1 |
greg |
2.1 |
#ifndef lint |
2 |
greg |
2.3 |
static const char RCSid[] = "$Id: ezxml.c,v 2.2 2011/06/01 00:29:40 greg Exp $"; |
3 |
greg |
2.1 |
#endif |
4 |
|
|
/* ezxml.c |
5 |
|
|
* |
6 |
|
|
* Copyright 2004-2006 Aaron Voisine <[email protected]> |
7 |
|
|
* |
8 |
|
|
* Permission is hereby granted, free of charge, to any person obtaining |
9 |
|
|
* a copy of this software and associated documentation files (the |
10 |
|
|
* "Software"), to deal in the Software without restriction, including |
11 |
|
|
* without limitation the rights to use, copy, modify, merge, publish, |
12 |
|
|
* distribute, sublicense, and/or sell copies of the Software, and to |
13 |
|
|
* permit persons to whom the Software is furnished to do so, subject to |
14 |
|
|
* the following conditions: |
15 |
|
|
* |
16 |
|
|
* The above copyright notice and this permission notice shall be included |
17 |
|
|
* in all copies or substantial portions of the Software. |
18 |
|
|
* |
19 |
|
|
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, |
20 |
|
|
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF |
21 |
|
|
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. |
22 |
|
|
* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY |
23 |
|
|
* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, |
24 |
|
|
* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE |
25 |
|
|
* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. |
26 |
|
|
*/ |
27 |
|
|
|
28 |
|
|
#include <stdlib.h> |
29 |
|
|
#include <stdio.h> |
30 |
|
|
#include <stdarg.h> |
31 |
|
|
#include <string.h> |
32 |
|
|
#include <ctype.h> |
33 |
|
|
#include <sys/types.h> |
34 |
|
|
#ifndef EZXML_NOMMAP |
35 |
|
|
#include <unistd.h> |
36 |
|
|
#include <sys/mman.h> |
37 |
greg |
2.2 |
#endif /* EZXML_NOMMAP */ |
38 |
greg |
2.1 |
#include <sys/stat.h> |
39 |
|
|
#include "ezxml.h" |
40 |
|
|
|
41 |
greg |
2.3 |
#ifdef _WIN32 |
42 |
|
|
#include <io.h> |
43 |
|
|
#define snprintf printf_s |
44 |
|
|
#define read _read |
45 |
|
|
#define open _open |
46 |
|
|
#define close _close |
47 |
|
|
#endif |
48 |
|
|
|
49 |
greg |
2.2 |
#define EZXML_WS "\t\r\n " /* whitespace */ |
50 |
|
|
#define EZXML_ERRL 128 /* maximum error string length */ |
51 |
greg |
2.1 |
|
52 |
|
|
typedef struct ezxml_root *ezxml_root_t; |
53 |
greg |
2.2 |
struct ezxml_root { /* additional data for the root tag */ |
54 |
|
|
struct ezxml xml; /* is a super-struct built on top of ezxml struct */ |
55 |
|
|
ezxml_t cur; /* current xml tree insertion point */ |
56 |
|
|
char *m; /* original xml string */ |
57 |
|
|
size_t len; /* length of allocated memory for mmap, -1 for malloc */ |
58 |
|
|
char *u; /* UTF-8 conversion of string if original was UTF-16 */ |
59 |
|
|
char *s; /* start of work area */ |
60 |
|
|
char *e; /* end of work area */ |
61 |
|
|
char **ent; /* general entities (ampersand sequences) */ |
62 |
|
|
char ***attr; /* default attributes */ |
63 |
|
|
char ***pi; /* processing instructions */ |
64 |
|
|
short standalone; /* non-zero if <?xml standalone="yes"?> */ |
65 |
|
|
char err[EZXML_ERRL]; /* error string */ |
66 |
greg |
2.1 |
}; |
67 |
|
|
|
68 |
greg |
2.2 |
char *EZXML_NIL[] = { NULL }; /* empty, null terminated array of strings */ |
69 |
greg |
2.1 |
|
70 |
greg |
2.2 |
/* returns the first child tag with the given name or NULL if not found */ |
71 |
greg |
2.1 |
ezxml_t ezxml_child(ezxml_t xml, const char *name) |
72 |
|
|
{ |
73 |
|
|
xml = (xml) ? xml->child : NULL; |
74 |
|
|
while (xml && strcmp(name, xml->name)) xml = xml->sibling; |
75 |
|
|
return xml; |
76 |
|
|
} |
77 |
|
|
|
78 |
greg |
2.2 |
/* returns the Nth tag with the same name in the same subsection or NULL if not */ |
79 |
|
|
/* found */ |
80 |
greg |
2.1 |
ezxml_t ezxml_idx(ezxml_t xml, int idx) |
81 |
|
|
{ |
82 |
|
|
for (; xml && idx; idx--) xml = xml->next; |
83 |
|
|
return xml; |
84 |
|
|
} |
85 |
|
|
|
86 |
greg |
2.2 |
/* returns the value of the requested tag attribute or NULL if not found */ |
87 |
greg |
2.1 |
const char *ezxml_attr(ezxml_t xml, const char *attr) |
88 |
|
|
{ |
89 |
|
|
int i = 0, j = 1; |
90 |
|
|
ezxml_root_t root = (ezxml_root_t)xml; |
91 |
|
|
|
92 |
|
|
if (! xml || ! xml->attr) return NULL; |
93 |
|
|
while (xml->attr[i] && strcmp(attr, xml->attr[i])) i += 2; |
94 |
greg |
2.2 |
if (xml->attr[i]) return xml->attr[i + 1]; /* found attribute */ |
95 |
greg |
2.1 |
|
96 |
greg |
2.2 |
while (root->xml.parent) root = (ezxml_root_t)root->xml.parent; /* root tag */ |
97 |
greg |
2.1 |
for (i = 0; root->attr[i] && strcmp(xml->name, root->attr[i][0]); i++); |
98 |
greg |
2.2 |
if (! root->attr[i]) return NULL; /* no matching default attributes */ |
99 |
greg |
2.1 |
while (root->attr[i][j] && strcmp(attr, root->attr[i][j])) j += 3; |
100 |
greg |
2.2 |
return (root->attr[i][j]) ? root->attr[i][j + 1] : NULL; /* found default */ |
101 |
greg |
2.1 |
} |
102 |
|
|
|
103 |
greg |
2.2 |
/* same as ezxml_get but takes an already initialized va_list */ |
104 |
greg |
2.1 |
ezxml_t ezxml_vget(ezxml_t xml, va_list ap) |
105 |
|
|
{ |
106 |
|
|
char *name = va_arg(ap, char *); |
107 |
|
|
int idx = -1; |
108 |
|
|
|
109 |
|
|
if (name && *name) { |
110 |
|
|
idx = va_arg(ap, int); |
111 |
|
|
xml = ezxml_child(xml, name); |
112 |
|
|
} |
113 |
|
|
return (idx < 0) ? xml : ezxml_vget(ezxml_idx(xml, idx), ap); |
114 |
|
|
} |
115 |
|
|
|
116 |
greg |
2.2 |
/* Traverses the xml tree to retrieve a specific subtag. Takes a variable */ |
117 |
|
|
/* length list of tag names and indexes. The argument list must be terminated */ |
118 |
|
|
/* by either an index of -1 or an empty string tag name. Example: */ |
119 |
|
|
/* title = ezxml_get(library, "shelf", 0, "book", 2, "title", -1); */ |
120 |
|
|
/* This retrieves the title of the 3rd book on the 1st shelf of library. */ |
121 |
|
|
/* Returns NULL if not found. */ |
122 |
greg |
2.1 |
ezxml_t ezxml_get(ezxml_t xml, ...) |
123 |
|
|
{ |
124 |
|
|
va_list ap; |
125 |
|
|
ezxml_t r; |
126 |
|
|
|
127 |
|
|
va_start(ap, xml); |
128 |
|
|
r = ezxml_vget(xml, ap); |
129 |
|
|
va_end(ap); |
130 |
|
|
return r; |
131 |
|
|
} |
132 |
|
|
|
133 |
