|
| 1 | +/* Parser interface for DOM-based parser (libxml) rather than |
| 2 | + stream-based SAX-type parser */ |
| 3 | + |
| 4 | +#include"postgres.h" |
| 5 | +#include"fmgr.h" |
| 6 | + |
| 7 | +/* libxml includes */ |
| 8 | + |
| 9 | +#include<libxml/xpath.h> |
| 10 | +#include<libxml/tree.h> |
| 11 | +#include<libxml/xmlmemory.h> |
| 12 | + |
| 13 | +/* declarations */ |
| 14 | + |
| 15 | +staticvoid*pgxml_palloc(size_tsize); |
| 16 | +staticvoid*pgxml_repalloc(void*ptr,size_tsize); |
| 17 | +staticvoidpgxml_pfree(void*ptr); |
| 18 | +staticchar*pgxml_pstrdup(constchar*string); |
| 19 | + |
| 20 | +staticvoidpgxml_parser_init(); |
| 21 | + |
| 22 | +staticxmlChar*pgxmlNodeSetToText(xmlNodeSetPtrnodeset,xmlDocPtrdoc, |
| 23 | +xmlChar*toptagname,xmlChar*septagname, |
| 24 | +intformat); |
| 25 | + |
| 26 | +staticxmlChar*pgxml_texttoxmlchar(text*textstring); |
| 27 | + |
| 28 | + |
| 29 | +Datumpgxml_parse(PG_FUNCTION_ARGS); |
| 30 | +Datumpgxml_xpath(PG_FUNCTION_ARGS); |
| 31 | + |
| 32 | +/* memory handling passthrough functions (e.g. palloc, pstrdup are |
| 33 | + currently macros, and the others might become so...) */ |
| 34 | + |
| 35 | +staticvoid*pgxml_palloc(size_tsize) |
| 36 | +{ |
| 37 | +returnpalloc(size); |
| 38 | +} |
| 39 | + |
| 40 | +staticvoid*pgxml_repalloc(void*ptr,size_tsize) |
| 41 | +{ |
| 42 | +returnrepalloc(ptr,size); |
| 43 | +} |
| 44 | + |
| 45 | +staticvoidpgxml_pfree(void*ptr) |
| 46 | +{ |
| 47 | +returnpfree(ptr); |
| 48 | +} |
| 49 | + |
| 50 | +staticchar*pgxml_pstrdup(constchar*string) |
| 51 | +{ |
| 52 | +returnpstrdup(string); |
| 53 | +} |
| 54 | + |
| 55 | +staticvoidpgxml_parser_init() |
| 56 | +{ |
| 57 | +/* This code should also set parser settings from user-supplied |
| 58 | + info. Quite how these settings are made is another matter :) */ |
| 59 | + |
| 60 | +xmlMemSetup(pgxml_pfree,pgxml_palloc,pgxml_repalloc,pgxml_pstrdup); |
| 61 | +xmlInitParser(); |
| 62 | + |
| 63 | +} |
| 64 | + |
| 65 | + |
| 66 | +/* Returns true if document is well-formed */ |
| 67 | + |
| 68 | +PG_FUNCTION_INFO_V1(pgxml_parse); |
| 69 | + |
| 70 | +Datum |
| 71 | +pgxml_parse(PG_FUNCTION_ARGS) |
| 72 | +{ |
| 73 | +/* called as pgxml_parse(document) */ |
| 74 | +xmlDocPtrdoctree; |
| 75 | +text*t=PG_GETARG_TEXT_P(0);/*document buffer */ |
| 76 | +int32docsize=VARSIZE(t)-VARHDRSZ; |
| 77 | + |
| 78 | +pgxml_parser_init(); |
| 79 | + |
| 80 | +doctree=xmlParseMemory((char*)VARDATA(t),docsize); |
| 81 | +if (doctree==NULL) { |
| 82 | +/* xmlCleanupParser(); */ |
| 83 | +PG_RETURN_BOOL(false);/* i.e. not well-formed */ |
