| 1 |
#ifndef lint |
| 2 |
static const char RCSid[] = "$Id: ezxml.c,v 2.6 2014/10/26 17:35:53 greg Exp $"; |
| 3 |
#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 |
#endif /* EZXML_NOMMAP */ |
| 38 |
#include <sys/stat.h> |
| 39 |
#include "ezxml.h" |
| 40 |
|
| 41 |
#ifdef _WIN32 |
| 42 |
#include <io.h> |
| 43 |
#define read _read |
| 44 |
#define open _open |
| 45 |
#define close _close |
| 46 |
#endif |
| 47 |
|
| 48 |
#define EZXML_WS "\t\r\n " /* whitespace */ |
| 49 |
#define EZXML_ERRL 128 /* maximum error string length */ |
| 50 |
|
| 51 |
typedef struct ezxml_root *ezxml_root_t; |
| 52 |
struct ezxml_root { /* additional data for the root tag */ |
| 53 |
struct ezxml xml; /* is a super-struct built on top of ezxml struct */ |
| 54 |
ezxml_t cur; /* current xml tree insertion point */ |
| 55 |
char *m; /* original xml string */ |
| 56 |
size_t len; /* length of allocated memory for mmap, -1 for malloc */ |
| 57 |
char *u; /* UTF-8 conversion of string if original was UTF-16 */ |
| 58 |
char *s; /* start of work area */ |
| 59 |
char *e; /* end of work area */ |
| 60 |
char **ent; /* general entities (ampersand sequences) */ |
| 61 |
char ***attr; /* default attributes */ |
| 62 |
char ***pi; /* processing instructions */ |
| 63 |
short standalone; /* non-zero if <?xml standalone="yes"?> */ |
| 64 |
char err[EZXML_ERRL]; /* error string */ |
| 65 |
}; |
| 66 |
|
| 67 |
char *EZXML_NIL[] = { NULL }; /* empty, null terminated array of strings */ |
| 68 |
|
| 69 |
/* returns the first child tag with the given name or NULL if not found */ |
| 70 |
ezxml_t ezxml_child(ezxml_t xml, const char *name) |
| 71 |
{ |
| 72 |
xml = (xml) ? xml->child : NULL; |
| 73 |
while (xml && strcmp(name, xml->name)) xml = xml->sibling; |
| 74 |
return xml; |
| 75 |
} |
| 76 |
|
| 77 |
/* returns the Nth tag with the same name in the same subsection or NULL if not */ |
| 78 |
/* found */ |
| 79 |
ezxml_t ezxml_idx(ezxml_t xml, int idx) |
| 80 |
{ |
| 81 |
for (; xml && idx; idx--) xml = xml->next; |
| 82 |
return xml; |
| 83 |
} |
| 84 |
|
| 85 |
/* returns the value of the requested tag attribute or NULL if not found */ |
| 86 |
const char *ezxml_attr(ezxml_t xml, const char *attr) |
| 87 |
{ |
| 88 |
int i = 0, j = 1; |
| 89 |
ezxml_root_t root = (ezxml_root_t)xml; |
| 90 |
|
| 91 |
if (! xml || ! xml->attr) return NULL; |
| 92 |
while (xml->attr[i] && strcmp(attr, xml->attr[i])) i += 2; |
| 93 |
if (xml->attr[i]) return xml->attr[i + 1]; /* found attribute */ |
| 94 |
|
| 95 |
while (root->xml.parent) root = (ezxml_root_t)root->xml.parent; /* root tag */ |
| 96 |
for (i = 0; root->attr[i] && strcmp(xml->name, root->attr[i][0]); i++); |
| 97 |
if (! root->attr[i]) return NULL; /* no matching default attributes */ |
| 98 |
while (root->attr[i][j] && strcmp(attr, root->attr[i][j])) j += 3; |
| 99 |
return (root->attr[i][j]) ? root->attr[i][j + 1] : NULL; /* found default */ |
| 100 |
} |
| 101 |
|
| 102 |
/* same as ezxml_get but takes an already initialized va_list */ |
| 103 |
ezxml_t ezxml_vget(ezxml_t xml, va_list ap) |
| 104 |
{ |
| 105 |
char *name = va_arg(ap, char *); |
| 106 |
int idx = -1; |
| 107 |
|
| 108 |
if (name && *name) { |
| 109 |
idx = va_arg(ap, int); |
| 110 |
xml = ezxml_child(xml, name); |
| 111 |
} |
| 112 |
return (idx < 0) ? xml : ezxml_vget(ezxml_idx(xml, idx), ap); |
| 113 |
} |
| 114 |
|
| 115 |
/* Traverses the xml tree to retrieve a specific subtag. Takes a variable */ |
| 116 |
/* length list of tag names and indexes. The argument list must be terminated */ |
| 117 |
/* by either an index of -1 or an empty string tag name. Example: */ |
| 118 |
/* title = ezxml_get(library, "shelf", 0, "book", 2, "title", -1); */ |
| 119 |
/* This retrieves the title of the 3rd book on the 1st shelf of library. */ |
| 120 |
/* Returns NULL if not found. */ |
| 121 |
ezxml_t ezxml_get(ezxml_t xml, ...) |
| 122 |
{ |
| 123 |
va_list ap; |
| 124 |
ezxml_t r; |
| 125 |
|
| 126 |
va_start(ap, xml); |
| 127 |
r = ezxml_vget(xml, ap); |
| 128 |
va_end(ap); |
| 129 |
return r; |
| 130 |
} |
| 131 |
|
| 132 |
/* returns a null terminated array of processing instructions for the given */ |
| 133 |
/* target */ |
| 134 |
const char **ezxml_pi(ezxml_t xml, const char *target) |
| 135 |
{ |
| 136 |
ezxml_root_t root = (ezxml_root_t)xml; |
| 137 |
int i = 0; |
| 138 |
|
| 139 |
if (! root) return (const char **)EZXML_NIL; |
| 140 |
while (root->xml.parent) root = (ezxml_root_t)root->xml.parent; /* root tag */ |
| 141 |
while (root->pi[i] && strcmp(target, root->pi[i][0])) i++; /* find target */ |
| 142 |
return (const char **)((root->pi[i]) ? root->pi[i] + 1 : EZXML_NIL); |
| 143 |
} |
| 144 |
|
| 145 |
/* set an error string and return root */ |
| 146 |
ezxml_t ezxml_err(ezxml_root_t root, char *s, const char *err, ...) |
| 147 |
{ |
| 148 |
va_list ap; |
| 149 |
int line = 1; |
| 150 |
char *t, fmt[EZXML_ERRL]; |
| 151 |
|
| 152 |
for (t = root->s; t < s; t++) if (*t == '\n') line++; |
| 153 |
sprintf(fmt, "[error near line %d]: %s", line, err); |
| 154 |
|
| 155 |
va_start(ap, err); |
| 156 |
vsnprintf(root->err, EZXML_ERRL, fmt, ap); |
| 157 |
va_end(ap); |
| 158 |
|
| 159 |
return &root->xml; |
| 160 |
} |
| 161 |
|
| 162 |