greg |
2.2 |
/* returns a null terminated array of processing instructions for the given */ |
134 |
|
|
/* target */ |
135 |
greg |
2.1 |
const char **ezxml_pi(ezxml_t xml, const char *target) |
136 |
|
|
{ |
137 |
|
|
ezxml_root_t root = (ezxml_root_t)xml; |
138 |
|
|
int i = 0; |
139 |
|
|
|
140 |
|
|
if (! root) return (const char **)EZXML_NIL; |
141 |
greg |
2.2 |
while (root->xml.parent) root = (ezxml_root_t)root->xml.parent; /* root tag */ |
142 |
|
|
while (root->pi[i] && strcmp(target, root->pi[i][0])) i++; /* find target */ |
143 |
greg |
2.1 |
return (const char **)((root->pi[i]) ? root->pi[i] + 1 : EZXML_NIL); |
144 |
|
|
} |
145 |
|
|
|
146 |
greg |
2.2 |
/* set an error string and return root */ |
147 |
greg |
2.1 |
ezxml_t ezxml_err(ezxml_root_t root, char *s, const char *err, ...) |
148 |
|
|
{ |
149 |
|
|
va_list ap; |
150 |
|
|
int line = 1; |
151 |
|
|
char *t, fmt[EZXML_ERRL]; |
152 |
|
|
|
153 |
|
|
for (t = root->s; t < s; t++) if (*t == '\n') line++; |
154 |
|
|
snprintf(fmt, EZXML_ERRL, "[error near line %d]: %s", line, err); |
155 |
|
|
|
156 |
|
|
va_start(ap, err); |
157 |
|
|
vsnprintf(root->err, EZXML_ERRL, fmt, ap); |
158 |
|
|
va_end(ap); |
159 |
|
|
|
160 |
|
|
return &root->xml; |
161 |
|
|
} |
162 |
|
|
|
163 |
greg |
2.2 |
/* Recursively decodes entity and character references and normalizes new lines */ |
164 |
|
|
/* ent is a null terminated array of alternating entity names and values. set t */ |
165 |
|
|
/* to '&' for general entity decoding, '%' for parameter entity decoding, 'c' */ |
166 |
|
|
/* for cdata sections, ' ' for attribute normalization, or '*' for non-cdata */ |
167 |
|
|
/* attribute normalization. Returns s, or if the decoded string is longer than */ |
168 |
|
|
/* s, returns a malloced string that must be freed. */ |
169 |
greg |
2.1 |
char *ezxml_decode(char *s, char **ent, char t) |
170 |
|
|
{ |
171 |
|
|
char *e, *r = s, *m = s; |
172 |
|
|
long b, c, d, l; |
173 |
|
|
|
174 |
greg |
2.2 |
for (; *s; s++) { /* normalize line endings */ |
175 |
greg |
2.1 |
while (*s == '\r') { |
176 |
|
|
*(s++) = '\n'; |
177 |
|
|
if (*s == '\n') memmove(s, (s + 1), strlen(s)); |
178 |
|
|
} |
179 |
|
|
} |
180 |
|
|
|
181 |
|
|
for (s = r; ; ) { |
182 |
|
|
while (*s && *s != '&' && (*s != '%' || t != '%') && !isspace(*s)) s++; |
183 |
|
|
|
184 |
|
|
if (! *s) break; |
185 |
greg |
2.2 |
else if (t != 'c' && ! strncmp(s, "&#", 2)) { /* character reference */ |
186 |
|
|
if (s[2] == 'x') c = strtol(s + 3, &e, 16); /* base 16 */ |
187 |
|
|
else c = strtol(s + 2, &e, 10); /* base 10 */ |
188 |
|
|
if (! c || *e != ';') { s++; continue; } /* not a character ref */ |
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 |
greg |
2.1 |
} |
197 |
|
|
|
198 |
|
|
memmove(s, strchr(s, ';') + 1, strlen(strchr(s, ';'))); |
199 |
|
|
} |
200 |
|
|
else if ((*s == '&' && (t == '&' || t == ' ' || t == '*')) || |
201 |
greg |
2.2 |
(*s == '%' && t == '%')) { /* entity reference */ |
202 |
greg |
2.1 |
for (b = 0; ent[b] && strncmp(s + 1, ent[b], strlen(ent[b])); |
203 |
greg |
2.2 |
b += 2); /* find entity in entity list */ |
204 |
greg |
2.1 |
|
205 |
greg |
2.2 |
if (ent[b++]) { /* found a match */ |
206 |
greg |
2.1 |
if ((c = strlen(ent[b])) - 1 > (e = strchr(s, ';')) - s) { |
207 |
greg |
2.2 |
l = (d = (s - r)) + c + strlen(e); /* new length */ |
208 |
greg |
2.1 |
r = (r == m) ? strcpy(malloc(l), r) : realloc(r, l); |
209 |
greg |
2.2 |
e = strchr((s = r + d), ';'); /* fix up pointers */ |
210 |
greg |
2.1 |
} |
211 |
|
|
|
212 |
greg |
2.2 |
memmove(s + c, e + 1, strlen(e)); /* shift rest of string */ |
213 |
|
|
strncpy(s, ent[b], c); /* copy in replacement text */ |
214 |
greg |
2.1 |
} |
215 |
greg |
2.2 |
else s++; /* not a known entity */ |
216 |
greg |
2.1 |
} |
217 |
|
|
else if ((t == ' ' || t == '*') && isspace(*s)) *(s++) = ' '; |
218 |
greg |
2.2 |
else s++; /* no decoding needed */ |
219 |
greg |
2.1 |
} |
220 |
|
|
|
221 |
greg |
2.2 |
if (t == '*') { /* normalize spaces for non-cdata attributes */ |
222 |
greg |
2.1 |
for (s = r; *s; s++) { |
223 |
|
|
if ((l = strspn(s, " "))) memmove(s, s + l, strlen(s + l) + 1); |
224 |
|
|
while (*s && *s != ' ') s++; |
225 |
|
|
} |
226 |
greg |
2.2 |
if (--s >= r && *s == ' ') *s = '\0'; /* trim any trailing space */ |
227 |
greg |
2.1 |
} |
228 |
|
|
return r; |
229 |
|
|
} |
230 |
|
|
|
231 |
greg |
2.2 |
/* called when parser finds start of new tag */ |
232 |
greg |
2.1 |
void ezxml_open_tag(ezxml_root_t root, char *name, char **attr) |
233 |
|
|
{ |
234 |
|
|
ezxml_t xml = root->cur; |
235 |
|
|
|
236 |
|
|
if (xml->name) xml = ezxml_add_child(xml, name, strlen(xml->txt)); |
237 |
greg |
2.2 |
else xml->name = name; /* first open tag */ |
238 |
greg |
2.1 |
|
239 |
|
|
xml->attr = attr; |
240 |
greg |
2.2 |
root->cur = xml; /* update tag insertion point */ |
241 |
greg |
2.1 |
} |
242 |
|
|
|
243 |
greg |
2.2 |
/* called when parser finds character content between open and closing tag */ |
244 |
greg |
2.1 |
void ezxml_char_content(ezxml_root_t root, char *s, size_t len, char t) |
245 |
|
|
{ |
246 |
|
|
ezxml_t xml = root->cur; |
247 |
|
|
char *m = s; |
248 |
|
|
size_t l; |
249 |
|
|
|
250 |
greg |
2.2 |
if (! xml || ! xml->name || ! len) return; /* sanity check */ |
251 |
greg |
2.1 |
|
252 |
greg |
2.2 |
s[len] = '\0'; /* null terminate text (calling functions anticipate this) */ |
253 |
greg |
2.1 |
len = strlen(s = ezxml_decode(s, root->ent, t)) + 1; |
254 |
|
|
|
255 |
greg |
2.2 |
if (! *(xml->txt)) xml->txt = s; /* initial character content */ |
256 |
|
|
else { /* allocate our own memory and make a copy */ |
257 |
|
|
xml->txt = (xml->flags & EZXML_TXTM) /* allocate some space */ |
258 |
greg |
2.1 |
? realloc(xml->txt, (l = strlen(xml->txt)) + len) |
259 |
|
|
: strcpy(malloc((l = strlen(xml->txt)) + len), xml->txt); |
260 |
greg |
2.2 |
strcpy(xml->txt + l, s); /* add new char content */ |
261 |
|
|
if (s != m) free(s); /* free s if it was malloced by ezxml_decode() */ |
262 |
greg |
2.1 |
} |
263 |
|
|
|
264 |
|
|
if (xml->txt != m) ezxml_set_flag(xml, EZXML_TXTM); |
265 |
|
|
} |
266 |
|
|
|
267 |
greg |
