root / logic / trunk / src / mxml / mxml-file.c @ 86
History | View | Annotate | Download (65.1 KB)
1 | 49 | Janez1 | /*
|
---|---|---|---|
2 | * "$Id: mxml-file.c 391 2009-05-17 05:20:52Z mike $"
|
||
3 | *
|
||
4 | * File loading code for Mini-XML, a small XML-like file parsing library.
|
||
5 | *
|
||
6 | * Copyright 2003-2009 by Michael Sweet.
|
||
7 | *
|
||
8 | * This program is free software; you can redistribute it and/or
|
||
9 | * modify it under the terms of the GNU Library General Public
|
||
10 | * License as published by the Free Software Foundation; either
|
||
11 | * version 2, or (at your option) any later version.
|
||
12 | *
|
||
13 | * This program is distributed in the hope that it will be useful,
|
||
14 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
||
15 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
||
16 | * GNU General Public License for more details.
|
||
17 | *
|
||
18 | * Contents:
|
||
19 | *
|
||
20 | * mxmlLoadFd() - Load a file descriptor into an XML node tree.
|
||
21 | * mxmlLoadFile() - Load a file into an XML node tree.
|
||
22 | * mxmlLoadString() - Load a string into an XML node tree.
|
||
23 | * mxmlSaveAllocString() - Save an XML node tree to an allocated string.
|
||
24 | * mxmlSaveFd() - Save an XML tree to a file descriptor.
|
||
25 | * mxmlSaveFile() - Save an XML tree to a file.
|
||
26 | * mxmlSaveString() - Save an XML node tree to a string.
|
||
27 | * mxmlSAXLoadFd() - Load a file descriptor into an XML node tree
|
||
28 | * using a SAX callback.
|
||
29 | * mxmlSAXLoadFile() - Load a file into an XML node tree
|
||
30 | * using a SAX callback.
|
||
31 | * mxmlSAXLoadString() - Load a string into an XML node tree
|
||
32 | * using a SAX callback.
|
||
33 | * mxmlSetCustomHandlers() - Set the handling functions for custom data.
|
||
34 | * mxmlSetErrorCallback() - Set the error message callback.
|
||
35 | * mxmlSetWrapMargin() - Set the the wrap margin when saving XML data.
|
||
36 | * mxml_add_char() - Add a character to a buffer, expanding as needed.
|
||
37 | * mxml_fd_getc() - Read a character from a file descriptor.
|
||
38 | * mxml_fd_putc() - Write a character to a file descriptor.
|
||
39 | * mxml_fd_read() - Read a buffer of data from a file descriptor.
|
||
40 | * mxml_fd_write() - Write a buffer of data to a file descriptor.
|
||
41 | * mxml_file_getc() - Get a character from a file.
|
||
42 | * mxml_file_putc() - Write a character to a file.
|
||
43 | * mxml_get_entity() - Get the character corresponding to an entity...
|
||
44 | * mxml_load_data() - Load data into an XML node tree.
|
||
45 | * mxml_parse_element() - Parse an element for any attributes...
|
||
46 | * mxml_string_getc() - Get a character from a string.
|
||
47 | * mxml_string_putc() - Write a character to a string.
|
||
48 | * mxml_write_name() - Write a name string.
|
||
49 | * mxml_write_node() - Save an XML node to a file.
|
||
50 | * mxml_write_string() - Write a string, escaping & and < as needed.
|
||
51 | * mxml_write_ws() - Do whitespace callback...
|
||
52 | */
|
||
53 | |||
54 | /*
|
||
55 | * Include necessary headers...
|
||
56 | */
|
||
57 | |||
58 | #ifndef WIN32
|
||
59 | # include <unistd.h> |
||
60 | #endif /* !WIN32 */ |
||
61 | #include "mxml-private.h" |
||
62 | |||
63 | |||
64 | /*
|
||
65 | * Character encoding...
|
||
66 | */
|
||
67 | |||
68 | #define ENCODE_UTF8 0 /* UTF-8 */ |
||
69 | #define ENCODE_UTF16BE 1 /* UTF-16 Big-Endian */ |
||
70 | #define ENCODE_UTF16LE 2 /* UTF-16 Little-Endian */ |
||
71 | |||
72 | |||
73 | /*
|
||
74 | * Macro to test for a bad XML character...
|
||
75 | */
|
||
76 | |||
77 | #define mxml_bad_char(ch) ((ch) < ' ' && (ch) != '\n' && (ch) != '\r' && (ch) != '\t') |
||
78 | |||
79 | |||
80 | /*
|
||
81 | * Types and structures...
|
||
82 | */
|
||
83 | |||
84 | typedef int (*_mxml_getc_cb_t)(void *, int *); |
||
85 | typedef int (*_mxml_putc_cb_t)(int, void *); |
||
86 | |||
87 | typedef struct _mxml_fdbuf_s /**** File descriptor buffer ****/ |
||
88 | { |
||
89 | int fd; /* File descriptor */ |
||
90 | unsigned char *current, /* Current position in buffer */ |
||
91 | *end, /* End of buffer */
|
||
92 | buffer[8192]; /* Character buffer */ |
||
93 | } _mxml_fdbuf_t; |
||
94 | |||
95 | |||
96 | /*
|
||
97 | * Local functions...
|
||
98 | */
|
||
99 | |||
100 | static int mxml_add_char(int ch, char **ptr, char **buffer, |
||
101 | int *bufsize);
|
||
102 | static int mxml_fd_getc(void *p, int *encoding); |
||
103 | static int mxml_fd_putc(int ch, void *p); |
||
104 | static int mxml_fd_read(_mxml_fdbuf_t *buf); |
||
105 | static int mxml_fd_write(_mxml_fdbuf_t *buf); |
||
106 | static int mxml_file_getc(void *p, int *encoding); |
||
107 | static int mxml_file_putc(int ch, void *p); |
||
108 | static int mxml_get_entity(mxml_node_t *parent, void *p, |
||
109 | int *encoding,
|
||
110 | _mxml_getc_cb_t getc_cb); |
||
111 | static inline int mxml_isspace(int ch) |
||
112 | { |
||
113 | return (ch == ' ' || ch == '\t' || ch == '\r' || |
||
114 | ch == '\n');
|
||
115 | } |
||
116 | static mxml_node_t *mxml_load_data(mxml_node_t *top, void *p, |
||
117 | mxml_load_cb_t cb, |
||
118 | _mxml_getc_cb_t getc_cb, |
||
119 | mxml_sax_cb_t sax_cb, void *sax_data);
|
||
120 | static int mxml_parse_element(mxml_node_t *node, void *p, |
||
121 | int *encoding,
|
||
122 | _mxml_getc_cb_t getc_cb); |
||
123 | static int mxml_string_getc(void *p, int *encoding); |
||
124 | static int mxml_string_putc(int ch, void *p); |
||
125 | static int mxml_write_name(const char *s, void *p, |
||
126 | _mxml_putc_cb_t putc_cb); |
||
127 | static int mxml_write_node(mxml_node_t *node, void *p, |
||
128 | mxml_save_cb_t cb, int col,
|
||
129 | _mxml_putc_cb_t putc_cb, |
||
130 | _mxml_global_t *global); |
||
131 | static int mxml_write_string(const char *s, void *p, |
||
132 | _mxml_putc_cb_t putc_cb); |
||
133 | static int mxml_write_ws(mxml_node_t *node, void *p, |
||
134 | mxml_save_cb_t cb, int ws,
|
||
135 | int col, _mxml_putc_cb_t putc_cb);
|
||
136 | |||
137 | |||
138 | /*
|
||
139 | * 'mxmlLoadFd()' - Load a file descriptor into an XML node tree.
|
||
140 | *
|
||
141 | * The nodes in the specified file are added to the specified top node.
|
||
142 | * If no top node is provided, the XML file MUST be well-formed with a
|
||
143 | * single parent node like <?xml> for the entire file. The callback
|
||
144 | * function returns the value type that should be used for child nodes.
|
||
145 | * If MXML_NO_CALLBACK is specified then all child nodes will be either
|
||
146 | * MXML_ELEMENT or MXML_TEXT nodes.
|
||
147 | *
|
||
148 | * The constants MXML_INTEGER_CALLBACK, MXML_OPAQUE_CALLBACK,
|
||
149 | * MXML_REAL_CALLBACK, and MXML_TEXT_CALLBACK are defined for loading
|
||
150 | * child nodes of the specified type.
|
||
151 | */
|
||
152 | |||
153 | mxml_node_t * /* O - First node or NULL if the file could not be read. */
|
||
154 | mxmlLoadFd(mxml_node_t *top, /* I - Top node */
|
||
155 | int fd, /* I - File descriptor to read from */ |
||
156 | mxml_load_cb_t cb) /* I - Callback function or MXML_NO_CALLBACK */
|
||
157 | { |
||
158 | _mxml_fdbuf_t buf; /* File descriptor buffer */
|
||
159 | |||
160 | |||
161 | /*
|
||
162 | * Initialize the file descriptor buffer...
|
||
163 | */
|
||
164 | |||
165 | buf.fd = fd; |
||
166 | buf.current = buf.buffer; |
||
167 | buf.end = buf.buffer; |
||
168 | |||
169 | /*
|
||
170 | * Read the XML data...
|
||
171 | */
|
||
172 | |||
173 | return (mxml_load_data(top, &buf, cb, mxml_fd_getc, MXML_NO_CALLBACK, NULL)); |
||
174 | } |
||
175 | |||
176 | |||
177 | /*
|
||
178 | * 'mxmlLoadFile()' - Load a file into an XML node tree.
|
||
179 | *
|
||
180 | * The nodes in the specified file are added to the specified top node.
|
||
181 | * If no top node is provided, the XML file MUST be well-formed with a
|
||
182 | * single parent node like <?xml> for the entire file. The callback
|
||
183 | * function returns the value type that should be used for child nodes.
|
||
184 | * If MXML_NO_CALLBACK is specified then all child nodes will be either
|
||
185 | * MXML_ELEMENT or MXML_TEXT nodes.
|
||
186 | *
|
||
187 | * The constants MXML_INTEGER_CALLBACK, MXML_OPAQUE_CALLBACK,
|
||
188 | * MXML_REAL_CALLBACK, and MXML_TEXT_CALLBACK are defined for loading
|
||
189 | * child nodes of the specified type.
|
||
190 | */
|
||
191 | |||
192 | mxml_node_t * /* O - First node or NULL if the file could not be read. */
|
||
193 | mxmlLoadFile(mxml_node_t *top, /* I - Top node */
|
||
194 | FILE *fp, /* I - File to read from */
|
||
195 | mxml_load_cb_t cb) /* I - Callback function or MXML_NO_CALLBACK */
|
||
196 | { |
||
197 | /*
|
||
198 | * Read the XML data...
|
||
199 | */
|
||
200 | |||
201 | return (mxml_load_data(top, fp, cb, mxml_file_getc, MXML_NO_CALLBACK, NULL)); |
||
202 | } |
||
203 | |||
204 | |||
205 | /*
|
||
206 | * 'mxmlLoadString()' - Load a string into an XML node tree.
|
||
207 | *
|
||
208 | * The nodes in the specified string are added to the specified top node.
|
||
209 | * If no top node is provided, the XML string MUST be well-formed with a
|
||
210 | * single parent node like <?xml> for the entire string. The callback
|
||
211 | * function returns the value type that should be used for child nodes.
|
||
212 | * If MXML_NO_CALLBACK is specified then all child nodes will be either
|
||
213 | * MXML_ELEMENT or MXML_TEXT nodes.
|
||
214 | *
|
||
215 | * The constants MXML_INTEGER_CALLBACK, MXML_OPAQUE_CALLBACK,
|
||
216 | * MXML_REAL_CALLBACK, and MXML_TEXT_CALLBACK are defined for loading
|
||
217 | * child nodes of the specified type.
|
||
218 | */
|
||
219 | |||
220 | mxml_node_t * /* O - First node or NULL if the string has errors. */
|
||
221 | mxmlLoadString(mxml_node_t *top, /* I - Top node */
|
||
222 | const char *s, /* I - String to load */ |
||
223 | mxml_load_cb_t cb) /* I - Callback function or MXML_NO_CALLBACK */
|
||
224 | { |
||
225 | /*
|
||
226 | * Read the XML data...
|
||
227 | */
|
||
228 | |||
229 | return (mxml_load_data(top, (void *)&s, cb, mxml_string_getc, MXML_NO_CALLBACK, |
||
230 | NULL));
|
||
231 | } |
||
232 | |||
233 | |||
234 | /*
|
||
235 | * 'mxmlSaveAllocString()' - Save an XML node tree to an allocated string.
|
||
236 | *
|
||
237 | * This function returns a pointer to a string containing the textual
|
||
238 | * representation of the XML node tree. The string should be freed
|
||
239 | * using the free() function when you are done with it. NULL is returned
|
||
240 | * if the node would produce an empty string or if the string cannot be
|
||
241 | * allocated.
|
||
242 | *
|
||
243 | * The callback argument specifies a function that returns a whitespace
|
||
244 | * string or NULL before and after each element. If MXML_NO_CALLBACK
|
||
245 | * is specified, whitespace will only be added before MXML_TEXT nodes
|
||
246 | * with leading whitespace and before attribute names inside opening
|
||
247 | * element tags.
|
||
248 | */
|
||
249 | |||
250 | char * /* O - Allocated string or NULL */ |
||
251 | mxmlSaveAllocString( |
||
252 | mxml_node_t *node, /* I - Node to write */
|
||
253 | mxml_save_cb_t cb) /* I - Whitespace callback or MXML_NO_CALLBACK */
|
||
254 | { |
||
255 | int bytes; /* Required bytes */ |
||
256 | char buffer[8192]; /* Temporary buffer */ |
||
257 | char *s; /* Allocated string */ |
||
258 | |||
259 | |||
260 | /*
|
||
261 | * Write the node to the temporary buffer...
|
||
262 | */
|
||
263 | |||
264 | bytes = mxmlSaveString(node, buffer, sizeof(buffer), cb);
|
||
265 | |||
266 | if (bytes <= 0) |
||
267 | return (NULL); |
||
268 | |||
269 | if (bytes < (int)(sizeof(buffer) - 1)) |
||
270 | { |
||
271 | /*
|
||
272 | * Node fit inside the buffer, so just duplicate that string and
|
||
273 | * return...
|
||
274 | */
|
||
275 | |||
276 | return (strdup(buffer));
|
||
277 | } |
||
278 | |||
279 | /*
|
||
280 | * Allocate a buffer of the required size and save the node to the
|
||
281 | * new buffer...
|
||
282 | */
|
||
283 | |||
284 | if ((s = malloc(bytes + 1)) == NULL) |
||
285 | return (NULL); |
||
286 | |||
287 | mxmlSaveString(node, s, bytes + 1, cb);
|
||
288 | |||
289 | /*
|
||
290 | * Return the allocated string...
|
||
291 | */
|
||
292 | |||
293 | return (s);
|
||
294 | } |
||
295 | |||
296 | |||
297 | /*
|
||
298 | * 'mxmlSaveFd()' - Save an XML tree to a file descriptor.
|
||
299 | *
|
||
300 | * The callback argument specifies a function that returns a whitespace
|
||
301 | * string or NULL before and after each element. If MXML_NO_CALLBACK
|
||
302 | * is specified, whitespace will only be added before MXML_TEXT nodes
|
||
303 | * with leading whitespace and before attribute names inside opening
|
||
304 | * element tags.
