« back to all changes in this revision
Viewing changes to src/libezxml/ezxml.c
- browse files at revision 10
- compare with another revision
- download diff
- download tarball
- view history from revision 10
- Committer: Bazaar Package Importer
- Author(s): Luke Faraone
- Date: 2011-02-08 17:23:25 UTC
- mfrom: (1.3.3 upstream)
- Revision ID: james.westby@ubuntu.com-20110208172325-0qf3sxpsu37j5ez9
Tags: 2011.01.24-1
* New upstream version.
* Switch to DEP5 copyright.
* Augment CFLAGS to use the c99 standard.
* Don't install the now-removed AUTHORS file into docs.
* Drop dep on cmake.
* Switch to DEP5 copyright.
* Augment CFLAGS to use the c99 standard.
* Don't install the now-removed AUTHORS file into docs.
* Drop dep on cmake.
- files added:
- contrib/change-station-dmenu.sh
- src/libezxml
- src/libpiano
- src/libwaitress
- files removed:
- AUTHORS
- libezxml
- libezxml/src
- libpiano
- libpiano/src
- libwaitress
- libwaitress/src
- files modified:
- COPYING
added
removed
src/libezxml/ezxml.c
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
#define _BSD_SOURCE /* required by strdup() */
26
27
#include <stdlib.h>
28
#include <stdio.h>
29
#include <stdarg.h>
30
#include <string.h>
31
#include <ctype.h>
32
#include <unistd.h>
33
#include <sys/types.h>
34
#include <sys/stat.h>
35
#include "ezxml.h"
36
37
#define EZXML_WS "\t\r\n " // whitespace
38
#define EZXML_ERRL 128 // maximum error string length
39
40
typedef struct ezxml_root *ezxml_root_t;
41
struct ezxml_root { // additional data for the root tag
42
struct ezxml xml; // is a super-struct built on top of ezxml struct
43
ezxml_t cur; // current xml tree insertion point
44
char *m; // original xml string
45
size_t len; // length of allocated memory for mmap, -1 for malloc
46
char *u; // UTF-8 conversion of string if original was UTF-16
47
char *s; // start of work area
48
char *e; // end of work area
49
char **ent; // general entities (ampersand sequences)
50
char ***attr; // default attributes
51
char ***pi; // processing instructions
52
short standalone; // non-zero if <?xml standalone="yes"?>
53
char err[EZXML_ERRL]; // error string
54
};
55
56
char *EZXML_NIL[] = { NULL }; // empty, null terminated array of strings
57
58
// sets a flag for the given tag and returns the tag
59
static ezxml_t ezxml_set_flag(ezxml_t xml, short flag) {
60
if (xml) xml->flags |= flag;
61
return xml;
62
}
63
64
// inserts an existing tag into an ezxml structure
65
static ezxml_t ezxml_insert(ezxml_t xml, ezxml_t dest, size_t off)
66
{
67
ezxml_t cur, prev, head;
68
69
xml->next = xml->sibling = xml->ordered = NULL;
70
xml->off = off;
71
xml->parent = dest;
72
73
if ((head = dest->child)) { // already have sub tags
74
if (head->off <= off) { // not first subtag
75
for (cur = head; cur->ordered && cur->ordered->off <= off;
76
cur = cur->ordered);
77
xml->ordered = cur->ordered;
78
cur->ordered = xml;
79
}
80
else { // first subtag
81
xml->ordered = head;
82
dest->child = xml;
83
}
84
85
for (cur = head, prev = NULL; cur && strcmp(cur->name, xml->name);
86
prev = cur, cur = cur->sibling); // find tag type
87
if (cur && cur->off <= off) { // not first of type
88
while (cur->next && cur->next->off <= off) cur = cur->next;
89
xml->next = cur->next;
90
cur->next = xml;
91
}
92
else { // first tag of this type
93
if (prev && cur) prev->sibling = cur->sibling; // remove old first
94
xml->next = cur; // old first tag is now next
95
for (cur = head, prev = NULL; cur && cur->off <= off;
96
prev = cur, cur = cur->sibling); // new sibling insert point
97
xml->sibling = cur;
98
if (prev) prev->sibling = xml;
99
}
100
}
101
else dest->child = xml; // only sub tag
102
103
return xml;
104
}
105
106
// Adds a child tag. off is the offset of the child tag relative to the start
107
// of the parent tag's character content. Returns the child tag.