2.2 |
/* called when parser finds closing tag */ |
268 |
greg |
2.1 |
ezxml_t ezxml_close_tag(ezxml_root_t root, char *name, char *s) |
269 |
|
|
{ |
270 |
|
|
if (! root->cur || ! root->cur->name || strcmp(name, root->cur->name)) |
271 |
|
|
return ezxml_err(root, s, "unexpected closing tag </%s>", name); |
272 |
|
|
|
273 |
|
|
root->cur = root->cur->parent; |
274 |
|
|
return NULL; |
275 |
|
|
} |
276 |
|
|
|
277 |
greg |
2.2 |
/* checks for circular entity references, returns non-zero if no circular */ |
278 |
|
|
/* references are found, zero otherwise */ |
279 |
greg |
2.1 |
int ezxml_ent_ok(char *name, char *s, char **ent) |
280 |
|
|
{ |
281 |
|
|
int i; |
282 |
|
|
|
283 |
|
|
for (; ; s++) { |
284 |
greg |
2.2 |
while (*s && *s != '&') s++; /* find next entity reference */ |
285 |
greg |
2.1 |
if (! *s) return 1; |
286 |
greg |
2.2 |
if (! strncmp(s + 1, name, strlen(name))) return 0; /* circular ref. */ |
287 |
greg |
2.1 |
for (i = 0; ent[i] && strncmp(ent[i], s + 1, strlen(ent[i])); i += 2); |
288 |
|
|
if (ent[i] && ! ezxml_ent_ok(name, ent[i + 1], ent)) return 0; |
289 |
|
|
} |
290 |
|
|
} |
291 |
|
|
|
292 |
greg |
2.2 |
/* called when the parser finds a processing instruction */ |
293 |
greg |
2.1 |
void ezxml_proc_inst(ezxml_root_t root, char *s, size_t len) |
294 |
|
|
{ |
295 |
|
|
int i = 0, j = 1; |
296 |
|
|
char *target = s; |
297 |
|
|
|
298 |
greg |
2.2 |
s[len] = '\0'; /* null terminate instruction */ |
299 |
greg |
2.1 |
if (*(s += strcspn(s, EZXML_WS))) { |
300 |
greg |
2.2 |
*s = '\0'; /* null terminate target */ |
301 |
|
|
s += strspn(s + 1, EZXML_WS) + 1; /* skip whitespace after target */ |
302 |
greg |
2.1 |
} |
303 |
|
|
|
304 |
greg |
2.2 |
if (! strcmp(target, "xml")) { /* <?xml ... ?> */ |
305 |
greg |
2.1 |
if ((s = strstr(s, "standalone")) && ! strncmp(s + strspn(s + 10, |
306 |
|
|
EZXML_WS "='\"") + 10, "yes", 3)) root->standalone = 1; |
307 |
|
|
return; |
308 |
|
|
} |
309 |
|
|
|
310 |
greg |
2.2 |
if (! root->pi[0]) *(root->pi = malloc(sizeof(char **))) = NULL; /*first pi */ |
311 |
greg |
2.1 |
|
312 |
greg |
2.2 |
while (root->pi[i] && strcmp(target, root->pi[i][0])) i++; /* find target */ |
313 |
|
|
if (! root->pi[i]) { /* new target */ |
314 |
greg |
2.1 |
root->pi = realloc(root->pi, sizeof(char **) * (i + 2)); |
315 |
|
|
root->pi[i] = malloc(sizeof(char *) * 3); |
316 |
|
|
root->pi[i][0] = target; |
317 |
greg |
2.2 |
root->pi[i][1] = (char *)(root->pi[i + 1] = NULL); /* terminate pi list */ |
318 |
|
|
root->pi[i][2] = strdup(""); /* empty document position list */ |
319 |
greg |
2.1 |
} |
320 |
|
|
|
321 |
greg |
2.2 |
while (root->pi[i][j]) j++; /* find end of instruction list for this target */ |
322 |
greg |
2.1 |
root->pi[i] = realloc(root->pi[i], sizeof(char *) * (j + 3)); |
323 |
|
|
root->pi[i][j + 2] = realloc(root->pi[i][j + 1], j + 1); |
324 |
|
|
strcpy(root->pi[i][j + 2] + j - 1, (root->xml.name) ? ">" : "<"); |
325 |
greg |
2.2 |
root->pi[i][j + 1] = NULL; /* null terminate pi list for this target */ |
326 |
|
|
root->pi[i][j] = s; /* set instruction */ |
327 |
greg |
2.1 |
} |
328 |
|
|
|
329 |
greg |
2.2 |
/* called when the parser finds an internal doctype subset */ |
330 |
greg |
2.1 |
short ezxml_internal_dtd(ezxml_root_t root, char *s, size_t len) |
331 |
|
|
{ |
332 |
|
|
char q, *c, *t, *n = NULL, *v, **ent, **pe; |
333 |
|
|
int i, j; |
334 |
|
|
|
335 |
|
|
pe = memcpy(malloc(sizeof(EZXML_NIL)), EZXML_NIL, sizeof(EZXML_NIL)); |
336 |
|
|
|
337 |
|
|
for (s[len] = '\0'; s; ) { |
338 |
greg |
2.2 |
while (*s && *s != '<' && *s != '%') s++; /* find next declaration */ |
339 |
greg |
2.1 |
|
340 |
|
|
if (! *s) break; |
341 |
greg |
2.2 |
else if (! strncmp(s, "<!ENTITY", 8)) { /* parse entity definitions */ |
342 |
|
|
c = s += strspn(s + 8, EZXML_WS) + 8; /* skip white space separator */ |
343 |
|
|
n = s + strspn(s, EZXML_WS "%"); /* find name */ |
344 |
|
|
*(s = n + strcspn(n, EZXML_WS)) = ';'; /* append ; to name */ |
345 |
greg |
2.1 |
|
346 |
greg |
2.2 |
v = s + strspn(s + 1, EZXML_WS) + 1; /* find value */ |
347 |
|
|
if ((q = *(v++)) != '"' && q != '\'') { /* skip externals */ |
348 |
greg |
2.1 |
s = strchr(s, '>'); |
349 |
|
|
continue; |
350 |
|
|
} |
351 |
|
|
|
352 |
|
|
for (i = 0, ent = (*c == '%') ? pe : root->ent; ent[i]; i++); |
353 |
greg |
2.2 |
ent = realloc(ent, (i + 3) * sizeof(char *)); /* space for next ent */ |
354 |
greg |
2.1 |
if (*c == '%') pe = ent; |
355 |
|
|
else root->ent = ent; |
356 |
|
|
|
357 |
greg |
2.2 |
*(++s) = '\0'; /* null terminate name */ |
358 |
|
|
if ((s = strchr(v, q))) *(s++) = '\0'; /* null terminate value */ |
359 |
|
|
ent[i + 1] = ezxml_decode(v, pe, '%'); /* set value */ |
360 |
|
|
ent[i + 2] = NULL; /* null terminate entity list */ |
361 |
|
|
if (! ezxml_ent_ok(n, ent[i + 1], ent)) { /* circular reference */ |
362 |
greg |
2.1 |
if (ent[i + 1] != v) free(ent[i + 1]); |
363 |
|
|
ezxml_err(root, v, "circular entity declaration &%s", n); |
364 |
|
|
break; |
365 |
|
|
} |
366 |
greg |
2.2 |
else ent[i] = n; /* set entity name */ |
367 |
greg |
2.1 |
} |
368 |
greg |
2.2 |
else if (! strncmp(s, "<!ATTLIST", 9)) { /* parse default attributes */ |
369 |
|
|
t = s + strspn(s + 9, EZXML_WS) + 9; /* skip whitespace separator */ |
370 |
greg |
2.1 |
if (! *t) { ezxml_err(root, t, "unclosed <!ATTLIST"); break; } |
371 |
|
|
if (*(s = t + strcspn(t, EZXML_WS ">")) == '>') continue; |
372 |
greg |
2.2 |
else *s = '\0'; /* null terminate tag name */ |
373 |
greg |
2.1 |
for (i = 0; root->attr[i] && strcmp(n, root->attr[i][0]); i++); |
374 |
|
|
|
375 |
|
|
while (*(n = ++s + strspn(s, EZXML_WS)) && *n != '>') { |
376 |
greg |
2.2 |
if (*(s = n + strcspn(n, EZXML_WS))) *s = '\0'; /* attr name */ |
377 |
greg |
2.1 |
else { ezxml_err(root, t, "malformed <!ATTLIST"); break; } |
378 |
|
|
|
379 |
greg |
2.2 |
s += strspn(s + 1, EZXML_WS) + 1; /* find next token */ |
380 |
|
|
c = (strncmp(s, "CDATA", 5)) ? "*" : " "; /* is it cdata? */ |
381 |
greg |
2.1 |
if (! strncmp(s, "NOTATION", 8)) |
382 |
|
|
s += strspn(s + 8, EZXML_WS) + 8; |
383 |
|
|
s = (*s == '(') ? strchr(s, ')') : s + strcspn(s, EZXML_WS); |