|
||
305 | */
|
||
306 | |||
307 | int /* O - 0 on success, -1 on error. */ |
||
308 | mxmlSaveFd(mxml_node_t *node, /* I - Node to write */
|
||
309 | int fd, /* I - File descriptor to write to */ |
||
310 | mxml_save_cb_t cb) /* I - Whitespace callback or MXML_NO_CALLBACK */
|
||
311 | { |
||
312 | int col; /* Final column */ |
||
313 | _mxml_fdbuf_t buf; /* File descriptor buffer */
|
||
314 | _mxml_global_t *global = _mxml_global(); |
||
315 | /* Global data */
|
||
316 | |||
317 | |||
318 | /*
|
||
319 | * Initialize the file descriptor buffer...
|
||
320 | */
|
||
321 | |||
322 | buf.fd = fd; |
||
323 | buf.current = buf.buffer; |
||
324 | buf.end = buf.buffer + sizeof(buf.buffer);
|
||
325 | |||
326 | /*
|
||
327 | * Write the node...
|
||
328 | */
|
||
329 | |||
330 | if ((col = mxml_write_node(node, &buf, cb, 0, mxml_fd_putc, global)) < 0) |
||
331 | return (-1); |
||
332 | |||
333 | if (col > 0) |
||
334 | if (mxml_fd_putc('\n', &buf) < 0) |
||
335 | return (-1); |
||
336 | |||
337 | /*
|
||
338 | * Flush and return...
|
||
339 | */
|
||
340 | |||
341 | return (mxml_fd_write(&buf));
|
||
342 | } |
||
343 | |||
344 | |||
345 | /*
|
||
346 | * 'mxmlSaveFile()' - Save an XML tree to a file.
|
||
347 | *
|
||
348 | * The callback argument specifies a function that returns a whitespace
|
||
349 | * string or NULL before and after each element. If MXML_NO_CALLBACK
|
||
350 | * is specified, whitespace will only be added before MXML_TEXT nodes
|
||
351 | * with leading whitespace and before attribute names inside opening
|
||
352 | * element tags.
|
||
353 | */
|
||
354 | |||
355 | int /* O - 0 on success, -1 on error. */ |
||
356 | mxmlSaveFile(mxml_node_t *node, /* I - Node to write */
|
||
357 | FILE *fp, /* I - File to write to */
|
||
358 | mxml_save_cb_t cb) /* I - Whitespace callback or MXML_NO_CALLBACK */
|
||
359 | { |
||
360 | int col; /* Final column */ |
||
361 | _mxml_global_t *global = _mxml_global(); |
||
362 | /* Global data */
|
||
363 | |||
364 | |||
365 | /*
|
||
366 | * Write the node...
|
||
367 | */
|
||
368 | |||
369 | if ((col = mxml_write_node(node, fp, cb, 0, mxml_file_putc, global)) < 0) |
||
370 | return (-1); |
||
371 | |||
372 | if (col > 0) |
||
373 | if (putc('\n', fp) < 0) |
||
374 | return (-1); |
||
375 | |||
376 | /*
|
||
377 | * Return 0 (success)...
|
||
378 | */
|
||
379 | |||
380 | return (0); |
||
381 | } |
||
382 | |||
383 | |||
384 | /*
|
||
385 | * 'mxmlSaveString()' - Save an XML node tree to a string.
|
||
386 | *
|
||
387 | * This function returns the total number of bytes that would be
|
||
388 | * required for the string but only copies (bufsize - 1) characters
|
||
389 | * into the specified buffer.
|
||
390 | *
|
||
391 | * The callback argument specifies a function that returns a whitespace
|
||
392 | * string or NULL before and after each element. If MXML_NO_CALLBACK
|
||
393 | * is specified, whitespace will only be added before MXML_TEXT nodes
|
||
394 | * with leading whitespace and before attribute names inside opening
|
||
395 | * element tags.
|
||
396 | */
|
||
397 | |||
398 | int /* O - Size of string */ |
||
399 | mxmlSaveString(mxml_node_t *node, /* I - Node to write */
|
||
400 | char *buffer, /* I - String buffer */ |
||
401 | int bufsize, /* I - Size of string buffer */ |
||
402 | mxml_save_cb_t cb) /* I - Whitespace callback or MXML_NO_CALLBACK */
|
||
403 | { |
||
404 | int col; /* Final column */ |
||
405 | char *ptr[2]; /* Pointers for putc_cb */ |
||
406 | _mxml_global_t *global = _mxml_global(); |
||
407 | /* Global data */
|
||
408 | |||
409 | |||
410 | /*
|
||
411 | * Write the node...
|
||
412 | */
|
||
413 | |||
414 | ptr[0] = buffer;
|
||
415 | ptr[1] = buffer + bufsize;
|
||
416 | |||
417 | if ((col = mxml_write_node(node, ptr, cb, 0, mxml_string_putc, global)) < 0) |
||
418 | return (-1); |
||
419 | |||
420 | if (col > 0) |
||
421 | mxml_string_putc('\n', ptr);
|
||
422 | |||
423 | /*
|
||
424 | * Nul-terminate the buffer...
|
||
425 | */
|
||
426 | |||
427 | if (ptr[0] >= ptr[1]) |
||
428 | buffer[bufsize - 1] = '\0'; |
||
429 | else
|
||
430 | ptr[0][0] = '\0'; |
||
431 | |||
432 | /*
|
||
433 | * Return the number of characters...
|
||
434 | */
|
||
435 | |||
436 | return (ptr[0] - buffer); |
||
437 | } |
||
438 | |||
439 | |||
440 | /*
|
||
441 | * 'mxmlSAXLoadFd()' - Load a file descriptor into an XML node tree
|
||
442 | * using a SAX callback.
|
||
443 | *
|
||
444 | * The nodes in the specified file are added to the specified top node.
|
||
445 | * If no top node is provided, the XML file MUST be well-formed with a
|
||
446 | * single parent node like <?xml> for the entire file. The callback
|
||
447 | * function returns the value type that should be used for child nodes.
|
||
448 | * If MXML_NO_CALLBACK is specified then all child nodes will be either
|
||
449 | * MXML_ELEMENT or MXML_TEXT nodes.
|
||
450 | *
|
||
451 | * The constants MXML_INTEGER_CALLBACK, MXML_OPAQUE_CALLBACK,
|
||
452 | * MXML_REAL_CALLBACK, and MXML_TEXT_CALLBACK are defined for loading
|
||
453 | * child nodes of the specified type.
|
||
454 | *
|
||
455 | * The SAX callback must call mxmlRetain() for any nodes that need to
|
||
456 | * be kept for later use. Otherwise, nodes are deleted when the parent
|
||
457 | * node is closed or after each data, comment, CDATA, or directive node.
|
||
458 | *
|
||
459 | * @since Mini-XML 2.3@
|
||
460 | */
|
||
461 | |||
462 | mxml_node_t * /* O - First node or NULL if the file could not be read. */
|
||
463 | mxmlSAXLoadFd(mxml_node_t *top, /* I - Top node */
|
||
464 | int fd, /* I - File descriptor to read from */ |
||
465 | mxml_load_cb_t cb, /* I - Callback function or MXML_NO_CALLBACK */
|
||
466 | mxml_sax_cb_t sax_cb, /* I - SAX callback or MXML_NO_CALLBACK */
|
||
467 | void *sax_data) /* I - SAX user data */ |
||
468 | { |
||
469 | _mxml_fdbuf_t buf; /* File descriptor buffer */
|
||
470 | |||
471 | |||
472 | /*
|
||
473 | * Initialize the file descriptor buffer...
|
||
474 | */
|
||
475 | |||
476 | buf.fd = fd; |
||
477 | buf.current = buf.buffer; |
||
478 | buf.end = buf.buffer; |
||
479 | |||
480 | /*
|
||
481 | * Read the XML data...
|
||
482 | */
|
||
483 | |||
484 | return (mxml_load_data(top, &buf, cb, mxml_fd_getc, sax_cb, sax_data));
|
||
485 | } |
||
486 | |||
487 | |||
488 | /*
|
||
489 | * 'mxmlSAXLoadFile()' - Load a file into an XML node tree
|
||
490 | * using a SAX callback.
|
||
491 | *
|
||
492 | * The nodes in the specified file are added to the specified top node.
|
||
493 | * If no top node is provided, the XML file MUST be well-formed with a
|
||
494 | * single parent node like <?xml> for the entire file. The callback
|
||
495 | * function returns the value type that should be used for child nodes.
|
||
496 | * If MXML_NO_CALLBACK is specified then all child nodes will be either
|
||
497 | * MXML_ELEMENT or MXML_TEXT nodes.
|
||
498 | *
|
||
499 | * The constants MXML_INTEGER_CALLBACK, MXML_OPAQUE_CALLBACK,
|
||
500 | * MXML_REAL_CALLBACK, and MXML_TEXT_CALLBACK are defined for loading
|
||
501 | * child nodes of the specified type.
|
||
502 | *
|
||
503 | * The SAX callback must call mxmlRetain() for any nodes that need to
|
||
504 | * be kept for later use. Otherwise, nodes are deleted when the parent
|
||
505 | * node is closed or after each data, comment, CDATA, or directive node.
|
||
506 | *
|
||
507 | * @since Mini-XML 2.3@
|
||
508 | */
|
||
509 | |||
510 | mxml_node_t * /* O - First node or NULL if the file could not be read. */
|
||
511 | mxmlSAXLoadFile( |
||
512 | mxml_node_t *top, /* I - Top node */
|
||
513 | FILE *fp, /* I - File to read from */
|
||
514 | mxml_load_cb_t cb, /* I - Callback function or MXML_NO_CALLBACK */
|
||
515 | mxml_sax_cb_t sax_cb, /* I - SAX callback or MXML_NO_CALLBACK */
|
||
516 | void *sax_data) /* I - SAX user data */ |
||
517 | { |
||
518 | /*
|
||
519 | * Read the XML data...
|
||
520 | */
|
||
521 | |||
522 | return (mxml_load_data(top, fp, cb, mxml_file_getc, sax_cb, sax_data));
|
||
523 | } |
||
524 | |||
525 | |||
526 | /*
|
||
527 | * 'mxmlSAXLoadString()' - Load a string into an XML node tree
|
||
528 | * using a SAX callback.
|
||
529 | *
|
||
530 | * The nodes in the specified string are added to the specified top node.
|
||
531 | * If no top node is provided, the XML string MUST be well-formed with a
|
||
532 | * single parent node like <?xml> for the entire string. The callback
|
||
533 | * function returns the value type that should be used for child nodes.
|
||
534 | * If MXML_NO_CALLBACK is specified then all child nodes will be either
|
||
535 | * MXML_ELEMENT or MXML_TEXT nodes.
|
||
536 | *
|
||
537 | * The constants MXML_INTEGER_CALLBACK, MXML_OPAQUE_CALLBACK,
|
||
538 | * MXML_REAL_CALLBACK, and MXML_TEXT_CALLBACK are defined for loading
|
||
539 | * child nodes of the specified type.
|
||
540 | *
|
||
541 | * The SAX callback must call mxmlRetain() for any nodes that need to
|
||
542 | * be kept for later use. Otherwise, nodes are deleted when the parent
|
||
543 | * node is closed or after each data, comment, CDATA, or directive node.
|
||
544 | *
|
||
545 | * @since Mini-XML 2.3@
|
||
546 | */
|
||
547 | |||
548 | mxml_node_t * /* O - First node or NULL if the string has errors. */
|
||
549 | mxmlSAXLoadString( |
||
550 | mxml_node_t *top, /* I - Top node */
|
||
551 | const char *s, /* I - String to load */ |
||
552 | mxml_load_cb_t cb, /* I - Callback function or MXML_NO_CALLBACK */
|
||
553 | mxml_sax_cb_t sax_cb, /* I - SAX callback or MXML_NO_CALLBACK */
|
||
554 | void *sax_data) /* I - SAX user data */ |
||
555 | { |
||
556 | /*
|
||
557 | * Read the XML data...
|
||
558 | */
|
||
559 | |||
560 | return (mxml_load_data(top, (void *)&s, cb, mxml_string_getc, sax_cb, sax_data)); |
||
561 | } |
||
562 | |||
563 | |||
564 | /*
|
||
565 | * 'mxmlSetCustomHandlers()' - Set the handling functions for custom data.
|
||
566 | *
|
||
567 | * The load function accepts a node pointer and a data string and must
|
||
568 | * return 0 on success and non-zero on error.
|
||
569 | *
|
||
570 | * The save function accepts a node pointer and must return a malloc'd
|
||
571 | * string on success and NULL on error.
|
||
572 | *
|
||
573 | */
|
||
574 | |||
575 | void
|
||
576 | mxmlSetCustomHandlers( |
||
577 | mxml_custom_load_cb_t load, /* I - Load function */
|
||
578 | mxml_custom_save_cb_t save) /* I - Save function */
|
||
579 | { |
||
580 | _mxml_global_t *global = _mxml_global(); |
||
581 | /* Global data */
|
||
582 | |||
583 | |||
584 | global->custom_load_cb = load; |
||
585 | global->custom_save_cb = save; |
||
586 | } |
||
587 | |||
588 | |||
589 | /*
|
||
590 | * 'mxmlSetErrorCallback()' - Set the error message callback.
|
||
591 | */
|
||
592 | |||
593 | void
|
||
594 | mxmlSetErrorCallback(mxml_error_cb_t cb)/* I - Error callback function */
|
||
595 | { |
||
596 | _mxml_global_t *global = _mxml_global(); |
||
597 | /* Global data */
|
||
598 | |||
599 | |||
600 | global->error_cb = cb; |
||
601 | } |
||
602 | |||
603 | |||
604 | /*
|
||
605 | * 'mxmlSetWrapMargin()' - Set the the wrap margin when saving XML data.
|
||
606 | *
|
||
607 | * Wrapping is disabled when "column" is 0.
|
||
608 | *
|
||
609 | * @since Mini-XML 2.3@
|
||
610 | */
|
||
611 | |||
612 | void
|
||
613 | mxmlSetWrapMargin(int column) /* I - Column for wrapping, 0 to disable wrapping */ |
||
614 | { |
||
615 | _mxml_global_t *global = _mxml_global(); |
||
616 | /* Global data */
|
||
617 | |||
618 | |||
619 | global->wrap = column; |
||
620 | } |
||
621 | |||
622 | |||
623 | /*
|
||
624 | * 'mxml_add_char()' - Add a character to a buffer, expanding as needed.
|
||
625 | */
|
||
626 | |||
627 | static int /* O - 0 on success, -1 on error */ |
||
628 | mxml_add_char(int ch, /* I - Character to add */ |
||
629 | char **bufptr, /* IO - Current position in buffer */ |
||
630 | char **buffer, /* IO - Current buffer */ |
||
631 | int *bufsize) /* IO - Current buffer size */ |
||
632 | { |
||
633 | char *newbuffer; /* New buffer value */ |
||
634 | |||
635 | |||
636 | if (*bufptr >= (*buffer + *bufsize - 4)) |
||
637 | { |
||
638 | /*
|
||
639 | * Increase the size of the buffer...
