();
3182 :
3183 9 : nspid = LookupExplicitNamespace(schemaname, false);
3184 :
3185 GIC 9 : xsd_schema_element_start(result, targetns);
3186 ECB :
3187 CBC 9 : SPI_connect();
3188 :
3189 9 : relid_list = schema_get_xml_visible_tables(nspid);
3190 ECB :
3191 GIC 9 : tupdesc_list = NIL;
3192 27 : foreach(cell, relid_list)
3193 ECB : {
3194 : Relation rel;
3195 :
3196 CBC 18 : rel = table_open(lfirst_oid(cell), AccessShareLock);
3197 GIC 18 : tupdesc_list = lappend(tupdesc_list, CreateTupleDescCopy(rel->rd_att));
3198 CBC 18 : table_close(rel, NoLock);
3199 : }
3200 :
3201 GIC 9 : appendStringInfoString(result,
3202 : map_sql_typecoll_to_xmlschema_types(tupdesc_list));
3203 ECB :
3204 GIC 9 : appendStringInfoString(result,
3205 ECB : map_sql_schema_to_xmlschema_types(nspid, relid_list,
3206 : nulls, tableforest, targetns));
3207 :
3208 CBC 9 : xsd_schema_element_end(result);
3209 :
3210 GIC 9 : SPI_finish();
3211 :
3212 CBC 9 : return result;
3213 ECB : }
3214 :
3215 :
3216 : Datum
3217 CBC 6 : schema_to_xmlschema(PG_FUNCTION_ARGS)
3218 : {
3219 GIC 6 : Name name = PG_GETARG_NAME(0);
3220 6 : bool nulls = PG_GETARG_BOOL(1);
3221 6 : bool tableforest = PG_GETARG_BOOL(2);
3222 CBC 6 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3223 :
3224 6 : PG_RETURN_XML_P(stringinfo_to_xmltype(schema_to_xmlschema_internal(NameStr(*name),
3225 ECB : nulls, tableforest, targetns)));
3226 : }
3227 :
3228 :
3229 : Datum
3230 GIC 3 : schema_to_xml_and_xmlschema(PG_FUNCTION_ARGS)
3231 : {
3232 CBC 3 : Name name = PG_GETARG_NAME(0);
3233 GIC 3 : bool nulls = PG_GETARG_BOOL(1);
3234 CBC 3 : bool tableforest = PG_GETARG_BOOL(2);
3235 GBC 3 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(3));
3236 : char *schemaname;
3237 ECB : Oid nspid;
3238 EUB : StringInfo xmlschema;
3239 :
3240 CBC 3 : schemaname = NameStr(*name);
3241 GIC 3 : nspid = LookupExplicitNamespace(schemaname, false);
3242 :
3243 CBC 3 : xmlschema = schema_to_xmlschema_internal(schemaname, nulls,
3244 ECB : tableforest, targetns);
3245 :
3246 CBC 3 : PG_RETURN_XML_P(stringinfo_to_xmltype(schema_to_xml_internal(nspid,
3247 : xmlschema->data, nulls,
3248 : tableforest, targetns, true)));
3249 : }
3250 :
3251 ECB :
3252 : /*
3253 : * Map SQL database to XML and/or XML Schema document; see SQL/XML:2008
3254 : * sections 9.16, 9.17.
3255 : */
3256 :
3257 : static StringInfo
3258 UIC 0 : database_to_xml_internal(const char *xmlschema, bool nulls,
3259 : bool tableforest, const char *targetns)
3260 ECB : {
3261 : StringInfo result;
3262 : List *nspid_list;
3263 EUB : ListCell *cell;
3264 : char *xmlcn;
3265 :
3266 UIC 0 : xmlcn = map_sql_identifier_to_xml_name(get_database_name(MyDatabaseId),
3267 ECB : true, false);
3268 UIC 0 : result = makeStringInfo();
3269 :
3270 LBC 0 : xmldata_root_element_start(result, xmlcn, xmlschema, targetns, true);
3271 UIC 0 : appendStringInfoChar(result, '\n');
3272 ECB :
3273 UIC 0 : if (xmlschema)
3274 0 : appendStringInfo(result, "%s\n\n", xmlschema);
3275 :
3276 0 : SPI_connect();
3277 ECB :
3278 UIC 0 : nspid_list = database_get_xml_visible_schemas();
3279 ECB :
3280 LBC 0 : foreach(cell, nspid_list)
3281 ECB : {
3282 LBC 0 : Oid nspid = lfirst_oid(cell);
3283 : StringInfo subres;
3284 :
3285 UIC 0 : subres = schema_to_xml_internal(nspid, NULL, nulls,
3286 ECB : tableforest, targetns, false);
3287 :
3288 UIC 0 : appendBinaryStringInfo(result, subres->data, subres->len);
3289 LBC 0 : appendStringInfoChar(result, '\n');
3290 : }
3291 ECB :
3292 UIC 0 : SPI_finish();
3293 :
3294 0 : xmldata_root_element_end(result, xmlcn);
3295 :
3296 0 : return result;
3297 : }
3298 ECB :
3299 :
3300 : Datum
3301 LBC 0 : database_to_xml(PG_FUNCTION_ARGS)
3302 ECB : {
3303 LBC 0 : bool nulls = PG_GETARG_BOOL(0);
3304 UIC 0 : bool tableforest = PG_GETARG_BOOL(1);
3305 0 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(2));
3306 :
3307 0 : PG_RETURN_XML_P(stringinfo_to_xmltype(database_to_xml_internal(NULL, nulls,
3308 : tableforest, targetns)));
3309 ECB : }
3310 :
3311 :
3312 EUB : static StringInfo
3313 UIC 0 : database_to_xmlschema_internal(bool nulls, bool tableforest,
3314 ECB : const char *targetns)
3315 EUB : {
3316 : List *relid_list;
3317 ECB : List *nspid_list;
3318 : List *tupdesc_list;
3319 : ListCell *cell;
3320 : StringInfo result;
3321 :
3322 LBC 0 : result = makeStringInfo();
3323 :
3324 UIC 0 : xsd_schema_element_start(result, targetns);
3325 :
3326 0 : SPI_connect();
3327 :
3328 0 : relid_list = database_get_xml_visible_tables();
3329 0 : nspid_list = database_get_xml_visible_schemas();
3330 :
3331 0 : tupdesc_list = NIL;
3332 0 : foreach(cell, relid_list)
3333 : {
3334 ECB : Relation rel;
3335 :
3336 UIC 0 : rel = table_open(lfirst_oid(cell), AccessShareLock);
3337 0 : tupdesc_list = lappend(tupdesc_list, CreateTupleDescCopy(rel->rd_att));
3338 0 : table_close(rel, NoLock);
3339 : }
3340 :
3341 0 : appendStringInfoString(result,
3342 ECB : map_sql_typecoll_to_xmlschema_types(tupdesc_list));
3343 :
3344 LBC 0 : appendStringInfoString(result,
3345 : map_sql_catalog_to_xmlschema_types(nspid_list, nulls, tableforest, targetns));
3346 ECB :
3347 LBC 0 : xsd_schema_element_end(result);
3348 :
3349 0 : SPI_finish();
3350 ECB :
3351 UIC 0 : return result;
3352 ECB : }
3353 :
3354 :
3355 : Datum
3356 LBC 0 : database_to_xmlschema(PG_FUNCTION_ARGS)
3357 : {
3358 0 : bool nulls = PG_GETARG_BOOL(0);
3359 UIC 0 : bool tableforest = PG_GETARG_BOOL(1);
3360 0 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(2));
3361 ECB :
3362 UIC 0 : PG_RETURN_XML_P(stringinfo_to_xmltype(database_to_xmlschema_internal(nulls,
3363 : tableforest, targetns)));
3364 ECB : }
3365 :
3366 :
3367 : Datum
3368 LBC 0 : database_to_xml_and_xmlschema(PG_FUNCTION_ARGS)
3369 : {
3370 0 : bool nulls = PG_GETARG_BOOL(0);
3371 UIC 0 : bool tableforest = PG_GETARG_BOOL(1);
3372 LBC 0 : const char *targetns = text_to_cstring(PG_GETARG_TEXT_PP(2));
3373 : StringInfo xmlschema;
3374 :
3375 UIC 0 : xmlschema = database_to_xmlschema_internal(nulls, tableforest, targetns);
3376 :
3377 LBC 0 : PG_RETURN_XML_P(stringinfo_to_xmltype(database_to_xml_internal(xmlschema->data,
3378 : nulls, tableforest, targetns)));
3379 ECB : }
3380 :
3381 :
3382 : /*
3383 : * Map a multi-part SQL name to an XML name; see SQL/XML:2008 section
3384 : * 9.2.
3385 : */
3386 : static char *
3387 CBC 192 : map_multipart_sql_identifier_to_xml_name(const char *a, const char *b, const char *c, const char *d)
3388 ECB : {
3389 : StringInfoData result;
3390 :
3391 GIC 192 : initStringInfo(&result);
3392 :
3393 192 : if (a)
3394 192 : appendStringInfoString(&result,
3395 192 : map_sql_identifier_to_xml_name(a, true, true));
3396 192 : if (b)
3397 192 : appendStringInfo(&result, ".%s",
3398 : map_sql_identifier_to_xml_name(b, true, true));
3399 CBC 192 : if (c)
3400 GIC 192 : appendStringInfo(&result, ".%s",
3401 ECB : map_sql_identifier_to_xml_name(c, true, true));
3402 GIC 192 : if (d)
3403 183 : appendStringInfo(&result, ".%s",
3404 ECB : map_sql_identifier_to_xml_name(d, true, true));
3405 :
3406 GIC 192 : return result.data;
3407 : }
3408 :
3409 :
3410 ECB : /*
3411 : * Map an SQL table to an XML Schema document; see SQL/XML:2008
3412 : * section 9.11.
3413 : *
3414 : * Map an SQL table to XML Schema data types; see SQL/XML:2008 section
3415 : * 9.9.