384 |
|
|
if (! s) { ezxml_err(root, t, "malformed <!ATTLIST"); break; } |
385 |
|
|
|
386 |
greg |
2.2 |
s += strspn(s, EZXML_WS ")"); /* skip white space separator */ |
387 |
greg |
2.1 |
if (! strncmp(s, "#FIXED", 6)) |
388 |
|
|
s += strspn(s + 6, EZXML_WS) + 6; |
389 |
greg |
2.2 |
if (*s == '#') { /* no default value */ |
390 |
greg |
2.1 |
s += strcspn(s, EZXML_WS ">") - 1; |
391 |
greg |
2.2 |
if (*c == ' ') continue; /* cdata is default, nothing to do */ |
392 |
greg |
2.1 |
v = NULL; |
393 |
|
|
} |
394 |
greg |
2.2 |
else if ((*s == '"' || *s == '\'') && /* default value */ |
395 |
greg |
2.1 |
(s = strchr(v = s + 1, *s))) *s = '\0'; |
396 |
|
|
else { ezxml_err(root, t, "malformed <!ATTLIST"); break; } |
397 |
|
|
|
398 |
greg |
2.2 |
if (! root->attr[i]) { /* new tag name */ |
399 |
greg |
2.1 |
root->attr = (! i) ? malloc(2 * sizeof(char **)) |
400 |
|
|
: realloc(root->attr, |
401 |
|
|
(i + 2) * sizeof(char **)); |
402 |
|
|
root->attr[i] = malloc(2 * sizeof(char *)); |
403 |
greg |
2.2 |
root->attr[i][0] = t; /* set tag name */ |
404 |
greg |
2.1 |
root->attr[i][1] = (char *)(root->attr[i + 1] = NULL); |
405 |
|
|
} |
406 |
|
|
|
407 |
greg |
2.2 |
for (j = 1; root->attr[i][j]; j += 3); /* find end of list */ |
408 |
greg |
2.1 |
root->attr[i] = realloc(root->attr[i], |
409 |
|
|
(j + 4) * sizeof(char *)); |
410 |
|
|
|
411 |
greg |
2.2 |
root->attr[i][j + 3] = NULL; /* null terminate list */ |
412 |
|
|
root->attr[i][j + 2] = c; /* is it cdata? */ |
413 |
greg |
2.1 |
root->attr[i][j + 1] = (v) ? ezxml_decode(v, root->ent, *c) |
414 |
|
|
: NULL; |
415 |
greg |
2.2 |
root->attr[i][j] = n; /* attribute name */ |
416 |
greg |
2.1 |
} |
417 |
|
|
} |
418 |
greg |
2.2 |
else if (! strncmp(s, "<!--", 4)) s = strstr(s + 4, "-->"); /* comments */ |
419 |
|
|
else if (! strncmp(s, "<?", 2)) { /* processing instructions */ |
420 |
greg |
2.1 |
if ((s = strstr(c = s + 2, "?>"))) |
421 |
|
|
ezxml_proc_inst(root, c, s++ - c); |
422 |
|
|
} |
423 |
greg |
2.2 |
else if (*s == '<') s = strchr(s, '>'); /* skip other declarations */ |
424 |
greg |
2.1 |
else if (*(s++) == '%' && ! root->standalone) break; |
425 |
|
|
} |
426 |
|
|
|
427 |
|
|
free(pe); |
428 |
|
|
return ! *root->err; |
429 |
|
|
} |
430 |
|
|
|
431 |
greg |
2.2 |
/* Converts a UTF-16 string to UTF-8. Returns a new string that must be freed */ |
432 |
|
|
/* or NULL if no conversion was needed. */ |
433 |
greg |
2.1 |
char *ezxml_str2utf8(char **s, size_t *len) |
434 |
|
|
{ |
435 |
|
|
char *u; |
436 |
|
|
size_t l = 0, sl, max = *len; |
437 |
|
|
long c, d; |
438 |
|
|
int b, be = (**s == '\xFE') ? 1 : (**s == '\xFF') ? 0 : -1; |
439 |
|
|
|
440 |
greg |
2.2 |
if (be == -1) return NULL; /* not UTF-16 */ |
441 |
greg |
2.1 |
|
442 |
|
|
u = malloc(max); |
443 |
|
|
for (sl = 2; sl < *len - 1; sl += 2) { |
444 |
greg |
2.2 |
c = (be) ? (((*s)[sl] & 0xFF) << 8) | ((*s)[sl + 1] & 0xFF) /*UTF-16BE */ |
445 |
|
|
: (((*s)[sl + 1] & 0xFF) << 8) | ((*s)[sl] & 0xFF); /*UTF-16LE */ |
446 |
|
|
if (c >= 0xD800 && c <= 0xDFFF && (sl += 2) < *len - 1) { /* high-half */ |
447 |
greg |
2.1 |
d = (be) ? (((*s)[sl] & 0xFF) << 8) | ((*s)[sl + 1] & 0xFF) |
448 |
|
|
: (((*s)[sl + 1] & 0xFF) << 8) | ((*s)[sl] & 0xFF); |
449 |
|
|
c = (((c & 0x3FF) << 10) | (d & 0x3FF)) + 0x10000; |
450 |
|
|
} |
451 |
|
|
|
452 |
|
|
while (l + 6 > max) u = realloc(u, max += EZXML_BUFSIZE); |
453 |
greg |
2.2 |
if (c < 0x80) u[l++] = c; /* US-ASCII subset */ |
454 |
|
|
else { /* multi-byte UTF-8 sequence */ |
455 |
|
|
for (b = 0, d = c; d; d /= 2) b++; /* bits in c */ |
456 |
|
|
b = (b - 2) / 5; /* bytes in payload */ |
457 |
|
|
u[l++] = (0xFF << (7 - b)) | (c >> (6 * b)); /* head */ |
458 |
|
|
while (b) u[l++] = 0x80 | ((c >> (6 * --b)) & 0x3F); /* payload */ |
459 |
greg |
2.1 |
} |
460 |
|
|
} |
461 |
|
|
return *s = realloc(u, *len = l); |
462 |
|
|
} |
463 |
|
|
|
464 |
greg |
2.2 |
/* frees a tag attribute list */ |
465 |
greg |
2.1 |
void ezxml_free_attr(char **attr) { |
466 |
|
|
int i = 0; |
467 |
|
|
char *m; |
468 |
|
|
|
469 |
greg |
2.2 |
if (! attr || attr == EZXML_NIL) return; /* nothing to free */ |
470 |
|
|
while (attr[i]) i += 2; /* find end of attribute list */ |
471 |
|
|
m = attr[i + 1]; /* list of which names and values are malloced */ |
472 |
greg |
2.1 |
for (i = 0; m[i]; i++) { |
473 |
|
|
if (m[i] & EZXML_NAMEM) free(attr[i * 2]); |
474 |
|
|
if (m[i] & EZXML_TXTM) free(attr[(i * 2) + 1]); |
475 |
|
|
} |
476 |
|
|
free(m); |
477 |
|
|
free(attr); |
478 |
|
|
} |
479 |
|
|
|
480 |
greg |
2.2 |
/* parse the given xml string and return an ezxml structure */ |
481 |
greg |
2.1 |
ezxml_t ezxml_parse_str(char *s, size_t len) |
482 |
|
|
{ |
483 |
|
|
ezxml_root_t root = (ezxml_root_t)ezxml_new(NULL); |
484 |
greg |
2.2 |
char q, e, *d, **attr, **a = NULL; /* initialize a to avoid compile warning */ |
485 |
greg |
2.1 |
int l, i, j; |
486 |
|
|
|
487 |
|
|
root->m = s; |
488 |
|
|
if (! len) return ezxml_err(root, NULL, "root tag missing"); |
489 |
greg |
2.2 |
root->u = ezxml_str2utf8(&s, &len); /* convert utf-16 to utf-8 */ |
490 |
|
|
root->e = (root->s = s) + len; /* record start and end of work area */ |
491 |
greg |
2.1 |
|
492 |
greg |
2.2 |
e = s[len - 1]; /* save end char */ |
493 |
|
|
s[len - 1] = '\0'; /* turn end char into null terminator */ |
494 |
greg |
2.1 |
|
495 |
greg |
2.2 |
while (*s && *s != '<') s++; /* find first tag */ |
496 |
greg |
2.1 |
if (! *s) return ezxml_err(root, s, "root tag missing"); |
497 |
|
|
|
498 |
|
|
for (; ; ) { |
499 |
|
|
attr = (char **)EZXML_NIL; |
500 |
|
|
d = ++s; |
501 |
|
|
|
502 |
greg |
2.2 |
if (isalpha(*s) || *s == '_' || *s == ':' || *s < '\0') { /* new tag */ |
503 |
greg |
2.1 |
if (! root->cur) |
504 |
|
|
return ezxml_err(root, d, "markup outside of root element"); |
505 |
|
|
|
506 |
|
|
s += strcspn(s, EZXML_WS "/>"); |
507 |
greg |
2.2 |
while (isspace(*s)) *(s++) = '\0'; /* null terminate tag name */ |
508 |
greg |
2.1 |
|
509 |
greg |
2.2 |
if (*s && *s != '/' && *s != '>') /* find tag in default attr list */ |
510 |
greg |
2.1 |
for (i = 0; (a = root->attr[i]) && strcmp(a[0], d); i++); |