108
static ezxml_t ezxml_add_child(ezxml_t xml, const char *name, size_t off)
109
{
110
ezxml_t child;
111
112
if (! xml) return NULL;
113
child = (ezxml_t)memset(malloc(sizeof(struct ezxml)), '\0',
114
sizeof(struct ezxml));
115
child->name = (char *)name;
116
child->attr = EZXML_NIL;
117
child->txt = "";
118
119
return ezxml_insert(child, xml, off);
120
}
121
122
// returns the first child tag with the given name or NULL if not found
123
ezxml_t ezxml_child(ezxml_t xml, const char *name)
124
{
125
xml = (xml) ? xml->child : NULL;
126
while (xml && strcmp(name, xml->name)) xml = xml->sibling;
127
return xml;
128
}
129
130
// returns a new empty ezxml structure with the given root tag name
131
static ezxml_t ezxml_new(const char *name)
132
{
133
static char *ent[] = { "lt;", "<", "gt;", ">", "quot;", """,
134
"apos;", "'", "amp;", "&", NULL };
135
ezxml_root_t root = (ezxml_root_t)memset(malloc(sizeof(struct ezxml_root)),
136
'\0', sizeof(struct ezxml_root));
137
root->xml.name = (char *)name;
138
root->cur = &root->xml;
139
strcpy(root->err, root->xml.txt = "");
140
root->ent = memcpy(malloc(sizeof(ent)), ent, sizeof(ent));
141
root->attr = root->pi = (char ***)(root->xml.attr = EZXML_NIL);
142
return &root->xml;
143
}
144
145
// returns the Nth tag with the same name in the same subsection or NULL if not
146
// found
147
ezxml_t ezxml_idx(ezxml_t xml, int idx)
148
{
149
for (; xml && idx; idx--) xml = xml->next;
150
return xml;
151
}
152
153
// returns the value of the requested tag attribute or NULL if not found
154
const char *ezxml_attr(ezxml_t xml, const char *attr)
155
{
156
int i = 0, j = 1;
157
ezxml_root_t root = (ezxml_root_t)xml;
158
159
if (! xml || ! xml->attr) return NULL;
160
while (xml->attr[i] && strcmp(attr, xml->attr[i])) i += 2;
161
if (xml->attr[i]) return xml->attr[i + 1]; // found attribute
162
163
while (root->xml.parent) root = (ezxml_root_t)root->xml.parent; // root tag
164
for (i = 0; root->attr[i] && strcmp(xml->name, root->attr[i][0]); i++);
165
if (! root->attr[i]) return NULL; // no matching default attributes
166
while (root->attr[i][j] && strcmp(attr, root->attr[i][j])) j += 3;
167
return (root->attr[i][j]) ? root->attr[i][j + 1] : NULL; // found default
168
}
169
170
// same as ezxml_get but takes an already initialized va_list
171
static ezxml_t ezxml_vget(ezxml_t xml, va_list ap)
172
{
173
char *name = va_arg(ap, char *);
174
int idx = -1;
175
176
if (name && *name) {
177
idx = va_arg(ap, int);
178
xml = ezxml_child(xml, name);
179
}
180
return (idx < 0) ? xml : ezxml_vget(ezxml_idx(xml, idx), ap);
181
}
182
183
// Traverses the xml tree to retrieve a specific subtag. Takes a variable
184
// length list of tag names and indexes. The argument list must be terminated
185
// by either an index of -1 or an empty string tag name. Example:
186
// title = ezxml_get(library, "shelf", 0, "book", 2, "title", -1);
187
// This retrieves the title of the 3rd book on the 1st shelf of library.
188
// Returns NULL if not found.
189
ezxml_t ezxml_get(ezxml_t xml, ...)