3416 : */
3417 : static const char *
3418 CBC 39 : map_sql_table_to_xmlschema(TupleDesc tupdesc, Oid relid, bool nulls,
3419 ECB : bool tableforest, const char *targetns)
3420 : {
3421 : int i;
3422 : char *xmltn;
3423 : char *tabletypename;
3424 : char *rowtypename;
3425 : StringInfoData result;
3426 :
3427 GIC 39 : initStringInfo(&result);
3428 :
3429 39 : if (OidIsValid(relid))
3430 : {
3431 : HeapTuple tuple;
3432 ECB : Form_pg_class reltuple;
3433 :
3434 CBC 27 : tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(relid));
3435 GIC 27 : if (!HeapTupleIsValid(tuple))
3436 LBC 0 : elog(ERROR, "cache lookup failed for relation %u", relid);
3437 GIC 27 : reltuple = (Form_pg_class) GETSTRUCT(tuple);
3438 ECB :
3439 GIC 27 : xmltn = map_sql_identifier_to_xml_name(NameStr(reltuple->relname),
3440 ECB : true, false);
3441 :
3442 CBC 27 : tabletypename = map_multipart_sql_identifier_to_xml_name("TableType",
3443 27 : get_database_name(MyDatabaseId),
3444 GIC 27 : get_namespace_name(reltuple->relnamespace),
3445 27 : NameStr(reltuple->relname));
3446 :
3447 CBC 27 : rowtypename = map_multipart_sql_identifier_to_xml_name("RowType",
3448 27 : get_database_name(MyDatabaseId),
3449 27 : get_namespace_name(reltuple->relnamespace),
3450 GIC 27 : NameStr(reltuple->relname));
3451 :
3452 CBC 27 : ReleaseSysCache(tuple);
3453 : }
3454 : else
3455 ECB : {
3456 GIC 12 : if (tableforest)
3457 6 : xmltn = "row";
3458 : else
3459 CBC 6 : xmltn = "table";
3460 :
3461 12 : tabletypename = "TableType";
3462 GIC 12 : rowtypename = "RowType";
3463 ECB : }
3464 :
3465 GIC 39 : xsd_schema_element_start(&result, targetns);
3466 :
3467 39 : appendStringInfoString(&result,
3468 CBC 39 : map_sql_typecoll_to_xmlschema_types(list_make1(tupdesc)));
3469 :
3470 39 : appendStringInfo(&result,
3471 ECB : "<xsd:complexType name=\"%s\">\n"
3472 : " <xsd:sequence>\n",
3473 : rowtypename);
3474 :
3475 CBC 162 : for (i = 0; i < tupdesc->natts; i++)
3476 : {
3477 GIC 123 : Form_pg_attribute att = TupleDescAttr(tupdesc, i);
3478 :
3479 123 : if (att->attisdropped)
3480 3 : continue;
3481 CBC 240 : appendStringInfo(&result,
3482 : " <xsd:element name=\"%s\" type=\"%s\"%s></xsd:element>\n",
3483 120 : map_sql_identifier_to_xml_name(NameStr(att->attname),
3484 ECB : true, false),
3485 : map_sql_type_to_xml_name(att->atttypid, -1),
3486 : nulls ? " nillable=\"true\"" : " minOccurs=\"0\"");
3487 : }
3488 :
3489 GIC 39 : appendStringInfoString(&result,
3490 : " </xsd:sequence>\n"
3491 ECB : "</xsd:complexType>\n\n");
3492 :
3493 GIC 39 : if (!tableforest)
3494 ECB : {
3495 GIC 21 : appendStringInfo(&result,
3496 : "<xsd:complexType name=\"%s\">\n"
3497 ECB : " <xsd:sequence>\n"
3498 : " <xsd:element name=\"row\" type=\"%s\" minOccurs=\"0\" maxOccurs=\"unbounded\"/>\n"
3499 : " </xsd:sequence>\n"
3500 : "</xsd:complexType>\n\n",
3501 : tabletypename, rowtypename);
3502 :
3503 GIC 21 : appendStringInfo(&result,
3504 : "<xsd:element name=\"%s\" type=\"%s\"/>\n\n",
3505 : xmltn, tabletypename);
3506 : }
3507 : else
3508 18 : appendStringInfo(&result,
3509 EUB : "<xsd:element name=\"%s\" type=\"%s\"/>\n\n",
3510 : xmltn, rowtypename);
3511 :
3512 GIC 39 : xsd_schema_element_end(&result);
3513 :
3514 39 : return result.data;
3515 : }
3516 :
3517 EUB :
3518 : /*
3519 : * Map an SQL schema to XML Schema data types; see SQL/XML:2008
3520 : * section 9.12.
3521 : */
3522 : static const char *
3523 GIC 9 : map_sql_schema_to_xmlschema_types(Oid nspid, List *relid_list, bool nulls,
3524 EUB : bool tableforest, const char *targetns)
3525 : {
3526 : char *dbname;
3527 : char *nspname;
3528 : char *xmlsn;
3529 : char *schematypename;
3530 : StringInfoData result;
3531 : ListCell *cell;
3532 :
3533 GBC 9 : dbname = get_database_name(MyDatabaseId);
3534 GIC 9 : nspname = get_namespace_name(nspid);
3535 :
3536 GBC 9 : initStringInfo(&result);
3537 :
3538 GIC 9 : xmlsn = map_sql_identifier_to_xml_name(nspname, true, false);
3539 EUB :
3540 GBC 9 : schematypename = map_multipart_sql_identifier_to_xml_name("SchemaType",
3541 : dbname,
3542 : nspname,
3543 EUB : NULL);
3544 :
3545 GBC 9 : appendStringInfo(&result,
3546 : "<xsd:complexType name=\"%s\">\n", schematypename);
3547 9 : if (!tableforest)
3548 GIC 3 : appendStringInfoString(&result,
3549 : " <xsd:all>\n");
3550 : else
3551 6 : appendStringInfoString(&result,
3552 EUB : " <xsd:sequence>\n");
3553 :
3554 GBC 27 : foreach(cell, relid_list)
3555 EUB : {
3556 GBC 18 : Oid relid = lfirst_oid(cell);
3557 GIC 18 : char *relname = get_rel_name(relid);
3558 GBC 18 : char *xmltn = map_sql_identifier_to_xml_name(relname, true, false);
3559 GIC 18 : char *tabletypename = map_multipart_sql_identifier_to_xml_name(tableforest ? "RowType" : "TableType",
3560 : dbname,
3561 : nspname,
3562 : relname);
3563 :
3564 GBC 18 : if (!tableforest)
3565 GIC 6 : appendStringInfo(&result,
3566 : " <xsd:element name=\"%s\" type=\"%s\"/>\n",
3567 : xmltn, tabletypename);
3568 : else
3569 12 : appendStringInfo(&result,
3570 : " <xsd:element name=\"%s\" type=\"%s\" minOccurs=\"0\" maxOccurs=\"unbounded\"/>\n",
3571 : xmltn, tabletypename);
3572 : }
3573 EUB :
3574 GIC 9 : if (!tableforest)
3575 GBC 3 : appendStringInfoString(&result,
3576 : " </xsd:all>\n");
3577 EUB : else
3578 GIC 6 : appendStringInfoString(&result,
3579 EUB : " </xsd:sequence>\n");
3580 GBC 9 : appendStringInfoString(&result,
3581 : "</xsd:complexType>\n\n");
3582 EUB :
3583 GBC 9 : appendStringInfo(&result,
3584 : "<xsd:element name=\"%s\" type=\"%s\"/>\n\n",
3585 : xmlsn, schematypename);
3586 :
3587 9 : return result.data;
3588 EUB : }
3589 :
3590 :
3591 : /*
3592 : * Map an SQL catalog to XML Schema data types; see SQL/XML:2008
3593 : * section 9.15.
3594 : */
3595 : static const char *
3596 UIC 0 : map_sql_catalog_to_xmlschema_types(List *nspid_list, bool nulls,
3597 : bool tableforest, const char *targetns)
3598 EUB : {
3599 : char *dbname;
3600 : char *xmlcn;
3601 : char *catalogtypename;
3602 : StringInfoData result;
3603 : ListCell *cell;
3604 :
3605 UIC 0 : dbname = get_database_name(MyDatabaseId);
3606 :
3607 UBC 0 : initStringInfo(&result);
3608 :
3609 0 : xmlcn = map_sql_identifier_to_xml_name(dbname, true, false);
3610 EUB :
3611 UBC 0 : catalogtypename = map_multipart_sql_identifier_to_xml_name("CatalogType",
3612 : dbname,
3613 EUB : NULL,
3614 : NULL);
3615 :
3616 UIC 0 : appendStringInfo(&result,
3617 : "<xsd:complexType name=\"%s\">\n", catalogtypename);
3618 0 : appendStringInfoString(&result,
3619 EUB : " <xsd:all>\n");
3620 :
3621 UBC 0 : foreach(cell, nspid_list)
3622 EUB : {
3623 UBC 0 : Oid nspid = lfirst_oid(cell);
3624 UIC 0 : char *nspname = get_namespace_name(nspid);
3625 0 : char *xmlsn = map_sql_identifier_to_xml_name(nspname, true, false);
3626 UBC 0 : char *schematypename = map_multipart_sql_identifier_to_xml_name("SchemaType",
3627 : dbname,
3628 EUB : nspname,
3629 : NULL);
3630 :
3631 UIC 0 : appendStringInfo(&result,
3632 : " <xsd:element name=\"%s\" type=\"%s\"/>\n",
3633 : xmlsn, schematypename);
3634 : }
3635 :
3636 0 : appendStringInfoString(&result,
3637 : " </xsd:all>\n");
3638 LBC 0 : appendStringInfoString(&result,
3639 : "</xsd:complexType>\n\n");
3640 :
3641 UIC 0 : appendStringInfo(&result,
3642 ECB : "<xsd:element name=\"%s\" type=\"%s\"/>\n\n",
3643 : xmlcn, catalogtypename);
3644 :
3645 LBC 0 : return result.data;
3646 ECB : }
3647 :
3648 :
3649 : /*
3650 : * Map an SQL data type to an XML name; see SQL/XML:2008 section 9.4.