/* Recursively decodes entity and character references and normalizes new lines */ |
| 163 |
/* ent is a null terminated array of alternating entity names and values. set t */ |
| 164 |
/* to '&' for general entity decoding, '%' for parameter entity decoding, 'c' */ |
| 165 |
/* for cdata sections, ' ' for attribute normalization, or '*' for non-cdata */ |
| 166 |
/* attribute normalization. Returns s, or if the decoded string is longer than */ |
| 167 |
/* s, returns a malloced string that must be freed. */ |
| 168 |
char *ezxml_decode(char *s, char **ent, char t) |
| 169 |
{ |
| 170 |
char *e, *r = s, *m = s; |
| 171 |
long b, c, d, l; |
| 172 |
|
| 173 |
for (; *s; s++) /* normalize line endings */ |
| 174 |
if (*s == '\r') { |
| 175 |
char *s2 = s+1; |
| 176 |
do { |
| 177 |
while (*s2 == '\r') |
| 178 |
++s2; |
| 179 |
*s++ = *s2; |
| 180 |
} while (*s2++); |
| 181 |
break; |
| 182 |
} |
| 183 |
|
| 184 |
for (s = r; ; ) { |
| 185 |
while (*s && *s != '&' && (*s != '%' || t != '%') && !isspace(*s)) s++; |
| 186 |
|
| 187 |
if (! *s) break; |
| 188 |
else if (t != 'c' && ! strncmp(s, "&#", 2)) { /* character reference */ |
| 189 |
if (s[2] == 'x') c = strtol(s + 3, &e, 16); /* base 16 */ |
| 190 |
else c = strtol(s + 2, &e, 10); /* base 10 */ |
| 191 |
if (! c || *e != ';') { s++; continue; } /* not a character ref */ |
| 192 |
|
| 193 |
if (c < 0x80) *(s++) = c; /* US-ASCII subset */ |
| 194 |
else { /* multi-byte UTF-8 sequence */ |
| 195 |
for (b = 0, d = c; d; d /= 2) b++; /* number of bits in c */ |
| 196 |
b = (b - 2) / 5; /* number of bytes in payload */ |
| 197 |
*(s++) = (0xFF << (7 - b)) | (c >> (6 * b)); /* head */ |
| 198 |
while (b) *(s++) = 0x80 | ((c >> (6 * --b)) & 0x3F); /* payload */ |
| 199 |
} |
| 200 |
|
| 201 |
memmove(s, strchr(s, ';') + 1, strlen(strchr(s, ';'))); |
| 202 |
} |
| 203 |
else if ((*s == '&' && (t == '&' || t == ' ' || t == '*')) || |
| 204 |
(*s == '%' && t == '%')) { /* entity reference */ |
| 205 |
for (b = 0; ent[b] && strncmp(s + 1, ent[b], strlen(ent[b])); |
| 206 |
b += 2); /* find entity in entity list */ |
| 207 |
|
| 208 |
if (ent[b++]) { /* found a match */ |
| 209 |
if ((c = strlen(ent[b])) - 1 > (e = strchr(s, ';')) - s) { |
| 210 |
l = (d = (s - r)) + c + strlen(e); /* new length */ |
| 211 |
r = (r == m) ? strcpy(malloc(l), r) : realloc(r, l); |
| 212 |
e = strchr((s = r + d), ';'); /* fix up pointers */ |
| 213 |
} |
| 214 |
|
| 215 |
memmove(s + c, e + 1, strlen(e)); /* shift rest of string */ |
| 216 |
strncpy(s, ent[b], c); /* copy in replacement text */ |
| 217 |
} |
| 218 |
else s++; /* not a known entity */ |
| 219 |
} |
| 220 |
else if ((t == ' ' || t == '*') && isspace(*s)) *(s++) = ' '; |
| 221 |
else s++; /* no decoding needed */ |
| 222 |
} |
| 223 |
|
| 224 |
if (t == '*') { /* normalize spaces for non-cdata attributes */ |
| 225 |
for (s = r; *s; s++) { |
| 226 |
if ((l = strspn(s, " "))) memmove(s, s + l, strlen(s + l) + 1); |
| 227 |
while (*s && *s != ' ') s++; |
| 228 |
} |
| 229 |
if (--s >= r && *s == ' ') *s = '\0'; /* trim any trailing space */ |
| 230 |
} |
| 231 |
return r; |
| 232 |
} |
| 233 |
|
| 234 |
/* called when parser finds start of new tag */ |
| 235 |
void ezxml_open_tag(ezxml_root_t root, char *name, char **attr) |
| 236 |
{ |
| 237 |
ezxml_t xml = root->cur; |
| 238 |
|
| 239 |
if (xml->name) xml = ezxml_add_child(xml, name, strlen(xml->txt)); |
| 240 |
else xml->name = name; /* first open tag */ |
| 241 |
|
| 242 |
xml->attr = attr; |
| 243 |
root->cur = xml; /* update tag insertion point */ |
| 244 |
} |
| 245 |
|
| 246 |
/* called when parser finds character content between open and closing tag */ |
| 247 |
void ezxml_char_content(ezxml_root_t root, char *s, size_t len, char t) |
| 248 |
{ |
| 249 |
ezxml_t xml = root->cur; |
| 250 |
char *m = s; |
| 251 |
size_t l; |
| 252 |
|
| 253 |
if (! xml || ! xml->name || ! len) return; /* sanity check */ |
| 254 |
|
| 255 |
s[len] = '\0'; /* null terminate text (calling functions anticipate this) */ |
| 256 |
len = strlen(s = ezxml_decode(s, root->ent, t)) + 1; |
| 257 |
|
| 258 |
if (! *(xml->txt)) xml->txt = s; /* initial character content */ |
| 259 |
else { /* allocate our own memory and make a copy */ |
| 260 |
xml->txt = (xml->flags & EZXML_TXTM) /* allocate some space */ |
| 261 |
? realloc(xml->txt, (l = strlen(xml->txt)) + len) |
| 262 |
: strcpy(malloc((l = strlen(xml->txt)) + len), xml->txt); |
| 263 |
strcpy(xml->txt + l, s); /* add new char content */ |
| 264 |
if (s != m) free(s); /* free s if it was malloced by ezxml_decode() */ |
| 265 |
} |
| 266 |
|
| 267 |
if (xml->txt != m) ezxml_set_flag(xml, EZXML_TXTM); |
| 268 |
} |
| 269 |
|
| 270 |
/* called when parser finds closing tag */ |
| 271 |
ezxml_t ezxml_close_tag(ezxml_root_t root, char *name, char *s) |
| 272 |
{ |
| 273 |
if (! root->cur || ! root->cur->name || strcmp(name, root->cur->name)) |
| 274 |
return ezxml_err(root, s, "unexpected closing tag </%s>", name); |
| 275 |
|
| 276 |
root->cur = root->cur->parent; |
| 277 |
return NULL; |
| 278 |
} |
| 279 |
|
| 280 |
/* checks for circular entity references, returns non-zero if no circular */ |
| 281 |
/* references are found, zero otherwise */ |
| 282 |
int ezxml_ent_ok(char *name, char *s, char **ent) |