190
{
191
va_list ap;
192
ezxml_t r;
193
194
va_start(ap, xml);
195
r = ezxml_vget(xml, ap);
196
va_end(ap);
197
return r;
198
}
199
200
// set an error string and return root
201
static ezxml_t ezxml_err(ezxml_root_t root, char *s, const char *err, ...)
202
{
203
va_list ap;
204
int line = 1;
205
char *t, fmt[EZXML_ERRL];
206
207
for (t = root->s; t < s; t++) if (*t == '\n') line++;
208
snprintf(fmt, EZXML_ERRL, "[error near line %d]: %s", line, err);
209
210
va_start(ap, err);
211
vsnprintf(root->err, EZXML_ERRL, fmt, ap);
212
va_end(ap);
213
214
return &root->xml;
215
}
216
217
// Recursively decodes entity and character references and normalizes new lines
218
// ent is a null terminated array of alternating entity names and values. set t
219
// to '&' for general entity decoding, '%' for parameter entity decoding, 'c'
220
// for cdata sections, ' ' for attribute normalization, or '*' for non-cdata
221
// attribute normalization. Returns s, or if the decoded string is longer than
222
// s, returns a malloced string that must be freed.
223
static char *ezxml_decode(char *s, char **ent, char t)
224
{
225
char *e, *r = s, *m = s;
226
long b, c, d, l;
227
228
for (; *s; s++) { // normalize line endings
229
while (*s == '\r') {
230
*(s++) = '\n';
231
if (*s == '\n') memmove(s, (s + 1), strlen(s));
232
}
233
}
234
235
for (s = r; ; ) {
236
while (*s && *s != '&' && (*s != '%' || t != '%') && !isspace(*s)) s++;
237
238
if (! *s) break;
239
else if (t != 'c' && ! strncmp(s, "&#", 2)) { // character reference
240
if (s[2] == 'x') c = strtol(s + 3, &e, 16); // base 16
241
else c = strtol(s + 2, &e, 10); // base 10
242
if (! c || *e != ';') { s++; continue; } // not a character ref
243
244
if (c < 0x80) *(s++) = c; // US-ASCII subset
245
else { // multi-byte UTF-8 sequence
246
for (b = 0, d = c; d; d /= 2) b++; // number of bits in c
247
b = (b - 2) / 5; // number of bytes in payload
248
*(s++) = (0xFF << (7 - b)) | (c >> (6 * b)); // head
249
while (b) *(s++) = 0x80 | ((c >> (6 * --b)) & 0x3F); // payload
250
}
251
252
memmove(s, strchr(s, ';') + 1, strlen(strchr(s, ';')));
253
}
254
else if ((*s == '&' && (t == '&' || t == ' ' || t == '*')) ||
255
(*s == '%' && t == '%')) { // entity reference
256
for (b = 0; ent[b] && strncmp(s + 1, ent[b], strlen(ent[b]));
257
b += 2); // find entity in entity list
258
259
if (ent[b++]) { // found a match
260
if ((c = strlen(ent[b])) - 1 > (e = strchr(s, ';')) - s) {
261
l = (d = (s - r)) + c + strlen(e); // new length
262
r = (r == m) ? strcpy(malloc(l), r) : realloc(r, l);
263
e = strchr((s = r + d), ';'); // fix up pointers
264
}
265
266
memmove(s + c, e + 1, strlen(e)); // shift rest of string
267
strncpy(s, ent[b], c); // copy in replacement text
268
}
269
else s++; // not a known entity
270
}
271
else if ((t == ' ' || t == '*') && isspace(*s)) *(s++) = ' ';
272
else s++; // no decoding needed
273
}
274
275
if (t == '*') { // normalize spaces for non-cdata attributes
276
for (s = r; *s; s++) {
277
if ((l = strspn(s, " "))) memmove(s, s + l, strlen(s + l) + 1);
278
while (*s && *s != ' ') s++;
279
}
280
if (--s >= r && *s == ' ') *s = '\0'; // trim any trailing space
281
}
282
return r;
283
}
284
285
// called when parser finds start of new tag
286
static void ezxml_open_tag(ezxml_root_t root, char *name, char **attr)
287
{
288
ezxml_t xml = root->cur;
289
290
if (xml->name) xml = ezxml_add_child(xml, name, strlen(xml->txt));
291
