1 /* ezxml.c
2 *
3 * Copyright 2004-2006 Aaron Voisine <aaron@voisine.org>
4 *
5 * Permission is hereby granted, free of charge, to any person obtaining
6 * a copy of this software and associated documentation files (the
7 * "Software"), to deal in the Software without restriction, including
8 * without limitation the rights to use, copy, modify, merge, publish,
9 * distribute, sublicense, and/or sell copies of the Software, and to
10 * permit persons to whom the Software is furnished to do so, subject to
11 * the following conditions:
12 *
13 * The above copyright notice and this permission notice shall be included
14 * in all copies or substantial portions of the Software.
15 *
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
17 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
18 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
19 * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
20 * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
21 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
22 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
23 */
24
25 #include <stdlib.h>
26 #include <stdio.h>
27 #include <stdarg.h>
28 #include <string.h>
29 #include <ctype.h>
30
31 #ifndef _WIN32
32 #include <unistd.h>
33 #else
34 #define EZXML_NOMMAP 1
35 #define snprintf _snprintf
36 #define vsnprintf _vsnprintf
37 #endif
38
39 #include <sys/types.h>
40 #ifndef EZXML_NOMMAP
41 #include <sys/mman.h>
42 #endif // EZXML_NOMMAP
43 #include <sys/stat.h>
44 #include "ezxml.h"
45
46 #define EZXML_WS "\t\r\n " // whitespace
47 #define EZXML_ERRL 128 // maximum error string length
48
49 typedef struct ezxml_root *ezxml_root_t;
50 struct ezxml_root { // additional data for the root tag
51 struct ezxml xml; // is a super-struct built on top of ezxml struct
52 ezxml_t cur; // current xml tree insertion point
53 char *m; // original xml string
54 size_t len; // length of allocated memory for mmap, -1 for malloc
55 char *u; // UTF-8 conversion of string if original was UTF-16
56 char *s; // start of work area
57 char *e; // end of work area
58 char **ent; // general entities (ampersand sequences)
59 char ***attr; // default attributes
60 char ***pi; // processing instructions
61 short standalone; // non-zero if <?xml standalone="yes"?>
62 char err[EZXML_ERRL]; // error string
63 };
64
65 char *EZXML_NIL[] = { NULL }; // empty, null terminated array of strings
66
67 // returns the first child tag with the given name or NULL if not found
ezxml_child(ezxml_t xml,const char * name)68 ezxml_t ezxml_child(ezxml_t xml, const char *name)
69 {
70 xml = (xml) ? xml->child : NULL;
71 while (xml && strcmp(name, xml->name)) xml = xml->sibling;
72 return xml;
73 }
74
75 // returns the Nth tag with the same name in the same subsection or NULL if not
76 // found
ezxml_idx(ezxml_t xml,int idx)77 ezxml_t ezxml_idx(ezxml_t xml, int idx)
78 {
79 for (; xml && idx; idx--) xml = xml->next;
80 return xml;
81 }
82
83 // returns the value of the requested tag attribute or NULL if not found
ezxml_attr(ezxml_t xml,const char * attr)84 const char *ezxml_attr(ezxml_t xml, const char *attr)
85 {
86 int i = 0, j = 1;
87 ezxml_root_t root = (ezxml_root_t)xml;
88
89 if (! xml || ! xml->attr) return NULL;
90 while (xml->attr[i] && strcmp(attr, xml->attr[i])) i += 2;
91 if (xml->attr[i]) return xml->attr[i + 1]; // found attribute
92
93 while (root->xml.parent) root = (ezxml_root_t)root->xml.parent; // root tag
94 for (i = 0; root->attr[i] && strcmp(xml->name, root->attr[i][0]); i++);
95 if (! root->attr[i]) return NULL; // no matching default attributes
96 while (root->attr[i][j] && strcmp(attr, root->attr[i][j])) j += 3;
97 return (root->attr[i][j]) ? root->attr[i][j + 1] : NULL; // found default
98 }
99
100 // same as ezxml_get but takes an already initialized va_list
ezxml_vget(ezxml_t xml,va_list ap)101 ezxml_t ezxml_vget(ezxml_t xml, va_list ap)
102 {
103 char *name = va_arg(ap, char *);
104 int idx = -1;
105
106 if (name && *name) {
107 idx = va_arg(ap, int);
108 xml = ezxml_child(xml, name);
109 }
110 return (idx < 0) ? xml : ezxml_vget(ezxml_idx(xml, idx), ap);
111 }
112
113 // Traverses the xml tree to retrieve a specific subtag. Takes a variable
114 // length list of tag names and indexes. The argument list must be terminated
115 // by either an index of -1 or an empty string tag name. Example:
116 // title = ezxml_get(library, "shelf", 0, "book", 2, "title", -1);
117 // This retrieves the title of the 3rd book on the 1st shelf of library.
118 // Returns NULL if not found.
ezxml_get(ezxml_t xml,...)119 ezxml_t ezxml_get(ezxml_t xml, ...)
120 {
121 va_list ap;
122 ezxml_t r;
123
124 va_start(ap, xml);
125 r = ezxml_vget(xml, ap);
126 va_end(ap);
127 return r;
128 }
129
130 // returns a null terminated array of processing instructions for the given
131 // target
ezxml_pi(ezxml_t xml,const char * target)132 const char **ezxml_pi(ezxml_t xml, const char *target)
133 {
134 ezxml_root_t root = (ezxml_root_t)xml;
135 int i = 0;
136
137 if (! root) return (const char **)EZXML_NIL;
138 while (root->xml.parent) root = (ezxml_root_t)root->xml.parent; // root tag
139 while (root->pi[i] && strcmp(target, root->pi[i][0])) i++; // find target
140 return (const char **)((root->pi[i]) ? root->pi[i] + 1 : EZXML_NIL);
141 }
142
143 // set an error string and return root
ezxml_err(ezxml_root_t root,char * s,const char * err,...)144 ezxml_t ezxml_err(ezxml_root_t root, char *s, const char *err, ...)
