Age Owner TLA Line data Source code
1 : /*
2 : * contrib/xml2/xpath.c
3 : *
4 : * Parser interface for DOM-based parser (libxml) rather than
5 : * stream-based SAX-type parser
6 : */
7 : #include "postgres.h"
8 :
9 : #include "access/htup_details.h"
10 : #include "executor/spi.h"
11 : #include "fmgr.h"
12 : #include "funcapi.h"
13 : #include "lib/stringinfo.h"
14 : #include "miscadmin.h"
15 : #include "utils/builtins.h"
16 : #include "utils/xml.h"
17 :
18 : /* libxml includes */
19 :
20 : #include <libxml/xpath.h>
21 : #include <libxml/tree.h>
22 : #include <libxml/xmlmemory.h>
23 : #include <libxml/xmlerror.h>
24 : #include <libxml/parserInternals.h>
25 :
6054 tgl 26 CBC 1 : PG_MODULE_MAGIC;
27 :
28 : /* exported for use by xslt_proc.c */
29 :
30 : PgXmlErrorContext *pgxml_parser_init(PgXmlStrictness strictness);
31 :
32 : /* workspace for pgxml_xpath() */
33 :
34 : typedef struct
35 : {
36 : xmlDocPtr doctree;
37 : xmlXPathContextPtr ctxt;
38 : xmlXPathObjectPtr res;
39 : } xpath_workspace;
40 :
41 : /* local declarations */
42 :
43 : static xmlChar *pgxmlNodeSetToText(xmlNodeSetPtr nodeset,
44 : xmlChar *toptagname, xmlChar *septagname,
45 : xmlChar *plainsep);
46 :
47 : static text *pgxml_result_to_text(xmlXPathObjectPtr res, xmlChar *toptag,
48 : xmlChar *septag, xmlChar *plainsep);
49 :
50 : static xmlChar *pgxml_texttoxmlchar(text *textstring);
51 :
52 : static xmlXPathObjectPtr pgxml_xpath(text *document, xmlChar *xpath,
53 : xpath_workspace *workspace);
54 :
55 : static void cleanup_workspace(xpath_workspace *workspace);
56 :
57 :
58 : /*
59 : * Initialize for xml parsing.
60 : *
61 : * As with the underlying pg_xml_init function, calls to this MUST be followed
62 : * by a PG_TRY block that guarantees that pg_xml_done is called.
63 : */
64 : PgXmlErrorContext *
4281 65 11 : pgxml_parser_init(PgXmlStrictness strictness)
66 : {
67 : PgXmlErrorContext *xmlerrcxt;
68 :
69 : /* Set up error handling (we share the core's error handler) */
70 11 : xmlerrcxt = pg_xml_init(strictness);
71 :
72 : /* Note: we're assuming an elog cannot be thrown by the following calls */
73 :
74 : /* Initialize libxml */
4788 75 11 : xmlInitParser();
76 :
6974 bruce 77 11 : xmlSubstituteEntitiesDefault(1);
78 11 : xmlLoadExtDtdDefaultValue = 1;
79 :
4281 tgl 80 11 : return xmlerrcxt;
81 : }
82 :
83 :
84 : /* Encodes special characters (<, >, &, " and \r) as XML entities */
85 :
6702 bruce 86 1 : PG_FUNCTION_INFO_V1(xml_encode_special_chars);
87 :
88 : Datum
6702 bruce 89 UBC 0 : xml_encode_special_chars(PG_FUNCTION_ARGS)
90 : {
2219 noah 91 0 : text *tin = PG_GETARG_TEXT_PP(0);
92 : text *tout;
93 : xmlChar *ts,
94 : *tt;
95 :
6702 bruce 96 0 : ts = pgxml_texttoxmlchar(tin);
97 :
98 0 : tt = xmlEncodeSpecialChars(NULL, ts);
99 :
100 0 : pfree(ts);
101 :
5453 tgl 102 0 : tout = cstring_to_text((char *) tt);
103 :
6702 bruce 104 0 : xmlFree(tt);
105 :
106 0 : PG_RETURN_TEXT_P(tout);
107 : }
108 :
109 : /*
110 : * Function translates a nodeset into a text representation
111 : *
112 : * iterates over each node in the set and calls xmlNodeDump to write it to
113 : * an xmlBuffer -from which an xmlChar * string is returned.