|
||
640 | */
|
||
641 | |||
642 | if (*bufsize < 1024) |
||
643 | (*bufsize) *= 2;
|
||
644 | else
|
||
645 | (*bufsize) += 1024;
|
||
646 | |||
647 | if ((newbuffer = realloc(*buffer, *bufsize)) == NULL) |
||
648 | { |
||
649 | free(*buffer); |
||
650 | |||
651 | mxml_error("Unable to expand string buffer to %d bytes!", *bufsize);
|
||
652 | |||
653 | return (-1); |
||
654 | } |
||
655 | |||
656 | *bufptr = newbuffer + (*bufptr - *buffer); |
||
657 | *buffer = newbuffer; |
||
658 | } |
||
659 | |||
660 | if (ch < 0x80) |
||
661 | { |
||
662 | /*
|
||
663 | * Single byte ASCII...
|
||
664 | */
|
||
665 | |||
666 | *(*bufptr)++ = ch; |
||
667 | } |
||
668 | else if (ch < 0x800) |
||
669 | { |
||
670 | /*
|
||
671 | * Two-byte UTF-8...
|
||
672 | */
|
||
673 | |||
674 | *(*bufptr)++ = 0xc0 | (ch >> 6); |
||
675 | *(*bufptr)++ = 0x80 | (ch & 0x3f); |
||
676 | } |
||
677 | else if (ch < 0x10000) |
||
678 | { |
||
679 | /*
|
||
680 | * Three-byte UTF-8...
|
||
681 | */
|
||
682 | |||
683 | *(*bufptr)++ = 0xe0 | (ch >> 12); |
||
684 | *(*bufptr)++ = 0x80 | ((ch >> 6) & 0x3f); |
||
685 | *(*bufptr)++ = 0x80 | (ch & 0x3f); |
||
686 | } |
||
687 | else
|
||
688 | { |
||
689 | /*
|
||
690 | * Four-byte UTF-8...
|
||
691 | */
|
||
692 | |||
693 | *(*bufptr)++ = 0xf0 | (ch >> 18); |
||
694 | *(*bufptr)++ = 0x80 | ((ch >> 12) & 0x3f); |
||
695 | *(*bufptr)++ = 0x80 | ((ch >> 6) & 0x3f); |
||
696 | *(*bufptr)++ = 0x80 | (ch & 0x3f); |
||
697 | } |
||
698 | |||
699 | return (0); |
||
700 | } |
||
701 | |||
702 | |||
703 | /*
|
||
704 | * 'mxml_fd_getc()' - Read a character from a file descriptor.
|
||
705 | */
|
||
706 | |||
707 | static int /* O - Character or EOF */ |
||
708 | mxml_fd_getc(void *p, /* I - File descriptor buffer */ |
||
709 | int *encoding) /* IO - Encoding */ |
||
710 | { |
||
711 | _mxml_fdbuf_t *buf; /* File descriptor buffer */
|
||
712 | int ch, /* Current character */ |
||
713 | temp; /* Temporary character */
|
||
714 | |||
715 | |||
716 | /*
|
||
717 | * Grab the next character in the buffer...
|
||
718 | */
|
||
719 | |||
720 | buf = (_mxml_fdbuf_t *)p; |
||
721 | |||
722 | if (buf->current >= buf->end)
|
||
723 | if (mxml_fd_read(buf) < 0) |
||
724 | return (EOF); |
||
725 | |||
726 | ch = *(buf->current)++; |
||
727 | |||
728 | switch (*encoding)
|
||
729 | { |
||
730 | case ENCODE_UTF8 :
|
||
731 | /*
|
||
732 | * Got a UTF-8 character; convert UTF-8 to Unicode and return...
|
||
733 | */
|
||
734 | |||
735 | if (!(ch & 0x80)) |
||
736 | { |
||
737 | #if DEBUG > 1 |
||
738 | printf("mxml_fd_getc: %c (0x%04x)\n", ch < ' ' ? '.' : ch, ch); |
||
739 | #endif /* DEBUG > 1 */ |
||
740 | |||
741 | if (mxml_bad_char(ch))
|
||
742 | { |
||
743 | mxml_error("Bad control character 0x%02x not allowed by XML standard!",
|
||
744 | ch); |
||
745 | return (EOF); |
||
746 | } |
||
747 | |||
748 | return (ch);
|
||
749 | } |
||
750 | else if (ch == 0xfe) |
||
751 | { |
||
752 | /*
|
||
753 | * UTF-16 big-endian BOM?
|
||
754 | */
|
||
755 | |||
756 | if (buf->current >= buf->end)
|
||
757 | if (mxml_fd_read(buf) < 0) |
||
758 | return (EOF); |
||
759 | |||
760 | ch = *(buf->current)++; |
||
761 | |||
762 | if (ch != 0xff) |
||
763 | return (EOF); |
||
764 | |||
765 | *encoding = ENCODE_UTF16BE; |
||
766 | |||
767 | return (mxml_fd_getc(p, encoding));
|
||
768 | } |
||
769 | else if (ch == 0xff) |
||
770 | { |
||
771 | /*
|
||
772 | * UTF-16 little-endian BOM?
|
||
773 | */
|
||
774 | |||
775 | if (buf->current >= buf->end)
|
||
776 | if (mxml_fd_read(buf) < 0) |
||
777 | return (EOF); |
||
778 | |||
779 | ch = *(buf->current)++; |
||
780 | |||
781 | if (ch != 0xfe) |
||
782 | return (EOF); |
||
783 | |||
784 | *encoding = ENCODE_UTF16LE; |
||
785 | |||
786 | return (mxml_fd_getc(p, encoding));
|
||
787 | } |
||
788 | else if ((ch & 0xe0) == 0xc0) |
||
789 | { |
||
790 | /*
|
||
791 | * Two-byte value...
|
||
792 | */
|
||
793 | |||
794 | if (buf->current >= buf->end)
|
||
795 | if (mxml_fd_read(buf) < 0) |
||
796 | return (EOF); |
||
797 | |||
798 | temp = *(buf->current)++; |
||
799 | |||
800 | if ((temp & 0xc0) != 0x80) |
||
801 | return (EOF); |
||
802 | |||
803 | ch = ((ch & 0x1f) << 6) | (temp & 0x3f); |
||
804 | |||
805 | if (ch < 0x80) |
||
806 | { |
||
807 | mxml_error("Invalid UTF-8 sequence for character 0x%04x!", ch);
|
||
808 | return (EOF); |
||
809 | } |
||
810 | } |
||
811 | else if ((ch & 0xf0) == 0xe0) |
||
812 | { |
||
813 | /*
|
||
814 | * Three-byte value...
|
||
815 | */
|
||
816 | |||
817 | if (buf->current >= buf->end)
|
||
818 | if (mxml_fd_read(buf) < 0) |
||
819 | return (EOF); |
||
820 | |||
821 | temp = *(buf->current)++; |
||
822 | |||
823 | if ((temp & 0xc0) != 0x80) |
||
824 | return (EOF); |
||
825 | |||
826 | ch = ((ch & 0x0f) << 6) | (temp & 0x3f); |
||
827 | |||
828 | if (buf->current >= buf->end)
|
||
829 | if (mxml_fd_read(buf) < 0) |
||
830 | return (EOF); |
||
831 | |||
832 | temp = *(buf->current)++; |
||
833 | |||
834 | if ((temp & 0xc0) != 0x80) |
||
835 | return (EOF); |
||
836 | |||
837 | ch = (ch << 6) | (temp & 0x3f); |
||
838 | |||
839 | if (ch < 0x800) |
||
840 | { |
||
841 | mxml_error("Invalid UTF-8 sequence for character 0x%04x!", ch);
|
||
842 | return (EOF); |
||
843 | } |
||
844 | |||
845 | /*
|
||
846 | * Ignore (strip) Byte Order Mark (BOM)...
|
||
847 | */
|
||
848 | |||
849 | if (ch == 0xfeff) |
||
850 | return (mxml_fd_getc(p, encoding));
|
||
851 | } |
||
852 | else if ((ch & 0xf8) == 0xf0) |
||
853 | { |
||
854 | /*
|
||
855 | * Four-byte value...
|
||
856 | */
|
||
857 | |||
858 | if (buf->current >= buf->end)
|
||
859 | if (mxml_fd_read(buf) < 0) |
||
860 | return (EOF); |
||
861 | |||
862 | temp = *(buf->current)++; |
||
863 | |||
864 | if ((temp & 0xc0) != 0x80) |
||
865 | return (EOF); |
||
866 | |||
867 | ch = ((ch & 0x07) << 6) | (temp & 0x3f); |
||
868 | |||
869 | if (buf->current >= buf->end)
|
||
870 | if (mxml_fd_read(buf) < 0) |
||
871 | return (EOF); |
||
872 | |||
873 | temp = *(buf->current)++; |
||
874 | |||
875 | if ((temp & 0xc0) != 0x80) |
||
876 | return (EOF); |
||
877 | |||
878 | ch = (ch << 6) | (temp & 0x3f); |
||
879 | |||
880 | if (buf->current >= buf->end)
|
||
881 | if (mxml_fd_read(buf) < 0) |
||
882 | return (EOF); |
||
883 | |||
884 | temp = *(buf->current)++; |
||
885 | |||
886 | if ((temp & 0xc0) != 0x80) |
||
887 | return (EOF); |
||
888 | |||
889 | ch = (ch << 6) | (temp & 0x3f); |
||
890 | |||
891 | if (ch < 0x10000) |
||
892 | { |
||
893 | mxml_error("Invalid UTF-8 sequence for character 0x%04x!", ch);
|
||
894 | return (EOF); |
||
895 | } |
||
896 | } |
||
897 | else
|
||
898 | return (EOF); |
||
899 | break;
|
||
900 | |||
901 | case ENCODE_UTF16BE :
|
||
902 | /*
|
||
903 | * Read UTF-16 big-endian char...
|
||
904 | */
|
||
905 | |||
906 | if (buf->current >= buf->end)
|
||
907 | if (mxml_fd_read(buf) < 0) |
||
908 | return (EOF); |
||
909 | |||
910 | temp = *(buf->current)++; |
||
911 | |||
912 | ch = (ch << 8) | temp;
|
||
913 | |||
914 | if (mxml_bad_char(ch))
|
||
915 | { |
||
916 | mxml_error("Bad control character 0x%02x not allowed by XML standard!",
|
||
917 | ch); |
||
918 | return (EOF); |
||
919 | } |
||
920 | else if (ch >= 0xd800 && ch <= 0xdbff) |
||
921 | { |
||
922 | /*
|
||
923 | * Multi-word UTF-16 char...
|
||
924 | */
|
||
925 | |||
926 | int lch;
|
||
927 | |||
928 | if (buf->current >= buf->end)
|
||
929 | if (mxml_fd_read(buf) < 0) |
||
930 | return (EOF); |
||
931 | |||
932 | lch = *(buf->current)++; |
||
933 | |||
934 | if (buf->current >= buf->end)
|
||
935 | if (mxml_fd_read(buf) < 0) |
||
936 | return (EOF); |
||
937 | |||
938 | temp = *(buf->current)++; |
||
939 | |||
940 | lch = (lch << 8) | temp;
|
||
941 | |||
942 | if (lch < 0xdc00 || lch >= 0xdfff) |
||
943 | return (EOF); |
||
944 | |||
945 | ch = (((ch & 0x3ff) << 10) | (lch & 0x3ff)) + 0x10000; |
||
946 | } |
||
947 | break;
|
||
948 | |||
949 | case ENCODE_UTF16LE :
|
||
950 | /*
|
||
951 | * Read UTF-16 little-endian char...
|
||
952 | */
|
||
953 | |||
954 | if (buf->current >= buf->end)
|
||
955 | if (mxml_fd_read(buf) < 0) |
||
956 | return (EOF); |
||
957 | |||
958 | temp = *(buf->current)++; |
||
959 | |||
960 | ch |= (temp << 8);
|
||
961 | |||
962 | if (mxml_bad_char(ch))
|
||
963 | { |
||
964 | mxml_error("Bad control character 0x%02x not allowed by XML standard!",
|
||
965 | ch); |
||
966 | return (EOF); |
||
967 | } |
||
968 | else if (ch >= 0xd800 && ch <= 0xdbff) |
||
969 | { |
||
970 | /*
|
||
971 | * Multi-word UTF-16 char...
|
||
972 | */
|
||
973 | |||
974 | int lch;
|
||
975 | |||
976 | if (buf->current >= buf->end)
|
||
977 | if (mxml_fd_read(buf) < 0) |
||
978 | return (EOF); |
||
979 | |||
980 | lch = *(buf->current)++; |
||
981 | |||
982 | if (buf->current >= buf->end)
|
||
983 | if (mxml_fd_read(buf) < 0) |
||
984 | return (EOF); |
||
985 | |||
986 | temp = *(buf->current)++; |
||
987 | |||
988 | lch |= (temp << 8);
|
||
989 | |||
990 | if (lch < 0xdc00 || lch >= 0xdfff) |
||
991 | return (EOF); |
||
992 | |||
993 | ch = (((ch & 0x3ff) << 10) | (lch & 0x3ff)) + 0x10000; |
||
994 | } |
||
995 | break;
|
||
996 | } |
||
997 | |||
998 | #if DEBUG > 1 |
||
999 | printf("mxml_fd_getc: %c (0x%04x)\n", ch < ' ' ? '.' : ch, ch); |
||
1000 | #endif /* DEBUG > 1 */ |
||
1001 | |||
1002 | return (ch);
|
||
1003 | } |
||
1004 | |||
1005 | |||
1006 | /*
|
||
1007 | * 'mxml_fd_putc()' - Write a character to a file descriptor.
|
||
1008 | */
|
||
1009 | |||
1010 | static int /* O - 0 on success, -1 on error */ |
||
1011 | mxml_fd_putc(int ch, /* I - Character */ |
||
1012 | void *p) /* I - File descriptor buffer */ |
||
1013 | { |
||
1014 | _mxml_fdbuf_t *buf; /* File descriptor buffer */
|
||
1015 | |||
1016 | |||
1017 | /*
|
||
1018 | * Flush the write buffer as needed...
|
||
1019 | */
|
||
1020 | |||
1021 | buf = (_mxml_fdbuf_t *)p; |
||
1022 | |||
1023 | if (buf->current >= buf->end)
|
||
1024 | if (mxml_fd_write(buf) < 0) |
||
1025 | return (-1); |
||
1026 | |||
1027 | *(buf->current)++ = ch; |
||
1028 | |||
1029 | /*
|
||
1030 | * Return successfully...
|
||
1031 | */
|
||
1032 | |||
1033 | return (0); |
||
1034 | } |
||
1035 | |||
1036 | |||
1037 | /*
|
||
1038 | * 'mxml_fd_read()' - Read a buffer of data from a file descriptor.
|
||
1039 | */
|
||
1040 | |||
1041 | static int /* O - 0 on success, -1 on error */ |
||
1042 | mxml_fd_read(_mxml_fdbuf_t *buf) /* I - File descriptor buffer */
|
||
1043 | { |
||
1044 | int bytes; /* Bytes read... */ |
||
1045 | |||
1046 | |||
1047 | /*
|
||
1048 | * Range check input...
|
||
1049 | */
|
||
1050 | |||
1051 | if (!buf)
|
||
1052 | return (-1); |
||
1053 | |||
1054 | /*
|
||
1055 | * Read from the file descriptor...
