2 /***************************************
3 $Header: /home/amb/routino/src/RCS/xmlparse.l,v 1.17 2010/05/25 18:24:20 amb Exp $
5 A simple generic XML parser where the structure comes from the function parameters.
6 Not intended to be fully conforming to XML staandard or a validating parser but
7 sufficient to parse OSM XML and simple program configuration files.
9 Part of the Routino routing software.
10 ******************/ /******************
11 This file Copyright 2010 Andrew M. Bishop
13 This program is free software: you can redistribute it and/or modify
14 it under the terms of the GNU Affero General Public License as published by
15 the Free Software Foundation, either version 3 of the License, or
16 (at your option) any later version.
18 This program is distributed in the hope that it will be useful,
19 but WITHOUT ANY WARRANTY; without even the implied warranty of
20 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
21 GNU Affero General Public License for more details.
23 You should have received a copy of the GNU Affero General Public License
24 along with this program. If not, see <http://www.gnu.org/licenses/>.
25 ***************************************/
40 #define LEX_TAG_BEGIN 1
41 #define LEX_XML_DECL_BEGIN 2
43 #define LEX_TAG_PUSH 4
44 #define LEX_XML_DECL_FINISH 6
45 #define LEX_TAG_FINISH 7
46 #define LEX_ATTR_KEY 8
47 #define LEX_ATTR_VAL 9
51 #define LEX_ERROR_TAG_START 101
52 #define LEX_ERROR_XML_DECL_START 102
53 #define LEX_ERROR_TAG 103
54 #define LEX_ERROR_XML_DECL 104
55 #define LEX_ERROR_ATTR 105
56 #define LEX_ERROR_END_TAG 106
57 #define LEX_ERROR_COMMENT 107
58 #define LEX_ERROR_CLOSE 108
59 #define LEX_ERROR_ATTR_VAL 109
60 #define LEX_ERROR_ENTITY_REF 110
61 #define LEX_ERROR_CHAR_REF 111
63 #define LEX_ERROR_UNEXP_TAG 201
64 #define LEX_ERROR_UNBALANCED 202
65 #define LEX_ERROR_NO_START 203
66 #define LEX_ERROR_UNEXP_ATT 204
67 #define LEX_ERROR_UNEXP_EOF 205
68 #define LEX_ERROR_XML_NOT_FIRST 206
70 #define LEX_ERROR_CALLBACK 255
73 /* Lexer definitions */
75 #define YY_SKIP_YYWRAP 1 /* Remove error with prototype of ..._yywrap */
77 /*+ Needed in lex but does nothing. +*/
81 /*+ Reset the current string. +*/
82 #define reset_string \
83 if(!string) string=(char*)malloc(16); \
87 /*+ append information to the current string. +*/
88 #define append_string(xx) \
90 if((stringused+newlen)>=stringlen) \
91 string=(char*)realloc((void*)string,stringlen=(stringused+newlen+16)); \
92 strcpy(string+stringused,xx); \
98 /* Lexer functions and variables */
100 extern int yylex(void);
102 static char *yylval=NULL;
104 static int xmlparse_options;
119 /* Grammar based on http://www.w3.org/TR/2004/REC-xml-20040204/ but for ASCII not Unicode. */
127 namechar ({letter}|{digit}|[-._:])
128 name ({letter}|[_:]){namechar}*
131 charref &#({digit}+|x{xdigit}+);
137 %x XML_DECL_START XML_DECL
144 /* Must use static variables since the parser returns often. */
145 static char *string=NULL;
146 static int stringlen=0,stringused=0;
147 static int after_attr=0;
151 /* Handle top level entities */
153 "<!--" { BEGIN(COMMENT); }
154 "<![CDATA[" { BEGIN(CDATA); }
155 "<!DOCTYPE" { BEGIN(DOCTYPE); doctype_depth=0; }
156 "</" { BEGIN(END_TAG1); }
157 "<?" { BEGIN(XML_DECL_START); }
158 "<" { BEGIN(TAG_START); }
159 ">" { return(LEX_ERROR_CLOSE); }
164 <COMMENT>"--->" { return(LEX_ERROR_COMMENT); }