114 : *
115 : * each representation is surrounded by <tagname> ... </tagname>
116 : *
117 : * plainsep is an ordinary (not tag) separator - if used, then nodes are
118 : * cast to string as output method
119 : */
120 : static xmlChar *
6974 121 0 : pgxmlNodeSetToText(xmlNodeSetPtr nodeset,
122 : xmlChar *toptagname,
123 : xmlChar *septagname,
124 : xmlChar *plainsep)
125 : {
126 : xmlBufferPtr buf;
127 : xmlChar *result;
128 : int i;
129 :
130 0 : buf = xmlBufferCreate();
131 :
132 0 : if ((toptagname != NULL) && (xmlStrlen(toptagname) > 0))
133 : {
134 0 : xmlBufferWriteChar(buf, "<");
135 0 : xmlBufferWriteCHAR(buf, toptagname);
136 0 : xmlBufferWriteChar(buf, ">");
137 : }
138 0 : if (nodeset != NULL)
139 : {
140 0 : for (i = 0; i < nodeset->nodeNr; i++)
141 : {
6797 142 0 : if (plainsep != NULL)
143 : {
144 0 : xmlBufferWriteCHAR(buf,
2118 tgl 145 0 : xmlXPathCastNodeToString(nodeset->nodeTab[i]));
146 :
147 : /* If this isn't the last entry, write the plain sep. */
6797 bruce 148 0 : if (i < (nodeset->nodeNr) - 1)
5747 tgl 149 0 : xmlBufferWriteChar(buf, (char *) plainsep);
150 : }
151 : else
152 : {
6797 bruce 153 0 : if ((septagname != NULL) && (xmlStrlen(septagname) > 0))
154 : {
155 0 : xmlBufferWriteChar(buf, "<");
156 0 : xmlBufferWriteCHAR(buf, septagname);
157 0 : xmlBufferWriteChar(buf, ">");
158 : }
159 0 : xmlNodeDump(buf,
160 0 : nodeset->nodeTab[i]->doc,
161 0 : nodeset->nodeTab[i],
162 : 1, 0);
163 :
164 0 : if ((septagname != NULL) && (xmlStrlen(septagname) > 0))
165 : {
166 0 : xmlBufferWriteChar(buf, "</");
167 0 : xmlBufferWriteCHAR(buf, septagname);
168 0 : xmlBufferWriteChar(buf, ">");
169 : }
170 : }
171 : }
172 : }
173 :
6974 174 0 : if ((toptagname != NULL) && (xmlStrlen(toptagname) > 0))
175 : {
176 0 : xmlBufferWriteChar(buf, "</");
177 0 : xmlBufferWriteCHAR(buf, toptagname);
178 0 : xmlBufferWriteChar(buf, ">");
179 : }
180 0 : result = xmlStrdup(buf->content);
181 0 : xmlBufferFree(buf);
182 0 : return result;
183 : }
184 :
185 :
186 : /* Translate a PostgreSQL "varlena" -i.e. a variable length parameter
187 : * into the libxml2 representation
188 : */
189 : static xmlChar *
190 0 : pgxml_texttoxmlchar(text *textstring)
191 : {
5453 tgl 192 0 : return (xmlChar *) text_to_cstring(textstring);
193 : }
194 :
195 : /* Publicly visible XPath functions */
196 :
197 : /*
198 : * This is a "raw" xpath function. Check that it returns child elements
199 : * properly
200 : */
6974 bruce 201 CBC 1 : PG_FUNCTION_INFO_V1(xpath_nodeset);
202 :
203 : Datum
6974 bruce 204 UBC 0 : xpath_nodeset(PG_FUNCTION_ARGS)
205 : {
2219 noah 206 0 : text *document = PG_GETARG_TEXT_PP(0);