3651 : */
3652 : static const char *
3653 CBC 405 : map_sql_type_to_xml_name(Oid typeoid, int typmod)
3654 ECB : {
3655 : StringInfoData result;
3656 :
3657 CBC 405 : initStringInfo(&result);
3658 :
3659 GIC 405 : switch (typeoid)
3660 : {
3661 15 : case BPCHAROID:
3662 15 : if (typmod == -1)
3663 15 : appendStringInfoString(&result, "CHAR");
3664 : else
3665 UIC 0 : appendStringInfo(&result, "CHAR_%d", typmod - VARHDRSZ);
3666 GIC 15 : break;
3667 27 : case VARCHAROID:
3668 27 : if (typmod == -1)
3669 CBC 27 : appendStringInfoString(&result, "VARCHAR");
3670 : else
3671 UIC 0 : appendStringInfo(&result, "VARCHAR_%d", typmod - VARHDRSZ);
3672 GIC 27 : break;
3673 15 : case NUMERICOID:
3674 15 : if (typmod == -1)
3675 15 : appendStringInfoString(&result, "NUMERIC");
3676 : else
3677 UIC 0 : appendStringInfo(&result, "NUMERIC_%d_%d",
3678 LBC 0 : ((typmod - VARHDRSZ) >> 16) & 0xffff,
3679 UIC 0 : (typmod - VARHDRSZ) & 0xffff);
3680 CBC 15 : break;
3681 GIC 87 : case INT4OID:
3682 87 : appendStringInfoString(&result, "INTEGER");
3683 87 : break;
3684 15 : case INT2OID:
3685 CBC 15 : appendStringInfoString(&result, "SMALLINT");
3686 15 : break;
3687 GBC 15 : case INT8OID:
3688 CBC 15 : appendStringInfoString(&result, "BIGINT");
3689 GIC 15 : break;
3690 CBC 15 : case FLOAT4OID:
3691 GIC 15 : appendStringInfoString(&result, "REAL");
3692 15 : break;
3693 LBC 0 : case FLOAT8OID:
3694 0 : appendStringInfoString(&result, "DOUBLE");
3695 0 : break;
3696 CBC 15 : case BOOLOID:
3697 GIC 15 : appendStringInfoString(&result, "BOOLEAN");
3698 CBC 15 : break;
3699 15 : case TIMEOID:
3700 15 : if (typmod == -1)
3701 15 : appendStringInfoString(&result, "TIME");
3702 : else
3703 LBC 0 : appendStringInfo(&result, "TIME_%d", typmod);
3704 GIC 15 : break;
3705 15 : case TIMETZOID:
3706 15 : if (typmod == -1)
3707 CBC 15 : appendStringInfoString(&result, "TIME_WTZ");
3708 ECB : else
3709 UIC 0 : appendStringInfo(&result, "TIME_WTZ_%d", typmod);
3710 CBC 15 : break;
3711 GIC 15 : case TIMESTAMPOID:
3712 CBC 15 : if (typmod == -1)
3713 15 : appendStringInfoString(&result, "TIMESTAMP");
3714 : else
3715 UIC 0 : appendStringInfo(&result, "TIMESTAMP_%d", typmod);
3716 CBC 15 : break;
3717 GIC 15 : case TIMESTAMPTZOID:
3718 CBC 15 : if (typmod == -1)
3719 15 : appendStringInfoString(&result, "TIMESTAMP_WTZ");
3720 : else
3721 LBC 0 : appendStringInfo(&result, "TIMESTAMP_WTZ_%d", typmod);
3722 GIC 15 : break;
3723 15 : case DATEOID:
3724 15 : appendStringInfoString(&result, "DATE");
3725 15 : break;
3726 CBC 15 : case XMLOID:
3727 GIC 15 : appendStringInfoString(&result, "XML");
3728 CBC 15 : break;
3729 GIC 111 : default:
3730 ECB : {
3731 : HeapTuple tuple;
3732 : Form_pg_type typtuple;
3733 :
3734 CBC 111 : tuple = SearchSysCache1(TYPEOID, ObjectIdGetDatum(typeoid));
3735 GIC 111 : if (!HeapTupleIsValid(tuple))
3736 UIC 0 : elog(ERROR, "cache lookup failed for type %u", typeoid);
3737 GIC 111 : typtuple = (Form_pg_type) GETSTRUCT(tuple);
3738 :
3739 111 : appendStringInfoString(&result,
3740 CBC 111 : map_multipart_sql_identifier_to_xml_name((typtuple->typtype == TYPTYPE_DOMAIN) ? "Domain" : "UDT",
3741 GIC 111 : get_database_name(MyDatabaseId),
3742 111 : get_namespace_name(typtuple->typnamespace),
3743 111 : NameStr(typtuple->typname)));
3744 ECB :
3745 GIC 111 : ReleaseSysCache(tuple);
3746 ECB : }
3747 : }
3748 :
3749 GIC 405 : return result.data;
3750 : }
3751 :
3752 :
3753 : /*
3754 ECB : * Map a collection of SQL data types to XML Schema data types; see
3755 : * SQL/XML:2008 section 9.7.
3756 : */
3757 : static const char *
3758 GIC 48 : map_sql_typecoll_to_xmlschema_types(List *tupdesc_list)
3759 ECB : {
3760 GIC 48 : List *uniquetypes = NIL;
3761 : int i;
3762 : StringInfoData result;
3763 ECB : ListCell *cell0;
3764 :
3765 : /* extract all column types used in the set of TupleDescs */
3766 GIC 105 : foreach(cell0, tupdesc_list)
3767 : {
3768 57 : TupleDesc tupdesc = (TupleDesc) lfirst(cell0);
3769 :
3770 351 : for (i = 0; i < tupdesc->natts; i++)
3771 : {
3772 294 : Form_pg_attribute att = TupleDescAttr(tupdesc, i);
3773 :
3774 CBC 294 : if (att->attisdropped)
3775 GIC 12 : continue;
3776 282 : uniquetypes = list_append_unique_oid(uniquetypes, att->atttypid);
3777 : }
3778 : }
3779 :
3780 : /* add base types of domains */
3781 321 : foreach(cell0, uniquetypes)
3782 : {
3783 273 : Oid typid = lfirst_oid(cell0);
3784 CBC 273 : Oid basetypid = getBaseType(typid);
3785 ECB :
3786 GIC 273 : if (basetypid != typid)
3787 CBC 12 : uniquetypes = list_append_unique_oid(uniquetypes, basetypid);
3788 : }
3789 ECB :
3790 : /* Convert to textual form */
3791 CBC 48 : initStringInfo(&result);
3792 :
3793 GIC 321 : foreach(cell0, uniquetypes)
3794 : {
3795 273 : appendStringInfo(&result, "%s\n",
3796 ECB : map_sql_type_to_xmlschema_type(lfirst_oid(cell0),
3797 : -1));
3798 : }
3799 :
3800 GIC 48 : return result.data;
3801 : }
3802 ECB :
3803 :
3804 : /*
3805 : * Map an SQL data type to a named XML Schema data type; see
3806 : * SQL/XML:2008 sections 9.5 and 9.6.
3807 : *
3808 : * (The distinction between 9.5 and 9.6 is basically that 9.6 adds
3809 : * a name attribute, which this function does. The name-less version
3810 : * 9.5 doesn't appear to be required anywhere.)
3811 : */
3812 : static const char *
3813 GIC 273 : map_sql_type_to_xmlschema_type(Oid typeoid, int typmod)
3814 : {
3815 ECB : StringInfoData result;
3816 CBC 273 : const char *typename = map_sql_type_to_xml_name(typeoid, typmod);
3817 :
3818 GIC 273 : initStringInfo(&result);
3819 :
3820 CBC 273 : if (typeoid == XMLOID)
3821 : {
3822 GIC 12 : appendStringInfoString(&result,
3823 : "<xsd:complexType mixed=\"true\">\n"
3824 : " <xsd:sequence>\n"
3825 ECB : " <xsd:any name=\"element\" minOccurs=\"0\" maxOccurs=\"unbounded\" processContents=\"skip\"/>\n"
3826 : " </xsd:sequence>\n"
3827 : "</xsd:complexType>\n");
3828 : }
3829 : else
3830 : {
3831 CBC 261 : appendStringInfo(&result,
3832 : "<xsd:simpleType name=\"%s\">\n", typename);
3833 :
3834 261 : switch (typeoid)
3835 : {
3836 GIC 69 : case BPCHAROID:
3837 : case VARCHAROID:
3838 ECB : case TEXTOID:
3839 GIC 69 : appendStringInfoString(&result,
3840 : " <xsd:restriction base=\"xsd:string\">\n");
3841 69 : if (typmod != -1)
3842 UIC 0 : appendStringInfo(&result,
3843 : " <xsd:maxLength value=\"%d\"/>\n",
3844 : typmod - VARHDRSZ);
3845 GIC 69 : appendStringInfoString(&result, " </xsd:restriction>\n");
3846 69 : break;
3847 EUB :
3848 GIC 12 : case BYTEAOID:
3849 12 : appendStringInfo(&result,
3850 : " <xsd:restriction base=\"xsd:%s\">\n"
3851 : " </xsd:restriction>\n",
3852 12 : xmlbinary == XMLBINARY_BASE64 ? "base64Binary" : "hexBinary");
3853 12 : break;
3854 :
3855 12 : case NUMERICOID:
3856 GBC 12 : if (typmod != -1)
3857 UIC 0 : appendStringInfo(&result,
3858 EUB : " <xsd:restriction base=\"xsd:decimal\">\n"
3859 : " <xsd:totalDigits value=\"%d\"/>\n"
3860 : " <xsd:fractionDigits value=\"%d\"/>\n"
3861 : " </xsd:restriction>\n",
3862 UBC 0 : ((typmod - VARHDRSZ) >> 16) & 0xffff,
3863 UIC 0 : (typmod - VARHDRSZ) & 0xffff);
3864 GIC 12 : break;
3865 :
3866 12 : case INT2OID:
3867 GBC 12 : appendStringInfo(&result,
3868 : " <xsd:restriction base=\"xsd:short\">\n"
3869 EUB : " <xsd:maxInclusive value=\"%d\"/>\n"
3870 : " <xsd:minInclusive value=\"%d\"/>\n"
3871 : " </xsd:restriction>\n",
3872 : SHRT_MAX, SHRT_MIN);
3873 GIC 12 : break;
3874 EUB :
3875 GBC 48 : case INT4OID:
3876 48 : appendStringInfo(&result,
3877 EUB : " <xsd:restriction base=\"xsd:int\">\n"
3878 : " <xsd:maxInclusive value=\"%d\"/>\n"
3879 : " <xsd:minInclusive value=\"%d\"/>\n"
3880 : " </xsd:restriction>\n",
3881 : INT_MAX, INT_MIN);
3882 GBC 48 : break;
3883 :
3884 GIC 12 : case INT8OID:
3885 12 : appendStringInfo(&result,
3886 : " <xsd:restriction base=\"xsd:long\">\n"
3887 EUB : " <xsd:maxInclusive value=\"" INT64_FORMAT "\"/>\n"
3888 : " <xsd:minInclusive value=\"" INT64_FORMAT "\"/>\n"
3889 : " </xsd:restriction>\n",
3890 : PG_INT64_MAX,
3891 : PG_INT64_MIN);
3892 GBC 12 : break;
3893 :
3894 GIC 12 : case FLOAT4OID:
3895 12 : appendStringInfoString(&result,
3896 EUB : " <xsd:restriction base=\"xsd:float\"></xsd:restriction>\n");
3897 GIC 12 : break;
3898 :
3899 UIC 0 : case FLOAT8OID:
3900 0 : appendStringInfoString(&result,
3901 : " <xsd:restriction base=\"xsd:double\"></xsd:restriction>\n");
3902 0 : break;
3903 :
3904 CBC 12 : case BOOLOID:
3905 GIC 12 : appendStringInfoString(&result,
3906 : " <xsd:restriction base=\"xsd:boolean\"></xsd:restriction>\n");
3907 12 : break;
3908 ECB :
3909 GIC 24 : case TIMEOID:
3910 ECB : case TIMETZOID:
3911 : {
3912 CBC 24 : const char *tz = (typeoid == TIMETZOID ? "(\\+|-)\\p{Nd}{2}:\\p{Nd}{2}" : "");
3913 ECB :
3914 CBC 24 : if (typmod == -1)
3915 GIC 24 : appendStringInfo(&result,
3916 EUB : " <xsd:restriction base=\"xsd:time\">\n"
3917 ECB : " <xsd:pattern value=\"\\p{Nd}{2}:\\p{Nd}{2}:\\p{Nd}{2}(.\\p{Nd}+)?%s\"/>\n"
3918 : " </xsd:restriction>\n", tz);
3919 LBC 0 : else if (typmod == 0)
3920 0 : appendStringInfo(&result,
3921 : " <xsd:restriction base=\"xsd:time\">\n"
3922 EUB : " <xsd:pattern value=\"\\p{Nd}{2}:\\p{Nd}{2}:\\p{Nd}{2}%s\"/>\n"
3923 ECB : " </xsd:restriction>\n", tz);
3924 : else
3925 LBC 0 : appendStringInfo(&result,
3926 ECB : " <xsd:restriction base=\"xsd:time\">\n"
3927 : " <xsd:pattern value=\"\\p{Nd}{2}:\\p{Nd}{2}:\\p{Nd}{2}.\\p{Nd}{%d}%s\"/>\n"
3928 EUB : " </xsd:restriction>\n", typmod - VARHDRSZ, tz);
3929 GBC 24 : break;
3930 EUB : }
3931 ECB :
3932 CBC 24 : case TIMESTAMPOID:
3933 ECB : case TIMESTAMPTZOID:
3934 : {
3935 CBC 24 : const char *tz = (typeoid == TIMESTAMPTZOID ? "(\\+|-)\\p{Nd}{2}:\\p{Nd}{2}" : "");
3936 ECB :
3937 CBC 24 : if (typmod == -1)
3938 24 : appendStringInfo(&result,
3939 ECB : " <xsd:restriction base=\"xsd:dateTime\">\n"
3940 : " <xsd:pattern value=\"\\p{Nd}{4}-\\p{Nd}{2}-\\p{Nd}{2}T\\p{Nd}{2}:\\p{Nd}{2}:\\p{Nd}{2}(.\\p{Nd}+)?%s\"/>\n"
3941 : " </xsd:restriction>\n", tz);
3942 LBC 0 : else if (typmod == 0)
3943 0 : appendStringInfo(&result,
3944 EUB : " <xsd:restriction base=\"xsd:dateTime\">\n"
3945 : " <xsd:pattern value=\"\\p{Nd}{4}-\\p{Nd}{2}-\\p{Nd}{2}T\\p{Nd}{2}:\\p{Nd}{2}:\\p{Nd}{2}%s\"/>\n"
3946 : " </xsd:restriction>\n", tz);
3947 ECB : else
3948 LBC 0 : appendStringInfo(&result,
3949 ECB : " <xsd:restriction base=\"xsd:dateTime\">\n"
3950 : " <xsd:pattern value=\"\\p{Nd}{4}-\\p{Nd}{2}-\\p{Nd}{2}T\\p{Nd}{2}:\\p{Nd}{2}:\\p{Nd}{2}.\\p{Nd}{%d}%s\"/>\n"
3951 : " </xsd:restriction>\n", typmod - VARHDRSZ, tz);
3952 CBC 24 : break;
3953 : }
3954 EUB :
3955 CBC 12 : case DATEOID:
3956 12 : appendStringInfoString(&result,
3957 ECB : " <xsd:restriction base=\"xsd:date\">\n"
3958 : " <xsd:pattern value=\"\\p{Nd}{4}-\\p{Nd}{2}-\\p{Nd}{2}\"/>\n"
3959 : " </xsd:restriction>\n");
3960 GBC 12 : break;
3961 ECB :
3962 CBC 12 : default:
3963 12 : if (get_typtype(typeoid) == TYPTYPE_DOMAIN)
3964 ECB : {
3965 : Oid base_typeoid;
3966 GBC 12 : int32 base_typmod = -1;
3967 ECB :
3968 CBC 12 : base_typeoid = getBaseTypeAndTypmod(typeoid, &base_typmod);
3969 ECB :
3970 CBC 12 : appendStringInfo(&result,
3971 : " <xsd:restriction base=\"%s\"/>\n",
3972 EUB : map_sql_type_to_xml_name(base_typeoid, base_typmod));
3973 ECB : }
3974 CBC 12 : break;
3975 ECB : }
3976 CBC 261 : appendStringInfoString(&result, "</xsd:simpleType>\n");
3977 ECB : }
3978 :
3979 CBC 273 : return result.data;
3980 ECB : }
3981 :
3982 :
3983 : /*
3984 : * Map an SQL row to an XML element, taking the row from the active
3985 : * SPI cursor. See also SQL/XML:2008 section 9.10.
3986 : */
3987 EUB : static void
3988 CBC 156 : SPI_sql_row_to_xmlelement(uint64 rownum, StringInfo result, char *tablename,
3989 : bool nulls, bool tableforest,
3990 ECB : const char *targetns, bool top_level)
3991 : {
3992 : int i;
3993 : char *xmltn;
3994 :
3995 GIC 156 : if (tablename)
3996 CBC 114 : xmltn = map_sql_identifier_to_xml_name(tablename, true, false);
3997 : else
3998 : {
3999 GIC 42 : if (tableforest)
4000 CBC 18 : xmltn = "row";
4001 : else
4002 GIC 24 : xmltn = "table";
4003 : }
4004 :
4005 156 : if (tableforest)
4006 81 : xmldata_root_element_start(result, xmltn, NULL, targetns, top_level);
4007 : else
4008 75 : appendStringInfoString(result, "<row>\n");
4009 ECB :
4010 GIC 636 : for (i = 1; i <= SPI_tuptable->tupdesc->natts; i++)
4011 ECB : {
4012 : char *colname;
4013 : Datum colval;
4014 : bool isnull;
4015 :
4016 GIC 480 : colname = map_sql_identifier_to_xml_name(SPI_fname(SPI_tuptable->tupdesc, i),
4017 ECB : true, false);
4018 GIC 480 : colval = SPI_getbinval(SPI_tuptable->vals[rownum],
4019 CBC 480 : SPI_tuptable->tupdesc,
4020 : i,
4021 ECB : &isnull);
4022 GIC 480 : if (isnull)
4023 ECB : {
4024 GIC 57 : if (nulls)
4025 CBC 30 : appendStringInfo(result, " <%s xsi:nil=\"true\"/>\n", colname);
4026 ECB : }
4027 : else
4028 GIC 423 : appendStringInfo(result, " <%s>%s</%s>\n",
4029 : colname,
4030 : map_sql_value_to_xml_value(colval,
4031 423 : SPI_gettypeid(SPI_tuptable->tupdesc, i), true),
4032 ECB : colname);
4033 : }
4034 :
4035 CBC 156 : if (tableforest)
4036 : {
4037 81 : xmldata_root_element_end(result, xmltn);
4038 81 : appendStringInfoChar(result, '\n');
4039 : }
4040 : else
4041 GIC 75 : appendStringInfoString(result, "</row>\n\n");
4042 CBC 156 : }
4043 :
4044 ECB :
4045 : /*
4046 : * XPath related functions
4047 : */
4048 :
4049 : #ifdef USE_LIBXML
4050 :
4051 : /*
4052 : * Convert XML node to text.
4053 : *
4054 : * For attribute and text nodes, return the escaped text. For anything else,
4055 : * dump the whole subtree.
4056 : */
4057 : static text *
4058 GIC 96 : xml_xmlnodetoxmltype(xmlNodePtr cur, PgXmlErrorContext *xmlerrcxt)
4059 : {
4060 96 : xmltype *result = NULL;
4061 :
4062 96 : if (cur->type != XML_ATTRIBUTE_NODE && cur->type != XML_TEXT_NODE)
4063 81 : {
4064 CBC 81 : void (*volatile nodefree) (xmlNodePtr) = NULL;
4065 GIC 81 : volatile xmlBufferPtr buf = NULL;
4066 81 : volatile xmlNodePtr cur_copy = NULL;
4067 ECB :
4068 GIC 81 : PG_TRY();
4069 ECB : {
4070 : int bytes;
4071 :
4072 GIC 81 : buf = xmlBufferCreate();
4073 CBC 81 : if (buf == NULL || xmlerrcxt->err_occurred)
4074 UIC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4075 : "could not allocate xmlBuffer");
4076 :
4077 : /*
4078 : * Produce a dump of the node that we can serialize. xmlNodeDump
4079 : * does that, but the result of that function won't contain
4080 : * namespace definitions from ancestor nodes, so we first do a
4081 : * xmlCopyNode() which duplicates the node along with its required
4082 ECB : * namespace definitions.
4083 : *
4084 : * Some old libxml2 versions such as 2.7.6 produce partially
4085 : * broken XML_DOCUMENT_NODE nodes (unset content field) when
4086 : * copying them. xmlNodeDump of such a node works fine, but
4087 : * xmlFreeNode crashes; set us up to call xmlFreeDoc instead.