else xml->name = name; // first open tag
292
293
xml->attr = attr;
294
root->cur = xml; // update tag insertion point
295
}
296
297
// called when parser finds character content between open and closing tag
298
static void ezxml_char_content(ezxml_root_t root, char *s, size_t len, char t)
299
{
300
ezxml_t xml = root->cur;
301
char *m = s;
302
size_t l;
303
304
if (! xml || ! xml->name || ! len) return; // sanity check
305
306
s[len] = '\0'; // null terminate text (calling functions anticipate this)
307
len = strlen(s = ezxml_decode(s, root->ent, t)) + 1;
308
309
if (! *(xml->txt)) xml->txt = s; // initial character content
310
else { // allocate our own memory and make a copy
311
xml->txt = (xml->flags & EZXML_TXTM) // allocate some space
312
? realloc(xml->txt, (l = strlen(xml->txt)) + len)
313
: strcpy(malloc((l = strlen(xml->txt)) + len), xml->txt);
314
strcpy(xml->txt + l, s); // add new char content
315
if (s != m) free(s); // free s if it was malloced by ezxml_decode()
316
}
317
318
if (xml->txt != m) ezxml_set_flag(xml, EZXML_TXTM);
319
}
320
321
// called when parser finds closing tag
322
static ezxml_t ezxml_close_tag(ezxml_root_t root, char *name, char *s)
323
{
324
if (! root->cur || ! root->cur->name || strcmp(name, root->cur->name))
325
return ezxml_err(root, s, "unexpected closing tag </%s>", name);
326
327
root->cur = root->cur->parent;
328
return NULL;
329
}
330
331
// checks for circular entity references, returns non-zero if no circular
332
// references are found, zero otherwise
333
static int ezxml_ent_ok(char *name, char *s, char **ent)
334
{
335
int i;
336
337
for (; ; s++) {
338
while (*s && *s != '&') s++; // find next entity reference
339
if (! *s) return 1;
340
if (! strncmp(s + 1, name, strlen(name))) return 0; // circular ref.
341
for (i = 0; ent[i] && strncmp(ent[i], s + 1, strlen(ent[i])); i += 2);
342
if (ent[i] && ! ezxml_ent_ok(name, ent[i + 1], ent)) return 0;
343
}
344
}
345
346
// called when the parser finds a processing instruction
347
static void ezxml_proc_inst(ezxml_root_t root, char *s, size_t len)
348
{
349
int i = 0, j = 1;
350
char *target = s;
351
352
s[len] = '\0'; // null terminate instruction
353
if (*(s += strcspn(s, EZXML_WS))) {
354
*s = '\0'; // null terminate target
355
s += strspn(s + 1, EZXML_WS) + 1; // skip whitespace after target
356
}
357
358
if (! strcmp(target, "xml")) { // <?xml ... ?>
359
if ((s = strstr(s, "standalone")) && ! strncmp(s + strspn(s + 10,
360
EZXML_WS "='\"") + 10, "yes", 3)) root->standalone = 1;
361
return;
362
}
363
364
if (! root->pi[0]) *(root->pi = malloc(sizeof(char **))) = NULL; //first pi
365
366
while (root->pi[i] && strcmp(target, root->pi[i][0])) i++; // find target
367
if (! root->pi[i]) { // new target
368
root->pi = realloc(root->pi, sizeof(char **) * (i + 2));
369
root->pi[i] = malloc(sizeof(char *) * 3);
370
root->pi[i][0] = target;
371
root->pi[i][1] = (char *)(root->pi[i + 1] = NULL); // terminate pi list
372
root->pi[i][2] = strdup(""); // empty document position list
373
}
374
375
while (root->pi[i][j]) j++; // find end of instruction list for this target
376
root->pi[i] = realloc(root->pi[i], sizeof(char *) * (j + 3));
377
root->pi[i][j + 2] = realloc(root->pi[i][j + 1], j + 1);
378
strcpy(root->pi[i][j + 2] + j - 1, (root->xml.name) ? ">" : "<");
379
root->pi[i][j + 1] = NULL; // null terminate pi list for this target
380
root->pi[i][j] = s; // set instruction
381
}
382
383