2118 tgl 207 0 : text *xpathsupp = PG_GETARG_TEXT_PP(1); /* XPath expression */
2219 noah 208 0 : xmlChar *toptag = pgxml_texttoxmlchar(PG_GETARG_TEXT_PP(2));
209 0 : xmlChar *septag = pgxml_texttoxmlchar(PG_GETARG_TEXT_PP(3));
210 : xmlChar *xpath;
211 : text *xpres;
212 : xmlXPathObjectPtr res;
213 : xpath_workspace workspace;
214 :
4517 tgl 215 0 : xpath = pgxml_texttoxmlchar(xpathsupp);
216 :
217 0 : res = pgxml_xpath(document, xpath, &workspace);
218 :
219 0 : xpres = pgxml_result_to_text(res, toptag, septag, NULL);
220 :
221 0 : cleanup_workspace(&workspace);
222 :
6752 neilc 223 0 : pfree(xpath);
224 :
6797 bruce 225 0 : if (xpres == NULL)
226 0 : PG_RETURN_NULL();
6974 227 0 : PG_RETURN_TEXT_P(xpres);
228 : }
229 :
230 : /*
231 : * The following function is almost identical, but returns the elements in
232 : * a list.
233 : */
6974 bruce 234 CBC 1 : PG_FUNCTION_INFO_V1(xpath_list);
235 :
236 : Datum
6974 bruce 237 UBC 0 : xpath_list(PG_FUNCTION_ARGS)
238 : {
2219 noah 239 0 : text *document = PG_GETARG_TEXT_PP(0);
2118 tgl 240 0 : text *xpathsupp = PG_GETARG_TEXT_PP(1); /* XPath expression */
2219 noah 241 0 : xmlChar *plainsep = pgxml_texttoxmlchar(PG_GETARG_TEXT_PP(2));
242 : xmlChar *xpath;
243 : text *xpres;
244 : xmlXPathObjectPtr res;
245 : xpath_workspace workspace;
246 :
4517 tgl 247 0 : xpath = pgxml_texttoxmlchar(xpathsupp);
248 :
249 0 : res = pgxml_xpath(document, xpath, &workspace);
250 :
251 0 : xpres = pgxml_result_to_text(res, NULL, NULL, plainsep);
252 :
253 0 : cleanup_workspace(&workspace);
254 :
6752 neilc 255 0 : pfree(xpath);
256 :
6797 bruce 257 0 : if (xpres == NULL)
258 0 : PG_RETURN_NULL();
6974 259 0 : PG_RETURN_TEXT_P(xpres);
260 : }
261 :
262 :
6974 bruce 263 CBC 2 : PG_FUNCTION_INFO_V1(xpath_string);
264 :
265 : Datum
266 1 : xpath_string(PG_FUNCTION_ARGS)
267 : {
2219 noah 268 1 : text *document = PG_GETARG_TEXT_PP(0);
2118 tgl 269 1 : text *xpathsupp = PG_GETARG_TEXT_PP(1); /* XPath expression */
270 : xmlChar *xpath;
271 : int32 pathsize;
272 : text *xpres;
273 : xmlXPathObjectPtr res;
274 : xpath_workspace workspace;
275 :
2219 noah 276 1 : pathsize = VARSIZE_ANY_EXHDR(xpathsupp);
277 :
278 : /*
279 : * We encapsulate the supplied path with "string()" = 8 chars + 1 for NUL
280 : * at end
281 : */
282 : /* We could try casting to string using the libxml function? */
283 :
6797 bruce 284 1 : xpath = (xmlChar *) palloc(pathsize + 9);
2997 tgl 285 1 : memcpy((char *) xpath, "string(", 7);
2219 noah 286 1 : memcpy((char *) (xpath + 7), VARDATA_ANY(xpathsupp), pathsize);
6797 bruce 287 1 : xpath[pathsize + 7] = ')';
288 1 : xpath[pathsize + 8] = '\0';
289 :