165 <COMMENT>"-->" { BEGIN(INITIAL); }
166 <COMMENT>"--"[^->]+ { }
172 <CDATA>"]]>" { BEGIN(INITIAL); }
178 <DOCTYPE>"<" { doctype_depth++; }
179 <DOCTYPE>">" { if(doctype_depth==0) BEGIN(INITIAL); else doctype_depth--; }
182 /* XML Declaration start */
184 <XML_DECL_START>{name} { BEGIN(XML_DECL); yylval=yytext; return(LEX_XML_DECL_BEGIN); }
185 <XML_DECL_START>.|\n { return(LEX_ERROR_XML_DECL_START); }
189 <XML_DECL>"?>" { BEGIN(INITIAL); return(LEX_XML_DECL_FINISH); }
191 <XML_DECL>{name} { after_attr=XML_DECL; BEGIN(ATTR_KEY); yylval=yytext; return(LEX_ATTR_KEY); }
192 <XML_DECL>.|\n { return(LEX_ERROR_XML_DECL); }
196 <TAG_START>{name} { BEGIN(TAG); yylval=yytext; return(LEX_TAG_BEGIN); }
197 <TAG_START>.|\n { return(LEX_ERROR_TAG_START); }
201 <END_TAG1>{name} { BEGIN(END_TAG2); yylval=yytext; return(LEX_TAG_POP); }
202 <END_TAG1>.|\n { return(LEX_ERROR_END_TAG); }
204 <END_TAG2>">" { BEGIN(INITIAL); }
205 <END_TAG2>.|\n { return(LEX_ERROR_END_TAG); }
209 <TAG>"/>" { BEGIN(INITIAL); return(LEX_TAG_FINISH); }
210 <TAG>">" { BEGIN(INITIAL); return(LEX_TAG_PUSH); }
212 <TAG>{name} { after_attr=TAG; BEGIN(ATTR_KEY); yylval=yytext; return(LEX_ATTR_KEY); }
213 <TAG>.|\n { return(LEX_ERROR_TAG); }
217 <ATTR_KEY>= { BEGIN(ATTR_VAL); }
218 <ATTR_KEY>.|\n { return(LEX_ERROR_ATTR); }
220 <ATTR_VAL>\" { BEGIN(DQUOTED); reset_string; }
221 <ATTR_VAL>\' { BEGIN(SQUOTED); reset_string; }
222 <ATTR_VAL>.|\n { return(LEX_ERROR_ATTR); }
226 <DQUOTED>\" { BEGIN(after_attr); yylval=string; return(LEX_ATTR_VAL); }
227 <DQUOTED>{entityref} { if(xmlparse_options&XMLPARSE_RETURN_ATTR_ENCODED) {append_string(yytext);}
228 else { const char *str=ParseXML_Decode_Entity_Ref(yytext); if(str) {append_string(str);} else {yylval=yytext; return(LEX_ERROR_ENTITY_REF);} } }
229 <DQUOTED>{charref} { if(xmlparse_options&XMLPARSE_RETURN_ATTR_ENCODED) {append_string(yytext);}
230 else { const char *str=ParseXML_Decode_Char_Ref(yytext); if(str) {append_string(str);} else {yylval=yytext; return(LEX_ERROR_CHAR_REF);} } }
231 <DQUOTED>[<>&] { yylval=yytext; return(LEX_ERROR_ATTR_VAL); }
232 <DQUOTED>[^<>&\"]+ { append_string(yytext); }
234 <SQUOTED>\' { BEGIN(after_attr); yylval=string; return(LEX_ATTR_VAL); }
235 <SQUOTED>{entityref} { if(xmlparse_options&XMLPARSE_RETURN_ATTR_ENCODED) {append_string(yytext);}
236 else { const char *str=ParseXML_Decode_Entity_Ref(yytext); if(str) {append_string(str);} else {yylval=yytext; return(LEX_ERROR_ENTITY_REF);} } }
237 <SQUOTED>{charref} { if(xmlparse_options&XMLPARSE_RETURN_ATTR_ENCODED) {append_string(yytext);}
238 else { const char *str=ParseXML_Decode_Char_Ref(yytext); if(str) {append_string(str);} else {yylval=yytext; return(LEX_ERROR_CHAR_REF);} } }
239 <SQUOTED>[<>&] { yylval=yytext; return(LEX_ERROR_ATTR_VAL); }
240 <SQUOTED>[^<>&\']+ { append_string(yytext); }
244 <<EOF>> { free(string); string=NULL; stringlen=stringused=0; BEGIN(INITIAL); return(LEX_EOF); }
249 /*++++++++++++++++++++++++++++++++++++++
250 A function to call the callback function with the parameters needed.
252 int call_callback Returns 1 if the callback returned with an error.
254 const char *name The name of the tag.
256 int (*callback)() The callback function.
258 int type The type of tag (start and/or end).
260 int nattributes The number of attributes collected.
262 char *attributes[XMLPARSE_MAX_ATTRS] The list of attributes.