145 {
146 va_list ap;
147 int line = 1;
148 char *t, fmt[EZXML_ERRL];
149
150 for (t = root->s; t < s; t++) if (*t == '\n') line++;
151 snprintf(fmt, EZXML_ERRL, "[error near line %d]: %s", line, err);
152
153 va_start(ap, err);
154 vsnprintf(root->err, EZXML_ERRL, fmt, ap);
155 va_end(ap);
156
157 return &root->xml;
158 }
159
160 // Recursively decodes entity and character references and normalizes new lines
161 // ent is a null terminated array of alternating entity names and values. set t
162 // to '&' for general entity decoding, '%' for parameter entity decoding, 'c'
163 // for cdata sections, ' ' for attribute normalization, or '*' for non-cdata
164 // attribute normalization. Returns s, or if the decoded string is longer than
165 // s, returns a malloced string that must be freed.
ezxml_decode(char * s,char ** ent,char t)166 char *ezxml_decode(char *s, char **ent, char t)
167 {
168 char *e, *r = s, *m = s;
169 long b, c, d, l;
170
171 for (; *s; s++) { // normalize line endings
172 while (*s == '\r') {
173 *(s++) = '\n';
174 if (*s == '\n') memmove(s, (s + 1), strlen(s));
175 }
176 }
177
178 for (s = r; ; ) {
179 while (*s && *s != '&' && (*s != '%' || t != '%') && !isspace(*s)) s++;
180
181 if (! *s) break;
182 else if (t != 'c' && ! strncmp(s, "&#", 2)) { // character reference
183 if (s[2] == 'x') c = strtol(s + 3, &e, 16); // base 16
184 else c = strtol(s + 2, &e, 10); // base 10
185 if (! c || *e != ';') {
186 s++; // not a character ref
187 continue;
188 }
189
190 if (c < 0x80) *(s++) = c; // US-ASCII subset
191 else { // multi-byte UTF-8 sequence
192 for (b = 0, d = c; d; d /= 2) b++; // number of bits in c
193 b = (b - 2) / 5; // number of bytes in payload
194 *(s++) = (0xFF << (7 - b)) | (c >> (6 * b)); // head
195 while (b) *(s++) = 0x80 | ((c >> (6 * --b)) & 0x3F); // payload
196 }
197
198 memmove(s, strchr(s, ';') + 1, strlen(strchr(s, ';')));
199 } else if ((*s == '&' && (t == '&' || t == ' ' || t == '*')) ||
200 (*s == '%' && t == '%')) { // entity reference
201 for (b = 0; ent[b] && strncmp(s + 1, ent[b], strlen(ent[b]));
202 b += 2); // find entity in entity list
203
204 if (ent[b++]) { // found a match
205 if ((c = strlen(ent[b])) - 1 > (e = strchr(s, ';')) - s) {
206 l = (d = (s - r)) + c + strlen(e); // new length
207 r = (r == m) ? strcpy(malloc(l), r) : realloc(r, l);
208 e = strchr((s = r + d), ';'); // fix up pointers
209 }
210
211 memmove(s + c, e + 1, strlen(e)); // shift rest of string
212 strncpy(s, ent[b], c); // copy in replacement text
213 } else s++; // not a known entity
214 } else if ((t == ' ' || t == '*') && isspace(*s)) *(s++) = ' ';
215 else s++; // no decoding needed
216 }
217
218 if (t == '*') { // normalize spaces for non-cdata attributes
219 for (s = r; *s; s++) {
220 if ((l = strspn(s, " "))) memmove(s, s + l, strlen(s + l) + 1);
221 while (*s && *s != ' ') s++;
222 }
223 if (--s >= r && *s == ' ') *s = '\0'; // trim any trailing space
224 }
225 return r;
226 }
227
228 // called when parser finds start of new tag
ezxml_open_tag(ezxml_root_t root,char * name,char ** attr)229 void ezxml_open_tag(ezxml_root_t root, char *name, char **attr)
230 {
231 ezxml_t xml = root->cur;
232
233 if (xml->name) xml = ezxml_add_child(xml, name, strlen(xml->txt));
234 else xml->name = name; // first open tag
235
236 xml->attr = attr;
237 root->cur = xml; // update tag insertion point
238 }
239
240 // called when parser finds character content between open and closing tag
ezxml_char_content(ezxml_root_t root,char * s,size_t len,char t)241 void ezxml_char_content(ezxml_root_t root, char *s, size_t len, char t)
242 {
243 ezxml_t xml = root->cur;
244 char *m = s;
245 size_t l;
246
247 if (! xml || ! xml->name || ! len) return; // sanity check
248
249 s[len] = '\0'; // null terminate text (calling functions anticipate this)
250 len = strlen(s = ezxml_decode(s, root->ent, t)) + 1;
251
252 if (! *(xml->txt)) xml->txt = s; // initial character content
253 else { // allocate our own memory and make a copy
254 xml->txt = (xml->flags & EZXML_TXTM) // allocate some space
255 ? realloc(xml->txt, (l = strlen(xml->txt)) + len)
256 : strcpy(malloc((l = strlen(xml->txt)) + len), xml->txt);
257 strcpy(xml->txt + l, s); // add new char content
258 if (s != m) free(s); // free s if it was malloced by ezxml_decode()
259 }
260
261 if (xml->txt != m) ezxml_set_flag(xml, EZXML_TXTM);
262 }
263
264 // called when parser finds closing tag
ezxml_close_tag(ezxml_root_t root,char * name,char * s)265 ezxml_t ezxml_close_tag(ezxml_root_t root, char *name, char *s)
266 {
267 if (! root->cur || ! root->cur->name || strcmp(name, root->cur->name))
268 return ezxml_err(root, s, "unexpected closing tag </%s>", name);
269
270 root->cur = root->cur->parent;
271 return NULL;
272 }
273
274 // checks for circular entity references, returns non-zero if no circular
275 // references are found, zero otherwise
ezxml_ent_ok(char * name,char * s,char ** ent)276 int ezxml_ent_ok(char *name, char *s, char **ent)
277 {
278 int i;
279
280 for (; ; s++) {
281 while (*s && *s != '&') s++; // find next entity reference
282 if (! *s) return 1;
283 if (! strncmp(s + 1, name, strlen(name))) return 0; // circular ref.