// called when the parser finds an internal doctype subset
384
static short ezxml_internal_dtd(ezxml_root_t root, char *s, size_t len)
385
{
386
char q, *c, *t, *n = NULL, *v, **ent, **pe;
387
int i, j;
388
389
pe = memcpy(malloc(sizeof(EZXML_NIL)), EZXML_NIL, sizeof(EZXML_NIL));
390
391
for (s[len] = '\0'; s; ) {
392
while (*s && *s != '<' && *s != '%') s++; // find next declaration
393
394
if (! *s) break;
395
else if (! strncmp(s, "<!ENTITY", 8)) { // parse entity definitions
396
c = s += strspn(s + 8, EZXML_WS) + 8; // skip white space separator
397
n = s + strspn(s, EZXML_WS "%"); // find name
398
*(s = n + strcspn(n, EZXML_WS)) = ';'; // append ; to name
399
400
v = s + strspn(s + 1, EZXML_WS) + 1; // find value
401
if ((q = *(v++)) != '"' && q != '\'') { // skip externals
402
s = strchr(s, '>');
403
continue;
404
}
405
406
for (i = 0, ent = (*c == '%') ? pe : root->ent; ent[i]; i++);
407
ent = realloc(ent, (i + 3) * sizeof(char *)); // space for next ent
408
if (*c == '%') pe = ent;
409
else root->ent = ent;
410
411
*(++s) = '\0'; // null terminate name
412
if ((s = strchr(v, q))) *(s++) = '\0'; // null terminate value
413
ent[i + 1] = ezxml_decode(v, pe, '%'); // set value
414
ent[i + 2] = NULL; // null terminate entity list
415
if (! ezxml_ent_ok(n, ent[i + 1], ent)) { // circular reference
416
if (ent[i + 1] != v) free(ent[i + 1]);
417
ezxml_err(root, v, "circular entity declaration &%s", n);
418
break;
419
}
420
else ent[i] = n; // set entity name
421
}
422
else if (! strncmp(s, "<!ATTLIST", 9)) { // parse default attributes
423
t = s + strspn(s + 9, EZXML_WS) + 9; // skip whitespace separator
424
if (! *t) { ezxml_err(root, t, "unclosed <!ATTLIST"); break; }
425
if (*(s = t + strcspn(t, EZXML_WS ">")) == '>') continue;
426
else *s = '\0'; // null terminate tag name
427
for (i = 0; root->attr[i] && strcmp(n, root->attr[i][0]); i++);
428
429
++s; // ansi cpr
430
while (*(n = s + strspn(s, EZXML_WS)) && *n != '>') {
431
if (*(s = n + strcspn(n, EZXML_WS))) *s = '\0'; // attr name
432
else { ezxml_err(root, t, "malformed <!ATTLIST"); break; }
433
434
s += strspn(s + 1, EZXML_WS) + 1; // find next token
435
c = (strncmp(s, "CDATA", 5)) ? "*" : " "; // is it cdata?
436
if (! strncmp(s, "NOTATION", 8))
437
s += strspn(s + 8, EZXML_WS) + 8;
438
s = (*s == '(') ? strchr(s, ')') : s + strcspn(s, EZXML_WS);
439
if (! s) { ezxml_err(root, t, "malformed <!ATTLIST"); break; }
440
441
s += strspn(s, EZXML_WS ")"); // skip white space separator
442
if (! strncmp(s, "#FIXED", 6))
443
s += strspn(s + 6, EZXML_WS) + 6;
444
if (*s == '#') { // no default value
445
s += strcspn(s, EZXML_WS ">") - 1;
446
if (*c == ' ') continue; // cdata is default, nothing to do
447
v = NULL;
448
}
449
else if ((*s == '"' || *s == '\'') && // default value
450
(s = strchr(v = s + 1, *s))) *s = '\0';
451
else { ezxml_err(root, t, "malformed <!ATTLIST"); break; }
452
453
if (! root->attr[i]) { // new tag name
454
root->attr = (! i) ? malloc(2 * sizeof(char **))
455
: realloc(root->attr,
456
(i + 2) * sizeof(char **));
457
root->attr[i] = malloc(2 * sizeof(char *));
458
root->attr[i][0] = t; // set tag name
459
root->attr[i][1] = (char *)(root->attr[i + 1] = NULL);
460
}
461
462
for (j = 1; root->attr[i][j]; j += 3); // find end of list
463
root->attr[i] = realloc(root->attr[i],
464
(j + 4) * sizeof(char *));
465
466
root->attr[i][j + 3] = NULL; // null terminate list
467
root->attr[i][j + 2] = c; // is it cdata?