| 283 |
{ |
| 284 |
int i; |
| 285 |
|
| 286 |
for (; ; s++) { |
| 287 |
while (*s && *s != '&') s++; /* find next entity reference */ |
| 288 |
if (! *s) return 1; |
| 289 |
if (! strncmp(s + 1, name, strlen(name))) return 0; /* circular ref. */ |
| 290 |
for (i = 0; ent[i] && strncmp(ent[i], s + 1, strlen(ent[i])); i += 2); |
| 291 |
if (ent[i] && ! ezxml_ent_ok(name, ent[i + 1], ent)) return 0; |
| 292 |
} |
| 293 |
} |
| 294 |
|
| 295 |
/* called when the parser finds a processing instruction */ |
| 296 |
void ezxml_proc_inst(ezxml_root_t root, char *s, size_t len) |
| 297 |
{ |
| 298 |
int i = 0, j = 1; |
| 299 |
char *target = s; |
| 300 |
|
| 301 |
s[len] = '\0'; /* null terminate instruction */ |
| 302 |
if (*(s += strcspn(s, EZXML_WS))) { |
| 303 |
*s = '\0'; /* null terminate target */ |
| 304 |
s += strspn(s + 1, EZXML_WS) + 1; /* skip whitespace after target */ |
| 305 |
} |
| 306 |
|
| 307 |
if (! strcmp(target, "xml")) { /* <?xml ... ?> */ |
| 308 |
if ((s = strstr(s, "standalone")) && ! strncmp(s + strspn(s + 10, |
| 309 |
EZXML_WS "='\"") + 10, "yes", 3)) root->standalone = 1; |
| 310 |
return; |
| 311 |
} |
| 312 |
|
| 313 |
if (! root->pi[0]) *(root->pi = malloc(sizeof(char **))) = NULL; /*first pi */ |
| 314 |
|
| 315 |
while (root->pi[i] && strcmp(target, root->pi[i][0])) i++; /* find target */ |
| 316 |
if (! root->pi[i]) { /* new target */ |
| 317 |
root->pi = realloc(root->pi, sizeof(char **) * (i + 2)); |
| 318 |
root->pi[i] = malloc(sizeof(char *) * 3); |
| 319 |
root->pi[i][0] = target; |
| 320 |
root->pi[i][1] = (char *)(root->pi[i + 1] = NULL); /* terminate pi list */ |
| 321 |
root->pi[i][2] = strdup(""); /* empty document position list */ |
| 322 |
} |
| 323 |
|
| 324 |
while (root->pi[i][j]) j++; /* find end of instruction list for this target */ |
| 325 |
root->pi[i] = realloc(root->pi[i], sizeof(char *) * (j + 3)); |
| 326 |
root->pi[i][j + 2] = realloc(root->pi[i][j + 1], j + 1); |
| 327 |
strcpy(root->pi[i][j + 2] + j - 1, (root->xml.name) ? ">" : "<"); |
| 328 |
root->pi[i][j + 1] = NULL; /* null terminate pi list for this target */ |
| 329 |
root->pi[i][j] = s; /* set instruction */ |
| 330 |
} |
| 331 |
|
| 332 |
/* called when the parser finds an internal doctype subset */ |
| 333 |
short ezxml_internal_dtd(ezxml_root_t root, char *s, size_t len) |
| 334 |
{ |
| 335 |
char q, *c, *t, *n = NULL, *v, **ent, **pe; |
| 336 |
int i, j; |
| 337 |
|
| 338 |
pe = memcpy(malloc(sizeof(EZXML_NIL)), EZXML_NIL, sizeof(EZXML_NIL)); |
| 339 |
|
| 340 |
for (s[len] = '\0'; s; ) { |
| 341 |
while (*s && *s != '<' && *s != '%') s++; /* find next declaration */ |
| 342 |
|
| 343 |
if (! *s) break; |
| 344 |
else if (! strncmp(s, "<!ENTITY", 8)) { /* parse entity definitions */ |
| 345 |
c = s += strspn(s + 8, EZXML_WS) + 8; /* skip white space separator */ |
| 346 |
n = s + strspn(s, EZXML_WS "%"); /* find name */ |
| 347 |
*(s = n + strcspn(n, EZXML_WS)) = ';'; /* append ; to name */ |
| 348 |
|
| 349 |
v = s + strspn(s + 1, EZXML_WS) + 1; /* find value */ |
| 350 |
if ((q = *(v++)) != '"' && q != '\'') { /* skip externals */ |
| 351 |
s = strchr(s, '>'); |
| 352 |
continue; |
| 353 |
} |
| 354 |
|
| 355 |
for (i = 0, ent = (*c == '%') ? pe : root->ent; ent[i]; i++); |
| 356 |
ent = realloc(ent, (i + 3) * sizeof(char *)); /* space for next ent */ |
| 357 |
if (*c == '%') pe = ent; |
| 358 |
else root->ent = ent; |
| 359 |
|
| 360 |
*(++s) = '\0'; /* null terminate name */ |
| 361 |
if ((s = strchr(v, q))) *(s++) = '\0'; /* null terminate value */ |
| 362 |
ent[i + 1] = ezxml_decode(v, pe, '%'); /* set value */ |
| 363 |
ent[i + 2] = NULL; /* null terminate entity list */ |
| 364 |
if (! ezxml_ent_ok(n, ent[i + 1], ent)) { /* circular reference */ |
| 365 |
if (ent[i + 1] != v) free(ent[i + 1]); |
| 366 |
ezxml_err(root, v, "circular entity declaration &%s", n); |
| 367 |
break; |
| 368 |
} |
| 369 |
else ent[i] = n; /* set entity name */ |
| 370 |
} |
| 371 |
else if (! strncmp(s, "<!ATTLIST", 9)) { /* parse default attributes */ |
| 372 |
t = s + strspn(s + 9, EZXML_WS) + 9; /* skip whitespace separator */ |
| 373 |
if (! *t) { ezxml_err(root, t, "unclosed <!ATTLIST"); break; } |
| 374 |
if (*(s = t + strcspn(t, EZXML_WS ">")) == '>') continue; |
| 375 |
else *s = '\0'; /* null terminate tag name */ |
| 376 |
for (i = 0; root->attr[i] && strcmp(n, root->attr[i][0]); i++); |
| 377 |
++s; |
| 378 |
while (*(n = s + strspn(s, EZXML_WS)) && *n != '>') { |
| 379 |
if (*(s = n + strcspn(n, EZXML_WS))) *s = '\0'; /* attr name */ |
| 380 |
else { ezxml_err(root, t, "malformed <!ATTLIST"); break; } |
| 381 |
|
| 382 |
s += strspn(s + 1, EZXML_WS) + 1; /* find next token */ |
| 383 |
c = (strncmp(s, "CDATA", 5)) ? "*" : " "; /* is it cdata? */ |
| 384 |
if (! strncmp(s, "NOTATION", 8)) |
| 385 |
s += strspn(s + 8, EZXML_WS) + 8; |
| 386 |
s = (*s == '(') ? strchr(s, ')') : s + strcspn(s, EZXML_WS); |
| 387 |
if (! s) { ezxml_err(root, t, "malformed <!ATTLIST"); break; } |
| 388 |
|
| 389 |
s += strspn(s, EZXML_WS ")"); /* skip white space separator */ |
| 390 |
if (! strncmp(s, "#FIXED", 6)) |
| 391 |
s += strspn(s + 6, EZXML_WS) + 6; |
| 392 |
if (*s == '#') { /* no default value */ |
| 393 |
s += strcspn(s, EZXML_WS ">") - 1; |