4517 tgl 290 1 : res = pgxml_xpath(document, xpath, &workspace);
291 :
292 1 : xpres = pgxml_result_to_text(res, NULL, NULL, NULL);
293 :
294 1 : cleanup_workspace(&workspace);
295 :
6752 neilc 296 1 : pfree(xpath);
297 :
6797 bruce 298 1 : if (xpres == NULL)
299 1 : PG_RETURN_NULL();
6974 bruce 300 UBC 0 : PG_RETURN_TEXT_P(xpres);
301 : }
302 :
303 :
6974 bruce 304 CBC 1 : PG_FUNCTION_INFO_V1(xpath_number);
305 :
306 : Datum
6974 bruce 307 UBC 0 : xpath_number(PG_FUNCTION_ARGS)
308 : {
2219 noah 309 0 : text *document = PG_GETARG_TEXT_PP(0);
2118 tgl 310 0 : text *xpathsupp = PG_GETARG_TEXT_PP(1); /* XPath expression */
311 : xmlChar *xpath;
312 : float4 fRes;
313 : xmlXPathObjectPtr res;
314 : xpath_workspace workspace;
315 :
6974 bruce 316 0 : xpath = pgxml_texttoxmlchar(xpathsupp);
317 :
4517 tgl 318 0 : res = pgxml_xpath(document, xpath, &workspace);
319 :
6752 neilc 320 0 : pfree(xpath);
321 :
6974 bruce 322 0 : if (res == NULL)
6797 323 0 : PG_RETURN_NULL();
324 :
6974 325 0 : fRes = xmlXPathCastToNumber(res);
326 :
4517 tgl 327 0 : cleanup_workspace(&workspace);
328 :
6974 bruce 329 0 : if (xmlXPathIsNaN(fRes))
6797 330 0 : PG_RETURN_NULL();
331 :
6974 332 0 : PG_RETURN_FLOAT4(fRes);
333 : }
334 :
335 :
6974 bruce 336 CBC 1 : PG_FUNCTION_INFO_V1(xpath_bool);
337 :
338 : Datum
6974 bruce 339 UBC 0 : xpath_bool(PG_FUNCTION_ARGS)
340 : {
2219 noah 341 0 : text *document = PG_GETARG_TEXT_PP(0);
2118 tgl 342 0 : text *xpathsupp = PG_GETARG_TEXT_PP(1); /* XPath expression */
343 : xmlChar *xpath;
344 : int bRes;
345 : xmlXPathObjectPtr res;
346 : xpath_workspace workspace;
347 :
6974 bruce 348 0 : xpath = pgxml_texttoxmlchar(xpathsupp);
349 :
4517 tgl 350 0 : res = pgxml_xpath(document, xpath, &workspace);
351 :
6752 neilc 352 0 : pfree(xpath);
353 :
6974 bruce 354 0 : if (res == NULL)
6797 355 0 : PG_RETURN_BOOL(false);
356 :
6974 357 0 : bRes = xmlXPathCastToBoolean(res);
358 :
4517 tgl 359 0 : cleanup_workspace(&workspace);
360 :
6974 bruce 361 0 : PG_RETURN_BOOL(bRes);
362 : }
363 :
364 :
365 :
366 : /* Core function to evaluate XPath query */
367 :
368 : static xmlXPathObjectPtr
4517 tgl 369 CBC 1 : pgxml_xpath(text *document, xmlChar *xpath, xpath_workspace *workspace)
370 : {
2219 noah 371 1 : int32 docsize = VARSIZE_ANY_EXHDR(document);
372 : PgXmlErrorContext *xmlerrcxt;
373 : xmlXPathCompExprPtr comppath;
374 :
4517 tgl 375 1 : workspace->doctree = NULL;
376 1 : workspace->ctxt = NULL;
377 1 : workspace->res = NULL;
378 :
4281 379 1 : xmlerrcxt = pgxml_parser_init(PG_XML_STRICTNESS_LEGACY);
380 :
381 1 : PG_TRY();
382 : {
2219 noah 383 1 : workspace->doctree = xmlParseMemory((char *) VARDATA_ANY(document),