4088 : */
4089 GIC 81 : cur_copy = xmlCopyNode(cur, 1);
4090 CBC 81 : if (cur_copy == NULL || xmlerrcxt->err_occurred)
4091 UIC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4092 ECB : "could not copy node");
4093 GBC 162 : nodefree = (cur_copy->type == XML_DOCUMENT_NODE) ?
4094 GIC 81 : (void (*) (xmlNodePtr)) xmlFreeDoc : xmlFreeNode;
4095 :
4096 CBC 81 : bytes = xmlNodeDump(buf, NULL, cur_copy, 0, 0);
4097 81 : if (bytes == -1 || xmlerrcxt->err_occurred)
4098 UIC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4099 ECB : "could not dump node");
4100 :
4101 GIC 81 : result = xmlBuffer_to_xmltype(buf);
4102 : }
4103 LBC 0 : PG_FINALLY();
4104 ECB : {
4105 GIC 81 : if (nodefree)
4106 CBC 81 : nodefree(cur_copy);
4107 81 : if (buf)
4108 GBC 81 : xmlBufferFree(buf);
4109 : }
4110 GIC 81 : PG_END_TRY();
4111 : }
4112 : else
4113 EUB : {
4114 : xmlChar *str;
4115 ECB :
4116 GIC 15 : str = xmlXPathCastNodeToString(cur);
4117 CBC 15 : PG_TRY();
4118 ECB : {
4119 : /* Here we rely on XML having the same representation as TEXT */
4120 GIC 15 : char *escaped = escape_xml((char *) str);
4121 :
4122 15 : result = (xmltype *) cstring_to_text(escaped);
4123 15 : pfree(escaped);
4124 ECB : }
4125 UIC 0 : PG_FINALLY();
4126 ECB : {
4127 CBC 15 : xmlFree(str);
4128 : }
4129 GIC 15 : PG_END_TRY();
4130 : }
4131 :
4132 96 : return result;
4133 ECB : }
4134 :
4135 : /*
4136 : * Convert an XML XPath object (the result of evaluating an XPath expression)
4137 : * to an array of xml values, which are appended to astate. The function
4138 : * result value is the number of elements in the array.
4139 : *
4140 : * If "astate" is NULL then we don't generate the array value, but we still
4141 : * return the number of elements it would have had.
4142 : *
4143 : * Nodesets are converted to an array containing the nodes' textual
4144 : * representations. Primitive values (float, double, string) are converted
4145 : * to a single-element array containing the value's string representation.
4146 : */
4147 : static int
4148 CBC 270 : xml_xpathobjtoxmlarray(xmlXPathObjectPtr xpathobj,
4149 : ArrayBuildState *astate,
4150 EUB : PgXmlErrorContext *xmlerrcxt)
4151 : {
4152 GIC 270 : int result = 0;
4153 EUB : Datum datum;
4154 : Oid datumtype;
4155 ECB : char *result_str;
4156 :
4157 GIC 270 : switch (xpathobj->type)
4158 ECB : {
4159 GIC 249 : case XPATH_NODESET:
4160 CBC 249 : if (xpathobj->nodesetval != NULL)
4161 : {
4162 GIC 177 : result = xpathobj->nodesetval->nodeNr;
4163 CBC 177 : if (astate != NULL)
4164 : {
4165 ECB : int i;
4166 :
4167 GIC 84 : for (i = 0; i < result; i++)
4168 : {
4169 45 : datum = PointerGetDatum(xml_xmlnodetoxmltype(xpathobj->nodesetval->nodeTab[i],
4170 EUB : xmlerrcxt));
4171 GBC 45 : (void) accumArrayResult(astate, datum, false,
4172 : XMLOID, CurrentMemoryContext);
4173 : }
4174 : }
4175 : }
4176 249 : return result;
4177 :
4178 GIC 6 : case XPATH_BOOLEAN:
4179 6 : if (astate == NULL)
4180 LBC 0 : return 1;
4181 GIC 6 : datum = BoolGetDatum(xpathobj->boolval);
4182 6 : datumtype = BOOLOID;
4183 CBC 6 : break;
4184 :
4185 GIC 9 : case XPATH_NUMBER:
4186 CBC 9 : if (astate == NULL)
4187 GIC 6 : return 1;
4188 CBC 3 : datum = Float8GetDatum(xpathobj->floatval);
4189 3 : datumtype = FLOAT8OID;
4190 GIC 3 : break;
4191 :
4192 6 : case XPATH_STRING:
4193 GBC 6 : if (astate == NULL)
4194 UBC 0 : return 1;
4195 GIC 6 : datum = CStringGetDatum((char *) xpathobj->stringval);
4196 6 : datumtype = CSTRINGOID;
4197 6 : break;
4198 :
4199 UBC 0 : default:
4200 UIC 0 : elog(ERROR, "xpath expression result type %d is unsupported",
4201 : xpathobj->type);
4202 : return 0; /* keep compiler quiet */
4203 ECB : }
4204 :
4205 : /* Common code for scalar-value cases */
4206 CBC 15 : result_str = map_sql_value_to_xml_value(datum, datumtype, true);
4207 15 : datum = PointerGetDatum(cstring_to_xmltype(result_str));
4208 GIC 15 : (void) accumArrayResult(astate, datum, false,
4209 : XMLOID, CurrentMemoryContext);
4210 15 : return 1;
4211 ECB : }
4212 :
4213 :
4214 : /*
4215 : * Common code for xpath() and xmlexists()
4216 : *
4217 : * Evaluate XPath expression and return number of nodes in res_nitems
4218 : * and array of XML values in astate. Either of those pointers can be
4219 : * NULL if the corresponding result isn't wanted.
4220 : *
4221 : * It is up to the user to ensure that the XML passed is in fact
4222 : * an XML document - XPath doesn't work easily on fragments without
4223 : * a context node being known.
4224 : */
4225 : static void
4226 GIC 279 : xpath_internal(text *xpath_expr_text, xmltype *data, ArrayType *namespaces,
4227 ECB : int *res_nitems, ArrayBuildState *astate)
4228 : {
4229 : PgXmlErrorContext *xmlerrcxt;
4230 CBC 279 : volatile xmlParserCtxtPtr ctxt = NULL;
4231 GIC 279 : volatile xmlDocPtr doc = NULL;
4232 279 : volatile xmlXPathContextPtr xpathctx = NULL;
4233 279 : volatile xmlXPathCompExprPtr xpathcomp = NULL;
4234 279 : volatile xmlXPathObjectPtr xpathobj = NULL;
4235 : char *datastr;
4236 : int32 len;
4237 : int32 xpath_len;
4238 : xmlChar *string;
4239 ECB : xmlChar *xpath_expr;
4240 GIC 279 : size_t xmldecl_len = 0;
4241 : int i;
4242 : int ndim;
4243 : Datum *ns_names_uris;
4244 : bool *ns_names_uris_nulls;
4245 : int ns_count;
4246 ECB :
4247 : /*
4248 : * Namespace mappings are passed as text[]. If an empty array is passed
4249 : * (ndim = 0, "0-dimensional"), then there are no namespace mappings.
4250 : * Else, a 2-dimensional array with length of the second axis being equal
4251 : * to 2 should be passed, i.e., every subarray contains 2 elements, the
4252 : * first element defining the name, the second one the URI. Example:
4253 : * ARRAY[ARRAY['myns', 'http://example.com'], ARRAY['myns2',
4254 : * 'http://example2.com']].
4255 : */
4256 CBC 279 : ndim = namespaces ? ARR_NDIM(namespaces) : 0;
4257 279 : if (ndim != 0)
4258 : {
4259 ECB : int *dims;
4260 :
4261 CBC 63 : dims = ARR_DIMS(namespaces);
4262 :
4263 GIC 63 : if (ndim != 2 || dims[1] != 2)
4264 UIC 0 : ereport(ERROR,
4265 : (errcode(ERRCODE_DATA_EXCEPTION),
4266 : errmsg("invalid array for XML namespace mapping"),
4267 ECB : errdetail("The array must be two-dimensional with length of the second axis equal to 2.")));
4268 :
4269 CBC 63 : Assert(ARR_ELEMTYPE(namespaces) == TEXTOID);
4270 ECB :
4271 GNC 63 : deconstruct_array_builtin(namespaces, TEXTOID,
4272 : &ns_names_uris, &ns_names_uris_nulls,
4273 : &ns_count);
4274 :
4275 CBC 63 : Assert((ns_count % 2) == 0); /* checked above */
4276 63 : ns_count /= 2; /* count pairs only */
4277 : }
4278 : else
4279 ECB : {
4280 GIC 216 : ns_names_uris = NULL;
4281 216 : ns_names_uris_nulls = NULL;
4282 CBC 216 : ns_count = 0;
4283 : }
4284 :
4285 GIC 279 : datastr = VARDATA(data);
4286 CBC 279 : len = VARSIZE(data) - VARHDRSZ;
4287 GIC 279 : xpath_len = VARSIZE_ANY_EXHDR(xpath_expr_text);
4288 CBC 279 : if (xpath_len == 0)
4289 3 : ereport(ERROR,
4290 : (errcode(ERRCODE_DATA_EXCEPTION),
4291 : errmsg("empty XPath expression")));
4292 ECB :
4293 CBC 276 : string = pg_xmlCharStrndup(datastr, len);
4294 GIC 276 : xpath_expr = pg_xmlCharStrndup(VARDATA_ANY(xpath_expr_text), xpath_len);
4295 :
4296 : /*
4297 : * In a UTF8 database, skip any xml declaration, which might assert
4298 : * another encoding. Ignore parse_xml_decl() failure, letting
4299 : * xmlCtxtReadMemory() report parse errors. Documentation disclaims
4300 : * xpath() support for non-ASCII data in non-UTF8 databases, so leave
4301 : * those scenarios bug-compatible with historical behavior.