| 394 |
if (*c == ' ') continue; /* cdata is default, nothing to do */ |
| 395 |
v = NULL; |
| 396 |
} |
| 397 |
else if ((*s == '"' || *s == '\'') && /* default value */ |
| 398 |
(s = strchr(v = s + 1, *s))) *s = '\0'; |
| 399 |
else { ezxml_err(root, t, "malformed <!ATTLIST"); break; } |
| 400 |
|
| 401 |
if (! root->attr[i]) { /* new tag name */ |
| 402 |
root->attr = (! i) ? malloc(2 * sizeof(char **)) |
| 403 |
: realloc(root->attr, |
| 404 |
(i + 2) * sizeof(char **)); |
| 405 |
root->attr[i] = malloc(2 * sizeof(char *)); |
| 406 |
root->attr[i][0] = t; /* set tag name */ |
| 407 |
root->attr[i][1] = (char *)(root->attr[i + 1] = NULL); |
| 408 |
} |
| 409 |
|
| 410 |
for (j = 1; root->attr[i][j]; j += 3); /* find end of list */ |
| 411 |
root->attr[i] = realloc(root->attr[i], |
| 412 |
(j + 4) * sizeof(char *)); |
| 413 |
|
| 414 |
root->attr[i][j + 3] = NULL; /* null terminate list */ |
| 415 |
root->attr[i][j + 2] = c; /* is it cdata? */ |
| 416 |
root->attr[i][j + 1] = (v) ? ezxml_decode(v, root->ent, *c) |
| 417 |
: NULL; |
| 418 |
root->attr[i][j] = n; /* attribute name */ |
| 419 |
++s; |
| 420 |
} |
| 421 |
} |
| 422 |
else if (! strncmp(s, "<!--", 4)) s = strstr(s + 4, "-->"); /* comments */ |
| 423 |
else if (! strncmp(s, "<?", 2)) { /* processing instructions */ |
| 424 |
if ((s = strstr(c = s + 2, "?>"))) |
| 425 |
ezxml_proc_inst(root, c, s++ - c); |
| 426 |
} |
| 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. */ |
| 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 */ |
| 469 |
void ezxml_free_attr(char **attr) { |
| 470 |
int i = 0; |
| 471 |
char *m; |
| 472 |
|
| 473 |
if (! attr || attr == EZXML_NIL) return; /* nothing to free */ |
| 474 |
while (attr[i]) i += 2; /* find end of attribute list */ |
| 475 |
m = attr[i + 1]; /* list of which names and values are malloced */ |
| 476 |
for (i = 0; m[i]; i++) { |
| 477 |
if (m[i] & EZXML_NAMEM) free(attr[i * 2]); |
| 478 |
if (m[i] & EZXML_TXTM) free(attr[(i * 2) + 1]); |
| 479 |
} |
| 480 |
free(m); |
| 481 |
free(attr); |
| 482 |
} |
| 483 |
|
| 484 |
/* parse the given xml string and return an ezxml structure */ |
| 485 |
ezxml_t ezxml_parse_str(char *s, size_t len) |
| 486 |
{ |
| 487 |
ezxml_root_t root = (ezxml_root_t)ezxml_new(NULL); |
| 488 |
char q, e, *d, **attr, **a = NULL; /* initialize a to avoid compile warning */ |
| 489 |
int l, i, j; |
| 490 |
|
| 491 |
root->m = s; |
| 492 |
if (! len) return ezxml_err(root, NULL, "root tag missing"); |
| 493 |
root->u = ezxml_str2utf8(&s, &len); /* convert utf-16 to utf-8 */ |
| 494 |
root->e = (root->s = s) + len; /* record start and end of work area */ |
| 495 |
|
| 496 |
e = s[len - 1]; /* save end char */ |
| 497 |
s[len - 1] = '\0'; /* turn end char into null terminator */ |
| 498 |
|
| 499 |
while (*s && *s != '<') s++; /* find first tag */ |
| 500 |
if (! *s) return ezxml_err(root, s, "root tag missing"); |
| 501 |
|
| 502 |
for (; ; ) { |
| 503 |
attr = (char **)EZXML_NIL; |
| 504 |
d = ++s; |
| 505 |
|
| 506 |
if (isalpha(*s) || *s == '_' || *s == ':' || *s < '\0') { /* new tag */ |
| 507 |
if (! root->cur) |
| 508 |
return ezxml_err(root, d, "markup outside of root element"); |
| 509 |
|
| 510 |
s += strcspn(s, EZXML_WS "/>"); |
| 511 |
while (isspace(*s)) *(s++) = '\0'; /* null terminate tag name */ |
| 512 |
|
| 513 |
if (*s && *s != '/' && *s != '>') /* find tag in default attr list */ |
| 514 |
for (i = 0; (a = root->attr[i]) && strcmp(a[0], d); i++); |
| 515 |
|
| 516 |
for (l = 0; *s && *s != '/' && *s != '>'; l += 2) { /* new attrib */ |
| 517 |
attr = (l) ? realloc(attr, (l + 4) * sizeof(char *)) |
| 518 |
: malloc(4 * sizeof(char *)); /* allocate space */ |
| 519 |
attr[l + 3] = (l) ? realloc(attr[l + 1], (l / 2) + 2) |
| 520 |
: malloc(2); /* mem for list of maloced vals */ |
| 521 |
strcpy(attr[l + 3] + (l / 2), " "); /* value is not malloced */ |
| 522 |
attr[l + 2] = NULL; /* null terminate list */ |
| 523 |
attr[l + 1] = ""; /* temporary attribute value */ |
| 524 |
attr[l] = s; /* set attribute name */ |
| 525 |
|
| 526 |
s += strcspn(s, EZXML_WS "=/>"); |
| 527 |
if (*s == '=' || isspace(*s)) { |
| 528 |
*(s++) = '\0'; /* null terminate tag attribute name */ |
| 529 |
q = *(s += strspn(s, EZXML_WS "=")); |
| 530 |
if (q == '"' || q == '\'') { /* attribute value */ |
| 531 |
attr[l + 1] = ++s; |
| 532 |
while (*s && *s != q) s++; |
| 533 |
if (*s) *(s++) = '\0'; /* null terminate attribute val */ |
| 534 |
else { |
| 535 |
ezxml_free_attr(attr); |
| 536 |
return ezxml_err(root, d, "missing %c", q); |
| 537 |
} |
| 538 |
|
| 539 |
for (j = 1; a && a[j] && strcmp(a[j], attr[l]); j +=3); |
| 540 |
attr[l + 1] = ezxml_decode(attr[l + 1], root->ent, (a |
| 541 |
&& a[j]) ? *a[j + 2] : ' '); |
| 542 |
if (attr[l + 1] < d || attr[l + 1] > s) |
| 543 |
attr[l + 3][l / 2] = EZXML_TXTM; /* value malloced */ |
| 544 |
} |
| 545 |
} |
| 546 |
while (isspace(*s)) s++; |
| 547 |
} |
| 548 |
|
| 549 |
if (*s == '/') { /* self closing tag */ |
| 550 |
*(s++) = '\0'; |
| 551 |
if ((*s && *s != '>') || (! *s && e != '>')) { |
| 552 |
if (l) ezxml_free_attr(attr); |
| 553 |
return ezxml_err(root, d, "missing >"); |
| 554 |
} |
| 555 |
ezxml_open_tag(root, d, attr); |
| 556 |
