Mercurial > repo

comparison interps/cfunge/cfunge-src/lib/genx/genx.c @ 996:859f9b4339e6

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
<Gregor> tar xf egobot.tar.xz
author HackBot
date Sun, 09 Dec 2012 19:30:08 +0000
parents
children
comparison
equal deleted inserted replaced
995:6883f5911eb7 996:859f9b4339e6
1 /*
2 * Copyright (c) 2004 by Tim Bray and Sun Microsystems. For copying
3 * permission, see http://www.tbray.org/ongoing/genx/COPYING
4 */
5
6 #include "../../src/global.h"
7
8 #define GENX_VERSION "beta5"
9 #define GENX_INTERNAL
10
11 #include "genx.h"
12
13 #if !defined(CFUN_NO_FLOATS) && !defined(CFUN_NO_TURT)
14
15 #include <stdlib.h>
16 #include <stdio.h>
17 #include <string.h>
18 #include <stdbool.h>
19
20 #define Boolean bool
21 #define True 1
22 #define False 0
23 #define STRLEN_XMLNS_COLON 6
24
25 // typedef void * (* genxAlloc)(void * userData, size_t bytes);
26 // typedef void (* genxDealloc)(void * userData, void * data);
27
28 /**
29 * writer state
30 */
31 typedef enum {
32 SEQUENCE_NO_DOC,
33 SEQUENCE_PRE_DOC,
34 SEQUENCE_POST_DOC,
35 SEQUENCE_START_TAG,
36 SEQUENCE_ATTRIBUTES,
37 SEQUENCE_CONTENT
38 } writerSequence;
39
40 /**
41 * generic pointer list
42 */
43 typedef struct genx_plist {
44 genxWriter writer;
45 ssize_t count;
46 size_t space;
47 void * * pointers;
48 } plist;
49
50 /**
51 * text collector, for attribute values
52 */
53 typedef struct genx_collector {
54 utf8 buf;
55 size_t used;
56 size_t space;
57 } collector;
58
59 /*******************************
60 * Structs with opaquely-exposed handles
61 */
62
63 /**
64 * This one's tricky, to handle stacking namespaces
65 * 'declaration' is the current attribute which would be used to
66 * declare the currently-effective prefix
67 * 'defDeclaration' is a appropriate declaration when this is being
68 * used with the default prefix as passed to genxDeclareNamespace
69 * baroque is true if this namespace has been used with more than one
70 * prefix, or is the default namespace but has been unset
71 */
72 struct genxNamespace_rec {
73 genxWriter writer;
74 utf8 name;
75 genxAttribute declaration;
76 genxAttribute defaultDecl;
77 int declCount;
78 Boolean baroque:1;
79 };
80
81 struct genxElement_rec {
82 genxWriter writer;
83 utf8 type;
84 genxNamespace ns;
85 };
86
87 typedef enum {
88 ATTR_NSDECL,
89 ATTR_NAKED,
90 ATTR_PREFIXED
91 } attrType;
92
93 struct genxAttribute_rec {
94 genxWriter writer;
95 utf8 name;
96 genxNamespace ns;
97 collector value;
98 int provided; /**< provided for current element? */
99 attrType atype;
100 };
101
102 /**
103 * genx's sandbox
104 */
105 struct genxWriter_rec {
106 FILE * file;
107 genxSender * sender;
108 genxStatus status;
109 writerSequence sequence;
110 char xmlChars[0x10000];
111 void * userData;
112 utf8 empty;
113 int nextPrefix;
114 Boolean defaultNsDeclared:1;
115 genxAttribute xmlnsEquals;
116 genxElement nowStarting;
117 plist namespaces;
118 plist elements;
119 plist attributes;
120 plist prefixes;
121 plist stack;
122 struct genxAttribute_rec arec;
123 const char * etext[100];
124 // genxAlloc alloc;
125 // genxDealloc dealloc;
126 };
127
128 /*******************************
129 * Forward declarations
130 */
131 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
132 static genxAttribute declareAttribute(genxWriter w, genxNamespace ns,
133 constUtf8 name, constUtf8 valuestr,
134 genxStatus * statusP);
135 FUNGE_ATTR_FAST
136 static genxStatus addNamespace(genxNamespace ns, constUtf8 prefix);