|
||
1056 | */
|
||
1057 | |||
1058 | while ((bytes = read(buf->fd, buf->buffer, sizeof(buf->buffer))) < 0) |
||
1059 | #ifdef EINTR
|
||
1060 | if (errno != EAGAIN && errno != EINTR)
|
||
1061 | #else
|
||
1062 | if (errno != EAGAIN)
|
||
1063 | #endif /* EINTR */ |
||
1064 | return (-1); |
||
1065 | |||
1066 | if (bytes == 0) |
||
1067 | return (-1); |
||
1068 | |||
1069 | /*
|
||
1070 | * Update the pointers and return success...
|
||
1071 | */
|
||
1072 | |||
1073 | buf->current = buf->buffer; |
||
1074 | buf->end = buf->buffer + bytes; |
||
1075 | |||
1076 | return (0); |
||
1077 | } |
||
1078 | |||
1079 | |||
1080 | /*
|
||
1081 | * 'mxml_fd_write()' - Write a buffer of data to a file descriptor.
|
||
1082 | */
|
||
1083 | |||
1084 | static int /* O - 0 on success, -1 on error */ |
||
1085 | mxml_fd_write(_mxml_fdbuf_t *buf) /* I - File descriptor buffer */
|
||
1086 | { |
||
1087 | int bytes; /* Bytes written */ |
||
1088 | unsigned char *ptr; /* Pointer into buffer */ |
||
1089 | |||
1090 | |||
1091 | /*
|
||
1092 | * Range check...
|
||
1093 | */
|
||
1094 | |||
1095 | if (!buf)
|
||
1096 | return (-1); |
||
1097 | |||
1098 | /*
|
||
1099 | * Return 0 if there is nothing to write...
|
||
1100 | */
|
||
1101 | |||
1102 | if (buf->current == buf->buffer)
|
||
1103 | return (0); |
||
1104 | |||
1105 | /*
|
||
1106 | * Loop until we have written everything...
|
||
1107 | */
|
||
1108 | |||
1109 | for (ptr = buf->buffer; ptr < buf->current; ptr += bytes)
|
||
1110 | if ((bytes = write(buf->fd, ptr, buf->current - ptr)) < 0) |
||
1111 | return (-1); |
||
1112 | |||
1113 | /*
|
||
1114 | * All done, reset pointers and return success...
|
||
1115 | */
|
||
1116 | |||
1117 | buf->current = buf->buffer; |
||
1118 | |||
1119 | return (0); |
||
1120 | } |
||
1121 | |||
1122 | |||
1123 | /*
|
||
1124 | * 'mxml_file_getc()' - Get a character from a file.
|
||
1125 | */
|
||
1126 | |||
1127 | static int /* O - Character or EOF */ |
||
1128 | mxml_file_getc(void *p, /* I - Pointer to file */ |
||
1129 | int *encoding) /* IO - Encoding */ |
||
1130 | { |
||
1131 | int ch, /* Character from file */ |
||
1132 | temp; /* Temporary character */
|
||
1133 | FILE *fp; /* Pointer to file */
|
||
1134 | |||
1135 | |||
1136 | /*
|
||
1137 | * Read a character from the file and see if it is EOF or ASCII...
|
||
1138 | */
|
||
1139 | |||
1140 | fp = (FILE *)p; |
||
1141 | ch = getc(fp); |
||
1142 | |||
1143 | if (ch == EOF) |
||
1144 | return (EOF); |
||
1145 | |||
1146 | switch (*encoding)
|
||
1147 | { |
||
1148 | case ENCODE_UTF8 :
|
||
1149 | /*
|
||
1150 | * Got a UTF-8 character; convert UTF-8 to Unicode and return...
|
||
1151 | */
|
||
1152 | |||
1153 | if (!(ch & 0x80)) |
||
1154 | { |
||
1155 | if (mxml_bad_char(ch))
|
||
1156 | { |
||
1157 | mxml_error("Bad control character 0x%02x not allowed by XML standard!",
|
||
1158 | ch); |
||
1159 | return (EOF); |
||
1160 | } |
||
1161 | |||
1162 | #if DEBUG > 1 |
||
1163 | printf("mxml_file_getc: %c (0x%04x)\n", ch < ' ' ? '.' : ch, ch); |
||
1164 | #endif /* DEBUG > 1 */ |
||
1165 | |||
1166 | return (ch);
|
||
1167 | } |
||
1168 | else if (ch == 0xfe) |
||
1169 | { |
||
1170 | /*
|
||
1171 | * UTF-16 big-endian BOM?
|
||
1172 | */
|
||
1173 | |||
1174 | ch = getc(fp); |
||
1175 | if (ch != 0xff) |
||
1176 | return (EOF); |
||
1177 | |||
1178 | *encoding = ENCODE_UTF16BE; |
||
1179 | |||
1180 | return (mxml_file_getc(p, encoding));
|
||
1181 | } |
||
1182 | else if (ch == 0xff) |
||
1183 | { |
||
1184 | /*
|
||
1185 | * UTF-16 little-endian BOM?
|
||
1186 | */
|
||
1187 | |||
1188 | ch = getc(fp); |
||
1189 | if (ch != 0xfe) |
||
1190 | return (EOF); |
||
1191 | |||
1192 | *encoding = ENCODE_UTF16LE; |
||
1193 | |||
1194 | return (mxml_file_getc(p, encoding));
|
||
1195 | } |
||
1196 | else if ((ch & 0xe0) == 0xc0) |
||
1197 | { |
||
1198 | /*
|
||
1199 | * Two-byte value...
|
||
1200 | */
|
||
1201 | |||
1202 | if ((temp = getc(fp)) == EOF || (temp & 0xc0) != 0x80) |
||
1203 | return (EOF); |
||
1204 | |||
1205 | ch = ((ch & 0x1f) << 6) | (temp & 0x3f); |
||
1206 | |||
1207 | if (ch < 0x80) |
||
1208 | { |
||
1209 | mxml_error("Invalid UTF-8 sequence for character 0x%04x!", ch);
|
||
1210 | return (EOF); |
||
1211 | } |
||
1212 | } |
||
1213 | else if ((ch & 0xf0) == 0xe0) |
||
1214 | { |
||
1215 | /*
|
||
1216 | * Three-byte value...
|
||
1217 | */
|
||
1218 | |||
1219 | if ((temp = getc(fp)) == EOF || (temp & 0xc0) != 0x80) |
||
1220 | return (EOF); |
||
1221 | |||
1222 | ch = ((ch & 0x0f) << 6) | (temp & 0x3f); |
||
1223 | |||
1224 | if ((temp = getc(fp)) == EOF || (temp & 0xc0) != 0x80) |
||
1225 | return (EOF); |
||
1226 | |||
1227 | ch = (ch << 6) | (temp & 0x3f); |
||
1228 | |||
1229 | if (ch < 0x800) |
||
1230 | { |
||
1231 | mxml_error("Invalid UTF-8 sequence for character 0x%04x!", ch);
|
||
1232 | return (EOF); |
||
1233 | } |
||
1234 | |||
1235 | /*
|
||
1236 | * Ignore (strip) Byte Order Mark (BOM)...
|
||
1237 | */
|
||
1238 | |||
1239 | if (ch == 0xfeff) |
||
1240 | return (mxml_file_getc(p, encoding));
|
||
1241 | } |
||
1242 | else if ((ch & 0xf8) == 0xf0) |
||
1243 | { |
||
1244 | /*
|
||
1245 | * Four-byte value...
|
||
1246 | */
|
||
1247 | |||
1248 | if ((temp = getc(fp)) == EOF || (temp & 0xc0) != 0x80) |
||
1249 | return (EOF); |
||
1250 | |||
1251 | ch = ((ch & 0x07) << 6) | (temp & 0x3f); |
||
1252 | |||
1253 | if ((temp = getc(fp)) == EOF || (temp & 0xc0) != 0x80) |
||
1254 | return (EOF); |
||
1255 | |||
1256 | ch = (ch << 6) | (temp & 0x3f); |
||
1257 | |||
1258 | if ((temp = getc(fp)) == EOF || (temp & 0xc0) != 0x80) |
||
1259 | return (EOF); |
||
1260 | |||
1261 | ch = (ch << 6) | (temp & 0x3f); |
||
1262 | |||
1263 | if (ch < 0x10000) |
||
1264 | { |
||
1265 | mxml_error("Invalid UTF-8 sequence for character 0x%04x!", ch);
|
||
1266 | return (EOF); |
||
1267 | } |
||
1268 | } |
||
1269 | else
|
||
1270 | return (EOF); |
||
1271 | break;
|
||
1272 | |||
1273 | case ENCODE_UTF16BE :
|
||
1274 | /*
|
||
1275 | * Read UTF-16 big-endian char...
|
||
1276 | */
|
||
1277 | |||
1278 | ch = (ch << 8) | getc(fp);
|
||
1279 | |||
1280 | if (mxml_bad_char(ch))
|
||
1281 | { |
||
1282 | mxml_error("Bad control character 0x%02x not allowed by XML standard!",
|
||
1283 | ch); |
||
1284 | return (EOF); |
||
1285 | } |
||
1286 | else if (ch >= 0xd800 && ch <= 0xdbff) |
||
1287 | { |
||
1288 | /*
|
||
1289 | * Multi-word UTF-16 char...
|
||
1290 | */
|
||
1291 | |||
1292 | int lch = (getc(fp) << 8) | getc(fp); |
||
1293 | |||
1294 | if (lch < 0xdc00 || lch >= 0xdfff) |
||
1295 | return (EOF); |
||
1296 | |||
1297 | ch = (((ch & 0x3ff) << 10) | (lch & 0x3ff)) + 0x10000; |
||
1298 | } |
||
1299 | break;
|
||
1300 | |||
1301 | case ENCODE_UTF16LE :
|
||
1302 | /*
|
||
1303 | * Read UTF-16 little-endian char...
|
||
1304 | */
|
||
1305 | |||
1306 | ch |= (getc(fp) << 8);
|
||
1307 | |||
1308 | if (mxml_bad_char(ch))
|
||
1309 | { |
||
1310 | mxml_error("Bad control character 0x%02x not allowed by XML standard!",
|
||
1311 | ch); |
||
1312 | return (EOF); |
||
1313 | } |
||
1314 | else if (ch >= 0xd800 && ch <= 0xdbff) |
||
1315 | { |
||
1316 | /*
|
||
1317 | * Multi-word UTF-16 char...
|
||
1318 | */
|
||
1319 | |||
1320 | int lch = getc(fp) | (getc(fp) << 8); |
||
1321 | |||
1322 | if (lch < 0xdc00 || lch >= 0xdfff) |
||
1323 | return (EOF); |
||
1324 | |||
1325 | ch = (((ch & 0x3ff) << 10) | (lch & 0x3ff)) + 0x10000; |
||
1326 | } |
||
1327 | break;
|
||
1328 | } |
||
1329 | |||
1330 | #if DEBUG > 1 |
||
1331 | printf("mxml_file_getc: %c (0x%04x)\n", ch < ' ' ? '.' : ch, ch); |
||
1332 | #endif /* DEBUG > 1 */ |
||
1333 | |||
1334 | return (ch);
|
||
1335 | } |
||
1336 | |||
1337 | |||
1338 | /*
|
||
1339 | * 'mxml_file_putc()' - Write a character to a file.
|
||
1340 | */
|
||
1341 | |||
1342 | static int /* O - 0 on success, -1 on failure */ |
||
1343 | mxml_file_putc(int ch, /* I - Character to write */ |
||
1344 | void *p) /* I - Pointer to file */ |
||
1345 | { |
||
1346 | return (putc(ch, (FILE *)p) == EOF ? -1 : 0); |
||
1347 | } |
||
1348 | |||
1349 | |||
1350 | /*
|
||
1351 | * 'mxml_get_entity()' - Get the character corresponding to an entity...
|
||
1352 | */
|
||
1353 | |||
1354 | static int /* O - Character value or EOF on error */ |
||
1355 | mxml_get_entity(mxml_node_t *parent, /* I - Parent node */
|
||
1356 | void *p, /* I - Pointer to source */ |
||
1357 | int *encoding, /* IO - Character encoding */ |
||
1358 | int (*getc_cb)(void *, int *)) |
||
1359 | /* I - Get character function */
|
||
1360 | { |
||
1361 | int ch; /* Current character */ |
||
1362 | char entity[64], /* Entity string */ |
||
1363 | *entptr; /* Pointer into entity */
|
||
1364 | |||
1365 | |||
1366 | entptr = entity; |
||
1367 | |||
1368 | while ((ch = (*getc_cb)(p, encoding)) != EOF) |
||
1369 | if (ch > 126 || (!isalnum(ch) && ch != '#')) |
||
1370 | break;
|
||
1371 | else if (entptr < (entity + sizeof(entity) - 1)) |
||
1372 | *entptr++ = ch; |
||
1373 | else
|
||
1374 | { |
||
1375 | mxml_error("Entity name too long under parent <%s>!",
|
||
1376 | parent ? parent->value.element.name : "null");
|
||
1377 | break;
|
||
1378 | } |
||
1379 | |||
1380 | *entptr = '\0';
|
||
1381 | |||
1382 | if (ch != ';') |
||
1383 | { |
||
1384 | mxml_error("Character entity \"%s\" not terminated under parent <%s>!",
|
||
1385 | entity, parent ? parent->value.element.name : "null");
|
||
1386 | return (EOF); |
||
1387 | } |
||
1388 | |||
1389 | if (entity[0] == '#') |
||
1390 | { |
||
1391 | if (entity[1] == 'x') |
||
1392 | ch = strtol(entity + 2, NULL, 16); |
||
1393 | else
|
||
1394 | ch = strtol(entity + 1, NULL, 10); |
||
1395 | } |
||
1396 | else if ((ch = mxmlEntityGetValue(entity)) < 0) |
||
1397 | mxml_error("Entity name \"%s;\" not supported under parent <%s>!",
|
||
1398 | entity, parent ? parent->value.element.name : "null");
|
||
1399 | |||
1400 | if (mxml_bad_char(ch))
|
||
1401 | { |
||
1402 | mxml_error("Bad control character 0x%02x under parent <%s> not allowed by XML standard!",
|
||
1403 | ch, parent ? parent->value.element.name : "null");
|
||
1404 | return (EOF); |
||
1405 | } |
||
1406 | |||
1407 | return (ch);
|
||
1408 | } |
||
1409 | |||
1410 | |||
1411 | /*
|
||
1412 | * 'mxml_load_data()' - Load data into an XML node tree.