284 for (i = 0; ent[i] && strncmp(ent[i], s + 1, strlen(ent[i])); i += 2);
285 if (ent[i] && ! ezxml_ent_ok(name, ent[i + 1], ent)) return 0;
286 }
287 }
288
289 // called when the parser finds a processing instruction
ezxml_proc_inst(ezxml_root_t root,char * s,size_t len)290 void ezxml_proc_inst(ezxml_root_t root, char *s, size_t len)
291 {
292 int i = 0, j = 1;
293 char *target = s;
294
295 s[len] = '\0'; // null terminate instruction
296 if (*(s += strcspn(s, EZXML_WS))) {
297 *s = '\0'; // null terminate target
298 s += strspn(s + 1, EZXML_WS) + 1; // skip whitespace after target
299 }
300
301 if (! strcmp(target, "xml")) { // <?xml ... ?>
302 if ((s = strstr(s, "standalone")) && ! strncmp(s + strspn(s + 10,
303 EZXML_WS "='\"") + 10, "yes", 3)) root->standalone = 1;
304 return;
305 }
306
307 if (! root->pi[0]) *(root->pi = malloc(sizeof(char **))) = NULL; //first pi
308
309 while (root->pi[i] && strcmp(target, root->pi[i][0])) i++; // find target
310 if (! root->pi[i]) { // new target
311 root->pi = realloc(root->pi, sizeof(char **) * (i + 2));
312 root->pi[i] = malloc(sizeof(char *) * 3);
313 root->pi[i][0] = target;
314 root->pi[i][1] = (char *)(root->pi[i + 1] = NULL); // terminate pi list
315 root->pi[i][2] = strdup(""); // empty document position list
316 }
317
318 while (root->pi[i][j]) j++; // find end of instruction list for this target
319 root->pi[i] = realloc(root->pi[i], sizeof(char *) * (j + 3));
320 root->pi[i][j + 2] = realloc(root->pi[i][j + 1], j + 1);
321 strcpy(root->pi[i][j + 2] + j - 1, (root->xml.name) ? ">" : "<");
322 root->pi[i][j + 1] = NULL; // null terminate pi list for this target
323 root->pi[i][j] = s; // set instruction
324 }
325
326 // called when the parser finds an internal doctype subset
ezxml_internal_dtd(ezxml_root_t root,char * s,size_t len)327 short ezxml_internal_dtd(ezxml_root_t root, char *s, size_t len)
328 {
329 char q, *c, *t, *n = NULL, *v, **ent, **pe;
330 int i, j;
331
332 pe = memcpy(malloc(sizeof(EZXML_NIL)), EZXML_NIL, sizeof(EZXML_NIL));
333
334 for (s[len] = '\0'; s; ) {
335 while (*s && *s != '<' && *s != '%') s++; // find next declaration
336
337 if (! *s) break;
338 else if (! strncmp(s, "<!ENTITY", 8)) { // parse entity definitions
339 c = s += strspn(s + 8, EZXML_WS) + 8; // skip white space separator
340 n = s + strspn(s, EZXML_WS "%"); // find name
341 *(s = n + strcspn(n, EZXML_WS)) = ';'; // append ; to name
342
343 v = s + strspn(s + 1, EZXML_WS) + 1; // find value
344 if ((q = *(v++)) != '"' && q != '\'') { // skip externals
345 s = strchr(s, '>');
346 continue;
347 }
348
349 for (i = 0, ent = (*c == '%') ? pe : root->ent; ent[i]; i++);
350 ent = realloc(ent, (i + 3) * sizeof(char *)); // space for next ent
351 if (*c == '%') pe = ent;
352 else root->ent = ent;
353
354 *(++s) = '\0'; // null terminate name
355 if ((s = strchr(v, q))) *(s++) = '\0'; // null terminate value
356 ent[i + 1] = ezxml_decode(v, pe, '%'); // set value
357 ent[i + 2] = NULL; // null terminate entity list
358 if (! ezxml_ent_ok(n, ent[i + 1], ent)) { // circular reference
359 if (ent[i + 1] != v) free(ent[i + 1]);
360 ezxml_err(root, v, "circular entity declaration &%s", n);
361 break;
362 } else ent[i] = n; // set entity name
363 } else if (! strncmp(s, "<!ATTLIST", 9)) { // parse default attributes
364 t = s + strspn(s + 9, EZXML_WS) + 9; // skip whitespace separator
365 if (! *t) {
366 ezxml_err(root, t, "unclosed <!ATTLIST");
367 break;
368 }
369 if (*(s = t + strcspn(t, EZXML_WS ">")) == '>') continue;
370 else *s = '\0'; // null terminate tag name
371 for (i = 0; root->attr[i] && strcmp(n, root->attr[i][0]); i++);
372
373 while (*(n = s + 1 + strspn(s + 1, EZXML_WS)) && *n != '>') {
374 if (*(s = n + strcspn(n, EZXML_WS))) *s = '\0'; // attr name
375 else {
376 ezxml_err(root, t, "malformed <!ATTLIST");
377 break;
378 }
379
380 s += strspn(s + 1, EZXML_WS) + 1; // find next token
381 c = (strncmp(s, "CDATA", 5)) ? "*" : " "; // is it cdata?
382 if (! strncmp(s, "NOTATION", 8))
383 s += strspn(s + 8, EZXML_WS) + 8;
384 s = (*s == '(') ? strchr(s, ')') : s + strcspn(s, EZXML_WS);
385 if (! s) {
386 ezxml_err(root, t, "malformed <!ATTLIST");
387 break;
388 }
389
390 s += strspn(s, EZXML_WS ")"); // skip white space separator
391 if (! strncmp(s, "#FIXED", 6))
392 s += strspn(s + 6, EZXML_WS) + 6;
393 if (*s == '#') { // no default value
394 s += strcspn(s, EZXML_WS ">") - 1;
395 if (*c == ' ') continue; // cdata is default, nothing to do
396 v = NULL;
397 } else if ((*s == '"' || *s == '\'') && // default value
398 (s = strchr(v = s + 1, *s))) *s = '\0';
399 else {
400 ezxml_err(root, t, "malformed <!ATTLIST");
401 break;
402 }
403
404 if (! root->attr[i]) { // new tag name
405 root->attr = (! i) ? malloc(2 * sizeof(char **))
406 : realloc(root->attr,
407 (i + 2) * sizeof(char **));
408 root->attr[i] = malloc(2 * sizeof(char *));
409 root->attr[i][0] = t; // set tag name
410 root->attr[i][1] = (char *)(root->attr[i + 1] = NULL);
411 }
412
413 for (j = 1; root->attr[i][j]; j += 3); // find end of list
414 root->attr[i] = realloc(root->attr[i],
415 (j + 4) * sizeof(char *));
416
417 root->attr[i][j + 3] = NULL; // null terminate list
418 root->attr[i][j + 2] = c; // is it cdata?