137 FUNGE_ATTR_FAST
138 static genxStatus unsetDefaultNamespace(genxWriter w);
139 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
140 static genxStatus addAttribute(genxAttribute a, constUtf8 valuestr);
141
142 /*******************************
143 * End of declarations
144 */
145
146 /*******************************
147 * private memory utilities
148 */
149 #if 0
150 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED FUNGE_ATTR_MALLOC
151 static inline void * allocate(size_t bytes)
152 {
153 return (void *) malloc(bytes);
154 }
155
156 FUNGE_ATTR_FAST
157 static inline void deallocate(void * data)
158 {
159 free(data);
160 }
161 #endif
162 #define allocate(___bytes) malloc(___bytes)
163 #define deallocate(___ptr) free(___ptr)
164
165 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
166 static utf8 copy(constUtf8 restrict from)
167 {
168 utf8 restrict temp;
169 size_t len = strlen((const char *) from);
170
171 if ((temp = (utf8) allocate(len + 1)) == NULL)
172 return NULL;
173 memcpy(temp, from, len);
174 temp[len] = '\0';
175 return temp;
176 }
177
178 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
179 static genxStatus initCollector(collector * restrict c)
180 {
181 c->space = 100;
182 if ((c->buf = (utf8) allocate(c->space)) == NULL)
183 return GENX_ALLOC_FAILED;
184 c->used = 0;
185 return GENX_SUCCESS;
186 }
187
188 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
189 static genxStatus growCollector(collector * c, size_t size)
190 {
191 utf8 newSpace;
192
193 c->space = size * 2;
194 if ((newSpace = (utf8) allocate(c->space)) == NULL)
195 return GENX_ALLOC_FAILED;
196
197 strncpy((char *) newSpace, (const char *) c->buf, c->used);
198 newSpace[c->used] = 0;
199 deallocate(c->buf);
200 c->buf = newSpace;
201 return GENX_SUCCESS;
202 }
203
204 FUNGE_ATTR_FAST static inline void startCollect(collector * c)
205 {
206 c->used = 0;
207 }
208 FUNGE_ATTR_FAST static inline void endCollect(collector * c)
209 {
210 c->buf[c->used] = 0;
211 }
212
213 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
214 static genxStatus collectString(genxWriter w, collector * restrict c, constUtf8 restrict string)
215 {
216 size_t sl = strlen((const char *) string);
217
218 if (sl >= c->space)
219 if ((w->status = growCollector(c, sl)) != GENX_SUCCESS)
220 return GENX_ALLOC_FAILED;
221
222 strcpy((char *) c->buf, (const char *) string);
223 return GENX_SUCCESS;
224 }
225
226 #define collectPiece(w,c,d,size) {if (((c)->used+(size))>=(c)->space){if (((w)->status=growCollector(c,(c)->used+(size)))!=GENX_SUCCESS) return (w)->status;}strncpy((char *)(c)->buf+(c)->used,d,size);(c)->used+=size;}
227
228 /*******************************
229 * private list utilities
230 */
231 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
232 static genxStatus initPlist(genxWriter w, plist * restrict pl)
233 {
234 pl->writer = w;
235 pl->count = 0;
236 pl->space = 10;
237 pl->pointers = (void * *) allocate(pl->space * sizeof(void *));
238 if (pl->pointers == NULL)
239 return GENX_ALLOC_FAILED;
240
241 return GENX_SUCCESS;
242 }
243
244 /**
245 * make room in a plist
246 */
247 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
248 static Boolean checkExpand(plist * restrict pl)
249 {
250 void ** restrict newlist;
251
252 if ((size_t)pl->count < pl->space)
253 return True;
254
255 pl->space *= 2;
256 newlist = (void **) allocate(pl->space * sizeof(void *));
257 if (newlist == NULL)
258 return False;
259 // This is to allow vectorising.