ezxml_close_tag(root, d, s); |
| 557 |
} |
| 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 |
} |
| 563 |
else { |
| 564 |
if (l) ezxml_free_attr(attr); |
| 565 |
return ezxml_err(root, d, "missing >"); |
| 566 |
} |
| 567 |
} |
| 568 |
else if (*s == '/') { /* close tag */ |
| 569 |
s += strcspn(d = s + 1, EZXML_WS ">") + 1; |
| 570 |
if (! (q = *s) && e != '>') return ezxml_err(root, d, "missing >"); |
| 571 |
*s = '\0'; /* temporarily null terminate tag name */ |
| 572 |
if (ezxml_close_tag(root, d, s)) return &root->xml; |
| 573 |
if (isspace(*s = q)) s += strspn(s, EZXML_WS); |
| 574 |
} |
| 575 |
else if (! strncmp(s, "!--", 3)) { /* xml comment */ |
| 576 |
if (! (s = strstr(s + 3, "--")) || (*(s += 2) != '>' && *s) || |
| 577 |
(! *s && e != '>')) return ezxml_err(root, d, "unclosed <!--"); |
| 578 |
} |
| 579 |
else if (! strncmp(s, "![CDATA[", 8)) { /* cdata */ |
| 580 |
if ((s = strstr(s, "]]>"))) |
| 581 |
ezxml_char_content(root, d + 8, (s += 2) - d - 10, 'c'); |
| 582 |
else return ezxml_err(root, d, "unclosed <![CDATA["); |
| 583 |
} |
| 584 |
else if (! strncmp(s, "!DOCTYPE", 8)) { /* dtd */ |
| 585 |
for (l = 0; *s && ((! l && *s != '>') || (l && (*s != ']' || |
| 586 |
*(s + strspn(s + 1, EZXML_WS) + 1) != '>'))); |
| 587 |
l = (*s == '[') ? 1 : l) s += strcspn(s + 1, "[]>") + 1; |
| 588 |
if (! *s && e != '>') |
| 589 |
return ezxml_err(root, d, "unclosed <!DOCTYPE"); |
| 590 |
d = (l) ? strchr(d, '[') + 1 : d; |
| 591 |
if (l && ! ezxml_internal_dtd(root, d, s++ - d)) return &root->xml; |
| 592 |
} |
| 593 |
else if (*s == '?') { /* <?...?> processing instructions */ |
| 594 |
do { s = strchr(s, '?'); } while (s && *(++s) && *s != '>'); |
| 595 |
if (! s || (! *s && e != '>')) |
| 596 |
return ezxml_err(root, d, "unclosed <?"); |
| 597 |
else ezxml_proc_inst(root, d + 1, s - d - 2); |
| 598 |
} |
| 599 |
else return ezxml_err(root, d, "unexpected <"); |
| 600 |
|
| 601 |
if (! s || ! *s) break; |
| 602 |
*s = '\0'; |
| 603 |
d = ++s; |
| 604 |
if (*s && *s != '<') { /* tag character content */ |
| 605 |
while (*s && *s != '<') s++; |
| 606 |
if (*s) ezxml_char_content(root, d, s - d, '&'); |
| 607 |
else break; |
| 608 |
} |
| 609 |
else if (! *s) break; |
| 610 |
} |
| 611 |
|
| 612 |
if (! root->cur) return &root->xml; |
| 613 |
else if (! root->cur->name) return ezxml_err(root, d, "root tag missing"); |
| 614 |
else return ezxml_err(root, d, "unclosed tag <%s>", root->cur->name); |
| 615 |
} |
| 616 |
|
| 617 |
/* Wrapper for ezxml_parse_str() that accepts a file stream. Reads the entire */ |
| 618 |
/* stream into memory and then parses it. For xml files, use ezxml_parse_file() */ |
| 619 |
/* or ezxml_parse_fd() */ |
| 620 |
ezxml_t ezxml_parse_fp(FILE *fp) |
| 621 |
{ |
| 622 |
ezxml_root_t root; |
| 623 |
size_t l, len = 0; |
| 624 |
char *s; |
| 625 |
|
| 626 |
if (! (s = malloc(EZXML_BUFSIZE))) return NULL; |
| 627 |
do { |
| 628 |
len += (l = fread((s + len), 1, EZXML_BUFSIZE, fp)); |
| 629 |
if (l == EZXML_BUFSIZE) s = realloc(s, len + EZXML_BUFSIZE); |
| 630 |
} while (s && l == EZXML_BUFSIZE); |
| 631 |
|
| 632 |
if (! s) return NULL; |
| 633 |
root = (ezxml_root_t)ezxml_parse_str(s, len); |
| 634 |
root->len = -1; /* so we know to free s in ezxml_free() */ |
| 635 |
return &root->xml; |
| 636 |
} |
| 637 |
|
| 638 |
/* A wrapper for ezxml_parse_str() that accepts a file descriptor. First */ |
| 639 |
/* attempts to mem map the file. Failing that, reads the file into memory. */ |
| 640 |
/* Returns NULL on failure. */ |
| 641 |
ezxml_t ezxml_parse_fd(int fd) |
| 642 |
{ |
| 643 |
ezxml_root_t root; |
| 644 |
struct stat st; |
| 645 |
size_t l; |
| 646 |
void *m; |
| 647 |
|
| 648 |
if (fd < 0) return NULL; |
| 649 |
fstat(fd, &st); |
| 650 |
|
| 651 |
#ifndef EZXML_NOMMAP |
| 652 |
l = (st.st_size + sysconf(_SC_PAGESIZE) - 1) & ~(sysconf(_SC_PAGESIZE) -1); |
| 653 |
if ((m = mmap(NULL, l, PROT_READ|PROT_WRITE, MAP_PRIVATE, fd, 0)) |
| 654 |
!= MAP_FAILED) { |
| 655 |
madvise(m, l, MADV_SEQUENTIAL); /* optimize for sequential access */ |
| 656 |
root = (ezxml_root_t)ezxml_parse_str(m, st.st_size); |
| 657 |
madvise(m, root->len = l, MADV_NORMAL); /* put it back to normal */ |
| 658 |
} |
| 659 |
else { /* mmap failed, read file into memory */ |
| 660 |
#endif /* EZXML_NOMMAP */ |
| 661 |
l = read(fd, m = malloc(st.st_size), st.st_size); |
| 662 |
root = (ezxml_root_t)ezxml_parse_str(m, l); |
| 663 |
root->len = -1; /* so we know to free s in ezxml_free() */ |
| 664 |
#ifndef EZXML_NOMMAP |
| 665 |
} |
| 666 |
#endif /* EZXML_NOMMAP */ |
| 667 |
return &root->xml; |
| 668 |
} |
| 669 |
|
| 670 |
/* a wrapper for ezxml_parse_fd that accepts a file name */ |
| 671 |
ezxml_t ezxml_parse_file(const char *file) |
| 672 |
{ |
| 673 |
int fd = open(file, O_RDONLY, 0); |
| 674 |
ezxml_t xml = ezxml_parse_fd(fd); |
| 675 |
|
| 676 |
if (fd >= 0) close(fd); |
| 677 |
return xml; |
| 678 |
} |
| 679 |
|
| 680 |
/* Encodes ampersand sequences appending the results to *dst, reallocating *dst */ |
| 681 |
/* if length excedes max. a is non-zero for attribute encoding. Returns *dst */ |
| 682 |
char *ezxml_ampencode(const char *s, size_t len, char **dst, size_t *dlen, |
| 683 |
size_t *max, short a) |
| 684 |
{ |
| 685 |