|
||
1413 | */
|
||
1414 | |||
1415 | static mxml_node_t * /* O - First node or NULL if the file could not be read. */ |
||
1416 | mxml_load_data( |
||
1417 | mxml_node_t *top, /* I - Top node */
|
||
1418 | void *p, /* I - Pointer to data */ |
||
1419 | mxml_load_cb_t cb, /* I - Callback function or MXML_NO_CALLBACK */
|
||
1420 | _mxml_getc_cb_t getc_cb, /* I - Read function */
|
||
1421 | mxml_sax_cb_t sax_cb, /* I - SAX callback or MXML_NO_CALLBACK */
|
||
1422 | void *sax_data) /* I - SAX user data */ |
||
1423 | { |
||
1424 | mxml_node_t *node, /* Current node */
|
||
1425 | *first, /* First node added */
|
||
1426 | *parent; /* Current parent node */
|
||
1427 | int ch, /* Character from file */ |
||
1428 | whitespace; /* Non-zero if whitespace seen */
|
||
1429 | char *buffer, /* String buffer */ |
||
1430 | *bufptr; /* Pointer into buffer */
|
||
1431 | int bufsize; /* Size of buffer */ |
||
1432 | mxml_type_t type; /* Current node type */
|
||
1433 | int encoding; /* Character encoding */ |
||
1434 | _mxml_global_t *global = _mxml_global(); |
||
1435 | /* Global data */
|
||
1436 | static const char * const types[] = /* Type strings... */ |
||
1437 | { |
||
1438 | "MXML_ELEMENT", /* XML element with attributes */ |
||
1439 | "MXML_INTEGER", /* Integer value */ |
||
1440 | "MXML_OPAQUE", /* Opaque string */ |
||
1441 | "MXML_REAL", /* Real value */ |
||
1442 | "MXML_TEXT", /* Text fragment */ |
||
1443 | "MXML_CUSTOM" /* Custom data */ |
||
1444 | }; |
||
1445 | |||
1446 | |||
1447 | /*
|
||
1448 | * Read elements and other nodes from the file...
|
||
1449 | */
|
||
1450 | |||
1451 | if ((buffer = malloc(64)) == NULL) |
||
1452 | { |
||
1453 | mxml_error("Unable to allocate string buffer!");
|
||
1454 | return (NULL); |
||
1455 | } |
||
1456 | |||
1457 | bufsize = 64;
|
||
1458 | bufptr = buffer; |
||
1459 | parent = top; |
||
1460 | first = NULL;
|
||
1461 | whitespace = 0;
|
||
1462 | encoding = ENCODE_UTF8; |
||
1463 | |||
1464 | if (cb && parent)
|
||
1465 | type = (*cb)(parent); |
||
1466 | else
|
||
1467 | type = MXML_TEXT; |
||
1468 | |||
1469 | while ((ch = (*getc_cb)(p, &encoding)) != EOF) |
||
1470 | { |
||
1471 | if ((ch == '<' || |
||
1472 | (mxml_isspace(ch) && type != MXML_OPAQUE && type != MXML_CUSTOM)) && |
||
1473 | bufptr > buffer) |
||
1474 | { |
||
1475 | /*
|
||
1476 | * Add a new value node...
|
||
1477 | */
|
||
1478 | |||
1479 | *bufptr = '\0';
|
||
1480 | |||
1481 | switch (type)
|
||
1482 | { |
||
1483 | case MXML_INTEGER :
|
||
1484 | node = mxmlNewInteger(parent, strtol(buffer, &bufptr, 0));
|
||
1485 | break;
|
||
1486 | |||
1487 | case MXML_OPAQUE :
|
||
1488 | node = mxmlNewOpaque(parent, buffer); |
||
1489 | break;
|
||
1490 | |||
1491 | case MXML_REAL :
|
||
1492 | node = mxmlNewReal(parent, strtod(buffer, &bufptr)); |
||
1493 | break;
|
||
1494 | |||
1495 | case MXML_TEXT :
|
||
1496 | node = mxmlNewText(parent, whitespace, buffer); |
||
1497 | break;
|
||
1498 | |||
1499 | case MXML_CUSTOM :
|
||
1500 | if (global->custom_load_cb)
|
||
1501 | { |
||
1502 | /*
|
||
1503 | * Use the callback to fill in the custom data...
|
||
1504 | */
|
||
1505 | |||
1506 | node = mxmlNewCustom(parent, NULL, NULL); |
||
1507 | |||
1508 | if ((*global->custom_load_cb)(node, buffer))
|
||
1509 | { |
||
1510 | mxml_error("Bad custom value '%s' in parent <%s>!",
|
||
1511 | buffer, parent ? parent->value.element.name : "null");
|
||
1512 | mxmlDelete(node); |
||
1513 | node = NULL;
|
||
1514 | } |
||
1515 | break;
|
||
1516 | } |
||
1517 | |||
1518 | default : /* Ignore... */ |
||
1519 | node = NULL;
|
||
1520 | break;
|
||
1521 | } |
||
1522 | |||
1523 | if (*bufptr)
|
||
1524 | { |
||
1525 | /*
|
||
1526 | * Bad integer/real number value...
|
||
1527 | */
|
||
1528 | |||
1529 | mxml_error("Bad %s value '%s' in parent <%s>!",
|
||
1530 | type == MXML_INTEGER ? "integer" : "real", buffer, |
||
1531 | parent ? parent->value.element.name : "null");
|
||
1532 | break;
|
||
1533 | } |
||
1534 | |||
1535 | bufptr = buffer; |
||
1536 | whitespace = mxml_isspace(ch) && type == MXML_TEXT; |
||
1537 | |||
1538 | if (!node && type != MXML_IGNORE)
|
||
1539 | { |
||
1540 | /*
|
||
1541 | * Print error and return...
|
||
1542 | */
|
||
1543 | |||
1544 | mxml_error("Unable to add value node of type %s to parent <%s>!",
|
||
1545 | types[type], parent ? parent->value.element.name : "null");
|
||
1546 | goto error;
|
||
1547 | } |
||
1548 | |||
1549 | if (sax_cb)
|
||
1550 | { |
||
1551 | (*sax_cb)(node, MXML_SAX_DATA, sax_data); |
||
1552 | |||
1553 | if (!mxmlRelease(node))
|
||
1554 | node = NULL;
|
||
1555 | } |
||
1556 | |||
1557 | if (!first && node)
|
||
1558 | first = node; |
||
1559 | } |
||
1560 | else if (mxml_isspace(ch) && type == MXML_TEXT) |
||
1561 | whitespace = 1;
|
||
1562 | |||
1563 | /*
|
||
1564 | * Add lone whitespace node if we have an element and existing
|
||
1565 | * whitespace...
|
||
1566 | */
|
||
1567 | |||
1568 | if (ch == '<' && whitespace && type == MXML_TEXT) |
||
1569 | { |
||
1570 | node = mxmlNewText(parent, whitespace, "");
|
||
1571 | |||
1572 | if (sax_cb)
|
||
1573 | { |
||
1574 | (*sax_cb)(node, MXML_SAX_DATA, sax_data); |
||
1575 | |||
1576 | if (!mxmlRelease(node))
|
||
1577 | node = NULL;
|
||
1578 | } |
||
1579 | |||
1580 | if (!first && node)
|
||
1581 | first = node; |
||
1582 | |||
1583 | whitespace = 0;
|
||
1584 | } |
||
1585 | |||
1586 | if (ch == '<') |
||
1587 | { |
||
1588 | /*
|
||
1589 | * Start of open/close tag...
|
||
1590 | */
|
||
1591 | |||
1592 | bufptr = buffer; |
||
1593 | |||
1594 | while ((ch = (*getc_cb)(p, &encoding)) != EOF) |
||
1595 | if (mxml_isspace(ch) || ch == '>' || (ch == '/' && bufptr > buffer)) |
||
1596 | break;
|
||
1597 | else if (ch == '<') |
||
1598 | { |
||
1599 | mxml_error("Bare < in element!");
|
||
1600 | goto error;
|
||
1601 | } |
||
1602 | else if (ch == '&') |
||
1603 | { |
||
1604 | if ((ch = mxml_get_entity(parent, p, &encoding, getc_cb)) == EOF) |
||
1605 | goto error;
|
||
1606 | |||
1607 | if (mxml_add_char(ch, &bufptr, &buffer, &bufsize))
|
||
1608 | goto error;
|
||
1609 | } |
||
1610 | else if (mxml_add_char(ch, &bufptr, &buffer, &bufsize)) |
||
1611 | goto error;
|
||
1612 | else if (((bufptr - buffer) == 1 && buffer[0] == '?') || |
||
1613 | ((bufptr - buffer) == 3 && !strncmp(buffer, "!--", 3)) || |
||
1614 | ((bufptr - buffer) == 8 && !strncmp(buffer, "![CDATA[", 8))) |
||
1615 | break;
|
||
1616 | |||
1617 | *bufptr = '\0';
|
||
1618 | |||
1619 | if (!strcmp(buffer, "!--")) |
||
1620 | { |
||
1621 | /*
|
||
1622 | * Gather rest of comment...
|
||
1623 | */
|
||
1624 | |||
1625 | while ((ch = (*getc_cb)(p, &encoding)) != EOF) |
||
1626 | { |
||
1627 | if (ch == '>' && bufptr > (buffer + 4) && |
||
1628 | bufptr[-3] != '-' && bufptr[-2] == '-' && bufptr[-1] == '-') |
||
1629 | break;
|
||
1630 | else if (mxml_add_char(ch, &bufptr, &buffer, &bufsize)) |
||
1631 | goto error;
|
||
1632 | } |
||
1633 | |||
1634 | /*
|
||
1635 | * Error out if we didn't get the whole comment...
|
||
1636 | */
|
||
1637 | |||
1638 | if (ch != '>') |
||
1639 | { |
||
1640 | /*
|
||
1641 | * Print error and return...
|
||
1642 | */
|
||
1643 | |||
1644 | mxml_error("Early EOF in comment node!");
|
||
1645 | goto error;
|
||
1646 | } |
||
1647 | |||
1648 | |||
1649 | /*
|
||
1650 | * Otherwise add this as an element under the current parent...
|
||
1651 | */
|
||
1652 | |||
1653 | *bufptr = '\0';
|
||
1654 | |||
1655 | if ((node = mxmlNewElement(parent, buffer)) == NULL) |
||
1656 | { |
||
1657 | /*
|
||
1658 | * Just print error for now...
|
||
1659 | */
|
||
1660 | |||
1661 | mxml_error("Unable to add comment node to parent <%s>!",
|
||
1662 | parent ? parent->value.element.name : "null");
|
||
1663 | break;
|
||
1664 | } |
||
1665 | |||
1666 | if (sax_cb)
|
||
1667 | { |
||
1668 | (*sax_cb)(node, MXML_SAX_COMMENT, sax_data); |
||
1669 | |||
1670 | if (!mxmlRelease(node))
|
||
1671 | node = NULL;
|
||
1672 | } |
||
1673 | |||
1674 | if (node && !first)
|
||
1675 | first = node; |
||
1676 | } |
||
1677 | else if (!strcmp(buffer, "![CDATA[")) |
||
1678 | { |
||
1679 | /*
|
||
1680 | * Gather CDATA section...
|
||
1681 | */
|
||
1682 | |||
1683 | while ((ch = (*getc_cb)(p, &encoding)) != EOF) |
||
1684 | { |
||
1685 | if (ch == '>' && !strncmp(bufptr - 2, "]]", 2)) |
||
1686 | break;
|
||
1687 | else if (mxml_add_char(ch, &bufptr, &buffer, &bufsize)) |
||
1688 | goto error;
|
||
1689 | } |
||
1690 | |||
1691 | /*
|
||
1692 | * Error out if we didn't get the whole comment...
|
||
1693 | */
|
||
1694 | |||
1695 | if (ch != '>') |
||
1696 | { |
||
1697 | /*
|
||
1698 | * Print error and return...
|
||
1699 | */
|
||
1700 | |||
1701 | mxml_error("Early EOF in CDATA node!");
|
||
1702 | goto error;
|
||
1703 | } |
||
1704 | |||
1705 | |||
1706 | /*
|
||
1707 | * Otherwise add this as an element under the current parent...
|
||
1708 | */
|
||
1709 | |||
1710 | *bufptr = '\0';
|
||
1711 | |||
1712 | if ((node = mxmlNewElement(parent, buffer)) == NULL) |
||
1713 | { |
||
1714 | /*
|
||
1715 | * Print error and return...
|
||
1716 | */
|
||
1717 | |||
1718 | mxml_error("Unable to add CDATA node to parent <%s>!",
|
||
1719 | parent ? parent->value.element.name : "null");
|
||
1720 | goto error;
|
||
1721 | } |
||
1722 | |||
1723 | if (sax_cb)
|
||
1724 | { |
||
1725 | (*sax_cb)(node, MXML_SAX_CDATA, sax_data); |
||
1726 | |||
1727 | if (!mxmlRelease(node))
|
||
1728 | node = NULL;
|
||
1729 | } |
||
1730 | |||
1731 | if (node && !first)
|
||
1732 | first = node; |
||
1733 | } |
||
1734 | else if (buffer[0] == '?') |
||
1735 | { |
||
1736 | /*
|
||
1737 | * Gather rest of processing instruction...
|
||
1738 | */
|
||
1739 | |||
1740 | while ((ch = (*getc_cb)(p, &encoding)) != EOF) |
||
1741 | { |
||
1742 | if (ch == '>' && bufptr > buffer && bufptr[-1] == '?') |
||
1743 | break;
|
||
1744 | else if (mxml_add_char(ch, &bufptr, &buffer, &bufsize)) |
||
1745 | goto error;
|
||
1746 | } |
||
1747 | |||
1748 | /*
|
||
1749 | * Error out if we didn't get the whole processing instruction...
|
||
1750 | */
|
||
1751 | |||
1752 | if (ch != '>') |
||
1753 | { |
||
1754 | /*
|
||
1755 | * Print error and return...
|
||
1756 | */
|
||
1757 | |||
1758 | mxml_error("Early EOF in processing instruction node!");
|
||
1759 | goto error;
|
||
1760 | } |
||
1761 | |||
1762 | /*
|
||
1763 | * Otherwise add this as an element under the current parent...
|
||
1764 | */
|
||
1765 | |||
1766 | *bufptr = '\0';
|
||
1767 | |||
1768 | if ((node = mxmlNewElement(parent, buffer)) == NULL) |
||
1769 | { |
||
1770 | /*
|
||
1771 | * Print error and return...
|
||
1772 | */
|
||
1773 | |||
1774 | mxml_error("Unable to add processing instruction node to parent <%s>!",
|
||
1775 | parent ? parent->value.element.name : "null");
|
||
1776 | goto error;
|
||
1777 | } |
||
1778 | |||
1779 | if (sax_cb)
|
||
1780 | { |
||
1781 | (*sax_cb)(node, MXML_SAX_DIRECTIVE, sax_data); |
||
1782 | |||
1783 | if (!mxmlRelease(node))
|
||
1784 | node = NULL;
|
||
1785 | } |
||
1786 | |||
1787 | if (node)
|
||
1788 | { |
||
1789 | if (!first)
|
||
1790 | first = node; |
||
1791 | |||
1792 | if (!parent)
|
||
1793 | { |
||
1794 | parent = node; |
||
1795 | |||
1796 | if (cb)
|
||
1797 | type = (*cb)(parent); |
||
1798 | } |
||
1799 | } |
||
1800 | } |
||
1801 | else if (buffer[0] == '!') |
||
1802 | { |
||
1803 | /*
|
||
1804 | * Gather rest of declaration...
|
||
1805 | */
|
||
1806 | |||
1807 | do
|
||
1808 | { |
||
1809 | if (ch == '>') |
||
1810 | break;
|
||
1811 | else
|
||
1812 | { |
||
1813 | if (ch == '&') |
||
1814 | if ((ch = mxml_get_entity(parent, p, &encoding, getc_cb)) == EOF) |
||
1815 | goto error;
|
||
1816 | |||
1817 | if (mxml_add_char(ch, &bufptr, &buffer, &bufsize))
|
||
1818 | goto error;
|
||
1819 | } |
||
1820 | } |
||
1821 | while ((ch = (*getc_cb)(p, &encoding)) != EOF); |
||
1822 | |||
1823 | /*
|
||
1824 | * Error out if we didn't get the whole declaration...