419 root->attr[i][j + 1] = (v) ? ezxml_decode(v, root->ent, *c)
420 : NULL;
421 root->attr[i][j] = n; // attribute name
422 }
423 } else if (! strncmp(s, "<!--", 4)) s = strstr(s + 4, "-->"); // comments
424 else if (! strncmp(s, "<?", 2)) { // processing instructions
425 if ((s = strstr(c = s + 2, "?>")))
426 ezxml_proc_inst(root, c, s++ - c);
427 } else if (*s == '<') s = strchr(s, '>'); // skip other declarations
428 else if (*(s++) == '%' && ! root->standalone) break;
429 }
430
431 free(pe);
432 return ! *root->err;
433 }
434
435 // Converts a UTF-16 string to UTF-8. Returns a new string that must be freed
436 // or NULL if no conversion was needed.
ezxml_str2utf8(char ** s,size_t * len)437 char *ezxml_str2utf8(char **s, size_t *len)
438 {
439 char *u;
440 size_t l = 0, sl, max = *len;
441 long c, d;
442 int b, be = (**s == '\xFE') ? 1 : (**s == '\xFF') ? 0 : -1;
443
444 if (be == -1) return NULL; // not UTF-16
445
446 u = malloc(max);
447 for (sl = 2; sl < *len - 1; sl += 2) {
448 c = (be) ? (((*s)[sl] & 0xFF) << 8) | ((*s)[sl + 1] & 0xFF) //UTF-16BE
449 : (((*s)[sl + 1] & 0xFF) << 8) | ((*s)[sl] & 0xFF); //UTF-16LE
450 if (c >= 0xD800 && c <= 0xDFFF && (sl += 2) < *len - 1) { // high-half
451 d = (be) ? (((*s)[sl] & 0xFF) << 8) | ((*s)[sl + 1] & 0xFF)
452 : (((*s)[sl + 1] & 0xFF) << 8) | ((*s)[sl] & 0xFF);
453 c = (((c & 0x3FF) << 10) | (d & 0x3FF)) + 0x10000;
454 }
455
456 while (l + 6 > max) u = realloc(u, max += EZXML_BUFSIZE);
457 if (c < 0x80) u[l++] = c; // US-ASCII subset
458 else { // multi-byte UTF-8 sequence
459 for (b = 0, d = c; d; d /= 2) b++; // bits in c
460 b = (b - 2) / 5; // bytes in payload
461 u[l++] = (0xFF << (7 - b)) | (c >> (6 * b)); // head
462 while (b) u[l++] = 0x80 | ((c >> (6 * --b)) & 0x3F); // payload
463 }
464 }
465 return *s = realloc(u, *len = l);
466 }
467
468 // frees a tag attribute list
ezxml_free_attr(char ** attr)469 void ezxml_free_attr(char **attr)
470 {
471 int i = 0;
472 char *m;
473
474 if (! attr || attr == EZXML_NIL) return; // nothing to free
475 while (attr[i]) i += 2; // find end of attribute list
476 m = attr[i + 1]; // list of which names and values are malloced
477 for (i = 0; m[i]; i++) {
478 if (m[i] & EZXML_NAMEM) free(attr[i * 2]);
479 if (m[i] & EZXML_TXTM) free(attr[(i * 2) + 1]);
480 }
481 free(m);
482 free(attr);
483 }
484
485 // parse the given xml string and return an ezxml structure
ezxml_parse_str(char * s,size_t len)486 ezxml_t ezxml_parse_str(char *s, size_t len)
487 {
488 ezxml_root_t root = (ezxml_root_t)ezxml_new(NULL);
489 char q, e, *d, **attr, **a = NULL; // initialize a to avoid compile warning
490 int l, i, j;
491
492 root->m = s;
493 if (! len) return ezxml_err(root, NULL, "root tag missing");
494 root->u = ezxml_str2utf8(&s, &len); // convert utf-16 to utf-8
495 root->e = (root->s = s) + len; // record start and end of work area
496
497 e = s[len - 1]; // save end char
498 s[len - 1] = '\0'; // turn end char into null terminator
499
500 while (*s && *s != '<') s++; // find first tag
501 if (! *s) return ezxml_err(root, s, "root tag missing");
502
503 for (; ; ) {
504 attr = (char **)EZXML_NIL;
505 d = ++s;
506
507 if (isalpha(*s) || *s == '_' || *s == ':' || *s < '\0') { // new tag
508 if (! root->cur)
509 return ezxml_err(root, d, "markup outside of root element");
510
511 s += strcspn(s, EZXML_WS "/>");
512 while (isspace(*s)) *(s++) = '\0'; // null terminate tag name
513
514 if (*s && *s != '/' && *s != '>') // find tag in default attr list
515 for (i = 0; (a = root->attr[i]) && strcmp(a[0], d); i++);
516
517 for (l = 0; *s && *s != '/' && *s != '>'; l += 2) { // new attrib
518 attr = (l) ? realloc(attr, (l + 4) * sizeof(char *))
519 : malloc(4 * sizeof(char *)); // allocate space
520 attr[l + 3] = (l) ? realloc(attr[l + 1], (l / 2) + 2)
521 : malloc(2); // mem for list of maloced vals
522 strcpy(attr[l + 3] + (l / 2), " "); // value is not malloced
523 attr[l + 2] = NULL; // null terminate list
524 attr[l + 1] = ""; // temporary attribute value
525 attr[l] = s; // set attribute name
526