384 : docsize);
4281 tgl 385 1 : if (workspace->doctree != NULL)
386 : {
4281 tgl 387 UBC 0 : workspace->ctxt = xmlXPathNewContext(workspace->doctree);
388 0 : workspace->ctxt->node = xmlDocGetRootElement(workspace->doctree);
389 :
390 : /* compile the path */
391 0 : comppath = xmlXPathCompile(xpath);
392 0 : if (comppath == NULL)
393 0 : xml_ereport(xmlerrcxt, ERROR, ERRCODE_EXTERNAL_ROUTINE_EXCEPTION,
394 : "XPath Syntax Error");
395 :
396 : /* Now evaluate the path expression. */
397 0 : workspace->res = xmlXPathCompiledEval(comppath, workspace->ctxt);
398 :
399 0 : xmlXPathFreeCompExpr(comppath);
400 : }
401 : }
402 0 : PG_CATCH();
403 : {
4517 404 0 : cleanup_workspace(workspace);
405 :
4281 406 0 : pg_xml_done(xmlerrcxt, true);
407 :
408 0 : PG_RE_THROW();
409 : }
4281 tgl 410 CBC 1 : PG_END_TRY();
411 :
412 1 : if (workspace->res == NULL)
4517 413 1 : cleanup_workspace(workspace);
414 :
4281 415 1 : pg_xml_done(xmlerrcxt, false);
416 :
417 1 : return workspace->res;
418 : }
419 :
420 : /* Clean up after processing the result of pgxml_xpath() */
421 : static void
4517 422 2 : cleanup_workspace(xpath_workspace *workspace)
423 : {
424 2 : if (workspace->res)
4517 tgl 425 UBC 0 : xmlXPathFreeObject(workspace->res);
4517 tgl 426 CBC 2 : workspace->res = NULL;
427 2 : if (workspace->ctxt)
4517 tgl 428 UBC 0 : xmlXPathFreeContext(workspace->ctxt);
4517 tgl 429 CBC 2 : workspace->ctxt = NULL;
430 2 : if (workspace->doctree)
4517 tgl 431 UBC 0 : xmlFreeDoc(workspace->doctree);
4517 tgl 432 CBC 2 : workspace->doctree = NULL;
433 2 : }
434 :
435 : static text *
6797 bruce 436 1 : pgxml_result_to_text(xmlXPathObjectPtr res,
437 : xmlChar *toptag,
438 : xmlChar *septag,
439 : xmlChar *plainsep)
440 : {
441 : xmlChar *xpresstr;
442 : text *xpres;
443 :
444 1 : if (res == NULL)
445 1 : return NULL;
446 :
6974 bruce 447 UBC 0 : switch (res->type)
448 : {
449 0 : case XPATH_NODESET:
450 0 : xpresstr = pgxmlNodeSetToText(res->nodesetval,
451 : toptag,
452 : septag, plainsep);
453 0 : break;
454 :
455 0 : case XPATH_STRING:
456 0 : xpresstr = xmlStrdup(res->stringval);
457 0 : break;
458 :
459 0 : default:
6248 neilc 460 0 : elog(NOTICE, "unsupported XQuery result: %d", res->type);
5747 tgl 461 0 : xpresstr = xmlStrdup((const xmlChar *) "<unsupported/>");
462 : }
463 :
464 : /* Now convert this result back to text */
5453 465 0 : xpres = cstring_to_text((char *) xpresstr);
466 :
467 : /* Free various storage */
6974 bruce 468 0 : xmlFree(xpresstr);
469 :
470 0 : return xpres;
471 : }
472 :
473 : /*
474 : * xpath_table is a table function. It needs some tidying (as do the
475 : * other functions here!