4302 : */
4303 276 : if (GetDatabaseEncoding() == PG_UTF8)
4304 276 : parse_xml_decl(string, &xmldecl_len, NULL, NULL, NULL);
4305 :
4306 276 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
4307 :
4308 276 : PG_TRY();
4309 ECB : {
4310 GIC 276 : xmlInitParser();
4311 ECB :
4312 : /*
4313 : * redundant XML parsing (two parsings for the same value during one
4314 : * command execution are possible)
4315 : */
4316 CBC 276 : ctxt = xmlNewParserCtxt();
4317 276 : if (ctxt == NULL || xmlerrcxt->err_occurred)
4318 UIC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4319 ECB : "could not allocate parser context");
4320 GIC 552 : doc = xmlCtxtReadMemory(ctxt, (char *) string + xmldecl_len,
4321 276 : len - xmldecl_len, NULL, NULL, 0);
4322 276 : if (doc == NULL || xmlerrcxt->err_occurred)
4323 CBC 6 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INVALID_XML_DOCUMENT,
4324 ECB : "could not parse XML document");
4325 GBC 270 : xpathctx = xmlXPathNewContext(doc);
4326 GIC 270 : if (xpathctx == NULL || xmlerrcxt->err_occurred)
4327 UIC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4328 : "could not allocate XPath context");
4329 GIC 270 : xpathctx->node = (xmlNodePtr) doc;
4330 :
4331 : /* register namespaces, if any */
4332 270 : if (ns_count > 0)
4333 : {
4334 126 : for (i = 0; i < ns_count; i++)
4335 : {
4336 : char *ns_name;
4337 : char *ns_uri;
4338 :
4339 63 : if (ns_names_uris_nulls[i * 2] ||
4340 CBC 63 : ns_names_uris_nulls[i * 2 + 1])
4341 LBC 0 : ereport(ERROR,
4342 EUB : (errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
4343 : errmsg("neither namespace name nor URI may be null")));
4344 CBC 63 : ns_name = TextDatumGetCString(ns_names_uris[i * 2]);
4345 63 : ns_uri = TextDatumGetCString(ns_names_uris[i * 2 + 1]);
4346 GIC 63 : if (xmlXPathRegisterNs(xpathctx,
4347 ECB : (xmlChar *) ns_name,
4348 : (xmlChar *) ns_uri) != 0)
4349 UBC 0 : ereport(ERROR, /* is this an internal error??? */
4350 : (errmsg("could not register XML namespace with name \"%s\" and URI \"%s\"",
4351 : ns_name, ns_uri)));
4352 ECB : }
4353 : }
4354 EUB :
4355 GIC 270 : xpathcomp = xmlXPathCompile(xpath_expr);
4356 CBC 270 : if (xpathcomp == NULL || xmlerrcxt->err_occurred)
4357 LBC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
4358 ECB : "invalid XPath expression");
4359 :
4360 : /*
4361 : * Version 2.6.27 introduces a function named
4362 : * xmlXPathCompiledEvalToBoolean, which would be enough for xmlexists,
4363 : * but we can derive the existence by whether any nodes are returned,
4364 : * thereby preventing a library version upgrade and keeping the code
4365 : * the same.
4366 : */
4367 CBC 270 : xpathobj = xmlXPathCompiledEval(xpathcomp, xpathctx);
4368 270 : if (xpathobj == NULL || xmlerrcxt->err_occurred)
4369 UIC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
4370 : "could not create XPath object");
4371 ECB :
4372 : /*
4373 : * Extract the results as requested.
4374 : */
4375 GIC 270 : if (res_nitems != NULL)
4376 GBC 216 : *res_nitems = xml_xpathobjtoxmlarray(xpathobj, astate, xmlerrcxt);
4377 : else
4378 CBC 54 : (void) xml_xpathobjtoxmlarray(xpathobj, astate, xmlerrcxt);
4379 : }
4380 6 : PG_CATCH();
4381 : {
4382 GIC 6 : if (xpathobj)
4383 LBC 0 : xmlXPathFreeObject(xpathobj);
4384 GIC 6 : if (xpathcomp)
4385 UIC 0 : xmlXPathFreeCompExpr(xpathcomp);
4386 GIC 6 : if (xpathctx)
4387 UIC 0 : xmlXPathFreeContext(xpathctx);
4388 GIC 6 : if (doc)
4389 6 : xmlFreeDoc(doc);
4390 6 : if (ctxt)
4391 6 : xmlFreeParserCtxt(ctxt);
4392 :
4393 6 : pg_xml_done(xmlerrcxt, true);
4394 :
4395 6 : PG_RE_THROW();
4396 : }
4397 270 : PG_END_TRY();
4398 :
4399 CBC 270 : xmlXPathFreeObject(xpathobj);
4400 GIC 270 : xmlXPathFreeCompExpr(xpathcomp);
4401 270 : xmlXPathFreeContext(xpathctx);
4402 270 : xmlFreeDoc(doc);
4403 CBC 270 : xmlFreeParserCtxt(ctxt);
4404 :
4405 GIC 270 : pg_xml_done(xmlerrcxt, false);
4406 270 : }
4407 : #endif /* USE_LIBXML */
4408 ECB :
4409 : /*
4410 : * Evaluate XPath expression and return array of XML values.
4411 : *
4412 : * As we have no support of XQuery sequences yet, this function seems
4413 : * to be the most useful one (array of XML functions plays a role of
4414 : * some kind of substitution for XQuery sequences).
4415 : */
4416 : Datum
4417 GIC 63 : xpath(PG_FUNCTION_ARGS)
4418 ECB : {
4419 : #ifdef USE_LIBXML
4420 CBC 63 : text *xpath_expr_text = PG_GETARG_TEXT_PP(0);
4421 GIC 63 : xmltype *data = PG_GETARG_XML_P(1);
4422 CBC 63 : ArrayType *namespaces = PG_GETARG_ARRAYTYPE_P(2);
4423 : ArrayBuildState *astate;
4424 :
4425 GIC 63 : astate = initArrayResult(XMLOID, CurrentMemoryContext, true);
4426 63 : xpath_internal(xpath_expr_text, data, namespaces,
4427 ECB : NULL, astate);
4428 GNC 54 : PG_RETURN_DATUM(makeArrayResult(astate, CurrentMemoryContext));
4429 ECB : #else
4430 : NO_XML_SUPPORT();
4431 EUB : return 0;
4432 ECB : #endif
4433 : }
4434 :
4435 : /*
4436 : * Determines if the node specified by the supplied XPath exists
4437 : * in a given XML document, returning a boolean.
4438 : */
4439 : Datum
4440 CBC 99 : xmlexists(PG_FUNCTION_ARGS)
4441 ECB : {
4442 : #ifdef USE_LIBXML
4443 CBC 99 : text *xpath_expr_text = PG_GETARG_TEXT_PP(0);
4444 99 : xmltype *data = PG_GETARG_XML_P(1);
4445 EUB : int res_nitems;
4446 ECB :
4447 CBC 99 : xpath_internal(xpath_expr_text, data, NULL,
4448 ECB : &res_nitems, NULL);
4449 :
4450 GBC 99 : PG_RETURN_BOOL(res_nitems > 0);
4451 EUB : #else
4452 : NO_XML_SUPPORT();
4453 : return 0;
4454 : #endif
4455 : }
4456 :
4457 ECB : /*
4458 : * Determines if the node specified by the supplied XPath exists
4459 : * in a given XML document, returning a boolean. Differs from
4460 : * xmlexists as it supports namespaces and is not defined in SQL/XML.
4461 : */
4462 : Datum
4463 GIC 117 : xpath_exists(PG_FUNCTION_ARGS)
4464 : {
4465 : #ifdef USE_LIBXML
4466 117 : text *xpath_expr_text = PG_GETARG_TEXT_PP(0);
4467 117 : xmltype *data = PG_GETARG_XML_P(1);
4468 117 : ArrayType *namespaces = PG_GETARG_ARRAYTYPE_P(2);
4469 : int res_nitems;
4470 :
4471 117 : xpath_internal(xpath_expr_text, data, namespaces,
4472 : &res_nitems, NULL);
4473 :
4474 117 : PG_RETURN_BOOL(res_nitems > 0);
4475 : #else
4476 : NO_XML_SUPPORT();
4477 ECB : return 0;
4478 : #endif
4479 : }
4480 :
4481 : /*
4482 : * Functions for checking well-formed-ness
4483 : */
4484 :
4485 : #ifdef USE_LIBXML
4486 : static bool
4487 GIC 57 : wellformed_xml(text *data, XmlOptionType xmloption_arg)
4488 : {
4489 : xmlDocPtr doc;
4490 GNC 57 : ErrorSaveContext escontext = {T_ErrorSaveContext};
4491 :
4492 : /*
4493 : * We'll report "true" if no soft error is reported by xml_parse().
4494 : */
4495 57 : doc = xml_parse(data, xmloption_arg, true,
4496 : GetDatabaseEncoding(), NULL, NULL, (Node *) &escontext);
4497 GIC 57 : if (doc)
4498 30 : xmlFreeDoc(doc);
4499 ECB :
4500 GNC 57 : return !escontext.error_occurred;
4501 : }
4502 : #endif
4503 :
4504 ECB : Datum
4505 GIC 45 : xml_is_well_formed(PG_FUNCTION_ARGS)
4506 ECB : {
4507 EUB : #ifdef USE_LIBXML
4508 GIC 45 : text *data = PG_GETARG_TEXT_PP(0);
4509 :
4510 45 : PG_RETURN_BOOL(wellformed_xml(data, xmloption));
4511 : #else
4512 ECB : NO_XML_SUPPORT();
4513 : return 0;
4514 : #endif /* not USE_LIBXML */
4515 : }
4516 :
4517 : Datum
4518 CBC 6 : xml_is_well_formed_document(PG_FUNCTION_ARGS)
4519 ECB : {
4520 : #ifdef USE_LIBXML
4521 GIC 6 : text *data = PG_GETARG_TEXT_PP(0);
4522 :
4523 CBC 6 : PG_RETURN_BOOL(wellformed_xml(data, XMLOPTION_DOCUMENT));
4524 ECB : #else
4525 : NO_XML_SUPPORT();
4526 : return 0;
4527 : #endif /* not USE_LIBXML */
4528 : }
4529 :
4530 : Datum
4531 CBC 6 : xml_is_well_formed_content(PG_FUNCTION_ARGS)
4532 ECB : {
4533 : #ifdef USE_LIBXML
4534 GIC 6 : text *data = PG_GETARG_TEXT_PP(0);
4535 :
4536 CBC 6 : PG_RETURN_BOOL(wellformed_xml(data, XMLOPTION_CONTENT));
4537 ECB : #else
4538 : NO_XML_SUPPORT();
4539 : return 0;
4540 : #endif /* not USE_LIBXML */
4541 : }
4542 :
4543 : /*
4544 : * support functions for XMLTABLE
4545 : *
4546 : */
4547 : #ifdef USE_LIBXML
4548 :
4549 : /*
4550 : * Returns private data from executor state. Ensure validity by check with
4551 : * MAGIC number.