|
||
1825 | */
|
||
1826 | |||
1827 | if (ch != '>') |
||
1828 | { |
||
1829 | /*
|
||
1830 | * Print error and return...
|
||
1831 | */
|
||
1832 | |||
1833 | mxml_error("Early EOF in declaration node!");
|
||
1834 | goto error;
|
||
1835 | } |
||
1836 | |||
1837 | /*
|
||
1838 | * Otherwise add this as an element under the current parent...
|
||
1839 | */
|
||
1840 | |||
1841 | *bufptr = '\0';
|
||
1842 | |||
1843 | if ((node = mxmlNewElement(parent, buffer)) == NULL) |
||
1844 | { |
||
1845 | /*
|
||
1846 | * Print error and return...
|
||
1847 | */
|
||
1848 | |||
1849 | mxml_error("Unable to add declaration node to parent <%s>!",
|
||
1850 | parent ? parent->value.element.name : "null");
|
||
1851 | goto error;
|
||
1852 | } |
||
1853 | |||
1854 | if (sax_cb)
|
||
1855 | { |
||
1856 | (*sax_cb)(node, MXML_SAX_DIRECTIVE, sax_data); |
||
1857 | |||
1858 | if (!mxmlRelease(node))
|
||
1859 | node = NULL;
|
||
1860 | } |
||
1861 | |||
1862 | if (node)
|
||
1863 | { |
||
1864 | if (!first)
|
||
1865 | first = node; |
||
1866 | |||
1867 | if (!parent)
|
||
1868 | { |
||
1869 | parent = node; |
||
1870 | |||
1871 | if (cb)
|
||
1872 | type = (*cb)(parent); |
||
1873 | } |
||
1874 | } |
||
1875 | } |
||
1876 | else if (buffer[0] == '/') |
||
1877 | { |
||
1878 | /*
|
||
1879 | * Handle close tag...
|
||
1880 | */
|
||
1881 | |||
1882 | if (!parent || strcmp(buffer + 1, parent->value.element.name)) |
||
1883 | { |
||
1884 | /*
|
||
1885 | * Close tag doesn't match tree; print an error for now...
|
||
1886 | */
|
||
1887 | |||
1888 | mxml_error("Mismatched close tag <%s> under parent <%s>!",
|
||
1889 | buffer, parent ? parent->value.element.name : "(null)");
|
||
1890 | goto error;
|
||
1891 | } |
||
1892 | |||
1893 | /*
|
||
1894 | * Keep reading until we see >...
|
||
1895 | */
|
||
1896 | |||
1897 | while (ch != '>' && ch != EOF) |
||
1898 | ch = (*getc_cb)(p, &encoding); |
||
1899 | |||
1900 | node = parent; |
||
1901 | parent = parent->parent; |
||
1902 | |||
1903 | if (sax_cb)
|
||
1904 | { |
||
1905 | (*sax_cb)(node, MXML_SAX_ELEMENT_CLOSE, sax_data); |
||
1906 | |||
1907 | mxmlRelease(node); |
||
1908 | } |
||
1909 | |||
1910 | /*
|
||
1911 | * Ascend into the parent and set the value type as needed...
|
||
1912 | */
|
||
1913 | |||
1914 | if (cb && parent)
|
||
1915 | type = (*cb)(parent); |
||
1916 | } |
||
1917 | else
|
||
1918 | { |
||
1919 | /*
|
||
1920 | * Handle open tag...
|
||
1921 | */
|
||
1922 | |||
1923 | if ((node = mxmlNewElement(parent, buffer)) == NULL) |
||
1924 | { |
||
1925 | /*
|
||
1926 | * Just print error for now...
|
||
1927 | */
|
||
1928 | |||
1929 | mxml_error("Unable to add element node to parent <%s>!",
|
||
1930 | parent ? parent->value.element.name : "null");
|
||
1931 | goto error;
|
||
1932 | } |
||
1933 | |||
1934 | if (mxml_isspace(ch))
|
||
1935 | { |
||
1936 | if ((ch = mxml_parse_element(node, p, &encoding, getc_cb)) == EOF) |
||
1937 | goto error;
|
||
1938 | } |
||
1939 | else if (ch == '/') |
||
1940 | { |
||
1941 | if ((ch = (*getc_cb)(p, &encoding)) != '>') |
||
1942 | { |
||
1943 | mxml_error("Expected > but got '%c' instead for element <%s/>!",
|
||
1944 | ch, buffer); |
||
1945 | mxmlDelete(node); |
||
1946 | goto error;
|
||
1947 | } |
||
1948 | |||
1949 | ch = '/';
|
||
1950 | } |
||
1951 | |||
1952 | if (sax_cb)
|
||
1953 | (*sax_cb)(node, MXML_SAX_ELEMENT_OPEN, sax_data); |
||
1954 | |||
1955 | if (!first)
|
||
1956 | first = node; |
||
1957 | |||
1958 | if (ch == EOF) |
||
1959 | break;
|
||
1960 | |||
1961 | if (ch != '/') |
||
1962 | { |
||
1963 | /*
|
||
1964 | * Descend into this node, setting the value type as needed...
|
||
1965 | */
|
||
1966 | |||
1967 | parent = node; |
||
1968 | |||
1969 | if (cb && parent)
|
||
1970 | type = (*cb)(parent); |
||
1971 | } |
||
1972 | else if (sax_cb) |
||
1973 | { |
||
1974 | (*sax_cb)(node, MXML_SAX_ELEMENT_CLOSE, sax_data); |
||
1975 | |||
1976 | if (!mxmlRelease(node) && first == node)
|
||
1977 | first = NULL;
|
||
1978 | } |
||
1979 | } |
||
1980 | |||
1981 | bufptr = buffer; |
||
1982 | } |
||
1983 | else if (ch == '&') |
||
1984 | { |
||
1985 | /*
|
||
1986 | * Add character entity to current buffer...
|
||
1987 | */
|
||
1988 | |||
1989 | if ((ch = mxml_get_entity(parent, p, &encoding, getc_cb)) == EOF) |
||
1990 | goto error;
|
||
1991 | |||
1992 | if (mxml_add_char(ch, &bufptr, &buffer, &bufsize))
|
||
1993 | goto error;
|
||
1994 | } |
||
1995 | else if (type == MXML_OPAQUE || type == MXML_CUSTOM || !mxml_isspace(ch)) |
||
1996 | { |
||
1997 | /*
|
||
1998 | * Add character to current buffer...
|
||
1999 | */
|
||
2000 | |||
2001 | if (mxml_add_char(ch, &bufptr, &buffer, &bufsize))
|
||
2002 | goto error;
|
||
2003 | } |
||
2004 | } |
||
2005 | |||
2006 | /*
|
||
2007 | * Free the string buffer - we don't need it anymore...
|
||
2008 | */
|
||
2009 | |||
2010 | free(buffer); |
||
2011 | |||
2012 | /*
|
||
2013 | * Find the top element and return it...
|
||
2014 | */
|
||
2015 | |||
2016 | if (parent)
|
||
2017 | { |
||
2018 | node = parent; |
||
2019 | |||
2020 | while (parent->parent != top && parent->parent)
|
||
2021 | parent = parent->parent; |
||
2022 | |||
2023 | if (node != parent)
|
||
2024 | { |
||
2025 | mxml_error("Missing close tag </%s> under parent <%s>!",
|
||
2026 | node->value.element.name, |
||
2027 | node->parent ? node->parent->value.element.name : "(null)");
|
||
2028 | |||
2029 | mxmlDelete(first); |
||
2030 | |||
2031 | return (NULL); |
||
2032 | } |
||
2033 | } |
||
2034 | |||
2035 | if (parent)
|
||
2036 | return (parent);
|
||
2037 | else
|
||
2038 | return (first);
|
||
2039 | |||
2040 | /*
|
||
2041 | * Common error return...
|
||
2042 | */
|
||
2043 | |||
2044 | error:
|
||
2045 | |||
2046 | mxmlDelete(first); |
||
2047 | |||
2048 | free(buffer); |
||
2049 | |||
2050 | return (NULL); |
||
2051 | } |
||
2052 | |||
2053 | |||
2054 | /*
|
||
2055 | * 'mxml_parse_element()' - Parse an element for any attributes...
|
||
2056 | */
|
||
2057 | |||
2058 | static int /* O - Terminating character */ |
||
2059 | mxml_parse_element( |
||
2060 | mxml_node_t *node, /* I - Element node */
|
||
2061 | void *p, /* I - Data to read from */ |
||
2062 | int *encoding, /* IO - Encoding */ |
||
2063 | _mxml_getc_cb_t getc_cb) /* I - Data callback */
|
||
2064 | { |
||
2065 | int ch, /* Current character in file */ |
||
2066 | quote; /* Quoting character */
|
||
2067 | char *name, /* Attribute name */ |
||
2068 | *value, /* Attribute value */
|
||
2069 | *ptr; /* Pointer into name/value */
|
||
2070 | int namesize, /* Size of name string */ |
||
2071 | valsize; /* Size of value string */
|
||
2072 | |||
2073 | |||
2074 | /*
|
||
2075 | * Initialize the name and value buffers...
|
||
2076 | */
|
||
2077 | |||
2078 | if ((name = malloc(64)) == NULL) |
||
2079 | { |
||
2080 | mxml_error("Unable to allocate memory for name!");
|
||
2081 | return (EOF); |
||
2082 | } |
||
2083 | |||
2084 | namesize = 64;
|
||
2085 | |||
2086 | if ((value = malloc(64)) == NULL) |
||
2087 | { |
||
2088 | free(name); |
||
2089 | mxml_error("Unable to allocate memory for value!");
|
||
2090 | return (EOF); |
||
2091 | } |
||
2092 | |||
2093 | valsize = 64;
|
||
2094 | |||
2095 | /*
|
||
2096 | * Loop until we hit a >, /, ?, or EOF...
|
||
2097 | */
|
||
2098 | |||
2099 | while ((ch = (*getc_cb)(p, encoding)) != EOF) |
||
2100 | { |
||
2101 | #if DEBUG > 1 |
||
2102 | fprintf(stderr, "parse_element: ch='%c'\n", ch);
|
||
2103 | #endif /* DEBUG > 1 */ |
||
2104 | |||
2105 | /*
|
||
2106 | * Skip leading whitespace...
|
||
2107 | */
|
||
2108 | |||
2109 | if (mxml_isspace(ch))
|
||
2110 | continue;
|
||
2111 | |||
2112 | /*
|
||
2113 | * Stop at /, ?, or >...
|
||
2114 | */
|
||
2115 | |||
2116 | if (ch == '/' || ch == '?') |
||
2117 | { |
||
2118 | /*
|
||
2119 | * Grab the > character and print an error if it isn't there...
|
||
2120 | */
|
||
2121 | |||
2122 | quote = (*getc_cb)(p, encoding); |
||
2123 | |||
2124 | if (quote != '>') |
||
2125 | { |
||
2126 | mxml_error("Expected '>' after '%c' for element %s, but got '%c'!",
|
||
2127 | ch, node->value.element.name, quote); |
||
2128 | goto error;
|
||
2129 | } |
||
2130 | |||
2131 | break;
|
||
2132 | } |
||
2133 | else if (ch == '<') |
||
2134 | { |
||
2135 | mxml_error("Bare < in element %s!", node->value.element.name);
|
||
2136 | goto error;
|
||
2137 | } |
||
2138 | else if (ch == '>') |
||
2139 | break;
|
||
2140 | |||
2141 | /*
|
||
2142 | * Read the attribute name...
|
||
2143 | */
|
||
2144 | |||
2145 | name[0] = ch;
|
||
2146 | ptr = name + 1;
|
||
2147 | |||
2148 | if (ch == '\"' || ch == '\'') |
||
2149 | { |
||
2150 | /*
|
||
2151 | * Name is in quotes, so get a quoted string...
|
||
2152 | */
|
||
2153 | |||
2154 | quote = ch; |
||
2155 | |||
2156 | while ((ch = (*getc_cb)(p, encoding)) != EOF) |
||
2157 | { |
||
2158 | if (ch == '&') |
||
2159 | if ((ch = mxml_get_entity(node, p, encoding, getc_cb)) == EOF) |
||
2160 | goto error;
|
||
2161 | |||
2162 | if (mxml_add_char(ch, &ptr, &name, &namesize))
|
||
2163 | goto error;
|
||
2164 | |||
2165 | if (ch == quote)
|
||
2166 | break;
|
||
2167 | } |
||
2168 | } |
||
2169 | else
|
||
2170 | { |
||
2171 | /*
|
||
2172 | * Grab an normal, non-quoted name...
|
||
2173 | */
|
||
2174 | |||
2175 | while ((ch = (*getc_cb)(p, encoding)) != EOF) |
||
2176 | if (mxml_isspace(ch) || ch == '=' || ch == '/' || ch == '>' || |
||
2177 | ch == '?')
|
||
2178 | break;
|
||
2179 | else
|
||
2180 | { |
||
2181 | if (ch == '&') |
||
2182 | if ((ch = mxml_get_entity(node, p, encoding, getc_cb)) == EOF) |
||
2183 | goto error;
|
||
2184 | |||
2185 | if (mxml_add_char(ch, &ptr, &name, &namesize))
|
||
2186 | goto error;
|
||
2187 | } |
||
2188 | } |
||
2189 | |||
2190 | *ptr = '\0';
|
||
2191 | |||
2192 | if (mxmlElementGetAttr(node, name))
|
||
2193 | goto error;
|
||
2194 | |||
2195 | while (ch != EOF && mxml_isspace(ch)) |
||
2196 | ch = (*getc_cb)(p, encoding); |
||
2197 | |||
2198 | if (ch == '=') |
||
2199 | { |
||
2200 | /*
|
||
2201 | * Read the attribute value...
|
||
2202 | */
|
||
2203 | |||
2204 | while ((ch = (*getc_cb)(p, encoding)) != EOF && mxml_isspace(ch)); |
||
2205 | |||
2206 | if (ch == EOF) |
||
2207 | { |
||
2208 | mxml_error("Missing value for attribute '%s' in element %s!",
|
||
2209 | name, node->value.element.name); |
||
2210 | goto error;
|
||
2211 | } |
||
2212 | |||
2213 | if (ch == '\'' || ch == '\"') |
||
2214 | { |
||
2215 | /*
|
||
2216 | * Read quoted value...
|
||
2217 | */
|
||
2218 | |||
2219 | quote = ch; |
||
2220 | ptr = value; |
||
2221 | |||
2222 | while ((ch = (*getc_cb)(p, encoding)) != EOF) |
||
2223 | if (ch == quote)
|
||
2224 | break;
|
||
2225 | else
|
||
2226 | { |
||
2227 | if (ch == '&') |
||
2228 | if ((ch = mxml_get_entity(node, p, encoding, getc_cb)) == EOF) |
||
2229 | goto error;
|
||
2230 | |||
2231 | if (mxml_add_char(ch, &ptr, &value, &valsize))
|
||
2232 | goto error;
|
||
2233 | } |
||
2234 | |||
2235 | *ptr = '\0';
|
||
2236 | } |
||
2237 | else
|
||
2238 | { |
||
2239 | /*
|
||
2240 | * Read unquoted value...