const char *e; |
| 686 |
|
| 687 |
for (e = s + len; s != e; s++) { |
| 688 |
while (*dlen + 10 > *max) *dst = realloc(*dst, *max += EZXML_BUFSIZE); |
| 689 |
|
| 690 |
switch (*s) { |
| 691 |
case '\0': return *dst; |
| 692 |
case '&': *dlen += sprintf(*dst + *dlen, "&"); break; |
| 693 |
case '<': *dlen += sprintf(*dst + *dlen, "<"); break; |
| 694 |
case '>': *dlen += sprintf(*dst + *dlen, ">"); break; |
| 695 |
case '"': *dlen += sprintf(*dst + *dlen, (a) ? """ : "\""); break; |
| 696 |
case '\n': *dlen += sprintf(*dst + *dlen, (a) ? "
" : "\n"); break; |
| 697 |
case '\t': *dlen += sprintf(*dst + *dlen, (a) ? "	" : "\t"); break; |
| 698 |
case '\r': *dlen += sprintf(*dst + *dlen, "
"); break; |
| 699 |
default: (*dst)[(*dlen)++] = *s; |
| 700 |
} |
| 701 |
} |
| 702 |
return *dst; |
| 703 |
} |
| 704 |
|
| 705 |
/* Recursively converts each tag to xml appending it to *s. Reallocates *s if */ |
| 706 |
/* its length excedes max. start is the location of the previous tag in the */ |
| 707 |
/* parent tag's character content. Returns *s. */ |
| 708 |
char *ezxml_toxml_r(ezxml_t xml, char **s, size_t *len, size_t *max, |
| 709 |
size_t start, char ***attr) |
| 710 |
{ |
| 711 |
int i, j; |
| 712 |
char *txt = (xml->parent) ? xml->parent->txt : ""; |
| 713 |
size_t off = 0; |
| 714 |
|
| 715 |
/* parent character content up to this tag */ |
| 716 |
*s = ezxml_ampencode(txt + start, xml->off - start, s, len, max, 0); |
| 717 |
|
| 718 |
while (*len + strlen(xml->name) + 4 > *max) /* reallocate s */ |
| 719 |
*s = realloc(*s, *max += EZXML_BUFSIZE); |
| 720 |
|
| 721 |
*len += sprintf(*s + *len, "<%s", xml->name); /* open tag */ |
| 722 |
for (i = 0; xml->attr[i]; i += 2) { /* tag attributes */ |
| 723 |
if (ezxml_attr(xml, xml->attr[i]) != xml->attr[i + 1]) continue; |
| 724 |
while (*len + strlen(xml->attr[i]) + 7 > *max) /* reallocate s */ |
| 725 |
*s = realloc(*s, *max += EZXML_BUFSIZE); |
| 726 |
|
| 727 |
*len += sprintf(*s + *len, " %s=\"", xml->attr[i]); |
| 728 |
ezxml_ampencode(xml->attr[i + 1], -1, s, len, max, 1); |
| 729 |
*len += sprintf(*s + *len, "\""); |
| 730 |
} |
| 731 |
|
| 732 |
for (i = 0; attr[i] && strcmp(attr[i][0], xml->name); i++); |
| 733 |
for (j = 1; attr[i] && attr[i][j]; j += 3) { /* default attributes */ |
| 734 |
if (! attr[i][j + 1] || ezxml_attr(xml, attr[i][j]) != attr[i][j + 1]) |
| 735 |
continue; /* skip duplicates and non-values */ |
| 736 |
while (*len + strlen(attr[i][j]) + 7 > *max) /* reallocate s */ |
| 737 |
*s = realloc(*s, *max += EZXML_BUFSIZE); |
| 738 |
|
| 739 |
*len += sprintf(*s + *len, " %s=\"", attr[i][j]); |
| 740 |
ezxml_ampencode(attr[i][j + 1], -1, s, len, max, 1); |
| 741 |
*len += sprintf(*s + *len, "\""); |
| 742 |
} |
| 743 |
*len += sprintf(*s + *len, ">"); |
| 744 |
|
| 745 |
*s = (xml->child) ? ezxml_toxml_r(xml->child, s, len, max, 0, attr) /*child */ |
| 746 |
: ezxml_ampencode(xml->txt, -1, s, len, max, 0); /*data */ |
| 747 |
|
| 748 |
while (*len + strlen(xml->name) + 4 > *max) /* reallocate s */ |
| 749 |
*s = realloc(*s, *max += EZXML_BUFSIZE); |
| 750 |
|
| 751 |
*len += sprintf(*s + *len, "</%s>", xml->name); /* close tag */ |
| 752 |
|
| 753 |
while (txt[off] && off < xml->off) off++; /* make sure off is within bounds */ |
| 754 |
return (xml->ordered) ? ezxml_toxml_r(xml->ordered, s, len, max, off, attr) |
| 755 |
: ezxml_ampencode(txt + off, -1, s, len, max, 0); |
| 756 |
} |
| 757 |
|
| 758 |
/* Converts an ezxml structure back to xml. Returns a string of xml data that */ |
| 759 |
/* must be freed. */ |
| 760 |
char *ezxml_toxml(ezxml_t xml) |
| 761 |
{ |
| 762 |
ezxml_t p = (xml) ? xml->parent : NULL, o = (xml) ? xml->ordered : NULL; |
| 763 |
ezxml_root_t root = (ezxml_root_t)xml; |
| 764 |
size_t len = 0, max = EZXML_BUFSIZE; |
| 765 |
char *s = strcpy(malloc(max), ""), *t, *n; |
| 766 |
int i, j, k; |
| 767 |
|
| 768 |
if (! xml || ! xml->name) return realloc(s, len + 1); |
| 769 |
while (root->xml.parent) root = (ezxml_root_t)root->xml.parent; /* root tag */ |
| 770 |
|
| 771 |
for (i = 0; ! p && root->pi[i]; i++) { /* pre-root processing instructions */ |
| 772 |
for (k = 2; root->pi[i][k - 1]; k++); |
| 773 |
for (j = 1; (n = root->pi[i][j]); j++) { |
| 774 |
if (root->pi[i][k][j - 1] == '>') continue; /* not pre-root */ |
| 775 |
while (len + strlen(t = root->pi[i][0]) + strlen(n) + 7 > max) |
| 776 |
s = realloc(s, max += EZXML_BUFSIZE); |
| 777 |
len += sprintf(s + len, "<?%s%s%s?>\n", t, *n ? " " : "", n); |
| 778 |
} |
| 779 |
} |
| 780 |
|
| 781 |
xml->parent = xml->ordered = NULL; |
| 782 |
s = ezxml_toxml_r(xml, &s, &len, &max, 0, root->attr); |
| 783 |
xml->parent = p; |
| 784 |
xml->ordered = o; |
| 785 |
|
| 786 |
for (i = 0; ! p && root->pi[i]; i++) { /* post-root processing instructions */ |
| 787 |
for (k = 2; root->pi[i][k - 1]; k++); |
| 788 |
for (j = 1; (n = root->pi[i][j]); j++) { |
| 789 |
if (root->pi[i][k][j - 1] == '<') continue; /* not post-root */ |
| 790 |
while (len + strlen(t = root->pi[i][0]) + strlen(n) + 7 > max) |
| 791 |
s = realloc(s, max += EZXML_BUFSIZE); |
| 792 |
len += sprintf(s + len, "\n<?%s%s%s?>", t, *n ? " " : "", n); |
| 793 |
} |
| 794 |
} |
| 795 |
return realloc(s, len + 1); |
| 796 |
} |
| 797 |
|
| 798 |