527 s += strcspn(s, EZXML_WS "=/>");
528 if (*s == '=' || isspace(*s)) {
529 *(s++) = '\0'; // null terminate tag attribute name
530 q = *(s += strspn(s, EZXML_WS "="));
531 if (q == '"' || q == '\'') { // attribute value
532 attr[l + 1] = ++s;
533 while (*s && *s != q) s++;
534 if (*s) *(s++) = '\0'; // null terminate attribute val
535 else {
536 ezxml_free_attr(attr);
537 return ezxml_err(root, d, "missing %c", q);
538 }
539
540 for (j = 1; a && a[j] && strcmp(a[j], attr[l]); j +=3);
541 attr[l + 1] = ezxml_decode(attr[l + 1], root->ent, (a
542 && a[j]) ? *a[j + 2] : ' ');
543 if (attr[l + 1] < d || attr[l + 1] > s)
544 attr[l + 3][l / 2] = EZXML_TXTM; // value malloced
545 }
546 }
547 while (isspace(*s)) s++;
548 }
549
550 if (*s == '/') { // self closing tag
551 *(s++) = '\0';
552 if ((*s && *s != '>') || (! *s && e != '>')) {
553 if (l) ezxml_free_attr(attr);
554 return ezxml_err(root, d, "missing >");
555 }
556 ezxml_open_tag(root, d, attr);
557 ezxml_close_tag(root, d, s);
558 } else if ((q = *s) == '>' || (! *s && e == '>')) { // open tag
559 *s = '\0'; // temporarily null terminate tag name
560 ezxml_open_tag(root, d, attr);
561 *s = q;
562 } else {
563 if (l) ezxml_free_attr(attr);
564 return ezxml_err(root, d, "missing >");
565 }
566 } else if (*s == '/') { // close tag
567 s += strcspn(d = s + 1, EZXML_WS ">") + 1;
568 if (! (q = *s) && e != '>') return ezxml_err(root, d, "missing >");
569 *s = '\0'; // temporarily null terminate tag name
570 if (ezxml_close_tag(root, d, s)) return &root->xml;
571 if (isspace(*s = q)) s += strspn(s, EZXML_WS);
572 } else if (! strncmp(s, "!--", 3)) { // xml comment
573 if (! (s = strstr(s + 3, "--")) || (*(s += 2) != '>' && *s) ||
574 (! *s && e != '>')) return ezxml_err(root, d, "unclosed <!--");
575 } else if (! strncmp(s, "![CDATA[", 8)) { // cdata
576 if ((s = strstr(s, "]]>")))
577 ezxml_char_content(root, d + 8, (s += 2) - d - 10, 'c');
578 else return ezxml_err(root, d, "unclosed <![CDATA[");
579 } else if (! strncmp(s, "!DOCTYPE", 8)) { // dtd
580 for (l = 0; *s && ((! l && *s != '>') || (l && (*s != ']' ||
581 *(s + strspn(s + 1, EZXML_WS) + 1) != '>')));
582 l = (*s == '[') ? 1 : l) s += strcspn(s + 1, "[]>") + 1;
583 if (! *s && e != '>')
584 return ezxml_err(root, d, "unclosed <!DOCTYPE");
585 d = (l) ? strchr(d, '[') + 1 : d;
586 if (l && ! ezxml_internal_dtd(root, d, s++ - d)) return &root->xml;
587 } else if (*s == '?') { // <?...?> processing instructions
588 do {
589 s = strchr(s, '?');
590 } while (s && *(++s) && *s != '>');
591 if (! s || (! *s && e != '>'))
592 return ezxml_err(root, d, "unclosed <?");
593 else ezxml_proc_inst(root, d + 1, s - d - 2);
594 } else return ezxml_err(root, d, "unexpected <");
595
596 if (! s || ! *s) break;
597 *s = '\0';
598 d = ++s;
599 if (*s && *s != '<') { // tag character content
600 while (*s && *s != '<') s++;
601 if (*s) ezxml_char_content(root, d, s - d, '&');
602 else break;
603 } else if (! *s) break;
604 }
605
606 if (! root->cur) return &root->xml;
607 else if (! root->cur->name) return ezxml_err(root, d, "root tag missing");
608 else return ezxml_err(root, d, "unclosed tag <%s>", root->cur->name);
609 }
610
611 // Wrapper for ezxml_parse_str() that accepts a file stream. Reads the entire
612 // stream into memory and then parses it. For xml files, use ezxml_parse_file()
613 // or ezxml_parse_fd()
ezxml_parse_fp(FILE * fp)614 ezxml_t ezxml_parse_fp(FILE *fp)
615 {
616 ezxml_root_t root;
617 size_t l, len = 0;
618 char *s;
619
620 if (! (s = malloc(EZXML_BUFSIZE))) return NULL;
621 do {
622 len += (l = fread((s + len), 1, EZXML_BUFSIZE, fp));
623 if (l == EZXML_BUFSIZE) s = realloc(s, len + EZXML_BUFSIZE);
624 } while (s && l == EZXML_BUFSIZE);
625
626 if (! s) return NULL;
627 root = (ezxml_root_t)ezxml_parse_str(s, len);
628 root->len = -1; // so we know to free s in ezxml_free()
629 return &root->xml;
630 }
631
632 // A wrapper for ezxml_parse_str() that accepts a file descriptor. First
633 // attempts to mem map the file. Failing that, reads the file into memory.