|
||
2241 | */
|
||
2242 | |||
2243 | value[0] = ch;
|
||
2244 | ptr = value + 1;
|
||
2245 | |||
2246 | while ((ch = (*getc_cb)(p, encoding)) != EOF) |
||
2247 | if (mxml_isspace(ch) || ch == '=' || ch == '/' || ch == '>') |
||
2248 | break;
|
||
2249 | else
|
||
2250 | { |
||
2251 | if (ch == '&') |
||
2252 | if ((ch = mxml_get_entity(node, p, encoding, getc_cb)) == EOF) |
||
2253 | goto error;
|
||
2254 | |||
2255 | if (mxml_add_char(ch, &ptr, &value, &valsize))
|
||
2256 | goto error;
|
||
2257 | } |
||
2258 | |||
2259 | *ptr = '\0';
|
||
2260 | } |
||
2261 | |||
2262 | /*
|
||
2263 | * Set the attribute with the given string value...
|
||
2264 | */
|
||
2265 | |||
2266 | mxmlElementSetAttr(node, name, value); |
||
2267 | } |
||
2268 | else
|
||
2269 | { |
||
2270 | mxml_error("Missing value for attribute '%s' in element %s!",
|
||
2271 | name, node->value.element.name); |
||
2272 | goto error;
|
||
2273 | } |
||
2274 | |||
2275 | /*
|
||
2276 | * Check the end character...
|
||
2277 | */
|
||
2278 | |||
2279 | if (ch == '/' || ch == '?') |
||
2280 | { |
||
2281 | /*
|
||
2282 | * Grab the > character and print an error if it isn't there...
|
||
2283 | */
|
||
2284 | |||
2285 | quote = (*getc_cb)(p, encoding); |
||
2286 | |||
2287 | if (quote != '>') |
||
2288 | { |
||
2289 | mxml_error("Expected '>' after '%c' for element %s, but got '%c'!",
|
||
2290 | ch, node->value.element.name, quote); |
||
2291 | ch = EOF;
|
||
2292 | } |
||
2293 | |||
2294 | break;
|
||
2295 | } |
||
2296 | else if (ch == '>') |
||
2297 | break;
|
||
2298 | } |
||
2299 | |||
2300 | /*
|
||
2301 | * Free the name and value buffers and return...
|
||
2302 | */
|
||
2303 | |||
2304 | free(name); |
||
2305 | free(value); |
||
2306 | |||
2307 | return (ch);
|
||
2308 | |||
2309 | /*
|
||
2310 | * Common error return point...
|
||
2311 | */
|
||
2312 | |||
2313 | error:
|
||
2314 | |||
2315 | free(name); |
||
2316 | free(value); |
||
2317 | |||
2318 | return (EOF); |
||
2319 | } |
||
2320 | |||
2321 | |||
2322 | /*
|
||
2323 | * 'mxml_string_getc()' - Get a character from a string.
|
||
2324 | */
|
||
2325 | |||
2326 | static int /* O - Character or EOF */ |
||
2327 | mxml_string_getc(void *p, /* I - Pointer to file */ |
||
2328 | int *encoding) /* IO - Encoding */ |
||
2329 | { |
||
2330 | int ch; /* Character */ |
||
2331 | const char **s; /* Pointer to string pointer */ |
||
2332 | |||
2333 | |||
2334 | s = (const char **)p; |
||
2335 | |||
2336 | if ((ch = (*s)[0] & 255) != 0 || *encoding == ENCODE_UTF16LE) |
||
2337 | { |
||
2338 | /*
|
||
2339 | * Got character; convert UTF-8 to integer and return...
|
||
2340 | */
|
||
2341 | |||
2342 | (*s)++; |
||
2343 | |||
2344 | switch (*encoding)
|
||
2345 | { |
||
2346 | case ENCODE_UTF8 :
|
||
2347 | if (!(ch & 0x80)) |
||
2348 | { |
||
2349 | #if DEBUG > 1 |
||
2350 | printf("mxml_string_getc: %c (0x%04x)\n", ch < ' ' ? '.' : ch, ch); |
||
2351 | #endif /* DEBUG > 1 */ |
||
2352 | |||
2353 | if (mxml_bad_char(ch))
|
||
2354 | { |
||
2355 | mxml_error("Bad control character 0x%02x not allowed by XML standard!",
|
||
2356 | ch); |
||
2357 | return (EOF); |
||
2358 | } |
||
2359 | |||
2360 | return (ch);
|
||
2361 | } |
||
2362 | else if (ch == 0xfe) |
||
2363 | { |
||
2364 | /*
|
||
2365 | * UTF-16 big-endian BOM?
|
||
2366 | */
|
||
2367 | |||
2368 | if (((*s)[0] & 255) != 0xff) |
||
2369 | return (EOF); |
||
2370 | |||
2371 | *encoding = ENCODE_UTF16BE; |
||
2372 | (*s)++; |
||
2373 | |||
2374 | return (mxml_string_getc(p, encoding));
|
||
2375 | } |
||
2376 | else if (ch == 0xff) |
||
2377 | { |
||
2378 | /*
|
||
2379 | * UTF-16 little-endian BOM?
|
||
2380 | */
|
||
2381 | |||
2382 | if (((*s)[0] & 255) != 0xfe) |
||
2383 | return (EOF); |
||
2384 | |||
2385 | *encoding = ENCODE_UTF16LE; |
||
2386 | (*s)++; |
||
2387 | |||
2388 | return (mxml_string_getc(p, encoding));
|
||
2389 | } |
||
2390 | else if ((ch & 0xe0) == 0xc0) |
||
2391 | { |
||
2392 | /*
|
||
2393 | * Two-byte value...
|
||
2394 | */
|
||
2395 | |||
2396 | if (((*s)[0] & 0xc0) != 0x80) |
||
2397 | return (EOF); |
||
2398 | |||
2399 | ch = ((ch & 0x1f) << 6) | ((*s)[0] & 0x3f); |
||
2400 | |||
2401 | (*s)++; |
||
2402 | |||
2403 | if (ch < 0x80) |
||
2404 | { |
||
2405 | mxml_error("Invalid UTF-8 sequence for character 0x%04x!", ch);
|
||
2406 | return (EOF); |
||
2407 | } |
||
2408 | |||
2409 | #if DEBUG > 1 |
||
2410 | printf("mxml_string_getc: %c (0x%04x)\n", ch < ' ' ? '.' : ch, ch); |
||
2411 | #endif /* DEBUG > 1 */ |
||
2412 | |||
2413 | return (ch);
|
||
2414 | } |
||
2415 | else if ((ch & 0xf0) == 0xe0) |
||
2416 | { |
||
2417 | /*
|
||
2418 | * Three-byte value...
|
||
2419 | */
|
||
2420 | |||
2421 | if (((*s)[0] & 0xc0) != 0x80 || |
||
2422 | ((*s)[1] & 0xc0) != 0x80) |
||
2423 | return (EOF); |
||
2424 | |||
2425 | ch = ((((ch & 0x0f) << 6) | ((*s)[0] & 0x3f)) << 6) | ((*s)[1] & 0x3f); |
||
2426 | |||
2427 | (*s) += 2;
|
||
2428 | |||
2429 | if (ch < 0x800) |
||
2430 | { |
||
2431 | mxml_error("Invalid UTF-8 sequence for character 0x%04x!", ch);
|
||
2432 | return (EOF); |
||
2433 | } |
||
2434 | |||
2435 | /*
|
||
2436 | * Ignore (strip) Byte Order Mark (BOM)...
|
||
2437 | */
|
||
2438 | |||
2439 | if (ch == 0xfeff) |
||
2440 | return (mxml_string_getc(p, encoding));
|
||
2441 | |||
2442 | #if DEBUG > 1 |
||
2443 | printf("mxml_string_getc: %c (0x%04x)\n", ch < ' ' ? '.' : ch, ch); |
||
2444 | #endif /* DEBUG > 1 */ |
||
2445 | |||
2446 | return (ch);
|
||
2447 | } |
||
2448 | else if ((ch & 0xf8) == 0xf0) |
||
2449 | { |
||
2450 | /*
|
||
2451 | * Four-byte value...
|
||
2452 | */
|
||
2453 | |||
2454 | if (((*s)[0] & 0xc0) != 0x80 || |
||
2455 | ((*s)[1] & 0xc0) != 0x80 || |
||
2456 | ((*s)[2] & 0xc0) != 0x80) |
||
2457 | return (EOF); |
||
2458 | |||
2459 | ch = ((((((ch & 0x07) << 6) | ((*s)[0] & 0x3f)) << 6) | |
||
2460 | ((*s)[1] & 0x3f)) << 6) | ((*s)[2] & 0x3f); |
||
2461 | |||
2462 | (*s) += 3;
|
||
2463 | |||
2464 | if (ch < 0x10000) |
||
2465 | { |
||
2466 | mxml_error("Invalid UTF-8 sequence for character 0x%04x!", ch);
|
||
2467 | return (EOF); |
||
2468 | } |
||
2469 | |||
2470 | #if DEBUG > 1 |
||
2471 | printf("mxml_string_getc: %c (0x%04x)\n", ch < ' ' ? '.' : ch, ch); |
||
2472 | #endif /* DEBUG > 1 */ |
||
2473 | |||
2474 | return (ch);
|
||
2475 | } |
||
2476 | else
|
||
2477 | return (EOF); |
||
2478 | |||
2479 | case ENCODE_UTF16BE :
|
||
2480 | /*
|
||
2481 | * Read UTF-16 big-endian char...
|
||
2482 | */
|
||
2483 | |||
2484 | ch = (ch << 8) | ((*s)[0] & 255); |
||
2485 | (*s) ++; |
||
2486 | |||
2487 | if (mxml_bad_char(ch))
|
||
2488 | { |
||
2489 | mxml_error("Bad control character 0x%02x not allowed by XML standard!",
|
||
2490 | ch); |
||
2491 | return (EOF); |
||
2492 | } |
||
2493 | else if (ch >= 0xd800 && ch <= 0xdbff) |
||
2494 | { |
||
2495 | /*
|
||
2496 | * Multi-word UTF-16 char...
|
||
2497 | */
|
||
2498 | |||
2499 | int lch; /* Lower word */ |
||
2500 | |||
2501 | |||
2502 | if (!(*s)[0]) |
||
2503 | return (EOF); |
||
2504 | |||
2505 | lch = (((*s)[0] & 255) << 8) | ((*s)[1] & 255); |
||
2506 | (*s) += 2;
|
||
2507 | |||
2508 | if (lch < 0xdc00 || lch >= 0xdfff) |
||
2509 | return (EOF); |
||
2510 | |||
2511 | ch = (((ch & 0x3ff) << 10) | (lch & 0x3ff)) + 0x10000; |
||
2512 | } |
||
2513 | |||
2514 | #if DEBUG > 1 |
||
2515 | printf("mxml_string_getc: %c (0x%04x)\n", ch < ' ' ? '.' : ch, ch); |
||
2516 | #endif /* DEBUG > 1 */ |
||
2517 | |||
2518 | return (ch);
|
||
2519 | |||
2520 | case ENCODE_UTF16LE :
|
||
2521 | /*
|
||
2522 | * Read UTF-16 little-endian char...
|
||
2523 | */
|
||
2524 | |||
2525 | ch = ch | (((*s)[0] & 255) << 8); |
||
2526 | |||
2527 | if (!ch)
|
||
2528 | { |
||
2529 | (*s) --; |
||
2530 | return (EOF); |
||
2531 | } |
||
2532 | |||
2533 | (*s) ++; |
||
2534 | |||
2535 | if (mxml_bad_char(ch))
|
||
2536 | { |
||
2537 | mxml_error("Bad control character 0x%02x not allowed by XML standard!",
|
||
2538 | ch); |
||
2539 | return (EOF); |
||
2540 | } |
||
2541 | else if (ch >= 0xd800 && ch <= 0xdbff) |
||
2542 | { |
||
2543 | /*
|
||
2544 | * Multi-word UTF-16 char...
|
||
2545 | */
|
||
2546 | |||
2547 | int lch; /* Lower word */ |
||
2548 | |||
2549 | |||
2550 | if (!(*s)[1]) |
||
2551 | return (EOF); |
||
2552 | |||
2553 | lch = (((*s)[1] & 255) << 8) | ((*s)[0] & 255); |
||
2554 | (*s) += 2;
|
||
2555 | |||
2556 | if (lch < 0xdc00 || lch >= 0xdfff) |
||
2557 | return (EOF); |
||
2558 | |||
2559 | ch = (((ch & 0x3ff) << 10) | (lch & 0x3ff)) + 0x10000; |
||
2560 | } |
||
2561 | |||
2562 | #if DEBUG > 1 |
||
2563 | printf("mxml_string_getc: %c (0x%04x)\n", ch < ' ' ? '.' : ch, ch); |
||
2564 | #endif /* DEBUG > 1 */ |
||
2565 | |||
2566 | return (ch);
|
||
2567 | } |
||
2568 | } |
||
2569 | |||
2570 | return (EOF); |
||
2571 | } |
||
2572 | |||
2573 | |||
2574 | /*
|
||
2575 | * 'mxml_string_putc()' - Write a character to a string.
|
||
2576 | */
|
||
2577 | |||
2578 | static int /* O - 0 on success, -1 on failure */ |
||
2579 | mxml_string_putc(int ch, /* I - Character to write */ |
||
2580 | void *p) /* I - Pointer to string pointers */ |
||
2581 | { |
||
2582 | char **pp; /* Pointer to string pointers */ |
||
2583 | |||
2584 | |||
2585 | pp = (char **)p;
|
||
2586 | |||
2587 | if (pp[0] < pp[1]) |
||
2588 | pp[0][0] = ch; |
||
2589 | |||
2590 | pp[0] ++;
|
||
2591 | |||
2592 | return (0); |
||
2593 | } |
||
2594 | |||
2595 | |||
2596 | /*
|
||
2597 | * 'mxml_write_name()' - Write a name string.
|
||
2598 | */
|
||
2599 | |||
2600 | static int /* O - 0 on success, -1 on failure */ |
||
2601 | mxml_write_name(const char *s, /* I - Name to write */ |
||
2602 | void *p, /* I - Write pointer */ |
||
2603 | int (*putc_cb)(int, void *)) |
||
2604 | /* I - Write callback */
|
||
2605 | { |
||
2606 | char quote; /* Quote character */ |
||
2607 | const char *name; /* Entity name */ |
||
2608 | |||
2609 | |||
2610 | if (*s == '\"' || *s == '\'') |
||
2611 | { |
||
2612 | /*
|
||
2613 | * Write a quoted name string...
|
||
2614 | */
|
||
2615 | |||
2616 | if ((*putc_cb)(*s, p) < 0) |
||
2617 | return (-1); |
||
2618 | |||
2619 | quote = *s++; |
||
2620 | |||
2621 | while (*s && *s != quote)
|
||
2622 | { |
||
2623 | if ((name = mxmlEntityGetName(*s)) != NULL) |
||
2624 | { |
||
2625 | if ((*putc_cb)('&', p) < 0) |
||
2626 | return (-1); |
||
2627 | |||
2628 | while (*name)
|
||
2629 | { |
||
2630 | if ((*putc_cb)(*name, p) < 0) |
||
2631 | return (-1); |
||
2632 | |||
2633 | name ++; |
||
2634 | } |
||
2635 | |||
2636 | if ((*putc_cb)(';', p) < 0) |
||
2637 | return (-1); |
||
2638 | } |
||
2639 | else if ((*putc_cb)(*s, p) < 0) |
||
2640 | return (-1); |
||
2641 | |||
2642 | s ++; |
||
2643 | } |
||
2644 | |||
2645 | /*
|
||
2646 | * Write the end quote...