| 84 | + } |
| 85 | +/* xmlCleanupParser(); */ |
| 86 | +xmlFreeDoc(doctree); |
| 87 | +PG_RETURN_BOOL(true); |
| 88 | +} |
| 89 | + |
| 90 | +staticxmlChar |
| 91 | +*pgxmlNodeSetToText(xmlNodeSetPtrnodeset, |
| 92 | +xmlDocPtrdoc, |
| 93 | +xmlChar*toptagname, |
| 94 | +xmlChar*septagname, |
| 95 | +intformat) |
| 96 | +{ |
| 97 | +/* Function translates a nodeset into a text representation */ |
| 98 | +/* iterates over each node in the set and calls xmlNodeDump to write |
| 99 | + it to an xmlBuffer -from which an xmlChar * string is returned. */ |
| 100 | +/* each representation is surrounded by <tagname> ... </tagname> */ |
| 101 | +/* if format==0, add a newline between nodes?? */ |
| 102 | + |
| 103 | +xmlBufferPtrbuf; |
| 104 | +xmlChar*result; |
| 105 | +inti; |
| 106 | + |
| 107 | +buf=xmlBufferCreate(); |
| 108 | + |
| 109 | +if ((toptagname!=NULL)&& (xmlStrlen(toptagname)>0)) { |
| 110 | +xmlBufferWriteChar(buf,"<"); |
| 111 | +xmlBufferWriteCHAR(buf,toptagname); |
| 112 | +xmlBufferWriteChar(buf,">"); |
| 113 | + } |
| 114 | +if (nodeset!=NULL) { |
| 115 | +for (i=0;i<nodeset->nodeNr;i++) { |
| 116 | +if ((septagname!=NULL)&& (xmlStrlen(septagname)>0)) { |
| 117 | +xmlBufferWriteChar(buf,"<"); |
| 118 | +xmlBufferWriteCHAR(buf,septagname); |
| 119 | +xmlBufferWriteChar(buf,">"); |
| 120 | + } |
| 121 | +xmlNodeDump(buf,doc,nodeset->nodeTab[i],1,(format==2)); |
| 122 | + |
| 123 | +if ((septagname!=NULL)&& (xmlStrlen(septagname)>0)) { |
| 124 | +xmlBufferWriteChar(buf,"</"); |
| 125 | +xmlBufferWriteCHAR(buf,septagname); |
| 126 | +xmlBufferWriteChar(buf,">"); |
| 127 | + } |
| 128 | +if (format) { |
| 129 | +xmlBufferWriteChar(buf,"\n"); |
| 130 | + } |
| 131 | + } |
| 132 | + } |
| 133 | + |
| 134 | +if ((toptagname!=NULL)&& (xmlStrlen(toptagname)>0)) { |
| 135 | +xmlBufferWriteChar(buf,"</"); |
| 136 | +xmlBufferWriteCHAR(buf,toptagname); |
| 137 | +xmlBufferWriteChar(buf,">"); |
| 138 | + } |
| 139 | +result=xmlStrdup(buf->content); |
| 140 | +xmlBufferFree(buf); |
| 141 | +returnresult; |
| 142 | +} |
| 143 | + |
| 144 | +staticxmlChar*pgxml_texttoxmlchar(text*textstring) { |
| 145 | +xmlChar*res; |
| 146 | +int32txsize; |
| 147 | + |
| 148 | +txsize=VARSIZE(textstring)-VARHDRSZ; |
| 149 | +res= (xmlChar*)palloc(txsize+1); |
| 150 | +memcpy((char*)res,VARDATA(textstring),txsize); |
| 151 | +res[txsize]='\0'; |
| 152 | +returnres; |
| 153 | +} |
| 154 | + |
| 155 | + |
| 156 | +PG_FUNCTION_INFO_V1(pgxml_xpath); |
| 157 | + |
| 158 | +Datum |
| 159 | +pgxml_xpath(PG_FUNCTION_ARGS) |
| 160 | +{ |