634 // Returns NULL on failure.
ezxml_parse_fd(int fd)635 ezxml_t ezxml_parse_fd(int fd)
636 {
637 ezxml_root_t root;
638 struct stat st;
639 size_t l;
640 void *m;
641
642 if (fd < 0) return NULL;
643 fstat(fd, &st);
644
645 #ifndef EZXML_NOMMAP
646 l = (st.st_size + sysconf(_SC_PAGESIZE) - 1) & ~(sysconf(_SC_PAGESIZE) -1);
647 if ((m = mmap(NULL, l, PROT_READ | PROT_WRITE, MAP_PRIVATE, fd, 0)) !=
648 MAP_FAILED) {
649 madvise(m, l, MADV_SEQUENTIAL); // optimize for sequential access
650 root = (ezxml_root_t)ezxml_parse_str(m, st.st_size);
651 madvise(m, root->len = l, MADV_NORMAL); // put it back to normal
652 } else { // mmap failed, read file into memory
653 #endif // EZXML_NOMMAP
654 l = read(fd, m = malloc(st.st_size), st.st_size);
655 root = (ezxml_root_t)ezxml_parse_str(m, l);
656 root->len = -1; // so we know to free s in ezxml_free()
657 #ifndef EZXML_NOMMAP
658 }
659 #endif // EZXML_NOMMAP
660 return &root->xml;
661 }
662
663 // a wrapper for ezxml_parse_fd that accepts a file name
ezxml_parse_file(const char * file)664 ezxml_t ezxml_parse_file(const char *file)
665 {
666 int fd = open(file, O_RDONLY, 0);
667 ezxml_t xml = ezxml_parse_fd(fd);
668
669 if (fd >= 0) close(fd);
670 return xml;
671 }
672
673 // Encodes ampersand sequences appending the results to *dst, reallocating *dst
674 // if length excedes max. a is non-zero for attribute encoding. Returns *dst
ezxml_ampencode(const char * s,size_t len,char ** dst,size_t * dlen,size_t * max,short a)675 char *ezxml_ampencode(const char *s, size_t len, char **dst, size_t *dlen,
676 size_t *max, short a)
677 {
678 const char *e;
679
680 for (e = s + len; s != e; s++) {
681 while (*dlen + 10 > *max) *dst = realloc(*dst, *max += EZXML_BUFSIZE);
682
683 switch (*s) {
684 case '\0':
685 return *dst;
686 case '&':
687 *dlen += sprintf(*dst + *dlen, "&");
688 break;
689 case '<':
690 *dlen += sprintf(*dst + *dlen, "<");
691 break;
692 case '>':
693 *dlen += sprintf(*dst + *dlen, ">");
694 break;
695 case '"':
696 *dlen += sprintf(*dst + *dlen, (a) ? """ : "\"");
697 break;
698 case '\n':
699 *dlen += sprintf(*dst + *dlen, (a) ? "
" : "\n");
700 break;
701 case '\t':
702 *dlen += sprintf(*dst + *dlen, (a) ? "	" : "\t");
703 break;
704 case '\r':
705 *dlen += sprintf(*dst + *dlen, "
");
706 break;
707 default:
708 (*dst)[(*dlen)++] = *s;
709 }
710 }
711 return *dst;
712 }
713
714 // Recursively converts each tag to xml appending it to *s. Reallocates *s if
715 // its length excedes max. start is the location of the previous tag in the
716 // parent tag's character content. Returns *s.
ezxml_toxml_r(ezxml_t xml,char ** s,size_t * len,size_t * max,size_t start,char *** attr)717 char *ezxml_toxml_r(ezxml_t xml, char **s, size_t *len, size_t *max,
718 size_t start, char ***attr)
719 {
720 int i, j;
721 char *txt = (xml->parent) ? xml->parent->txt : "";
722 size_t off = 0;
723
724 // parent character content up to this tag
725 *s = ezxml_ampencode(txt + start, xml->off - start, s, len, max, 0);
726
727 while (*len + strlen(xml->name) + 4 > *max) // reallocate s
728 *s = realloc(*s, *max += EZXML_BUFSIZE);
729
730 *len += sprintf(*s + *len, "<%s", xml->name); // open tag
731 for (i = 0; xml->attr[i]; i += 2) { // tag attributes
732 if (ezxml_attr(xml, xml->attr[i]) != xml->attr[i + 1]) continue;
733 while (*len + strlen(xml->attr[i]) + 7 > *max) // reallocate s
734 *s = realloc(*s, *max += EZXML_BUFSIZE);
735
736 *len += sprintf(*s + *len, " %s=\"", xml->attr[i]);
737 ezxml_ampencode(xml->attr[i + 1], -1, s, len, max, 1);
738 *len += sprintf(*s + *len, "\"");
739 }
740
741 for (i = 0; attr[i] && strcmp(attr[i][0], xml->name); i++);
742 for (j = 1; attr[i] && attr[i][j]; j += 3) { // default attributes
743 if (! attr[i][j + 1] || ezxml_attr(xml, attr[i][j]) != attr[i][j + 1])
744 continue; // skip duplicates and non-values
745 while (*len + strlen(attr[i][j]) + 7 > *max) // reallocate s
746 *s = realloc(*s, *max += EZXML_BUFSIZE);
747
748 *len += sprintf(*s + *len, " %s=\"", attr[i][j]);
749 ezxml_ampencode(attr[i][j + 1], -1, s, len, max, 1);
750 *len += sprintf(*s + *len, "\"");
751 }
752 *len += sprintf(*s + *len, ">");
753
754 *s = (xml->child) ? ezxml_toxml_r(xml->child, s, len, max, 0, attr) //child
755 : ezxml_ampencode(xml->txt, -1, s, len, max, 0); //data
756
757 while (*len + strlen(xml->name) + 4 > *max) // reallocate s
758 *s = realloc(*s, *max += EZXML_BUFSIZE);
759
760 *len += sprintf(*s + *len, "</%s>", xml->name); // close tag
761
762 while (txt[off] && off < xml->off) off++; // make sure off is within bounds
763 return (xml->ordered) ? ezxml_toxml_r(xml->ordered, s, len, max, off, attr)
764 : ezxml_ampencode(txt + off, -1, s, len, max, 0);
765 }
766
767 // Converts an ezxml structure back to xml. Returns a string of xml data that
768 // must be freed.