468
root->attr[i][j + 1] = (v) ? ezxml_decode(v, root->ent, *c)
469
: NULL;
470
root->attr[i][j] = n; // attribute name
471
++s;
472
}
473
}
474
else if (! strncmp(s, "<!--", 4)) s = strstr(s + 4, "-->"); // comments
475
else if (! strncmp(s, "<?", 2)) { // processing instructions
476
if ((s = strstr(c = s + 2, "?>")))
477
ezxml_proc_inst(root, c, s++ - c);
478
}
479
else if (*s == '<') s = strchr(s, '>'); // skip other declarations
480
else if (*(s++) == '%' && ! root->standalone) break;
481
}
482
483
free(pe);
484
return ! *root->err;
485
}
486
487
// Converts a UTF-16 string to UTF-8. Returns a new string that must be freed
488
// or NULL if no conversion was needed.
489
static char *ezxml_str2utf8(char **s, size_t *len)
490
{
491
char *u;
492
size_t l = 0, sl, max = *len;
493
long c, d;
494
int b, be = (**s == '\xFE') ? 1 : (**s == '\xFF') ? 0 : -1;
495
496
if (be == -1) return NULL; // not UTF-16
497
498
u = malloc(max);
499
for (sl = 2; sl < *len - 1; sl += 2) {
500
c = (be) ? (((*s)[sl] & 0xFF) << 8) | ((*s)[sl + 1] & 0xFF) //UTF-16BE
501
: (((*s)[sl + 1] & 0xFF) << 8) | ((*s)[sl] & 0xFF); //UTF-16LE
502
if (c >= 0xD800 && c <= 0xDFFF && (sl += 2) < *len - 1) { // high-half
503
d = (be) ? (((*s)[sl] & 0xFF) << 8) | ((*s)[sl + 1] & 0xFF)
504
: (((*s)[sl + 1] & 0xFF) << 8) | ((*s)[sl] & 0xFF);
505
c = (((c & 0x3FF) << 10) | (d & 0x3FF)) + 0x10000;
506
}
507
508
while (l + 6 > max) u = realloc(u, max += EZXML_BUFSIZE);
509
if (c < 0x80) u[l++] = c; // US-ASCII subset
510
else { // multi-byte UTF-8 sequence
511
for (b = 0, d = c; d; d /= 2) b++; // bits in c
512
b = (b - 2) / 5; // bytes in payload
513
u[l++] = (0xFF << (7 - b)) | (c >> (6 * b)); // head
514
while (b) u[l++] = 0x80 | ((c >> (6 * --b)) & 0x3F); // payload
515
}
516
}
517
return *s = realloc(u, *len = l);
518
}
519
520
// frees a tag attribute list
521
static void ezxml_free_attr(char **attr) {
522
int i = 0;
523
char *m;
524
525
if (! attr || attr == EZXML_NIL) return; // nothing to free
526
while (attr[i]) i += 2; // find end of attribute list
527
m = attr[i + 1]; // list of which names and values are malloced
528
for (i = 0; m[i]; i++) {
529
if (m[i] & EZXML_NAMEM) free(attr[i * 2]);
530
if (m[i] & EZXML_TXTM) free(attr[(i * 2) + 1]);
531
}
532
free(m);
533
free(attr);
534
}
535
536
// parse the given xml string and return an ezxml structure
537
ezxml_t ezxml_parse_str(char *s, size_t len)
538
{
539
ezxml_root_t root = (ezxml_root_t)ezxml_new(NULL);
540
char q, e, *d, **attr, **a = NULL; // initialize a to avoid compile warning
541
int l, i, j;
542
543
root->m = s;
544
if (! len) return ezxml_err(root, NULL, "root tag missing");
545
root->u = ezxml_str2utf8(&s, &len); // convert utf-16 to utf-8
546
root->e = (root->s = s) + len; // record start and end of work area
547
548
e = s[len - 1]; // save end char
549
s[len - 1] = '\0'; // turn end char into null terminator
550
551
while (*s && *s != '<') s++; // find first tag