|
||
2647 | */
|
||
2648 | |||
2649 | if ((*putc_cb)(quote, p) < 0) |
||
2650 | return (-1); |
||
2651 | } |
||
2652 | else
|
||
2653 | { |
||
2654 | /*
|
||
2655 | * Write a non-quoted name string...
|
||
2656 | */
|
||
2657 | |||
2658 | while (*s)
|
||
2659 | { |
||
2660 | if ((*putc_cb)(*s, p) < 0) |
||
2661 | return (-1); |
||
2662 | |||
2663 | s ++; |
||
2664 | } |
||
2665 | } |
||
2666 | |||
2667 | return (0); |
||
2668 | } |
||
2669 | |||
2670 | |||
2671 | /*
|
||
2672 | * 'mxml_write_node()' - Save an XML node to a file.
|
||
2673 | */
|
||
2674 | |||
2675 | static int /* O - Column or -1 on error */ |
||
2676 | mxml_write_node(mxml_node_t *node, /* I - Node to write */
|
||
2677 | void *p, /* I - File to write to */ |
||
2678 | mxml_save_cb_t cb, /* I - Whitespace callback */
|
||
2679 | int col, /* I - Current column */ |
||
2680 | _mxml_putc_cb_t putc_cb,/* I - Output callback */
|
||
2681 | _mxml_global_t *global)/* I - Global data */
|
||
2682 | { |
||
2683 | int i, /* Looping var */ |
||
2684 | width; /* Width of attr + value */
|
||
2685 | mxml_attr_t *attr; /* Current attribute */
|
||
2686 | char s[255]; /* Temporary string */ |
||
2687 | |||
2688 | |||
2689 | while (node != NULL) |
||
2690 | { |
||
2691 | /*
|
||
2692 | * Print the node value...
|
||
2693 | */
|
||
2694 | |||
2695 | switch (node->type)
|
||
2696 | { |
||
2697 | case MXML_ELEMENT :
|
||
2698 | col = mxml_write_ws(node, p, cb, MXML_WS_BEFORE_OPEN, col, putc_cb); |
||
2699 | |||
2700 | if ((*putc_cb)('<', p) < 0) |
||
2701 | return (-1); |
||
2702 | if (node->value.element.name[0] == '?' || |
||
2703 | !strncmp(node->value.element.name, "!--", 3) || |
||
2704 | !strncmp(node->value.element.name, "![CDATA[", 8)) |
||
2705 | { |
||
2706 | /*
|
||
2707 | * Comments, CDATA, and processing instructions do not
|
||
2708 | * use character entities.
|
||
2709 | */
|
||
2710 | |||
2711 | const char *ptr; /* Pointer into name */ |
||
2712 | |||
2713 | |||
2714 | for (ptr = node->value.element.name; *ptr; ptr ++)
|
||
2715 | if ((*putc_cb)(*ptr, p) < 0) |
||
2716 | return (-1); |
||
2717 | } |
||
2718 | else if (mxml_write_name(node->value.element.name, p, putc_cb) < 0) |
||
2719 | return (-1); |
||
2720 | |||
2721 | col += strlen(node->value.element.name) + 1;
|
||
2722 | |||
2723 | for (i = node->value.element.num_attrs, attr = node->value.element.attrs;
|
||
2724 | i > 0;
|
||
2725 | i --, attr ++) |
||
2726 | { |
||
2727 | width = strlen(attr->name); |
||
2728 | |||
2729 | if (attr->value)
|
||
2730 | width += strlen(attr->value) + 3;
|
||
2731 | |||
2732 | if (global->wrap > 0 && (col + width) > global->wrap) |
||
2733 | { |
||
2734 | if ((*putc_cb)('\n', p) < 0) |
||
2735 | return (-1); |
||
2736 | |||
2737 | col = 0;
|
||
2738 | } |
||
2739 | else
|
||
2740 | { |
||
2741 | if ((*putc_cb)(' ', p) < 0) |
||
2742 | return (-1); |
||
2743 | |||
2744 | col ++; |
||
2745 | } |
||
2746 | |||
2747 | if (mxml_write_name(attr->name, p, putc_cb) < 0) |
||
2748 | return (-1); |
||
2749 | |||
2750 | if (attr->value)
|
||
2751 | { |
||
2752 | if ((*putc_cb)('=', p) < 0) |
||
2753 | return (-1); |
||
2754 | if ((*putc_cb)('\"', p) < 0) |
||
2755 | return (-1); |
||
2756 | if (mxml_write_string(attr->value, p, putc_cb) < 0) |
||
2757 | return (-1); |
||
2758 | if ((*putc_cb)('\"', p) < 0) |
||
2759 | return (-1); |
||
2760 | } |
||
2761 | |||
2762 | col += width; |
||
2763 | } |
||
2764 | |||
2765 | if (node->child)
|
||
2766 | { |
||
2767 | /*
|
||
2768 | * Write children...
|
||
2769 | */
|
||
2770 | |||
2771 | if ((*putc_cb)('>', p) < 0) |
||
2772 | return (-1); |
||
2773 | else
|
||
2774 | col ++; |
||
2775 | |||
2776 | col = mxml_write_ws(node, p, cb, MXML_WS_AFTER_OPEN, col, putc_cb); |
||
2777 | |||
2778 | if ((col = mxml_write_node(node->child, p, cb, col, putc_cb,
|
||
2779 | global)) < 0)
|
||
2780 | return (-1); |
||
2781 | |||
2782 | /*
|
||
2783 | * The ? and ! elements are special-cases and have no end tags...
|
||
2784 | */
|
||
2785 | |||
2786 | if (node->value.element.name[0] != '!' && |
||
2787 | node->value.element.name[0] != '?') |
||
2788 | { |
||
2789 | col = mxml_write_ws(node, p, cb, MXML_WS_BEFORE_CLOSE, col, putc_cb); |
||
2790 | |||
2791 | if ((*putc_cb)('<', p) < 0) |
||
2792 | return (-1); |
||
2793 | if ((*putc_cb)('/', p) < 0) |
||
2794 | return (-1); |
||
2795 | if (mxml_write_string(node->value.element.name, p, putc_cb) < 0) |
||
2796 | return (-1); |
||
2797 | if ((*putc_cb)('>', p) < 0) |
||
2798 | return (-1); |
||
2799 | |||
2800 | col += strlen(node->value.element.name) + 3;
|
||
2801 | |||
2802 | col = mxml_write_ws(node, p, cb, MXML_WS_AFTER_CLOSE, col, putc_cb); |
||
2803 | } |
||
2804 | } |
||
2805 | else if (node->value.element.name[0] == '!' || |
||
2806 | node->value.element.name[0] == '?') |
||
2807 | { |
||
2808 | /*
|
||
2809 | * The ? and ! elements are special-cases...
|
||
2810 | */
|
||
2811 | |||
2812 | if ((*putc_cb)('>', p) < 0) |
||
2813 | return (-1); |
||
2814 | else
|
||
2815 | col ++; |
||
2816 | |||
2817 | col = mxml_write_ws(node, p, cb, MXML_WS_AFTER_OPEN, col, putc_cb); |
||
2818 | } |
||
2819 | else
|
||
2820 | { |
||
2821 | if ((*putc_cb)(' ', p) < 0) |
||
2822 | return (-1); |
||
2823 | if ((*putc_cb)('/', p) < 0) |
||
2824 | return (-1); |
||
2825 | if ((*putc_cb)('>', p) < 0) |
||
2826 | return (-1); |
||
2827 | |||
2828 | col += 3;
|
||
2829 | |||
2830 | col = mxml_write_ws(node, p, cb, MXML_WS_AFTER_OPEN, col, putc_cb); |
||
2831 | } |
||
2832 | break;
|
||
2833 | |||
2834 | case MXML_INTEGER :
|
||
2835 | if (node->prev)
|
||
2836 | { |
||
2837 | if (global->wrap > 0 && col > global->wrap) |
||
2838 | { |
||
2839 | if ((*putc_cb)('\n', p) < 0) |
||
2840 | return (-1); |
||
2841 | |||
2842 | col = 0;
|
||
2843 | } |
||
2844 | else if ((*putc_cb)(' ', p) < 0) |
||
2845 | return (-1); |
||
2846 | else
|
||
2847 | col ++; |
||
2848 | } |
||
2849 | |||
2850 | sprintf(s, "%d", node->value.integer);
|
||
2851 | if (mxml_write_string(s, p, putc_cb) < 0) |
||
2852 | return (-1); |
||
2853 | |||
2854 | col += strlen(s); |
||
2855 | break;
|
||
2856 | |||
2857 | case MXML_OPAQUE :
|
||
2858 | if (mxml_write_string(node->value.opaque, p, putc_cb) < 0) |
||
2859 | return (-1); |
||
2860 | |||
2861 | col += strlen(node->value.opaque); |
||
2862 | break;
|
||
2863 | |||
2864 | case MXML_REAL :
|
||
2865 | if (node->prev)
|
||
2866 | { |
||
2867 | if (global->wrap > 0 && col > global->wrap) |
||
2868 | { |
||
2869 | if ((*putc_cb)('\n', p) < 0) |
||
2870 | return (-1); |
||
2871 | |||
2872 | col = 0;
|
||
2873 | } |
||
2874 | else if ((*putc_cb)(' ', p) < 0) |
||
2875 | return (-1); |
||
2876 | else
|
||
2877 | col ++; |
||
2878 | } |
||
2879 | |||
2880 | sprintf(s, "%f", node->value.real);
|
||
2881 | if (mxml_write_string(s, p, putc_cb) < 0) |
||
2882 | return (-1); |
||
2883 | |||
2884 | col += strlen(s); |
||
2885 | break;
|
||
2886 | |||
2887 | case MXML_TEXT :
|
||
2888 | if (node->value.text.whitespace && col > 0) |
||
2889 | { |
||
2890 | if (global->wrap > 0 && col > global->wrap) |
||
2891 | { |
||
2892 | if ((*putc_cb)('\n', p) < 0) |
||
2893 | return (-1); |
||
2894 | |||
2895 | col = 0;
|
||
2896 | } |
||
2897 | else if ((*putc_cb)(' ', p) < 0) |
||
2898 | return (-1); |
||
2899 | else
|
||
2900 | col ++; |
||
2901 | } |
||
2902 | |||
2903 | if (mxml_write_string(node->value.text.string, p, putc_cb) < 0) |
||
2904 | return (-1); |
||
2905 | |||
2906 | col += strlen(node->value.text.string); |
||
2907 | break;
|
||
2908 | |||
2909 | case MXML_CUSTOM :
|
||
2910 | if (global->custom_save_cb)
|
||
2911 | { |
||
2912 | char *data; /* Custom data string */ |
||
2913 | const char *newline; /* Last newline in string */ |
||
2914 | |||
2915 | |||
2916 | if ((data = (*global->custom_save_cb)(node)) == NULL) |
||
2917 | return (-1); |
||
2918 | |||
2919 | if (mxml_write_string(data, p, putc_cb) < 0) |
||
2920 | return (-1); |
||
2921 | |||
2922 | if ((newline = strrchr(data, '\n')) == NULL) |
||
2923 | col += strlen(data); |
||
2924 | else
|
||
2925 | col = strlen(newline); |
||
2926 | |||
2927 | free(data); |
||
2928 | break;
|
||
2929 | } |
||
2930 | |||
2931 | default : /* Should never happen */ |
||
2932 | return (-1); |
||
2933 | } |
||
2934 | |||
2935 | /*
|
||
2936 | * Next node...
|
||
2937 | */
|
||
2938 | |||
2939 | node = node->next; |
||
2940 | } |
||
2941 | |||
2942 | return (col);
|
||
2943 | } |
||
2944 | |||
2945 | |||
2946 | /*
|
||
2947 | * 'mxml_write_string()' - Write a string, escaping & and < as needed.
|
||
2948 | */
|
||
2949 | |||
2950 | static int /* O - 0 on success, -1 on failure */ |
||
2951 | mxml_write_string( |
||
2952 | const char *s, /* I - String to write */ |
||
2953 | void *p, /* I - Write pointer */ |
||
2954 | _mxml_putc_cb_t putc_cb) /* I - Write callback */
|
||
2955 | { |
||
2956 | const char *name; /* Entity name, if any */ |
||
2957 | |||
2958 | |||
2959 | while (*s)
|
||
2960 | { |
||
2961 | if ((name = mxmlEntityGetName(*s)) != NULL) |
||
2962 | { |
||
2963 | if ((*putc_cb)('&', p) < 0) |
||
2964 | return (-1); |
||
2965 | |||
2966 | while (*name)
|
||
2967 | { |
||
2968 | if ((*putc_cb)(*name, p) < 0) |
||
2969 | return (-1); |
||
2970 | name ++; |
||
2971 | } |
||
2972 | |||
2973 | if ((*putc_cb)(';', p) < 0) |
||
2974 | return (-1); |
||
2975 | } |
||
2976 | else if ((*putc_cb)(*s, p) < 0) |
||
2977 | return (-1); |
||
2978 | |||
2979 | s ++; |
||
2980 | } |
||
2981 | |||
2982 | return (0); |
||
2983 | } |
||
2984 | |||
2985 | |||
2986 | /*
|
||
2987 | * 'mxml_write_ws()' - Do whitespace callback...
|
||
2988 | */
|
||
2989 | |||
2990 | static int /* O - New column */ |
||
2991 | mxml_write_ws(mxml_node_t *node, /* I - Current node */
|
||
2992 | void *p, /* I - Write pointer */ |
||
2993 | mxml_save_cb_t cb, /* I - Callback function */
|
||
2994 | int ws, /* I - Where value */ |
||
2995 | int col, /* I - Current column */ |
||
2996 | _mxml_putc_cb_t putc_cb) /* I - Write callback */
|
||
2997 | { |
||
2998 | const char *s; /* Whitespace string */ |
||
2999 | |||
3000 | |||
3001 | if (cb && (s = (*cb)(node, ws)) != NULL) |
||
3002 | { |
||
3003 | while (*s)
|
||
3004 | { |
||
3005 | if ((*putc_cb)(*s, p) < 0) |
||
3006 | return (-1); |
||
3007 | else if (*s == '\n') |
||
3008 | col = 0;
|
||
3009 | else if (*s == '\t') |
||
3010 | { |
||
3011 | col += MXML_TAB; |
||
3012 | col = col - (col % MXML_TAB); |
||
3013 | } |
||
3014 | else
|
||
3015 | col ++; |
||
3016 | |||
3017 | s ++; |
||
3018 | } |
||
3019 | } |
||
3020 | |||
3021 | return (col);
|
||
3022 | } |
||
3023 | |||
3024 | |||
3025 | /*
|
||
3026 | * End of "$Id: mxml-file.c 391 2009-05-17 05:20:52Z mike $".
|
||
3027 | */ |