4552 : */
4553 : static inline XmlTableBuilderData *
4554 GIC 77370 : GetXmlTableBuilderPrivateData(TableFuncScanState *state, const char *fname)
4555 : {
4556 : XmlTableBuilderData *result;
4557 :
4558 77370 : if (!IsA(state, TableFuncScanState))
4559 LBC 0 : elog(ERROR, "%s called with invalid TableFuncScanState", fname);
4560 CBC 77370 : result = (XmlTableBuilderData *) state->opaque;
4561 GBC 77370 : if (result->magic != XMLTABLE_CONTEXT_MAGIC)
4562 UIC 0 : elog(ERROR, "%s called with invalid TableFuncScanState", fname);
4563 ECB :
4564 CBC 77370 : return result;
4565 ECB : }
4566 : #endif
4567 :
4568 : /*
4569 : * XmlTableInitOpaque
4570 EUB : * Fill in TableFuncScanState->opaque for XmlTable processor; initialize
4571 : * the XML parser.
4572 ECB : *
4573 : * Note: Because we call pg_xml_init() here and pg_xml_done() in
4574 : * XmlTableDestroyOpaque, it is critical for robustness that no other
4575 : * executor nodes run until this node is processed to completion. Caller
4576 : * must execute this to completion (probably filling a tuplestore to exhaust
4577 : * this node in a single pass) instead of using row-per-call mode.
4578 : */
4579 : static void
4580 GIC 132 : XmlTableInitOpaque(TableFuncScanState *state, int natts)
4581 : {
4582 ECB : #ifdef USE_LIBXML
4583 CBC 132 : volatile xmlParserCtxtPtr ctxt = NULL;
4584 EUB : XmlTableBuilderData *xtCxt;
4585 : PgXmlErrorContext *xmlerrcxt;
4586 :
4587 CBC 132 : xtCxt = palloc0(sizeof(XmlTableBuilderData));
4588 132 : xtCxt->magic = XMLTABLE_CONTEXT_MAGIC;
4589 132 : xtCxt->natts = natts;
4590 GIC 132 : xtCxt->xpathscomp = palloc0(sizeof(xmlXPathCompExprPtr) * natts);
4591 :
4592 GBC 132 : xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL);
4593 :
4594 GIC 132 : PG_TRY();
4595 : {
4596 132 : xmlInitParser();
4597 :
4598 CBC 132 : ctxt = xmlNewParserCtxt();
4599 132 : if (ctxt == NULL || xmlerrcxt->err_occurred)
4600 UBC 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4601 : "could not allocate parser context");
4602 : }
4603 UIC 0 : PG_CATCH();
4604 : {
4605 0 : if (ctxt != NULL)
4606 0 : xmlFreeParserCtxt(ctxt);
4607 :
4608 0 : pg_xml_done(xmlerrcxt, true);
4609 :
4610 LBC 0 : PG_RE_THROW();
4611 ECB : }
4612 GBC 132 : PG_END_TRY();
4613 :
4614 GIC 132 : xtCxt->xmlerrcxt = xmlerrcxt;
4615 132 : xtCxt->ctxt = ctxt;
4616 :
4617 132 : state->opaque = xtCxt;
4618 ECB : #else
4619 : NO_XML_SUPPORT();
4620 : #endif /* not USE_LIBXML */
4621 CBC 132 : }
4622 :
4623 ECB : /*
4624 : * XmlTableSetDocument
4625 : * Install the input document
4626 EUB : */
4627 ECB : static void
4628 GBC 132 : XmlTableSetDocument(TableFuncScanState *state, Datum value)
4629 ECB : {
4630 EUB : #ifdef USE_LIBXML
4631 ECB : XmlTableBuilderData *xtCxt;
4632 CBC 132 : xmltype *xmlval = DatumGetXmlP(value);
4633 ECB : char *str;
4634 : xmlChar *xstr;
4635 : int length;
4636 CBC 132 : volatile xmlDocPtr doc = NULL;
4637 GIC 132 : volatile xmlXPathContextPtr xpathcxt = NULL;
4638 ECB :
4639 GIC 132 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableSetDocument");
4640 ECB :
4641 : /*
4642 : * Use out function for casting to string (remove encoding property). See
4643 : * comment in xml_out.
4644 : */
4645 CBC 132 : str = xml_out_internal(xmlval, 0);
4646 ECB :
4647 GIC 132 : length = strlen(str);
4648 CBC 132 : xstr = pg_xmlCharStrndup(str, length);
4649 ECB :
4650 GIC 132 : PG_TRY();
4651 : {
4652 132 : doc = xmlCtxtReadMemory(xtCxt->ctxt, (char *) xstr, length, NULL, NULL, 0);
4653 132 : if (doc == NULL || xtCxt->xmlerrcxt->err_occurred)
4654 UIC 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_INVALID_XML_DOCUMENT,
4655 : "could not parse XML document");
4656 GIC 132 : xpathcxt = xmlXPathNewContext(doc);
4657 132 : if (xpathcxt == NULL || xtCxt->xmlerrcxt->err_occurred)
4658 UIC 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
4659 : "could not allocate XPath context");
4660 CBC 132 : xpathcxt->node = (xmlNodePtr) doc;
4661 : }
4662 UIC 0 : PG_CATCH();
4663 ECB : {
4664 LBC 0 : if (xpathcxt != NULL)
4665 0 : xmlXPathFreeContext(xpathcxt);
4666 UIC 0 : if (doc != NULL)
4667 0 : xmlFreeDoc(doc);
4668 ECB :
4669 LBC 0 : PG_RE_THROW();
4670 : }
4671 CBC 132 : PG_END_TRY();
4672 :
4673 GIC 132 : xtCxt->doc = doc;
4674 132 : xtCxt->xpathcxt = xpathcxt;
4675 : #else
4676 : NO_XML_SUPPORT();
4677 : #endif /* not USE_LIBXML */
4678 132 : }
4679 :
4680 : /*
4681 : * XmlTableSetNamespace
4682 : * Add a namespace declaration
4683 ECB : */
4684 : static void
4685 GIC 9 : XmlTableSetNamespace(TableFuncScanState *state, const char *name, const char *uri)
4686 ECB : {
4687 : #ifdef USE_LIBXML
4688 : XmlTableBuilderData *xtCxt;
4689 :
4690 CBC 9 : if (name == NULL)
4691 GIC 3 : ereport(ERROR,
4692 : (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
4693 ECB : errmsg("DEFAULT namespace is not supported")));
4694 GIC 6 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableSetNamespace");
4695 :
4696 6 : if (xmlXPathRegisterNs(xtCxt->xpathcxt,
4697 6 : pg_xmlCharStrndup(name, strlen(name)),
4698 6 : pg_xmlCharStrndup(uri, strlen(uri))))
4699 UIC 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_DATA_EXCEPTION,
4700 : "could not set XML namespace");
4701 : #else
4702 : NO_XML_SUPPORT();
4703 : #endif /* not USE_LIBXML */
4704 GIC 6 : }
4705 :
4706 ECB : /*
4707 : * XmlTableSetRowFilter
4708 : * Install the row-filter Xpath expression.
4709 : */
4710 : static void
4711 CBC 129 : XmlTableSetRowFilter(TableFuncScanState *state, const char *path)
4712 : {
4713 : #ifdef USE_LIBXML
4714 ECB : XmlTableBuilderData *xtCxt;
4715 : xmlChar *xstr;
4716 :
4717 CBC 129 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableSetRowFilter");
4718 :
4719 GIC 129 : if (*path == '\0')
4720 UIC 0 : ereport(ERROR,
4721 : (errcode(ERRCODE_DATA_EXCEPTION),
4722 : errmsg("row path filter must not be empty string")));
4723 :
4724 GIC 129 : xstr = pg_xmlCharStrndup(path, strlen(path));
4725 :
4726 129 : xtCxt->xpathcomp = xmlXPathCompile(xstr);
4727 129 : if (xtCxt->xpathcomp == NULL || xtCxt->xmlerrcxt->err_occurred)
4728 UIC 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_SYNTAX_ERROR,
4729 : "invalid XPath expression");
4730 ECB : #else
4731 : NO_XML_SUPPORT();
4732 : #endif /* not USE_LIBXML */
4733 CBC 129 : }
4734 :
4735 : /*
4736 : * XmlTableSetColumnFilter
4737 : * Install the column-filter Xpath expression, for the given column.
4738 ECB : */
4739 : static void
4740 CBC 387 : XmlTableSetColumnFilter(TableFuncScanState *state, const char *path, int colnum)
4741 ECB : {
4742 : #ifdef USE_LIBXML
4743 : XmlTableBuilderData *xtCxt;
4744 : xmlChar *xstr;
4745 :
4746 GNC 387 : Assert(PointerIsValid(path));
4747 :
4748 CBC 387 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableSetColumnFilter");
4749 :
4750 GIC 387 : if (*path == '\0')
4751 LBC 0 : ereport(ERROR,
4752 : (errcode(ERRCODE_DATA_EXCEPTION),
4753 ECB : errmsg("column path filter must not be empty string")));
4754 :
4755 GIC 387 : xstr = pg_xmlCharStrndup(path, strlen(path));
4756 :
4757 387 : xtCxt->xpathscomp[colnum] = xmlXPathCompile(xstr);
4758 387 : if (xtCxt->xpathscomp[colnum] == NULL || xtCxt->xmlerrcxt->err_occurred)
4759 UIC 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_DATA_EXCEPTION,
4760 : "invalid XPath expression");
4761 ECB : #else
4762 : NO_XML_SUPPORT();
4763 : #endif /* not USE_LIBXML */
4764 CBC 387 : }
4765 :
4766 ECB : /*
4767 : * XmlTableFetchRow
4768 : * Prepare the next "current" tuple for upcoming GetValue calls.
4769 : * Returns false if the row-filter expression returned no more rows.