476 : */
6974 bruce 477 CBC 2 : PG_FUNCTION_INFO_V1(xpath_table);
478 :
479 : Datum
6797 480 5 : xpath_table(PG_FUNCTION_ARGS)
481 : {
482 : /* Function parameters */
4788 tgl 483 5 : char *pkeyfield = text_to_cstring(PG_GETARG_TEXT_PP(0));
484 5 : char *xmlfield = text_to_cstring(PG_GETARG_TEXT_PP(1));
485 5 : char *relname = text_to_cstring(PG_GETARG_TEXT_PP(2));
486 5 : char *xpathset = text_to_cstring(PG_GETARG_TEXT_PP(3));
487 5 : char *condition = text_to_cstring(PG_GETARG_TEXT_PP(4));
488 :
489 : /* SPI (input tuple) support */
490 : SPITupleTable *tuptable;
491 : HeapTuple spi_tuple;
492 : TupleDesc spi_tupdesc;
493 :
494 :
6797 bruce 495 5 : ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo;
496 : AttInMetadata *attinmeta;
497 :
498 : char **values;
499 : xmlChar **xpaths;
500 : char *pos;
5747 tgl 501 5 : const char *pathsep = "|";
502 :
503 : int numpaths;
504 : int ret;
505 : uint64 proc;
506 : int j;
507 : int rownr; /* For issuing multiple rows from one original
508 : * document */
509 : bool had_values; /* To determine end of nodeset results */
510 : StringInfoData query_buf;
511 : PgXmlErrorContext *xmlerrcxt;
4281 512 5 : volatile xmlDocPtr doctree = NULL;
513 :
173 michael 514 5 : InitMaterializedSRF(fcinfo, MAT_SRF_USE_EXPECTED_DESC);
515 :
516 : /* must have at least one output column (for the pkey) */
397 517 5 : if (rsinfo->setDesc->natts < 1)
4788 tgl 518 UBC 0 : ereport(ERROR,
519 : (errcode(ERRCODE_SYNTAX_ERROR),
520 : errmsg("xpath_table must have at least one output column")));
521 :
522 : /*
523 : * At the moment we assume that the returned attributes make sense for the
524 : * XPath specified (i.e. we trust the caller). It's not fatal if they get
525 : * it wrong - the input function for the column type will raise an error
526 : * if the path result can't be converted into the correct binary
527 : * representation.
528 : */
529 :
397 michael 530 CBC 5 : attinmeta = TupleDescGetAttInMetadata(rsinfo->setDesc);
531 :
532 5 : values = (char **) palloc(rsinfo->setDesc->natts * sizeof(char *));
533 5 : xpaths = (xmlChar **) palloc(rsinfo->setDesc->natts * sizeof(xmlChar *));
534 :
535 : /*
536 : * Split XPaths. xpathset is a writable CString.