511 |
|
|
|
512 |
greg |
2.2 |
for (l = 0; *s && *s != '/' && *s != '>'; l += 2) { /* new attrib */ |
513 |
greg |
2.1 |
attr = (l) ? realloc(attr, (l + 4) * sizeof(char *)) |
514 |
greg |
2.2 |
: malloc(4 * sizeof(char *)); /* allocate space */ |
515 |
greg |
2.1 |
attr[l + 3] = (l) ? realloc(attr[l + 1], (l / 2) + 2) |
516 |
greg |
2.2 |
: malloc(2); /* mem for list of maloced vals */ |
517 |
|
|
strcpy(attr[l + 3] + (l / 2), " "); /* value is not malloced */ |
518 |
|
|
attr[l + 2] = NULL; /* null terminate list */ |
519 |
|
|
attr[l + 1] = ""; /* temporary attribute value */ |
520 |
|
|
attr[l] = s; /* set attribute name */ |
521 |
greg |
2.1 |
|
522 |
|
|
s += strcspn(s, EZXML_WS "=/>"); |
523 |
|
|
if (*s == '=' || isspace(*s)) { |
524 |
greg |
2.2 |
*(s++) = '\0'; /* null terminate tag attribute name */ |
525 |
greg |
2.1 |
q = *(s += strspn(s, EZXML_WS "=")); |
526 |
greg |
2.2 |
if (q == '"' || q == '\'') { /* attribute value */ |
527 |
greg |
2.1 |
attr[l + 1] = ++s; |
528 |
|
|
while (*s && *s != q) s++; |
529 |
greg |
2.2 |
if (*s) *(s++) = '\0'; /* null terminate attribute val */ |
530 |
greg |
2.1 |
else { |
531 |
|
|
ezxml_free_attr(attr); |
532 |
|
|
return ezxml_err(root, d, "missing %c", q); |
533 |
|
|
} |
534 |
|
|
|
535 |
|
|
for (j = 1; a && a[j] && strcmp(a[j], attr[l]); j +=3); |
536 |
|
|
attr[l + 1] = ezxml_decode(attr[l + 1], root->ent, (a |
537 |
|
|
&& a[j]) ? *a[j + 2] : ' '); |
538 |
|
|
if (attr[l + 1] < d || attr[l + 1] > s) |
539 |
greg |
2.2 |
attr[l + 3][l / 2] = EZXML_TXTM; /* value malloced */ |
540 |
greg |
2.1 |
} |
541 |
|
|
} |
542 |
|
|
while (isspace(*s)) s++; |
543 |
|
|
} |
544 |
|
|
|
545 |
greg |
2.2 |
if (*s == '/') { /* self closing tag */ |
546 |
greg |
2.1 |
*(s++) = '\0'; |
547 |
|
|
if ((*s && *s != '>') || (! *s && e != '>')) { |
548 |
|
|
if (l) ezxml_free_attr(attr); |
549 |
|
|
return ezxml_err(root, d, "missing >"); |
550 |
|
|
} |
551 |
|
|
ezxml_open_tag(root, d, attr); |
552 |
|
|
ezxml_close_tag(root, d, s); |
553 |
|
|
} |
554 |
greg |
2.2 |
else if ((q = *s) == '>' || (! *s && e == '>')) { /* open tag */ |
555 |
|
|
*s = '\0'; /* temporarily null terminate tag name */ |
556 |
greg |
2.1 |
ezxml_open_tag(root, d, attr); |
557 |
|
|
*s = q; |
558 |
|
|
} |
559 |
|
|
else { |
560 |
|
|
if (l) ezxml_free_attr(attr); |
561 |
|
|
return ezxml_err(root, d, "missing >"); |
562 |
|
|
} |
563 |
|
|
} |
564 |
greg |
2.2 |
else if (*s == '/') { /* close tag */ |
565 |
greg |
2.1 |
s += strcspn(d = s + 1, EZXML_WS ">") + 1; |
566 |
|
|
if (! (q = *s) && e != '>') return ezxml_err(root, d, "missing >"); |
567 |
greg |
2.2 |
*s = '\0'; /* temporarily null terminate tag name */ |
568 |
greg |
2.1 |
if (ezxml_close_tag(root, d, s)) return &root->xml; |
569 |
|
|
if (isspace(*s = q)) s += strspn(s, EZXML_WS); |
570 |
|
|
} |
571 |
greg |
2.2 |
else if (! strncmp(s, "!--", 3)) { /* xml comment */ |
572 |
greg |
2.1 |
if (! (s = strstr(s + 3, "--")) || (*(s += 2) != '>' && *s) || |
573 |
|
|
(! *s && e != '>')) return ezxml_err(root, d, "unclosed <!--"); |
574 |
|
|
} |
575 |
greg |
2.2 |
else if (! strncmp(s, "![CDATA[", 8)) { /* cdata */ |
576 |
greg |
2.1 |
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 |
|
|
} |
580 |
greg |
2.2 |
else if (! strncmp(s, "!DOCTYPE", 8)) { /* dtd */ |
581 |
greg |
2.1 |
for (l = 0; *s && ((! l && *s != '>') || (l && (*s != ']' || |
582 |
|
|
*(s + strspn(s + 1, EZXML_WS) + 1) != '>'))); |
583 |
|
|
l = (*s == '[') ? 1 : l) s += strcspn(s + 1, "[]>") + 1; |
584 |
|
|
if (! *s && e != '>') |
585 |
|
|
return ezxml_err(root, d, "unclosed <!DOCTYPE"); |
586 |
|
|
d = (l) ? strchr(d, '[') + 1 : d; |
587 |
|
|
if (l && ! ezxml_internal_dtd(root, d, s++ - d)) return &root->xml; |
588 |
|
|
} |
589 |
greg |
2.2 |
else if (*s == '?') { /* <?...?> processing instructions */ |
590 |
greg |
2.1 |
do { s = strchr(s, '?'); } 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 |
|
|
} |
595 |
|
|
else return ezxml_err(root, d, "unexpected <"); |
596 |
|
|
|
597 |
|
|
if (! s || ! *s) break; |
598 |
|
|
*s = '\0'; |
599 |
|
|
d = ++s; |
600 |
greg |
2.2 |
if (*s && *s != '<') { /* tag character content */ |
601 |
greg |
2.1 |
while (*s && *s != '<') s++; |
602 |
|
|
if (*s) ezxml_char_content(root, d, s - d, '&'); |
603 |
|
|
else break; |
604 |
|
|
} |
605 |
|
|
else if (! *s) break; |
606 |
|
|
} |
607 |
|
|
|
608 |
|
|
if (! root->cur) return &root->xml; |
609 |
|
|
else if (! root->cur->name) return ezxml_err(root, d, "root tag missing"); |
610 |
|
|
else return ezxml_err(root, d, "unclosed tag <%s>", root->cur->name); |
611 |
|
|
} |
612 |
|
|
|
613 |
greg |
2.2 |
/* Wrapper for ezxml_parse_str() that accepts a file stream. Reads the entire */ |
614 |
|
|
/* stream into memory and then parses it. For xml files, use ezxml_parse_file() */ |
615 |
|
|
/* or ezxml_parse_fd() */ |
616 |
greg |
2.1 |
ezxml_t ezxml_parse_fp(FILE *fp) |
617 |
|
|
{ |
618 |
|
|
ezxml_root_t root; |
619 |
|
|
size_t l, len = 0; |
620 |
|
|
char *s; |
621 |
|
|
|
622 |
|
|
if (! (s = malloc(EZXML_BUFSIZE))) return NULL; |
623 |
|
|
do { |
624 |
|
|
len += (l = fread((s + len), 1, EZXML_BUFSIZE, fp)); |
625 |
|
|
if (l == EZXML_BUFSIZE) s = realloc(s, len + EZXML_BUFSIZE); |
626 |
|
|
} while (s && l == EZXML_BUFSIZE); |
627 |
|
|
|
628 |
|
|
if (! s) return NULL; |
629 |
|
|
root = (ezxml_root_t)ezxml_parse_str(s, len); |
630 |
greg |
2.2 |
root->len = -1; /* so we know to free s in ezxml_free() */ |
631 |
greg |
2.1 |
return &root->xml; |
632 |
|
|
} |
633 |
|
|
|
634 |
greg |
2.2 |
/* A wrapper for ezxml_parse_str() that accepts a file descriptor. First */ |
635 |
|
|
/* attempts to mem map the file. Failing that, reads the file into memory. */ |
636 |
|
|
/* Returns NULL on failure. */ |
637 |
greg |
2.1 |
ezxml_t ezxml_parse_fd(int fd) |
638 |
|
|
{ |
639 |
|
|
ezxml_root_t root; |
640 |
|
|
struct stat st; |
641 |
|
|
size_t l; |
642 |
|
|
void *m; |
643 |
|
|
|
644 |
|
|
if (fd < 0) return NULL; |
645 |
|
|
fstat(fd, &st); |
646 |
|
|
|
647 |
|
|
#ifndef EZXML_NOMMAP |
648 |
|
|
l = (st.st_size + sysconf(_SC_PAGESIZE) - 1) & ~(sysconf(_SC_PAGESIZE) -1); |