4770 : */
4771 : static bool
4772 GIC 11121 : XmlTableFetchRow(TableFuncScanState *state)
4773 : {
4774 ECB : #ifdef USE_LIBXML
4775 : XmlTableBuilderData *xtCxt;
4776 :
4777 CBC 11121 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableFetchRow");
4778 :
4779 ECB : /* Propagate our own error context to libxml2 */
4780 GIC 11121 : xmlSetStructuredErrorFunc((void *) xtCxt->xmlerrcxt, xml_errorHandler);
4781 :
4782 11121 : if (xtCxt->xpathobj == NULL)
4783 : {
4784 129 : xtCxt->xpathobj = xmlXPathCompiledEval(xtCxt->xpathcomp, xtCxt->xpathcxt);
4785 129 : if (xtCxt->xpathobj == NULL || xtCxt->xmlerrcxt->err_occurred)
4786 UIC 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
4787 : "could not create XPath object");
4788 :
4789 GIC 129 : xtCxt->row_count = 0;
4790 : }
4791 :
4792 11121 : if (xtCxt->xpathobj->type == XPATH_NODESET)
4793 : {
4794 11121 : if (xtCxt->xpathobj->nodesetval != NULL)
4795 : {
4796 11121 : if (xtCxt->row_count++ < xtCxt->xpathobj->nodesetval->nodeNr)
4797 CBC 10998 : return true;
4798 : }
4799 : }
4800 :
4801 123 : return false;
4802 EUB : #else
4803 ECB : NO_XML_SUPPORT();
4804 : return false;
4805 EUB : #endif /* not USE_LIBXML */
4806 : }
4807 ECB :
4808 : /*
4809 : * XmlTableGetValue
4810 : * Return the value for column number 'colnum' for the current row. If
4811 : * column -1 is requested, return representation of the whole row.
4812 : *
4813 : * This leaks memory, so be sure to reset often the context in which it's
4814 : * called.
4815 : */
4816 : static Datum
4817 GIC 65463 : XmlTableGetValue(TableFuncScanState *state, int colnum,
4818 : Oid typid, int32 typmod, bool *isnull)
4819 : {
4820 : #ifdef USE_LIBXML
4821 : XmlTableBuilderData *xtCxt;
4822 65463 : Datum result = (Datum) 0;
4823 ECB : xmlNodePtr cur;
4824 GIC 65463 : char *cstr = NULL;
4825 65463 : volatile xmlXPathObjectPtr xpathobj = NULL;
4826 ECB :
4827 GIC 65463 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableGetValue");
4828 :
4829 65463 : Assert(xtCxt->xpathobj &&
4830 ECB : xtCxt->xpathobj->type == XPATH_NODESET &&
4831 : xtCxt->xpathobj->nodesetval != NULL);
4832 :
4833 : /* Propagate our own error context to libxml2 */
4834 GIC 65463 : xmlSetStructuredErrorFunc((void *) xtCxt->xmlerrcxt, xml_errorHandler);
4835 ECB :
4836 GIC 65463 : *isnull = false;
4837 ECB :
4838 GIC 65463 : cur = xtCxt->xpathobj->nodesetval->nodeTab[xtCxt->row_count - 1];
4839 ECB :
4840 GIC 65463 : Assert(xtCxt->xpathscomp[colnum] != NULL);
4841 ECB :
4842 CBC 65463 : PG_TRY();
4843 EUB : {
4844 : /* Set current node as entry point for XPath evaluation */
4845 GIC 65463 : xtCxt->xpathcxt->node = cur;
4846 EUB :
4847 : /* Evaluate column path */
4848 GBC 65463 : xpathobj = xmlXPathCompiledEval(xtCxt->xpathscomp[colnum], xtCxt->xpathcxt);
4849 65463 : if (xpathobj == NULL || xtCxt->xmlerrcxt->err_occurred)
4850 UIC 0 : xml_ereport(xtCxt->xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR,
4851 EUB : "could not create XPath object");
4852 :
4853 : /*
4854 : * There are four possible cases, depending on the number of nodes
4855 ECB : * returned by the XPath expression and the type of the target column:
4856 : * a) XPath returns no nodes. b) The target type is XML (return all
4857 : * as XML). For non-XML return types: c) One node (return content).
4858 : * d) Multiple nodes (error).
4859 : */
4860 CBC 65463 : if (xpathobj->type == XPATH_NODESET)
4861 : {
4862 GIC 65448 : int count = 0;
4863 :
4864 CBC 65448 : if (xpathobj->nodesetval != NULL)
4865 GIC 65343 : count = xpathobj->nodesetval->nodeNr;
4866 :
4867 65448 : if (xpathobj->nodesetval == NULL || count == 0)
4868 : {
4869 11122 : *isnull = true;
4870 : }
4871 ECB : else
4872 : {
4873 GIC 54326 : if (typid == XMLOID)
4874 : {
4875 ECB : text *textstr;
4876 : StringInfoData str;
4877 :
4878 : /* Concatenate serialized values */
4879 CBC 36 : initStringInfo(&str);
4880 87 : for (int i = 0; i < count; i++)
4881 : {
4882 ECB : textstr =
4883 GIC 51 : xml_xmlnodetoxmltype(xpathobj->nodesetval->nodeTab[i],
4884 : xtCxt->xmlerrcxt);
4885 :
4886 51 : appendStringInfoText(&str, textstr);
4887 : }
4888 CBC 36 : cstr = str.data;
4889 : }
4890 ECB : else
4891 : {
4892 : xmlChar *str;
4893 :
4894 GIC 54290 : if (count > 1)
4895 CBC 3 : ereport(ERROR,
4896 ECB : (errcode(ERRCODE_CARDINALITY_VIOLATION),
4897 EUB : errmsg("more than one value returned by column XPath expression")));
4898 :
4899 CBC 54287 : str = xmlXPathCastNodeSetToString(xpathobj->nodesetval);
4900 54287 : cstr = str ? xml_pstrdup_and_free(str) : "";
4901 EUB : }
4902 : }
4903 ECB : }
4904 GIC 15 : else if (xpathobj->type == XPATH_STRING)
4905 EUB : {
4906 : /* Content should be escaped when target will be XML */
4907 GBC 9 : if (typid == XMLOID)
4908 3 : cstr = escape_xml((char *) xpathobj->stringval);
4909 EUB : else
4910 GBC 6 : cstr = (char *) xpathobj->stringval;
4911 : }
4912 6 : else if (xpathobj->type == XPATH_BOOLEAN)
4913 : {
4914 ECB : char typcategory;
4915 : bool typispreferred;
4916 : xmlChar *str;
4917 :
4918 : /* Allow implicit casting from boolean to numbers */
4919 GIC 3 : get_type_category_preferred(typid, &typcategory, &typispreferred);
4920 :
4921 CBC 3 : if (typcategory != TYPCATEGORY_NUMERIC)
4922 GIC 3 : str = xmlXPathCastBooleanToString(xpathobj->boolval);
4923 : else
4924 UIC 0 : str = xmlXPathCastNumberToString(xmlXPathCastBooleanToNumber(xpathobj->boolval));
4925 :
4926 GIC 3 : cstr = xml_pstrdup_and_free(str);
4927 : }
4928 CBC 3 : else if (xpathobj->type == XPATH_NUMBER)
4929 : {
4930 : xmlChar *str;
4931 :
4932 GIC 3 : str = xmlXPathCastNumberToString(xpathobj->floatval);
4933 CBC 3 : cstr = xml_pstrdup_and_free(str);
4934 ECB : }
4935 : else
4936 UIC 0 : elog(ERROR, "unexpected XPath object type %u", xpathobj->type);
4937 ECB :
4938 : /*
4939 : * By here, either cstr contains the result value, or the isnull flag
4940 : * has been set.
4941 : */
4942 GBC 65460 : Assert(cstr || *isnull);
4943 :
4944 GIC 65460 : if (!*isnull)
4945 54338 : result = InputFunctionCall(&state->in_functions[colnum],
4946 : cstr,
4947 CBC 54338 : state->typioparams[colnum],
4948 : typmod);
4949 : }
4950 GIC 3 : PG_FINALLY();
4951 : {
4952 65463 : if (xpathobj != NULL)
4953 65463 : xmlXPathFreeObject(xpathobj);
4954 ECB : }
4955 GIC 65463 : PG_END_TRY();
4956 :
4957 65460 : return result;
4958 : #else
4959 : NO_XML_SUPPORT();
4960 ECB : return 0;
4961 : #endif /* not USE_LIBXML */
4962 : }
4963 EUB :
4964 : /*
4965 : * XmlTableDestroyOpaque
4966 : * Release all libxml2 resources
4967 ECB : */
4968 : static void
4969 CBC 132 : XmlTableDestroyOpaque(TableFuncScanState *state)
4970 ECB : {
4971 EUB : #ifdef USE_LIBXML
4972 : XmlTableBuilderData *xtCxt;
4973 :
4974 GIC 132 : xtCxt = GetXmlTableBuilderPrivateData(state, "XmlTableDestroyOpaque");
4975 :
4976 ECB : /* Propagate our own error context to libxml2 */
4977 GIC 132 : xmlSetStructuredErrorFunc((void *) xtCxt->xmlerrcxt, xml_errorHandler);
4978 :
4979 132 : if (xtCxt->xpathscomp != NULL)
4980 : {
4981 : int i;
4982 :
4983 CBC 558 : for (i = 0; i < xtCxt->natts; i++)
4984 GIC 426 : if (xtCxt->xpathscomp[i] != NULL)
4985 387 : xmlXPathFreeCompExpr(xtCxt->xpathscomp[i]);
4986 : }
4987 :
4988 132 : if (xtCxt->xpathobj != NULL)
4989 CBC 129 : xmlXPathFreeObject(xtCxt->xpathobj);
4990 GIC 132 : if (xtCxt->xpathcomp != NULL)
4991 CBC 129 : xmlXPathFreeCompExpr(xtCxt->xpathcomp);
4992 GIC 132 : if (xtCxt->xpathcxt != NULL)
4993 CBC 132 : xmlXPathFreeContext(xtCxt->xpathcxt);
4994 GBC 132 : if (xtCxt->doc != NULL)
4995 GIC 132 : xmlFreeDoc(xtCxt->doc);
4996 132 : if (xtCxt->ctxt != NULL)
4997 132 : xmlFreeParserCtxt(xtCxt->ctxt);
4998 ECB :
4999 GIC 132 : pg_xml_done(xtCxt->xmlerrcxt, true);
5000 ECB :
5001 : /* not valid anymore */
5002 GBC 132 : xtCxt->magic = 0;
5003 GIC 132 : state->opaque = NULL;
5004 :
5005 : #else
5006 : NO_XML_SUPPORT();
5007 ECB : #endif /* not USE_LIBXML */
5008 GIC 132 : }
|