537 : *
538 : * Note that we stop splitting once we've done all needed for tupdesc
539 : */
6797 bruce 540 5 : numpaths = 0;
541 5 : pos = xpathset;
397 michael 542 7 : while (numpaths < (rsinfo->setDesc->natts - 1))
543 : {
4788 tgl 544 5 : xpaths[numpaths++] = (xmlChar *) pos;
6797 bruce 545 5 : pos = strstr(pos, pathsep);
546 5 : if (pos != NULL)
547 : {
548 2 : *pos = '\0';
549 2 : pos++;
550 : }
551 : else
4788 tgl 552 3 : break;
553 : }
554 :
555 : /* Now build query */
6248 neilc 556 5 : initStringInfo(&query_buf);
557 :
558 : /* Build initial sql statement */
559 5 : appendStringInfo(&query_buf, "SELECT %s, %s FROM %s WHERE %s",
560 : pkeyfield,
561 : xmlfield,
562 : relname,
563 : condition);
564 :
6797 bruce 565 5 : if ((ret = SPI_connect()) < 0)
6797 bruce 566 UBC 0 : elog(ERROR, "xpath_table: SPI_connect returned %d", ret);
567 :
6248 neilc 568 CBC 5 : if ((ret = SPI_exec(query_buf.data, 0)) != SPI_OK_SELECT)
4788 tgl 569 UBC 0 : elog(ERROR, "xpath_table: SPI execution failed for query %s",
570 : query_buf.data);
571 :
6797 bruce 572 CBC 5 : proc = SPI_processed;
573 5 : tuptable = SPI_tuptable;
574 5 : spi_tupdesc = tuptable->tupdesc;
575 :
576 : /*
577 : * Check that SPI returned correct result. If you put a comma into one of
578 : * the function parameters, this will catch it when the SPI query returns
579 : * e.g. 3 columns.
580 : */
581 5 : if (spi_tupdesc->natts != 2)
582 : {
6797 bruce 583 UBC 0 : ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
584 : errmsg("expression returning multiple columns is not valid in parameter list"),
585 : errdetail("Expected two columns in SPI result, got %d.", spi_tupdesc->natts)));
586 : }
587 :
588 : /*
589 : * Setup the parser. This should happen after we are done evaluating the
590 : * query, in case it calls functions that set up libxml differently.
591 : */
4281 tgl 592 CBC 5 : xmlerrcxt = pgxml_parser_init(PG_XML_STRICTNESS_LEGACY);
593 :
594 5 : PG_TRY();
595 : {
596 : /* For each row i.e. document returned from SPI */
597 : uint64 i;
598 :
3955 bruce 599 10 : for (i = 0; i < proc; i++)
600 : {
601 : char *pkey;
602 : char *xmldoc;
603 : xmlXPathContextPtr ctxt;
604 : xmlXPathObjectPtr res;
605 : xmlChar *resstr;
606 : xmlXPathCompExprPtr comppath;
607 : HeapTuple ret_tuple;
608 :
609 : /* Extract the row data as C Strings */
610 5 : spi_tuple = tuptable->vals[i];
611 5 : pkey = SPI_getvalue(spi_tuple, spi_tupdesc, 1);
612 5 : xmldoc = SPI_getvalue(spi_tuple, spi_tupdesc, 2);
613 :
614 : /*
615 : * Clear the values array, so that not-well-formed documents
616 : * return NULL in all columns. Note that this also means that
617 : * spare columns will be NULL.