/* free the memory allocated for the ezxml structure */ |
| 799 |
void ezxml_free(ezxml_t xml) |
| 800 |
{ |
| 801 |
ezxml_root_t root = (ezxml_root_t)xml; |
| 802 |
int i, j; |
| 803 |
char **a, *s; |
| 804 |
|
| 805 |
if (! xml) return; |
| 806 |
ezxml_free(xml->child); |
| 807 |
ezxml_free(xml->ordered); |
| 808 |
|
| 809 |
if (! xml->parent) { /* free root tag allocations */ |
| 810 |
for (i = 10; root->ent[i]; i += 2) /* 0 - 9 are default entites (<>&"') */ |
| 811 |
if ((s = root->ent[i + 1]) < root->s || s > root->e) free(s); |
| 812 |
free(root->ent); /* free list of general entities */ |
| 813 |
|
| 814 |
for (i = 0; (a = root->attr[i]); i++) { |
| 815 |
for (j = 1; a[j++]; j += 2) /* free malloced attribute values */ |
| 816 |
if (a[j] && (a[j] < root->s || a[j] > root->e)) free(a[j]); |
| 817 |
free(a); |
| 818 |
} |
| 819 |
if (root->attr[0]) free(root->attr); /* free default attribute list */ |
| 820 |
|
| 821 |
for (i = 0; root->pi[i]; i++) { |
| 822 |
for (j = 1; root->pi[i][j]; j++); |
| 823 |
free(root->pi[i][j + 1]); |
| 824 |
free(root->pi[i]); |
| 825 |
} |
| 826 |
if (root->pi[0]) free(root->pi); /* free processing instructions */ |
| 827 |
|
| 828 |
if (root->len == -1) free(root->m); /* malloced xml data */ |
| 829 |
#ifndef EZXML_NOMMAP |
| 830 |
else if (root->len) munmap(root->m, root->len); /* mem mapped xml data */ |
| 831 |
#endif /* EZXML_NOMMAP */ |
| 832 |
if (root->u) free(root->u); /* utf8 conversion */ |
| 833 |
} |
| 834 |
|
| 835 |
ezxml_free_attr(xml->attr); /* tag attributes */ |
| 836 |
if ((xml->flags & EZXML_TXTM)) free(xml->txt); /* character content */ |
| 837 |
if ((xml->flags & EZXML_NAMEM)) free(xml->name); /* tag name */ |
| 838 |
free(xml); |
| 839 |
} |
| 840 |
|
| 841 |
/* return parser error message or empty string if none */ |
| 842 |
const char *ezxml_error(ezxml_t xml) |
| 843 |
{ |
| 844 |
while (xml && xml->parent) xml = xml->parent; /* find root tag */ |
| 845 |
return (xml) ? ((ezxml_root_t)xml)->err : ""; |
| 846 |
} |
| 847 |
|
| 848 |
/* returns a new empty ezxml structure with the given root tag name */ |
| 849 |
ezxml_t ezxml_new(const char *name) |
| 850 |
{ |
| 851 |
static char *ent[] = { "lt;", "<", "gt;", ">", "quot;", """, |
| 852 |
"apos;", "'", "amp;", "&", NULL }; |
| 853 |
ezxml_root_t root = (ezxml_root_t)memset(malloc(sizeof(struct ezxml_root)), |
| 854 |
'\0', sizeof(struct ezxml_root)); |
| 855 |
root->xml.name = (char *)name; |
| 856 |
root->cur = &root->xml; |
| 857 |
strcpy(root->err, root->xml.txt = ""); |
| 858 |
root->ent = memcpy(malloc(sizeof(ent)), ent, sizeof(ent)); |
| 859 |
root->attr = root->pi = (char ***)(root->xml.attr = EZXML_NIL); |
| 860 |
return &root->xml; |
| 861 |
} |
| 862 |
|
| 863 |
/* inserts an existing tag into an ezxml structure */ |
| 864 |
ezxml_t ezxml_insert(ezxml_t xml, ezxml_t dest, size_t off) |
| 865 |
{ |
| 866 |
ezxml_t cur, prev, head; |
| 867 |
|
| 868 |
xml->next = xml->sibling = xml->ordered = NULL; |
| 869 |
xml->off = off; |
| 870 |
xml->parent = dest; |
| 871 |
|
| 872 |
if ((head = dest->child)) { /* already have sub tags */ |
| 873 |
if (head->off <= off) { /* not first subtag */ |
| 874 |
for (cur = head; cur->ordered && cur->ordered->off <= off; |
| 875 |
cur = cur->ordered); |
| 876 |
xml->ordered = cur->ordered; |
| 877 |
cur->ordered = xml; |
| 878 |
} |
| 879 |
else { /* first subtag */ |
| 880 |
xml->ordered = head; |
| 881 |
dest->child = xml; |
| 882 |
} |
| 883 |
|
| 884 |
for (cur = head, prev = NULL; cur && strcmp(cur->name, xml->name); |
| 885 |
prev = cur, cur = cur->sibling); /* find tag type */ |
| 886 |
if (cur && cur->off <= off) { /* not first of type */ |
| 887 |
while (cur->next && cur->next->off <= off) cur = cur->next; |
| 888 |
xml->next = cur->next; |
| 889 |
cur->next = xml; |
| 890 |
} |
| 891 |
else { /* first tag of this type */ |
| 892 |
if (prev && cur) prev->sibling = cur->sibling; /* remove old first */ |
| 893 |
xml->next = cur; /* old first tag is now next */ |
| 894 |
for (cur = head, prev = NULL; cur && cur->off <= off; |
| 895 |
prev = cur, cur = cur->sibling); /* new sibling insert point */ |
| 896 |
xml->sibling = cur; |
| 897 |
if (prev) prev->sibling = xml; |
| 898 |
} |
| 899 |
} |
| 900 |
else dest->child = xml; /* only sub tag */ |
| 901 |
|
| 902 |
return xml; |
| 903 |
} |
| 904 |
|
| 905 |
/* Adds a child tag. off is the offset of the child tag relative to the start */ |
| 906 |
/* of the parent tag's character content. Returns the child tag. */ |
| 907 |
ezxml_t ezxml_add_child(ezxml_t xml, const char *name, size_t off) |
| 908 |
{ |
| 909 |
ezxml_t child; |
| 910 |
|
| 911 |
if (! xml) return NULL; |
| 912 |
child = (ezxml_t)memset(malloc(sizeof(struct ezxml)), '\0', |
| 913 |
sizeof(struct ezxml)); |
| 914 |
child->name = (char *)name; |
| 915 |
child->attr = EZXML_NIL; |
| 916 |
child->txt = ""; |
| 917 |
|
| 918 |
return ezxml_insert(child, xml, off); |
| 919 |
} |
| 920 |
|
| 921 |
/* sets the character content for the given tag and returns the tag */ |
| 922 |
ezxml_t ezxml_set_txt(ezxml_t xml, const char *txt) |
| 923 |
{ |
| 924 |
if (! xml) return NULL; |
| 925 |
if (txt == xml->txt) return xml; |
| 926 |
if (xml->flags & EZXML_TXTM) free(xml->txt); /* existing txt was malloced */ |