263 ++++++++++++++++++++++++++++++++++++++*/
265 static inline int call_callback(const char *name,int (*callback)(),int type,int nattributes,char *attributes[XMLPARSE_MAX_ATTRS])
269 case 0: return (*callback)(name,type);
270 case 1: return (*callback)(name,type,attributes[0]);
271 case 2: return (*callback)(name,type,attributes[0],attributes[1]);
272 case 3: return (*callback)(name,type,attributes[0],attributes[1],attributes[2]);
273 case 4: return (*callback)(name,type,attributes[0],attributes[1],attributes[2],attributes[3]);
274 case 5: return (*callback)(name,type,attributes[0],attributes[1],attributes[2],attributes[3],attributes[4]);
275 case 6: return (*callback)(name,type,attributes[0],attributes[1],attributes[2],attributes[3],attributes[4],attributes[5]);
276 case 7: return (*callback)(name,type,attributes[0],attributes[1],attributes[2],attributes[3],attributes[4],attributes[5],attributes[6]);
277 case 8: return (*callback)(name,type,attributes[0],attributes[1],attributes[2],attributes[3],attributes[4],attributes[5],attributes[6],attributes[7]);
278 case 9: return (*callback)(name,type,attributes[0],attributes[1],attributes[2],attributes[3],attributes[4],attributes[5],attributes[6],attributes[7],attributes[8]);
279 case 10: return (*callback)(name,type,attributes[0],attributes[1],attributes[2],attributes[3],attributes[4],attributes[5],attributes[6],attributes[7],attributes[8],attributes[9]);
280 case 11: return (*callback)(name,type,attributes[0],attributes[1],attributes[2],attributes[3],attributes[4],attributes[5],attributes[6],attributes[7],attributes[8],attributes[9],attributes[10]);
281 case 12: return (*callback)(name,type,attributes[0],attributes[1],attributes[2],attributes[3],attributes[4],attributes[5],attributes[6],attributes[7],attributes[8],attributes[9],attributes[10],attributes[11]);
282 case 13: return (*callback)(name,type,attributes[0],attributes[1],attributes[2],attributes[3],attributes[4],attributes[5],attributes[6],attributes[7],attributes[8],attributes[9],attributes[10],attributes[11],attributes[12]);
283 case 14: return (*callback)(name,type,attributes[0],attributes[1],attributes[2],attributes[3],attributes[4],attributes[5],attributes[6],attributes[7],attributes[8],attributes[9],attributes[10],attributes[11],attributes[12],attributes[13]);
284 case 15: return (*callback)(name,type,attributes[0],attributes[1],attributes[2],attributes[3],attributes[4],attributes[5],attributes[6],attributes[7],attributes[8],attributes[9],attributes[10],attributes[11],attributes[12],attributes[13],attributes[14]);
285 case 16: return (*callback)(name,type,attributes[0],attributes[1],attributes[2],attributes[3],attributes[4],attributes[5],attributes[6],attributes[7],attributes[8],attributes[9],attributes[10],attributes[11],attributes[12],attributes[13],attributes[14],attributes[15]);
288 fprintf(stderr,"XML Parser: Error on line %d: too many attributes for tag '%s' source code needs changing.\n",yylineno,name);
294 /*++++++++++++++++++++++++++++++++++++++
295 Parse the XML and call the functions for each tag as seen.
297 int ParseXML Returns 0 if OK or something else in case of an error.
299 FILE *file The file to parse.
301 xmltag **tags The array of pointers to tags for the top level.
303 int options A list of XML Parser options OR-ed together.