649 |
|
|
if ((m = mmap(NULL, l, PROT_READ|PROT_WRITE, MAP_PRIVATE, fd, 0)) |
650 |
|
|
!= MAP_FAILED) { |
651 |
greg |
2.2 |
madvise(m, l, MADV_SEQUENTIAL); /* optimize for sequential access */ |
652 |
greg |
2.1 |
root = (ezxml_root_t)ezxml_parse_str(m, st.st_size); |
653 |
greg |
2.2 |
madvise(m, root->len = l, MADV_NORMAL); /* put it back to normal */ |
654 |
greg |
2.1 |
} |
655 |
greg |
2.2 |
else { /* mmap failed, read file into memory */ |
656 |
|
|
#endif /* EZXML_NOMMAP */ |
657 |
greg |
2.1 |
l = read(fd, m = malloc(st.st_size), st.st_size); |
658 |
|
|
root = (ezxml_root_t)ezxml_parse_str(m, l); |
659 |
greg |
2.2 |
root->len = -1; /* so we know to free s in ezxml_free() */ |
660 |
greg |
2.1 |
#ifndef EZXML_NOMMAP |
661 |
|
|
} |
662 |
greg |
2.2 |
#endif /* EZXML_NOMMAP */ |
663 |
greg |
2.1 |
return &root->xml; |
664 |
|
|
} |
665 |
|
|
|
666 |
greg |
2.2 |
/* a wrapper for ezxml_parse_fd that accepts a file name */ |
667 |
greg |
2.1 |
ezxml_t ezxml_parse_file(const char *file) |
668 |
|
|
{ |
669 |
|
|
int fd = open(file, O_RDONLY, 0); |
670 |
|
|
ezxml_t xml = ezxml_parse_fd(fd); |
671 |
|
|
|
672 |
|
|
if (fd >= 0) close(fd); |
673 |
|
|
return xml; |
674 |
|
|
} |
675 |
|
|
|
676 |
greg |
2.2 |
/* Encodes ampersand sequences appending the results to *dst, reallocating *dst */ |
677 |
|
|
/* if length excedes max. a is non-zero for attribute encoding. Returns *dst */ |
678 |
greg |
2.1 |
char *ezxml_ampencode(const char *s, size_t len, char **dst, size_t *dlen, |
679 |
|
|
size_t *max, short a) |
680 |
|
|
{ |
681 |
|
|
const char *e; |
682 |
|
|
|
683 |
|
|
for (e = s + len; s != e; s++) { |
684 |
|
|
while (*dlen + 10 > *max) *dst = realloc(*dst, *max += EZXML_BUFSIZE); |
685 |
|
|
|
686 |
|
|
switch (*s) { |
687 |
|
|
case '\0': return *dst; |
688 |
|
|
case '&': *dlen += sprintf(*dst + *dlen, "&"); break; |
689 |
|
|
case '<': *dlen += sprintf(*dst + *dlen, "<"); break; |
690 |
|
|
case '>': *dlen += sprintf(*dst + *dlen, ">"); break; |
691 |
|
|
case '"': *dlen += sprintf(*dst + *dlen, (a) ? """ : "\""); break; |
692 |
|
|
case '\n': *dlen += sprintf(*dst + *dlen, (a) ? "
" : "\n"); break; |
693 |
|
|
case '\t': *dlen += sprintf(*dst + *dlen, (a) ? "	" : "\t"); break; |
694 |
|
|
case '\r': *dlen += sprintf(*dst + *dlen, "
"); break; |
695 |
|
|
default: (*dst)[(*dlen)++] = *s; |
696 |
|
|
} |
697 |
|
|
} |
698 |
|
|
return *dst; |
699 |
|
|
} |
700 |
|
|
|
701 |
greg |
2.2 |
/* Recursively converts each tag to xml appending it to *s. Reallocates *s if */ |
702 |
|
|
/* its length excedes max. start is the location of the previous tag in the */ |
703 |
|
|
/* parent tag's character content. Returns *s. */ |
704 |
greg |
2.1 |
char *ezxml_toxml_r(ezxml_t xml, char **s, size_t *len, size_t *max, |
705 |
|
|
size_t start, char ***attr) |
706 |
|
|
{ |
707 |
|
|
int i, j; |
708 |
|
|
char *txt = (xml->parent) ? xml->parent->txt : ""; |
709 |
|
|
size_t off = 0; |
710 |
|
|
|
711 |
greg |
2.2 |
/* parent character content up to this tag */ |
712 |
greg |
2.1 |
*s = ezxml_ampencode(txt + start, xml->off - start, s, len, max, 0); |
713 |
|
|
|
714 |
greg |
2.2 |
while (*len + strlen(xml->name) + 4 > *max) /* reallocate s */ |
715 |
greg |
2.1 |
*s = realloc(*s, *max += EZXML_BUFSIZE); |
716 |
|
|
|
717 |
greg |
2.2 |
*len += sprintf(*s + *len, "<%s", xml->name); /* open tag */ |
718 |
|
|
for (i = 0; xml->attr[i]; i += 2) { /* tag attributes */ |
719 |
greg |
2.1 |
if (ezxml_attr(xml, xml->attr[i]) != xml->attr[i + 1]) continue; |
720 |
greg |
2.2 |
while (*len + strlen(xml->attr[i]) + 7 > *max) /* reallocate s */ |
721 |
greg |
2.1 |
*s = realloc(*s, *max += EZXML_BUFSIZE); |
722 |
|
|
|
723 |
|
|
*len += sprintf(*s + *len, " %s=\"", xml->attr[i]); |
724 |
|
|
ezxml_ampencode(xml->attr[i + 1], -1, s, len, max, 1); |
725 |
|
|
*len += sprintf(*s + *len, "\""); |
726 |
|
|
} |
727 |
|
|
|
728 |
|
|
for (i = 0; attr[i] && strcmp(attr[i][0], xml->name); i++); |
729 |
greg |
2.2 |
for (j = 1; attr[i] && attr[i][j]; j += 3) { /* default attributes */ |
730 |
greg |
2.1 |
if (! attr[i][j + 1] || ezxml_attr(xml, attr[i][j]) != attr[i][j + 1]) |
731 |
greg |
2.2 |
continue; /* skip duplicates and non-values */ |
732 |
|
|
while (*len + strlen(attr[i][j]) + 7 > *max) /* reallocate s */ |
733 |
greg |
2.1 |
*s = realloc(*s, *max += EZXML_BUFSIZE); |
734 |
|
|
|
735 |
|
|
*len += sprintf(*s + *len, " %s=\"", attr[i][j]); |
736 |
|
|
ezxml_ampencode(attr[i][j + 1], -1, s, len, max, 1); |
737 |
|
|
*len += sprintf(*s + *len, "\""); |
738 |
|
|
} |
739 |
|
|
*len += sprintf(*s + *len, ">"); |
740 |
|
|
|
741 |
greg |
2.2 |
*s = (xml->child) ? ezxml_toxml_r(xml->child, s, len, max, 0, attr) /*child */ |
742 |
|
|
: ezxml_ampencode(xml->txt, -1, s, len, max, 0); /*data */ |
743 |
greg |
2.1 |
|
744 |
greg |
2.2 |
while (*len + strlen(xml->name) + 4 > *max) /* reallocate s */ |
745 |
greg |
2.1 |
*s = realloc(*s, *max += EZXML_BUFSIZE); |
746 |
|
|
|
747 |
greg |
2.2 |
*len += sprintf(*s + *len, "</%s>", xml->name); /* close tag */ |
748 |
greg |
2.1 |
|
749 |
greg |
2.2 |
while (txt[off] && off < xml->off) off++; /* make sure off is within bounds */ |
750 |
greg |
2.1 |
return (xml->ordered) ? ezxml_toxml_r(xml->ordered, s, len, max, off, attr) |
751 |
|
|
: ezxml_ampencode(txt + off, -1, s, len, max, 0); |
752 |
|
|
} |
753 |
|
|
|
754 |
greg |
2.2 |
/* Converts an ezxml structure back to xml. Returns a string of xml data that */ |
755 |
|
|
/* must be freed. */ |
756 |
greg |
2.1 |
char *ezxml_toxml(ezxml_t xml) |
757 |
|
|
{ |
758 |
|
|
ezxml_t p = (xml) ? xml->parent : NULL, o = (xml) ? xml->ordered : NULL; |
759 |
|
|
ezxml_root_t root = (ezxml_root_t)xml; |
760 |
|
|
size_t len = 0, max = EZXML_BUFSIZE; |
761 |
|
|
char *s = strcpy(malloc(max), ""), *t, *n; |
762 |
|
|
int i, j, k; |
763 |
|
|
|
764 |
|
|
if (! xml || ! xml->name) return realloc(s, len + 1); |
765 |
greg |
2.2 |
while (root->xml.parent) root = (ezxml_root_t)root->xml.parent; /* root tag */ |
766 |
greg |
2.1 |
|
767 |
greg |
2.2 |
for (i = 0; ! p && root->pi[i]; i++) { /* pre-root processing instructions */ |
768 |
greg |