| 927 |
xml->flags &= ~EZXML_TXTM; |
| 928 |
xml->txt = (char *)txt; |
| 929 |
return xml; |
| 930 |
} |
| 931 |
|
| 932 |
/* add text to the current character content, allocating memory as needed (GW) */ |
| 933 |
ezxml_t ezxml_add_txt(ezxml_t xml, const char *txt) |
| 934 |
{ |
| 935 |
int len; |
| 936 |
if (! xml) return NULL; |
| 937 |
if (!*txt) return xml; |
| 938 |
len = strlen(xml->txt) + strlen(txt) + 1; |
| 939 |
if (xml->flags & EZXML_TXTM) { |
| 940 |
xml->txt = (char *)realloc(xml->txt, len); |
| 941 |
} else { |
| 942 |
xml->txt = strcpy((char *)malloc(len), xml->txt); |
| 943 |
xml->flags |= EZXML_TXTM; |
| 944 |
} |
| 945 |
strcat(xml->txt, txt); |
| 946 |
return xml; |
| 947 |
} |
| 948 |
|
| 949 |
/* Sets the given tag attribute or adds a new attribute if not found. A value */ |
| 950 |
/* of NULL will remove the specified attribute. Returns the tag given. */ |
| 951 |
ezxml_t ezxml_set_attr(ezxml_t xml, const char *name, const char *value) |
| 952 |
{ |
| 953 |
int l = 0, c; |
| 954 |
|
| 955 |
if (! xml) return NULL; |
| 956 |
while (xml->attr[l] && strcmp(xml->attr[l], name)) l += 2; |
| 957 |
if (! xml->attr[l]) { /* not found, add as new attribute */ |
| 958 |
if (! value) return xml; /* nothing to do */ |
| 959 |
if (xml->attr == EZXML_NIL) { /* first attribute */ |
| 960 |
xml->attr = malloc(4 * sizeof(char *)); |
| 961 |
xml->attr[1] = strdup(""); /* empty list of malloced names/vals */ |
| 962 |
} |
| 963 |
else xml->attr = realloc(xml->attr, (l + 4) * sizeof(char *)); |
| 964 |
|
| 965 |
xml->attr[l] = (char *)name; /* set attribute name */ |
| 966 |
xml->attr[l + 2] = NULL; /* null terminate attribute list */ |
| 967 |
xml->attr[l + 3] = realloc(xml->attr[l + 1], |
| 968 |
(c = strlen(xml->attr[l + 1])) + 2); |
| 969 |
strcpy(xml->attr[l + 3] + c, " "); /* set name/value as not malloced */ |
| 970 |
if (xml->flags & EZXML_DUP) xml->attr[l + 3][c] = EZXML_NAMEM; |
| 971 |
} |
| 972 |
else if (xml->flags & EZXML_DUP) free((char *)name); /* name was strduped */ |
| 973 |
|
| 974 |
for (c = l; xml->attr[c]; c += 2); /* find end of attribute list */ |
| 975 |
if (xml->attr[c + 1][l / 2] & EZXML_TXTM) free(xml->attr[l + 1]); /*old val */ |
| 976 |
if (xml->flags & EZXML_DUP) xml->attr[c + 1][l / 2] |= EZXML_TXTM; |
| 977 |
else xml->attr[c + 1][l / 2] &= ~EZXML_TXTM; |
| 978 |
|
| 979 |
if (value) xml->attr[l + 1] = (char *)value; /* set attribute value */ |
| 980 |
else { /* remove attribute */ |
| 981 |
if (xml->attr[c + 1][l / 2] & EZXML_NAMEM) free(xml->attr[l]); |
| 982 |
memmove(xml->attr + l, xml->attr + l + 2, (c - l + 2) * sizeof(char*)); |
| 983 |
xml->attr = realloc(xml->attr, (c + 2) * sizeof(char *)); |
| 984 |
memmove(xml->attr[c + 1] + (l / 2), xml->attr[c + 1] + (l / 2) + 1, |
| 985 |
(c / 2) - (l / 2)); /* fix list of which name/vals are malloced */ |
| 986 |
} |
| 987 |
xml->flags &= ~EZXML_DUP; /* clear strdup() flag */ |
| 988 |
return xml; |
| 989 |
} |
| 990 |
|
| 991 |
/* sets a flag for the given tag and returns the tag */ |
| 992 |
ezxml_t ezxml_set_flag(ezxml_t xml, short flag) |
| 993 |
{ |
| 994 |
if (xml) xml->flags |= flag; |
| 995 |
return xml; |
| 996 |
} |
| 997 |
|
| 998 |
/* removes a tag along with its subtags without freeing its memory */ |
| 999 |
ezxml_t ezxml_cut(ezxml_t xml) |
| 1000 |
{ |
| 1001 |
ezxml_t cur; |
| 1002 |
|
| 1003 |
if (! xml) return NULL; /* nothing to do */ |
| 1004 |
if (xml->next) xml->next->sibling = xml->sibling; /* patch sibling list */ |
| 1005 |
|
| 1006 |
if (xml->parent) { /* not root tag */ |
| 1007 |
cur = xml->parent->child; /* find head of subtag list */ |
| 1008 |
if (cur == xml) xml->parent->child = xml->ordered; /* first subtag */ |
| 1009 |
else { /* not first subtag */ |
| 1010 |
while (cur->ordered != xml) cur = cur->ordered; |
| 1011 |
cur->ordered = cur->ordered->ordered; /* patch ordered list */ |
| 1012 |
|
| 1013 |
cur = xml->parent->child; /* go back to head of subtag list */ |
| 1014 |
if (strcmp(cur->name, xml->name)) { /* not in first sibling list */ |
| 1015 |
while (strcmp(cur->sibling->name, xml->name)) |
| 1016 |
cur = cur->sibling; |
| 1017 |
if (cur->sibling == xml) { /* first of a sibling list */ |
| 1018 |
cur->sibling = (xml->next) ? xml->next |
| 1019 |
: cur->sibling->sibling; |
| 1020 |
} |
| 1021 |
else cur = cur->sibling; /* not first of a sibling list */ |
| 1022 |
} |
| 1023 |
|
| 1024 |
while (cur->next && cur->next != xml) cur = cur->next; |
| 1025 |
if (cur->next) cur->next = cur->next->next; /* patch next list */ |
| 1026 |
} |
| 1027 |
} |
| 1028 |
xml->ordered = xml->sibling = xml->next = NULL; |
| 1029 |
return xml; |
| 1030 |
} |
| 1031 |
|
| 1032 |
#ifdef EZXML_TEST /* test harness */ |
| 1033 |
int main(int argc, char **argv) |
| 1034 |
{ |
| 1035 |
ezxml_t xml; |
| 1036 |
char *s; |
| 1037 |
int i; |
| 1038 |
|
| 1039 |
if (argc != 2) return fprintf(stderr, "usage: %s xmlfile\n", argv[0]); |
| 1040 |
|
| 1041 |
xml = ezxml_parse_file(argv[1]); |
| 1042 |
printf("%s\n", (s = ezxml_toxml(xml))); |
| 1043 |
free(s); |
| 1044 |
i = fprintf(stderr, "%s", ezxml_error(xml)); |
| 1045 |
ezxml_free(xml); |
| 1046 |
return (i) ? 1 : 0; |
| 1047 |
} |
| 1048 |
#endif /* EZXML_TEST */ |