304 ++++++++++++++++++++++++++++++++++++++*/
306 int ParseXML(FILE *file,xmltag **tags,int options)
310 char *attributes[XMLPARSE_MAX_ATTRS]={NULL};
313 int stackdepth=0,stackused=0;
314 xmltag ***tags_stack=NULL;
315 xmltag **tag_stack=NULL;
318 /* The actual parser. */
320 xmlparse_options=options;
336 /* The start of a tag for an XML declaration */
338 case LEX_XML_DECL_BEGIN:
342 fprintf(stderr,"XML Parser: Error on line %d: XML declaration not before all other tags.\n",yylineno);
343 yychar=LEX_ERROR_XML_NOT_FIRST;
347 /* The start of a tag for an element */
354 if(!strcasecmp(yylval,tags[i]->name))
358 for(i=0;i<tag->nattributes;i++)
370 fprintf(stderr,"XML Parser: Error on line %d: unexpected tag '%s'.\n",yylineno,yylval);
371 yychar=LEX_ERROR_UNEXP_TAG;
376 /* The end of the start-tag for an element */
380 if(stackused==stackdepth)
382 tag_stack =(xmltag**) realloc((void*)tag_stack ,(stackdepth+=8)*sizeof(xmltag*));
383 tags_stack=(xmltag***)realloc((void*)tags_stack,(stackdepth+=8)*sizeof(xmltag**));
386 tag_stack [stackused]=tag;
387 tags_stack[stackused]=tags;
391 if(call_callback(tag->name,tag->callback,XMLPARSE_TAG_START,tag->nattributes,attributes))
392 yychar=LEX_ERROR_CALLBACK;
398 /* The end of the empty-element-tag for an XML declaration */
400 case LEX_XML_DECL_FINISH:
402 /* The end of the empty-element-tag for an element */
407 if(call_callback(tag->name,tag->callback,XMLPARSE_TAG_START|XMLPARSE_TAG_END,tag->nattributes,attributes))
408 yychar=LEX_ERROR_CALLBACK;
411 tag=tag_stack[stackused-1];
417 /* The end of the end-tag for an element */
422 tags=tags_stack[stackused];
423 tag =tag_stack [stackused];
425 if(strcmp(tag->name,yylval))
427 fprintf(stderr,"XML Parser: Error on line %d: end tag '</%s>' doesn't match start tag '<%s ...>'.\n",yylineno,yylval,tag->name);
428 yychar=LEX_ERROR_UNBALANCED;
433 fprintf(stderr,"XML Parser: Error on line %d: end tag '</%s>' seen but there was no start tag '<%s ...>'.\n",yylineno,yylval,yylval);
434 yychar=LEX_ERROR_NO_START;
437 for(i=0;i<tag->nattributes;i++)
445 if(call_callback(tag->name,tag->callback,XMLPARSE_TAG_END,tag->nattributes,attributes))
446 yychar=LEX_ERROR_CALLBACK;
449 tag=tag_stack[stackused-1];
455 /* An attribute key */
461 for(i=0;i<tag->nattributes;i++)
462 if(!strcasecmp(yylval,tag->attributes[i]))
471 if((options&XMLPARSE_UNKNOWN_ATTRIBUTES)==XMLPARSE_UNKNOWN_ATTR_ERROR ||
472 ((options&XMLPARSE_UNKNOWN_ATTRIBUTES)==XMLPARSE_UNKNOWN_ATTR_ERRNONAME && !strchr(yylval,':')))
474 fprintf(stderr,"XML Parser: Error on line %d: unexpected attribute '%s' for tag '%s'.\n",yylineno,yylval,tag->name);
475 yychar=LEX_ERROR_UNEXP_ATT;
477 else if((options&XMLPARSE_UNKNOWN_ATTRIBUTES)==XMLPARSE_UNKNOWN_ATTR_WARN)
478 fprintf(stderr,"XML Parser: Warning on line %d: unexpected attribute '%s' for tag '%s'.\n",yylineno,yylval,tag->name);
483 /* An attribute value */
487 if(tag->callback && attribute!=-1 && yylval)
488 attributes[attribute]=strcpy(malloc(strlen(yylval)+1),yylval);
498 fprintf(stderr,"XML Parser: Error on line %d: end of file seen without end tag '</%s>'.\n",yylineno,tag->name);
499 yychar=LEX_ERROR_UNEXP_EOF;
504 case LEX_ERROR_TAG_START:
505 fprintf(stderr,"XML Parser: Error on line %d: character '<' seen not at start of tag.\n",yylineno);
508 case LEX_ERROR_XML_DECL_START:
509 fprintf(stderr,"XML Parser: Error on line %d: characters '<?' seen not at start of XML declaration.\n",yylineno);
513 fprintf(stderr,"XML Parser: Error on line %d: invalid character seen inside tag '<%s...>'.\n",yylineno,tag->name);
516 case LEX_ERROR_XML_DECL:
517 fprintf(stderr,"XML Parser: Error on line %d: invalid character seen inside XML declaration '<?%s...>'.\n",yylineno,tag->name);
521 fprintf(stderr,"XML Parser: Error on line %d: invalid attribute definition seen in tag.\n",yylineno);
524 case LEX_ERROR_END_TAG:
525 fprintf(stderr,"XML Parser: Error on line %d: invalid character seen in end-tag.\n",yylineno);
528 case LEX_ERROR_COMMENT:
529 fprintf(stderr,"XML Parser: Error on line %d: invalid comment seen.\n",yylineno);
532 case LEX_ERROR_CLOSE:
533 fprintf(stderr,"XML Parser: Error on line %d: character '>' seen not at end of tag.\n",yylineno);
536 case LEX_ERROR_ATTR_VAL:
537 fprintf(stderr,"XML Parser: Error on line %d: invalid character '%s' seen in attribute value.\n",yylineno,yylval);
540 case LEX_ERROR_ENTITY_REF:
541 fprintf(stderr,"XML Parser: Error on line %d: invalid entity reference '%s' seen in attribute value.\n",yylineno,yylval);
544 case LEX_ERROR_CHAR_REF:
545 fprintf(stderr,"XML Parser: Error on line %d: invalid character reference '%s' seen in attribute value.\n",yylineno,yylval);
549 while(yychar>LEX_EOF && yychar<LEX_ERROR);
551 /* Delete the tagdata */
553 for(i=0;i<XMLPARSE_MAX_ATTRS;i++)