| 161 | +xmlDocPtrdoctree; |
| 162 | +xmlXPathContextPtrctxt; |
| 163 | +xmlXPathObjectPtrres; |
| 164 | +xmlChar*xpath,*xpresstr,*toptag,*septag; |
| 165 | +xmlXPathCompExprPtrcomppath; |
| 166 | + |
| 167 | +int32docsize,ressize; |
| 168 | +text*t,*xpres; |
| 169 | + |
| 170 | +t=PG_GETARG_TEXT_P(0);/*document buffer */ |
| 171 | +xpath=pgxml_texttoxmlchar(PG_GETARG_TEXT_P(1));/* XPath expression */ |
| 172 | +toptag=pgxml_texttoxmlchar(PG_GETARG_TEXT_P(2)); |
| 173 | +septag=pgxml_texttoxmlchar(PG_GETARG_TEXT_P(3)); |
| 174 | + |
| 175 | +docsize=VARSIZE(t)-VARHDRSZ; |
| 176 | + |
| 177 | +pgxml_parser_init(); |
| 178 | + |
| 179 | +doctree=xmlParseMemory((char*)VARDATA(t),docsize); |
| 180 | +if (doctree==NULL) {/* not well-formed */ |
| 181 | +PG_RETURN_NULL(); |
| 182 | + } |
| 183 | + |
| 184 | +ctxt=xmlXPathNewContext(doctree); |
| 185 | +ctxt->node=xmlDocGetRootElement(doctree); |
| 186 | + |
| 187 | +/* compile the path */ |
| 188 | +comppath=xmlXPathCompile(xpath); |
| 189 | +if (comppath==NULL) { |
| 190 | +elog(NOTICE,"XPath syntax error"); |
| 191 | +xmlFreeDoc(doctree); |
| 192 | +pfree((void*)xpath); |
| 193 | +PG_RETURN_NULL(); |
| 194 | + } |
| 195 | + |
| 196 | +/* Now evaluate the path expression. */ |
| 197 | +res=xmlXPathCompiledEval(comppath,ctxt); |
| 198 | +xmlXPathFreeCompExpr(comppath); |
| 199 | + |
| 200 | +if (res==NULL) { |
| 201 | +xmlFreeDoc(doctree); |
| 202 | +pfree((void*)xpath); |
| 203 | +PG_RETURN_NULL();/* seems appropriate */ |
| 204 | + } |
| 205 | +/* now we dump this node, ?surrounding by tags? */ |
| 206 | +/* To do this, we look first at the type */ |
| 207 | +switch(res->type) { |
| 208 | +caseXPATH_NODESET: |
| 209 | +xpresstr=pgxmlNodeSetToText(res->nodesetval, |
| 210 | +doctree, |
| 211 | +toptag,septag,0); |
| 212 | +break; |
| 213 | +caseXPATH_STRING: |
| 214 | +xpresstr=xmlStrdup(res->stringval); |
| 215 | +break; |
| 216 | +default: |
| 217 | +elog(NOTICE,"Unsupported XQuery result: %d",res->type); |
| 218 | +xpresstr=xmlStrdup("<unsupported/>"); |
| 219 | + } |
| 220 | + |
| 221 | + |
| 222 | +/* Now convert this result back to text */ |
| 223 | +ressize=strlen(xpresstr); |
| 224 | +xpres= (text*)palloc(ressize+VARHDRSZ); |
| 225 | +memcpy(VARDATA(xpres),xpresstr,ressize); |
| 226 | +VARATT_SIZEP(xpres)=ressize+VARHDRSZ; |
| 227 | + |
| 228 | +/* Free various storage */ |
| 229 | +xmlFreeDoc(doctree); |
| 230 | +pfree((void*)xpath); |
| 231 | +xmlFree(xpresstr); |
| 232 | + |
| 233 | +PG_RETURN_TEXT_P(xpres); |
| 234 | +} |
| 235 | + |
| 236 | + |
| 237 | + |
| 238 | + |