2.1 |
for (k = 2; root->pi[i][k - 1]; k++); |
769 |
|
|
for (j = 1; (n = root->pi[i][j]); j++) { |
770 |
greg |
2.2 |
if (root->pi[i][k][j - 1] == '>') continue; /* not pre-root */ |
771 |
greg |
2.1 |
while (len + strlen(t = root->pi[i][0]) + strlen(n) + 7 > max) |
772 |
|
|
s = realloc(s, max += EZXML_BUFSIZE); |
773 |
|
|
len += sprintf(s + len, "<?%s%s%s?>\n", t, *n ? " " : "", n); |
774 |
|
|
} |
775 |
|
|
} |
776 |
|
|
|
777 |
|
|
xml->parent = xml->ordered = NULL; |
778 |
|
|
s = ezxml_toxml_r(xml, &s, &len, &max, 0, root->attr); |
779 |
|
|
xml->parent = p; |
780 |
|
|
xml->ordered = o; |
781 |
|
|
|
782 |
greg |
2.2 |
for (i = 0; ! p && root->pi[i]; i++) { /* post-root processing instructions */ |
783 |
greg |
2.1 |
for (k = 2; root->pi[i][k - 1]; k++); |
784 |
|
|
for (j = 1; (n = root->pi[i][j]); j++) { |
785 |
greg |
2.2 |
if (root->pi[i][k][j - 1] == '<') continue; /* not post-root */ |
786 |
greg |
2.1 |
while (len + strlen(t = root->pi[i][0]) + strlen(n) + 7 > max) |
787 |
|
|
s = realloc(s, max += EZXML_BUFSIZE); |
788 |
|
|
len += sprintf(s + len, "\n<?%s%s%s?>", t, *n ? " " : "", n); |
789 |
|
|
} |
790 |
|
|
} |
791 |
|
|
return realloc(s, len + 1); |
792 |
|
|
} |
793 |
|
|
|
794 |
greg |
2.2 |
/* free the memory allocated for the ezxml structure */ |
795 |
greg |
2.1 |
void ezxml_free(ezxml_t xml) |
796 |
|
|
{ |
797 |
|
|
ezxml_root_t root = (ezxml_root_t)xml; |
798 |
|
|
int i, j; |
799 |
|
|
char **a, *s; |
800 |
|
|
|
801 |
|
|
if (! xml) return; |
802 |
|
|
ezxml_free(xml->child); |
803 |
|
|
ezxml_free(xml->ordered); |
804 |
|
|
|
805 |
greg |
2.2 |
if (! xml->parent) { /* free root tag allocations */ |
806 |
|
|
for (i = 10; root->ent[i]; i += 2) /* 0 - 9 are default entites (<>&"') */ |
807 |
greg |
2.1 |
if ((s = root->ent[i + 1]) < root->s || s > root->e) free(s); |
808 |
greg |
2.2 |
free(root->ent); /* free list of general entities */ |
809 |
greg |
2.1 |
|
810 |
|
|
for (i = 0; (a = root->attr[i]); i++) { |
811 |
greg |
2.2 |
for (j = 1; a[j++]; j += 2) /* free malloced attribute values */ |
812 |
greg |
2.1 |
if (a[j] && (a[j] < root->s || a[j] > root->e)) free(a[j]); |
813 |
|
|
free(a); |
814 |
|
|
} |
815 |
greg |
2.2 |
if (root->attr[0]) free(root->attr); /* free default attribute list */ |
816 |
greg |
2.1 |
|
817 |
|
|
for (i = 0; root->pi[i]; i++) { |
818 |
|
|
for (j = 1; root->pi[i][j]; j++); |
819 |
|
|
free(root->pi[i][j + 1]); |
820 |
|
|
free(root->pi[i]); |
821 |
|
|
} |
822 |
greg |
2.2 |
if (root->pi[0]) free(root->pi); /* free processing instructions */ |
823 |
greg |
2.1 |
|
824 |
greg |
2.2 |
if (root->len == -1) free(root->m); /* malloced xml data */ |
825 |
greg |
2.1 |
#ifndef EZXML_NOMMAP |
826 |
greg |
2.2 |
else if (root->len) munmap(root->m, root->len); /* mem mapped xml data */ |
827 |
|
|
#endif /* EZXML_NOMMAP */ |
828 |
|
|
if (root->u) free(root->u); /* utf8 conversion */ |
829 |
greg |
2.1 |
} |
830 |
|
|
|
831 |
greg |
2.2 |
ezxml_free_attr(xml->attr); /* tag attributes */ |
832 |
|
|
if ((xml->flags & EZXML_TXTM)) free(xml->txt); /* character content */ |
833 |
|
|
if ((xml->flags & EZXML_NAMEM)) free(xml->name); /* tag name */ |
834 |
greg |
2.1 |
free(xml); |
835 |
|
|
} |
836 |
|
|
|
837 |
greg |
2.2 |
/* return parser error message or empty string if none */ |
838 |
greg |
2.1 |
const char *ezxml_error(ezxml_t xml) |
839 |
|
|
{ |
840 |
greg |
2.2 |
while (xml && xml->parent) xml = xml->parent; /* find root tag */ |
841 |
greg |
2.1 |
return (xml) ? ((ezxml_root_t)xml)->err : ""; |
842 |
|
|
} |
843 |
|
|
|
844 |
greg |
2.2 |
/* returns a new empty ezxml structure with the given root tag name */ |
845 |
greg |
2.1 |
ezxml_t ezxml_new(const char *name) |
846 |
|
|
{ |
847 |
|
|
static char *ent[] = { "lt;", "<", "gt;", ">", "quot;", """, |
848 |
|
|
"apos;", "'", "amp;", "&", NULL }; |
849 |
|
|
ezxml_root_t root = (ezxml_root_t)memset(malloc(sizeof(struct ezxml_root)), |
850 |
|
|
'\0', sizeof(struct ezxml_root)); |
851 |
|
|
root->xml.name = (char *)name; |
852 |
|
|
root->cur = &root->xml; |
853 |
|
|
strcpy(root->err, root->xml.txt = ""); |
854 |
|
|
root->ent = memcpy(malloc(sizeof(ent)), ent, sizeof(ent)); |
855 |
|
|
root->attr = root->pi = (char ***)(root->xml.attr = EZXML_NIL); |
856 |
|
|
return &root->xml; |
857 |
|
|
} |
858 |
|
|
|
859 |
greg |
2.2 |
/* inserts an existing tag into an ezxml structure */ |
860 |
greg |
2.1 |
ezxml_t ezxml_insert(ezxml_t xml, ezxml_t dest, size_t off) |
861 |
|
|
{ |
862 |
|
|
ezxml_t cur, prev, head; |
863 |
|
|
|
864 |
|
|
xml->next = xml->sibling = xml->ordered = NULL; |
865 |
|
|
xml->off = off; |
866 |
|
|
xml->parent = dest; |
867 |
|
|
|
868 |
greg |
2.2 |
if ((head = dest->child)) { /* already have sub tags */ |
869 |
|
|
if (head->off <= off) { /* not first subtag */ |
870 |
greg |
2.1 |
for (cur = head; cur->ordered && cur->ordered->off <= off; |
871 |
|
|
cur = cur->ordered); |
872 |
|
|
xml->ordered = cur->ordered; |
873 |
|
|
cur->ordered = xml; |
874 |
|
|
} |
875 |
greg |
2.2 |
else { /* first subtag */ |
876 |
greg |
2.1 |
xml->ordered = head; |
877 |
|
|
dest->child = xml; |
878 |
|
|
} |
879 |
|
|
|
880 |
|
|
for (cur = head, prev = NULL; cur && strcmp(cur->name, xml->name); |
881 |
greg |
2.2 |
prev = cur, cur = cur->sibling); /* find tag type */ |
882 |
|
|
if (cur && cur->off <= off) { /* not first of type */ |
883 |
greg |
2.1 |
while (cur->next && cur->next->off <= off) cur = cur->next; |
884 |
|
|
xml->next = cur->next; |
885 |
|
|
cur->next = xml; |
886 |
|
|
} |
887 |
greg |
2.2 |
else { /* first tag of this type */ |
888 |
|
|
if (prev && cur) prev->sibling = cur->sibling; /* remove old first */ |
889 |
|
|
xml->next = cur; /* old first tag is now next */ |
890 |
greg |
2.1 |
for (cur = head, prev = NULL; cur && cur->off <= off; |
891 |
greg |
2.2 |
prev = cur, cur = cur->sibling); /* new sibling insert point */ |
892 |
greg |
2.1 |
xml->sibling = cur; |
893 |
|
|
if (prev) prev->sibling = xml; |
894 |
|
|
} |
895 |
|
|
} |
896 |
greg |
2.2 |
else dest->child = xml; /* only sub tag */ |
897 |
greg |
2.1 |
|
898 |
|
|
return xml; |
899 |
|
|
} |
900 |
|
|
|
901 |
greg |