618 : */
397 michael 619 15 : for (j = 0; j < rsinfo->setDesc->natts; j++)
3955 bruce 620 10 : values[j] = NULL;
621 :
622 : /* Insert primary key */
623 5 : values[0] = pkey;
624 :
625 : /* Parse the document */
626 5 : if (xmldoc)
627 5 : doctree = xmlParseMemory(xmldoc, strlen(xmldoc));
628 : else /* treat NULL as not well-formed */
3955 bruce 629 UBC 0 : doctree = NULL;
630 :
3955 bruce 631 CBC 5 : if (doctree == NULL)
632 : {
633 : /* not well-formed, so output all-NULL tuple */
3955 bruce 634 UBC 0 : ret_tuple = BuildTupleFromCStrings(attinmeta, values);
397 michael 635 0 : tuplestore_puttuple(rsinfo->setResult, ret_tuple);
3955 bruce 636 0 : heap_freetuple(ret_tuple);
637 : }
638 : else
639 : {
640 : /* New loop here - we have to deal with nodeset results */
3955 bruce 641 CBC 5 : rownr = 0;
642 :
643 : do
644 : {
645 : /* Now evaluate the set of xpaths. */
646 8 : had_values = false;
647 18 : for (j = 0; j < numpaths; j++)
648 : {
649 10 : ctxt = xmlXPathNewContext(doctree);
650 10 : ctxt->node = xmlDocGetRootElement(doctree);
651 :
652 : /* compile the path */
653 10 : comppath = xmlXPathCompile(xpaths[j]);
654 10 : if (comppath == NULL)
3955 bruce 655 UBC 0 : xml_ereport(xmlerrcxt, ERROR,
656 : ERRCODE_EXTERNAL_ROUTINE_EXCEPTION,
657 : "XPath Syntax Error");
658 :
659 : /* Now evaluate the path expression. */
3955 bruce 660 CBC 10 : res = xmlXPathCompiledEval(comppath, ctxt);
661 10 : xmlXPathFreeCompExpr(comppath);
662 :
663 10 : if (res != NULL)
664 : {
665 10 : switch (res->type)
666 : {
667 10 : case XPATH_NODESET:
668 : /* We see if this nodeset has enough nodes */
669 10 : if (res->nodesetval != NULL &&
670 10 : rownr < res->nodesetval->nodeNr)
671 : {
672 4 : resstr = xmlXPathCastNodeToString(res->nodesetval->nodeTab[rownr]);
673 4 : had_values = true;
674 : }
675 : else
676 6 : resstr = NULL;
677 :
678 10 : break;
679 :
3955 bruce 680 UBC 0 : case XPATH_STRING:
681 0 : resstr = xmlStrdup(res->stringval);
682 0 : break;
683 :
684 0 : default:
685 0 : elog(NOTICE, "unsupported XQuery result: %d", res->type);
686 0 : resstr = xmlStrdup((const xmlChar *) "<unsupported/>");
687 : }
688 :
689 : /*
690 : * Insert this into the appropriate column in the
691 : * result tuple.
692 : */
3955 bruce 693 CBC 10 : values[j + 1] = (char *) resstr;
694 : }
695 10 : xmlXPathFreeContext(ctxt);
696 : }
697 :
698 : /* Now add the tuple to the output, if there is one. */
699 8 : if (had_values)
700 : {
701 3 : ret_tuple = BuildTupleFromCStrings(attinmeta, values);
397 michael 702 3 : tuplestore_puttuple(rsinfo->setResult, ret_tuple);
3955 bruce 703 3 : heap_freetuple(ret_tuple);
704 : }
705 :
706 8 : rownr++;
707 8 : } while (had_values);
708 : }
709 :
710 5 : if (doctree != NULL)
711 5 : xmlFreeDoc(doctree);
712 5 : doctree = NULL;
713 :
714 5 : if (pkey)
715 5 : pfree(pkey);
716 5 : if (xmldoc)
717 5 : pfree(xmldoc);
718 : }
719 : }
4281 tgl 720 UBC 0 : PG_CATCH();
721 : {
722 0 : if (doctree != NULL)
723 0 : xmlFreeDoc(doctree);
724 :
725 0 : pg_xml_done(xmlerrcxt, true);
726 :
727 0 : PG_RE_THROW();
728 : }
4281 tgl 729 CBC 5 : PG_END_TRY();
730 :
731 5 : if (doctree != NULL)
4281 tgl 732 UBC 0 : xmlFreeDoc(doctree);
733 :
4281 tgl 734 CBC 5 : pg_xml_done(xmlerrcxt, false);
735 :
6797 bruce 736 5 : SPI_finish();
737 :
738 : /*
739 : * SFRM_Materialize mode expects us to return a NULL Datum. The actual
740 : * tuples are in our tuplestore and passed back through rsinfo->setResult.
741 : * rsinfo->setDesc is set to the tuple description that we actually used
742 : * to build our tuples with, so the caller can verify we did what it was
743 : * expecting.
744 : */
745 5 : return (Datum) 0;
746 : }
|