ezxml_toxml(ezxml_t xml)769 char *ezxml_toxml(ezxml_t xml)
770 {
771 ezxml_t p = (xml) ? xml->parent : NULL, o = (xml) ? xml->ordered : NULL;
772 ezxml_root_t root = (ezxml_root_t)xml;
773 size_t len = 0, max = EZXML_BUFSIZE;
774 char *s = strcpy(malloc(max), ""), *t, *n;
775 int i, j, k;
776
777 if (! xml || ! xml->name) return realloc(s, len + 1);
778 while (root->xml.parent) root = (ezxml_root_t)root->xml.parent; // root tag
779
780 for (i = 0; ! p && root->pi[i]; i++) { // pre-root processing instructions
781 for (k = 2; root->pi[i][k - 1]; k++);
782 for (j = 1; (n = root->pi[i][j]); j++) {
783 if (root->pi[i][k][j - 1] == '>') continue; // not pre-root
784 while (len + strlen(t = root->pi[i][0]) + strlen(n) + 7 > max)
785 s = realloc(s, max += EZXML_BUFSIZE);
786 len += sprintf(s + len, "<?%s%s%s?>\n", t, *n ? " " : "", n);
787 }
788 }
789
790 xml->parent = xml->ordered = NULL;
791 s = ezxml_toxml_r(xml, &s, &len, &max, 0, root->attr);
792 xml->parent = p;
793 xml->ordered = o;
794
795 for (i = 0; ! p && root->pi[i]; i++) { // post-root processing instructions
796 for (k = 2; root->pi[i][k - 1]; k++);
797 for (j = 1; (n = root->pi[i][j]); j++) {
798 if (root->pi[i][k][j - 1] == '<') continue; // not post-root
799 while (len + strlen(t = root->pi[i][0]) + strlen(n) + 7 > max)
800 s = realloc(s, max += EZXML_BUFSIZE);
801 len += sprintf(s + len, "\n<?%s%s%s?>", t, *n ? " " : "", n);
802 }
803 }
804 return realloc(s, len + 1);
805 }
806
807 // free the memory allocated for the ezxml structure
ezxml_free(ezxml_t xml)808 void ezxml_free(ezxml_t xml)
809 {
810 ezxml_root_t root = (ezxml_root_t)xml;
811 int i, j;
812 char **a, *s;
813
814 if (! xml) return;
815 ezxml_free(xml->child);
816 ezxml_free(xml->ordered);
817
818 if (! xml->parent) { // free root tag allocations
819 for (i = 10; root->ent[i]; i += 2) // 0 - 9 are default entites (<>&"')
820 if ((s = root->ent[i + 1]) < root->s || s > root->e) free(s);
821 free(root->ent); // free list of general entities
822
823 for (i = 0; (a = root->attr[i]); i++) {
824 for (j = 1; a[j++]; j += 2) // free malloced attribute values
825 if (a[j] && (a[j] < root->s || a[j] > root->e)) free(a[j]);
826 free(a);
827 }
828 if (root->attr[0]) free(root->attr); // free default attribute list
829
830 for (i = 0; root->pi[i]; i++) {
831 for (j = 1; root->pi[i][j]; j++);
832 free(root->pi[i][j + 1]);
833 free(root->pi[i]);
834 }
835 if (root->pi[0]) free(root->pi); // free processing instructions
836
837 if (root->len == -1) free(root->m); // malloced xml data
838 #ifndef EZXML_NOMMAP
839 else if (root->len) munmap(root->m, root->len); // mem mapped xml data
840 #endif // EZXML_NOMMAP
841 if (root->u) free(root->u); // utf8 conversion
842 }
843
844 ezxml_free_attr(xml->attr); // tag attributes
845 if ((xml->flags & EZXML_TXTM)) free(xml->txt); // character content
846 if ((xml->flags & EZXML_NAMEM)) free(xml->name); // tag name
847 free(xml);
848 }
849
850 // return parser error message or empty string if none
ezxml_error(ezxml_t xml)851 const char *ezxml_error(ezxml_t xml)
852 {
853 while (xml && xml->parent) xml = xml->parent; // find root tag
854 return (xml) ? ((ezxml_root_t)xml)->err : "";
855 }
856
857 // returns a new empty ezxml structure with the given root tag name
ezxml_new(const char * name)858 ezxml_t ezxml_new(const char *name)
859 {
860 static char *ent[] = { "lt;", "<", "gt;", ">", "quot;", """,
861 "apos;", "'", "amp;", "&", NULL
862 };
863 ezxml_root_t root = (ezxml_root_t)memset(malloc(sizeof(struct ezxml_root)),
864 '\0', sizeof(struct ezxml_root));
865 root->xml.name = (char *)name;
866 root->cur = &root->xml;
867 strcpy(root->err, root->xml.txt = "");
868 root->ent = memcpy(malloc(sizeof(ent)), ent, sizeof(ent));
869 root->attr = root->pi = (char ***)(root->xml.attr = EZXML_NIL);
870 return &root->xml;
871 }
872
873 // inserts an existing tag into an ezxml structure
ezxml_insert(ezxml_t xml,ezxml_t dest,size_t off)874 ezxml_t ezxml_insert(ezxml_t xml, ezxml_t dest, size_t off)
875 {
876 ezxml_t cur, prev, head;
877
878 xml->next = xml->sibling = xml->ordered = NULL;
879 xml->off = off;
880 xml->parent = dest;
881
882 if ((head = dest->child)) { // already have sub tags
883 if (head->off <= off) { // not first subtag
884 for (cur = head; cur->ordered && cur->ordered->off <= off;
885 cur = cur->ordered);
886 xml->ordered = cur->ordered;
887 cur->ordered = xml;
888 } else { // first subtag
889 xml->ordered = head;
890 dest->child = xml;
891 }
892
893 for (cur = head, prev = NULL; cur && strcmp(cur->name, xml->name);
894 prev = cur, cur = cur->sibling); // find tag type
895 if (cur && cur->off <= off) { // not first of type
896 while (cur->next && cur->next->off <= off) cur = cur->next;
897 xml->next = cur->next;
898 cur->next = xml;
899 } else { // first tag of this type
900 if (prev && cur) prev->sibling = cur->sibling; // remove old first
901 xml->next = cur; // old first tag is now next
902 for (cur = head, prev = NULL; cur && cur->off <= off;
903 prev = cur, cur = cur->sibling); // new sibling insert point
904 xml->sibling = cur;
905 if (prev) prev->sibling = xml;
906 }
907 } else dest->child = xml; // only sub tag
908
909 return xml;
910 }
911
912 // Adds a child tag. off is the offset of the child tag relative to the start
913 // of the parent tag's character content. Returns the child tag.