2.2 |
/* Adds a child tag. off is the offset of the child tag relative to the start */ |
902 |
|
|
/* of the parent tag's character content. Returns the child tag. */ |
903 |
greg |
2.1 |
ezxml_t ezxml_add_child(ezxml_t xml, const char *name, size_t off) |
904 |
|
|
{ |
905 |
|
|
ezxml_t child; |
906 |
|
|
|
907 |
|
|
if (! xml) return NULL; |
908 |
|
|
child = (ezxml_t)memset(malloc(sizeof(struct ezxml)), '\0', |
909 |
|
|
sizeof(struct ezxml)); |
910 |
|
|
child->name = (char *)name; |
911 |
|
|
child->attr = EZXML_NIL; |
912 |
|
|
child->txt = ""; |
913 |
|
|
|
914 |
|
|
return ezxml_insert(child, xml, off); |
915 |
|
|
} |
916 |
|
|
|
917 |
greg |
2.2 |
/* sets the character content for the given tag and returns the tag */ |
918 |
greg |
2.1 |
ezxml_t ezxml_set_txt(ezxml_t xml, const char *txt) |
919 |
|
|
{ |
920 |
|
|
if (! xml) return NULL; |
921 |
greg |
2.2 |
if (xml->flags & EZXML_TXTM) free(xml->txt); /* existing txt was malloced */ |
922 |
greg |
2.1 |
xml->flags &= ~EZXML_TXTM; |
923 |
|
|
xml->txt = (char *)txt; |
924 |
|
|
return xml; |
925 |
|
|
} |
926 |
|
|
|
927 |
greg |
2.2 |
/* Sets the given tag attribute or adds a new attribute if not found. A value */ |
928 |
|
|
/* of NULL will remove the specified attribute. Returns the tag given. */ |
929 |
greg |
2.1 |
ezxml_t ezxml_set_attr(ezxml_t xml, const char *name, const char *value) |
930 |
|
|
{ |
931 |
|
|
int l = 0, c; |
932 |
|
|
|
933 |
|
|
if (! xml) return NULL; |
934 |
|
|
while (xml->attr[l] && strcmp(xml->attr[l], name)) l += 2; |
935 |
greg |
2.2 |
if (! xml->attr[l]) { /* not found, add as new attribute */ |
936 |
|
|
if (! value) return xml; /* nothing to do */ |
937 |
|
|
if (xml->attr == EZXML_NIL) { /* first attribute */ |
938 |
greg |
2.1 |
xml->attr = malloc(4 * sizeof(char *)); |
939 |
greg |
2.2 |
xml->attr[1] = strdup(""); /* empty list of malloced names/vals */ |
940 |
greg |
2.1 |
} |
941 |
|
|
else xml->attr = realloc(xml->attr, (l + 4) * sizeof(char *)); |
942 |
|
|
|
943 |
greg |
2.2 |
xml->attr[l] = (char *)name; /* set attribute name */ |
944 |
|
|
xml->attr[l + 2] = NULL; /* null terminate attribute list */ |
945 |
greg |
2.1 |
xml->attr[l + 3] = realloc(xml->attr[l + 1], |
946 |
|
|
(c = strlen(xml->attr[l + 1])) + 2); |
947 |
greg |
2.2 |
strcpy(xml->attr[l + 3] + c, " "); /* set name/value as not malloced */ |
948 |
greg |
2.1 |
if (xml->flags & EZXML_DUP) xml->attr[l + 3][c] = EZXML_NAMEM; |
949 |
|
|
} |
950 |
greg |
2.2 |
else if (xml->flags & EZXML_DUP) free((char *)name); /* name was strduped */ |
951 |
greg |
2.1 |
|
952 |
greg |
2.2 |
for (c = l; xml->attr[c]; c += 2); /* find end of attribute list */ |
953 |
|
|
if (xml->attr[c + 1][l / 2] & EZXML_TXTM) free(xml->attr[l + 1]); /*old val */ |
954 |
greg |
2.1 |
if (xml->flags & EZXML_DUP) xml->attr[c + 1][l / 2] |= EZXML_TXTM; |
955 |
|
|
else xml->attr[c + 1][l / 2] &= ~EZXML_TXTM; |
956 |
|
|
|
957 |
greg |
2.2 |
if (value) xml->attr[l + 1] = (char *)value; /* set attribute value */ |
958 |
|
|
else { /* remove attribute */ |
959 |
greg |
2.1 |
if (xml->attr[c + 1][l / 2] & EZXML_NAMEM) free(xml->attr[l]); |
960 |
|
|
memmove(xml->attr + l, xml->attr + l + 2, (c - l + 2) * sizeof(char*)); |
961 |
|
|
xml->attr = realloc(xml->attr, (c + 2) * sizeof(char *)); |
962 |
|
|
memmove(xml->attr[c + 1] + (l / 2), xml->attr[c + 1] + (l / 2) + 1, |
963 |
greg |
2.2 |
(c / 2) - (l / 2)); /* fix list of which name/vals are malloced */ |
964 |
greg |
2.1 |
} |
965 |
greg |
2.2 |
xml->flags &= ~EZXML_DUP; /* clear strdup() flag */ |
966 |
greg |
2.1 |
return xml; |
967 |
|
|
} |
968 |
|
|
|
969 |
greg |
2.2 |
/* sets a flag for the given tag and returns the tag */ |
970 |
greg |
2.1 |
ezxml_t ezxml_set_flag(ezxml_t xml, short flag) |
971 |
|
|
{ |
972 |
|
|
if (xml) xml->flags |= flag; |
973 |
|
|
return xml; |
974 |
|
|
} |
975 |
|
|
|
976 |
greg |
2.2 |
/* removes a tag along with its subtags without freeing its memory */ |
977 |
greg |
2.1 |
ezxml_t ezxml_cut(ezxml_t xml) |
978 |
|
|
{ |
979 |
|
|
ezxml_t cur; |
980 |
|
|
|
981 |
greg |
2.2 |
if (! xml) return NULL; /* nothing to do */ |
982 |
|
|
if (xml->next) xml->next->sibling = xml->sibling; /* patch sibling list */ |
983 |
greg |
2.1 |
|
984 |
greg |
2.2 |
if (xml->parent) { /* not root tag */ |
985 |
|
|
cur = xml->parent->child; /* find head of subtag list */ |
986 |
|
|
if (cur == xml) xml->parent->child = xml->ordered; /* first subtag */ |
987 |
|
|
else { /* not first subtag */ |
988 |
greg |
2.1 |
while (cur->ordered != xml) cur = cur->ordered; |
989 |
greg |
2.2 |
cur->ordered = cur->ordered->ordered; /* patch ordered list */ |
990 |
greg |
2.1 |
|
991 |
greg |
2.2 |
cur = xml->parent->child; /* go back to head of subtag list */ |
992 |
|
|
if (strcmp(cur->name, xml->name)) { /* not in first sibling list */ |
993 |
greg |
2.1 |
while (strcmp(cur->sibling->name, xml->name)) |
994 |
|
|
cur = cur->sibling; |
995 |
greg |
2.2 |
if (cur->sibling == xml) { /* first of a sibling list */ |
996 |
greg |
2.1 |
cur->sibling = (xml->next) ? xml->next |
997 |
|
|
: cur->sibling->sibling; |
998 |
|
|
} |
999 |
greg |
2.2 |
else cur = cur->sibling; /* not first of a sibling list */ |
1000 |
greg |
2.1 |
} |
1001 |
|
|
|
1002 |
|
|
while (cur->next && cur->next != xml) cur = cur->next; |
1003 |
greg |
2.2 |
if (cur->next) cur->next = cur->next->next; /* patch next list */ |
1004 |
greg |
2.1 |
} |
1005 |
|
|
} |
1006 |
|
|
xml->ordered = xml->sibling = xml->next = NULL; |
1007 |
|
|
return xml; |
1008 |
|
|
} |
1009 |
|
|
|
1010 |
greg |
2.2 |
#ifdef EZXML_TEST /* test harness */ |
1011 |
greg |
2.1 |
int main(int argc, char **argv) |
1012 |
|
|
{ |
1013 |
|
|
ezxml_t xml; |
1014 |
|
|
char *s; |
1015 |
|
|
int i; |
1016 |
|
|
|
1017 |
|
|
if (argc != 2) return fprintf(stderr, "usage: %s xmlfile\n", argv[0]); |
1018 |
|
|
|
1019 |
|
|
xml = ezxml_parse_file(argv[1]); |
1020 |
|
|
printf("%s\n", (s = ezxml_toxml(xml))); |
1021 |
|
|
free(s); |
1022 |
|
|
i = fprintf(stderr, "%s", ezxml_error(xml)); |
1023 |
|
|
ezxml_free(xml); |
1024 |
|
|
return (i) ? 1 : 0; |
1025 |
|
|
} |
1026 |
greg |
2.2 |
#endif /* EZXML_TEST */ |