567 /*++++++++++++++++++++++++++++++++++++++
568 Return the current parser line number.
570 unsigned long ParseXML_LineNumber Returns the line number.
571 ++++++++++++++++++++++++++++++++++++++*/
573 unsigned long ParseXML_LineNumber(void)
579 /*++++++++++++++++++++++++++++++++++++++
580 Convert an XML entity reference into an ASCII string.
582 char *ParseXML_Decode_Entity_Ref Returns a pointer to the replacement decoded string.
584 const char *string The entity reference string.
585 ++++++++++++++++++++++++++++++++++++++*/
587 char *ParseXML_Decode_Entity_Ref(const char *string)
589 if(!strcmp(string,"&")) return("&");
590 if(!strcmp(string,"<")) return("<");
591 if(!strcmp(string,">")) return(">");
592 if(!strcmp(string,"'")) return("'");
593 if(!strcmp(string,""")) return("\"");
598 /*++++++++++++++++++++++++++++++++++++++
599 Convert an XML character reference into an ASCII string.
601 char *ParseXML_Decode_Char_Ref Returns a pointer to the replacement decoded string.
603 const char *string The character reference string.
604 ++++++++++++++++++++++++++++++++++++++*/
606 char *ParseXML_Decode_Char_Ref(const char *string)
608 static char result[2]=" ";
611 if(string[2]=='x') val=strtol(string+3,NULL,16);
612 else val=strtol(string+2,NULL,10);
623 /*++++++++++++++++++++++++++++++++++++++
624 Convert a string into something that is safe to output in an XML file.
626 char *ParseXML_Encode_Safe_XML Returns a pointer to the replacement encoded string (or the original if no change needed).
628 const char *string The string to convert.
629 ++++++++++++++++++++++++++++++++++++++*/
631 char *ParseXML_Encode_Safe_XML(const char *string)
633 static const char hexstring[17]="0123456789ABCDEF";
637 for(i=0;string[i];i++)
638 if(string[i]=='<' || string[i]=='>' || string[i]=='&' || string[i]=='\'' || string[i]=='"' || string[i]<32 || string[i]>126)
642 return((char*)string);
646 result=(char*)malloc(len+7);
647 strncpy(result,string,j=i);
651 for(;j<len && string[i];i++)
659 else if(string[i]=='>')
666 else if(string[i]=='&')
674 else if(string[i]=='\'')
683 else if(string[i]=='"')
692 else if(string[i]<32 || string[i]>126)
697 result[j++]=hexstring[(string[i]&0xf0)>>4];
698 result[j++]=hexstring[ string[i]&0x0f ];
702 result[j++]=string[i];
704 if(string[i]) /* Not finished */
707 result=(char*)realloc((void*)result,len+7);
718 /*++++++++++++++++++++++++++++++++++++++
719 Convert a string to a integer (checking that it really is a integer).
721 int ParseXML_GetInteger Returns 1 if a integer could be found or 0 otherwise.
723 const char *string The string to be parsed.
725 int *number Returns the number.
726 ++++++++++++++++++++++++++++++++++++++*/
728 int ParseXML_GetInteger(const char *string,int *number)
730 const char *p=string;
732 if(*p=='-' || *p=='+')
741 *number=atoi(string);
747 /*++++++++++++++++++++++++++++++++++++++
748 Convert a string to a floating point number (checking that it really is a number).
750 int ParseXML_GetFloating Returns 1 if a number could be found or 0 otherwise.
752 const char *string The string to be parsed.
754 int *number Returns the number.
755 ++++++++++++++++++++++++++++++++++++++*/
757 int ParseXML_GetFloating(const char *string,double *number)
759 const char *p=string;
761 if(*p=='-' || *p=='+')
764 while(isdigit(*p) || *p=='.')
767 if(*p=='e' || *p=='E')
771 if(*p=='-' || *p=='+')
781 *number=atof(string);