260 {
261 const ssize_t count = pl->count;
262 void ** restrict oldlist = pl->pointers;
263 for (ssize_t i = 0; i < count; i++)
264 newlist[i] = oldlist[i];
265 }
266 deallocate(pl->pointers);
267 pl->pointers = newlist;
268
269 return True;
270 }
271
272 /**
273 * stick something on the end of a plist
274 */
275 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
276 static genxStatus listAppend(plist * pl, void * pointer)
277 {
278 if (!checkExpand(pl))
279 return GENX_ALLOC_FAILED;
280
281 pl->pointers[pl->count++] = pointer;
282 return GENX_SUCCESS;
283 }
284
285 /**
286 * insert in place, shuffling up
287 */
288 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
289 static genxStatus listInsert(plist * pl, void * pointer, ssize_t at)
290 {
291 ssize_t i;
292
293 if (!checkExpand(pl))
294 return GENX_ALLOC_FAILED;
295
296 for (i = pl->count; i > at; i--)
297 pl->pointers[i] = pl->pointers[i - 1];
298 pl->count++;
299
300 pl->pointers[at] = pointer;
301 return GENX_SUCCESS;
302 }
303
304 /*******************************
305 * list lookups
306 */
307
308 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
309 static genxNamespace findNamespace(plist * pl, constUtf8 uri)
310 {
311 ssize_t i;
312 genxNamespace * nn = (genxNamespace *) pl->pointers;
313
314 for (i = 0; i < pl->count; i++)
315 if (strcmp((const char *) uri, (const char *) nn[i]->name) == 0)
316 return nn[i];
317
318 return NULL;
319 }
320
321 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
322 static genxElement findElement(plist * pl, constUtf8 xmlns, constUtf8 type)
323 {
324 ssize_t i;
325 genxElement * ee = (genxElement *) pl->pointers;
326
327 for (i = 0; i < pl->count; i++) {
328 if (xmlns == NULL) {
329 if (ee[i]->ns == NULL && strcmp((const char *) type,
330 (const char *) ee[i]->type) == 0)
331 return ee[i];
332 } else {
333 if (ee[i]->ns != NULL &&
334 strcmp((const char *) xmlns, (const char *) ee[i]->ns->name) == 0 &&
335 strcmp((const char *) type, (const char *) ee[i]->type) == 0)
336 return ee[i];
337 }
338 }
339
340 return NULL;
341 }
342
343 /**
344 * store & intern a prefix, after giving it the
345 * "xmlns:" prefix. Don't allow storing the same one twice unless 'force'
346 * is set.
347 */
348 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
349 static constUtf8 storePrefix(genxWriter w, constUtf8 prefix, Boolean force)
350 {
351 ssize_t high, low;
352 utf8 * pp = (utf8 *) w->prefixes.pointers;
353
354 if (prefix[0] == 0)
355 prefix = (constUtf8) "xmlns";
356 else {
357 unsigned char buf[1024];
358 snprintf((char *) buf, sizeof(buf), "xmlns:%s", (const char*)prefix);
359 prefix = buf;
360 }
361
362 high = w->prefixes.count; low = -1;
363 while (high - low > 1) {
364 ssize_t probe = (high + low) / 2;
365 if (strcmp((const char *) prefix, (const char *) pp[probe]) < 0)
366 high = probe;
367 else
368 low = probe;
369 }
370
371 /* already there? */
372 if (low != -1 && strcmp((const char *) prefix, (const char *) pp[low]) == 0) {
373 if (force)
374 return pp[low];
375
376 w->status = GENX_DUPLICATE_PREFIX;
377 return NULL;
378 }
379
380 /* copy & insert */
381 if ((prefix = copy(prefix)) == NULL) {
382 w->status = GENX_ALLOC_FAILED;
383 return NULL;
384 }
385
386 w->status = listInsert(&w->prefixes, (void *) prefix, high);
387 if (w->status != GENX_SUCCESS)
388 return NULL;
389
390 return (constUtf8) prefix;
391 }
392
393 /*******************************