552
if (! *s) return ezxml_err(root, s, "root tag missing");
553
554
for (; ; ) {
555
attr = (char **)EZXML_NIL;
556
d = ++s;
557
558
if (isalpha(*s) || *s == '_' || *s == ':' || *s < '\0') { // new tag
559
if (! root->cur)
560
return ezxml_err(root, d, "markup outside of root element");
561
562
s += strcspn(s, EZXML_WS "/>");
563
while (isspace(*s)) *(s++) = '\0'; // null terminate tag name
564
565
if (*s && *s != '/' && *s != '>') // find tag in default attr list
566
for (i = 0; (a = root->attr[i]) && strcmp(a[0], d); i++);
567
568
for (l = 0; *s && *s != '/' && *s != '>'; l += 2) { // new attrib
569
attr = (l) ? realloc(attr, (l + 4) * sizeof(char *))
570
: malloc(4 * sizeof(char *)); // allocate space
571
attr[l + 3] = (l) ? realloc(attr[l + 1], (l / 2) + 2)
572
: malloc(2); // mem for list of maloced vals
573
strcpy(attr[l + 3] + (l / 2), " "); // value is not malloced
574
attr[l + 2] = NULL; // null terminate list
575
attr[l + 1] = ""; // temporary attribute value
576
attr[l] = s; // set attribute name
577
578
s += strcspn(s, EZXML_WS "=/>");
579
if (*s == '=' || isspace(*s)) {
580
*(s++) = '\0'; // null terminate tag attribute name
581
q = *(s += strspn(s, EZXML_WS "="));
582
if (q == '"' || q == '\'') { // attribute value
583
attr[l + 1] = ++s;
584
while (*s && *s != q) s++;
585
if (*s) *(s++) = '\0'; // null terminate attribute val
586
else {
587
ezxml_free_attr(attr);
588
return ezxml_err(root, d, "missing %c", q);
589
}
590
591
for (j = 1; a && a[j] && strcmp(a[j], attr[l]); j +=3);
592
attr[l + 1] = ezxml_decode(attr[l + 1], root->ent, (a
593
&& a[j]) ? *a[j + 2] : ' ');
594
if (attr[l + 1] < d || attr[l + 1] > s)
595
attr[l + 3][l / 2] = EZXML_TXTM; // value malloced
596
}
597
}
598
while (isspace(*s)) s++;
599
}
600
601
if (*s == '/') { // self closing tag
602
*(s++) = '\0';
603
if ((*s && *s != '>') || (! *s && e != '>')) {
604
if (l) ezxml_free_attr(attr);
605
return ezxml_err(root, d, "missing >");
606
}
607
ezxml_open_tag(root, d, attr);
608
ezxml_close_tag(root, d, s);
609
}
610
else if ((q = *s) == '>' || (! *s && e == '>')) { // open tag
611
*s = '\0'; // temporarily null terminate tag name
612
ezxml_open_tag(root, d, attr);
613
*s = q;
614
}
615
else {
616
if (l) ezxml_free_attr(attr);
617
return ezxml_err(root, d, "missing >");
618
}
619
}
620
else if (*s == '/') { // close tag
621
s += strcspn(d = s + 1, EZXML_WS ">") + 1;
622
if (! (q = *s) && e != '>') return ezxml_err(root, d, "missing >");
623
*s = '\0'; // temporarily null terminate tag name
624
if (ezxml_close_tag(root, d, s)) return &root->xml;
625
if (isspace(*s = q)) s += strspn(s, EZXML_WS);
626
}
627
else if (! strncmp(s, "!--", 3)) { // xml comment
628
if (! (s = strstr(s + 3, "--")) || (*(s += 2) != '>' && *s) ||
629
(! *s && e != '>')) return ezxml_err(root, d, "unclosed <!--");
630
}
631
else if (! strncmp(s, "![CDATA[", 8)) { // cdata
632
if ((s = strstr(s, "]]>")))
633