ezxml_add_child(ezxml_t xml,const char * name,size_t off)914 ezxml_t ezxml_add_child(ezxml_t xml, const char *name, size_t off)
915 {
916 ezxml_t child;
917
918 if (! xml) return NULL;
919 child = (ezxml_t)memset(malloc(sizeof(struct ezxml)), '\0',
920 sizeof(struct ezxml));
921 child->name = (char *)name;
922 child->attr = EZXML_NIL;
923 child->txt = "";
924
925 return ezxml_insert(child, xml, off);
926 }
927
928 // sets the character content for the given tag and returns the tag
ezxml_set_txt(ezxml_t xml,const char * txt)929 ezxml_t ezxml_set_txt(ezxml_t xml, const char *txt)
930 {
931 if (! xml) return NULL;
932 if (xml->flags & EZXML_TXTM) free(xml->txt); // existing txt was malloced
933 xml->flags &= ~EZXML_TXTM;
934 xml->txt = (char *)txt;
935 return xml;
936 }
937
938 // Sets the given tag attribute or adds a new attribute if not found. A value
939 // of NULL will remove the specified attribute. Returns the tag given.
ezxml_set_attr(ezxml_t xml,const char * name,const char * value)940 ezxml_t ezxml_set_attr(ezxml_t xml, const char *name, const char *value)
941 {
942 int l = 0, c;
943
944 if (! xml) return NULL;
945 while (xml->attr[l] && strcmp(xml->attr[l], name)) l += 2;
946 if (! xml->attr[l]) { // not found, add as new attribute
947 if (! value) return xml; // nothing to do
948 if (xml->attr == EZXML_NIL) { // first attribute
949 xml->attr = malloc(4 * sizeof(char *));
950 xml->attr[1] = strdup(""); // empty list of malloced names/vals
951 } else xml->attr = realloc(xml->attr, (l + 4) * sizeof(char *));
952
953 xml->attr[l] = (char *)name; // set attribute name
954 xml->attr[l + 2] = NULL; // null terminate attribute list
955 xml->attr[l + 3] = realloc(xml->attr[l + 1],
956 (c = strlen(xml->attr[l + 1])) + 2);
957 strcpy(xml->attr[l + 3] + c, " "); // set name/value as not malloced
958 if (xml->flags & EZXML_DUP) xml->attr[l + 3][c] = EZXML_NAMEM;
959 } else if (xml->flags & EZXML_DUP) free((char *)name); // name was strduped
960
961 for (c = l; xml->attr[c]; c += 2); // find end of attribute list
962 if (xml->attr[c + 1][l / 2] & EZXML_TXTM) free(xml->attr[l + 1]); //old val
963 if (xml->flags & EZXML_DUP) xml->attr[c + 1][l / 2] |= EZXML_TXTM;
964 else xml->attr[c + 1][l / 2] &= ~EZXML_TXTM;
965
966 if (value) xml->attr[l + 1] = (char *)value; // set attribute value
967 else { // remove attribute
968 if (xml->attr[c + 1][l / 2] & EZXML_NAMEM) free(xml->attr[l]);
969 memmove(xml->attr + l, xml->attr + l + 2, (c - l + 2) * sizeof(char*));
970 xml->attr = realloc(xml->attr, (c + 2) * sizeof(char *));
971 memmove(xml->attr[c + 1] + (l / 2), xml->attr[c + 1] + (l / 2) + 1,
972 (c / 2) - (l / 2)); // fix list of which name/vals are malloced
973 }
974 xml->flags &= ~EZXML_DUP; // clear strdup() flag
975 return xml;
976 }
977
978 // sets a flag for the given tag and returns the tag
ezxml_set_flag(ezxml_t xml,short flag)979 ezxml_t ezxml_set_flag(ezxml_t xml, short flag)
980 {
981 if (xml) xml->flags |= flag;
982 return xml;
983 }
984
985 // removes a tag along with its subtags without freeing its memory
ezxml_cut(ezxml_t xml)986 ezxml_t ezxml_cut(ezxml_t xml)
987 {
988 ezxml_t cur;
989
990 if (! xml) return NULL; // nothing to do
991 if (xml->next) xml->next->sibling = xml->sibling; // patch sibling list
992
993 if (xml->parent) { // not root tag
994 cur = xml->parent->child; // find head of subtag list
995 if (cur == xml) xml->parent->child = xml->ordered; // first subtag
996 else { // not first subtag
997 while (cur->ordered != xml) cur = cur->ordered;
998 cur->ordered = cur->ordered->ordered; // patch ordered list
999
1000 cur = xml->parent->child; // go back to head of subtag list
1001 if (strcmp(cur->name, xml->name)) { // not in first sibling list
1002 while (strcmp(cur->sibling->name, xml->name))
1003 cur = cur->sibling;
1004 if (cur->sibling == xml) { // first of a sibling list
1005 cur->sibling = (xml->next) ? xml->next
1006 : cur->sibling->sibling;
1007 } else cur = cur->sibling; // not first of a sibling list
1008 }
1009
1010 while (cur->next && cur->next != xml) cur = cur->next;
1011 if (cur->next) cur->next = cur->next->next; // patch next list
1012 }
1013 }
1014 xml->ordered = xml->sibling = xml->next = NULL;
1015 return xml;
1016 }
1017
1018 #ifdef EZXML_TEST // test harness
main(int argc,char ** argv)1019 int main(int argc, char **argv)
1020 {
1021 ezxml_t xml;
1022 char *s;
1023 int i;
1024
1025 if (argc != 2) return fprintf(stderr, "usage: %s xmlfile\n", argv[0]);
1026
1027 xml = ezxml_parse_file(argv[1]);
1028 printf("%s\n", (s = ezxml_toxml(xml)));
1029 free(s);
1030 i = fprintf(stderr, "%s", ezxml_error(xml));
1031 ezxml_free(xml);
1032 return (i) ? 1 : 0;
1033 }
1034 #endif // EZXML_TEST
1035