ezxml_char_content(root, d + 8, (s += 2) - d - 10, 'c');
634
else return ezxml_err(root, d, "unclosed <![CDATA[");
635
}
636
else if (! strncmp(s, "!DOCTYPE", 8)) { // dtd
637
for (l = 0; *s && ((! l && *s != '>') || (l && (*s != ']' ||
638
*(s + strspn(s + 1, EZXML_WS) + 1) != '>')));
639
l = (*s == '[') ? 1 : l) s += strcspn(s + 1, "[]>") + 1;
640
if (! *s && e != '>')
641
return ezxml_err(root, d, "unclosed <!DOCTYPE");
642
d = (l) ? strchr(d, '[') + 1 : d;
643
if (l && ! ezxml_internal_dtd(root, d, s++ - d)) return &root->xml;
644
}
645
else if (*s == '?') { // <?...?> processing instructions
646
do { s = strchr(s, '?'); } while (s && *(++s) && *s != '>');
647
if (! s || (! *s && e != '>'))
648
return ezxml_err(root, d, "unclosed <?");
649
else ezxml_proc_inst(root, d + 1, s - d - 2);
650
}
651
else return ezxml_err(root, d, "unexpected <");
652
653
if (! s || ! *s) break;
654
*s = '\0';
655
d = ++s;
656
if (*s && *s != '<') { // tag character content
657
while (*s && *s != '<') s++;
658
if (*s) ezxml_char_content(root, d, s - d, '&');
659
else break;
660
}
661
else if (! *s) break;
662
}
663
664
if (! root->cur) return &root->xml;
665
else if (! root->cur->name) return ezxml_err(root, d, "root tag missing");
666
else return ezxml_err(root, d, "unclosed tag <%s>", root->cur->name);
667
}
668
669
// free the memory allocated for the ezxml structure
670
void ezxml_free(ezxml_t xml)
671
{
672
ezxml_root_t root = (ezxml_root_t)xml;
673
int i, j;
674
char **a, *s;
675
676
if (! xml) return;
677
ezxml_free(xml->child);
678
ezxml_free(xml->ordered);
679
680
if (! xml->parent) { // free root tag allocations
681
for (i = 10; root->ent[i]; i += 2) // 0 - 9 are default entites (<>&"')
682
if ((s = root->ent[i + 1]) < root->s || s > root->e) free(s);
683
free(root->ent); // free list of general entities
684
685
for (i = 0; (a = root->attr[i]); i++) {
686
for (j = 1; a[j++]; j += 2) // free malloced attribute values
687
if (a[j] && (a[j] < root->s || a[j] > root->e)) free(a[j]);
688
free(a);
689
}
690
if (root->attr[0]) free(root->attr); // free default attribute list
691
692
for (i = 0; root->pi[i]; i++) {
693
for (j = 1; root->pi[i][j]; j++);
694
free(root->pi[i][j + 1]);
695
free(root->pi[i]);
696
}
697
if (root->pi[0]) free(root->pi); // free processing instructions
698
699
if (root->len == -1) free(root->m); // malloced xml data
700
if (root->u) free(root->u); // utf8 conversion
701
}
702
703
ezxml_free_attr(xml->attr); // tag attributes
704
if ((xml->flags & EZXML_TXTM)) free(xml->txt); // character content
705
if ((xml->flags & EZXML_NAMEM)) free(xml->name); // tag name
706
free(xml);
707
}
708
709
// return parser error message or empty string if none
710
const char *ezxml_error(ezxml_t xml)
711
{
712
while (xml && xml->parent) xml = xml->parent; // find root tag
713
return (xml) ? ((ezxml_root_t)xml)->err : "";
714
}
715
Loggerhead is a web-based interface for Breezy
Version: 2.0.1