394 * UTF8 bit-banging
395 */
396
397 /**
398 * Retrieve the character pointed at, and advance the pointer; return -1 on
399 * error
400 */
401 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
402 int genxNextUnicodeChar(constUtf8 * sp)
403 {
404 constUtf8 s = (constUtf8) * sp;
405 int c;
406
407 if (*s == 0)
408 return -1;
409
410 if (*s < 0x80)
411 c = *s++;
412
413 /* all this encoding sanity-checking taken from section 3.10 of Unicode 4 */
414 else if (*s < 0xc2)
415 goto malformed;
416
417 /* 2-byte encodings, first byte c2 .. df */
418 else if (*s < 0xe0) {
419 c = (*s++ & 0x1f) << 6;
420
421 /*
422 * for this common idiom, if ((c & 0xc0) != 0x80) is slightly faster
423 * on MacOS (PPC)
424 */
425 if (*s < 0x80 || *s > 0xbf)
426 goto malformed;
427
428 c |= *s++ & 0x3f;
429 }
430
431 /* 3-byte encodings, first byte e0 .. ef */
432 else if (*s < 0xf0) {
433 int b0 = *s;
434 c = (*s++ & 0x0f) << 12;
435
436 if ((b0 == 0xe0 && (*s < 0xa0 || *s > 0xbf)) ||
437 (b0 < 0xed && (*s < 0x80 || *s > 0xbf)) ||
438 (b0 == 0xed && (*s < 0x80 || *s > 0x9f)) ||
439 (b0 > 0xed && (*s < 0x80 || *s > 0xbf)))
440 goto malformed;
441
442 c |= (*s++ & 0x3f) << 6;
443
444 if (*s < 0x80 || *s > 0xbf)
445 goto malformed;
446
447 c |= *s++ & 0x3f;
448 }
449
450 /* 4-byte encodings, first byte f0 .. f4 */
451 else if (*s < 0xf5) {
452 int b0 = *s;
453 c = (*s++ & 0x07) << 18;
454
455 if ((b0 == 0xf0 && (*s < 0x90 || *s > 0xbf)) ||
456 (b0 < 0xf4 && (*s < 0x80 || *s > 0xbf)) ||
457 (b0 >= 0xf4 && (*s < 0x80 || *s > 0x8f)))
458 goto malformed;
459
460 c |= (*s++ & 0x3f) << 12;
461
462 if (*s < 0x80 || *s > 0xbf)
463 goto malformed;
464
465 c |= (*s++ & 0x3f) << 6;
466
467 if (*s < 0x80 || *s > 0xbf)
468 goto malformed;
469
470 c |= *s++ & 0x3f;
471 } else
472 goto malformed;
473
474 *sp = s;
475 return c;
476
477 /*
478 * this is needed by scrubText, which wants to get the pointer moved
479 * past the problem area.
480 */
481 malformed:
482 if (*s)
483 ++s;
484 *sp = s;
485 return -1;
486 }
487
488 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
489 static inline Boolean isXMLChar(const genxWriter restrict w, int c)
490 {
491 if (c < 0)
492 return False;
493 else if (c < 0x10000)
494 return (int) w->xmlChars[c];
495 else
496 return (c <= 0x10ffff);
497 }
498
499 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
500 static inline Boolean isLetter(const genxWriter restrict w, int c)
501 {
502 if (c < 0 || c > 0xffff)
503 return False;
504 else
505 return w->xmlChars[c] & GENX_LETTER;
506 }
507
508 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
509 static inline Boolean isNameChar(const genxWriter restrict w, int c)
510 {
511 if (c < 0 || c > 0xffff)
512 return False;
513 else
514 return w->xmlChars[c] & GENX_NAMECHAR;
515 }
516
517 /*******************************
518 * Constructors, setters/getters
519 */
520
521 /**
522 * Construct a new genxWriter
523 */
524 FUNGE_ATTR_FAST
525 genxWriter genxNew(void)
526 {
527 genxWriter w;
528 genxNamespace xml;
529
530 //if (alloc)
531 // w = (genxWriter)(*alloc)(userData, sizeof(struct genxWriter_rec));
532 //else
533 w = (genxWriter) malloc(sizeof(struct genxWriter_rec));
534
535 if (w == NULL)
536 return NULL;
537
538 w->status = GENX_SUCCESS;
539 // w->alloc = NULL;
540 // w->dealloc = NULL;
541 w->userData = NULL;
542 w->sequence = SEQUENCE_NO_DOC;
543
544 if (initPlist(w, &w->namespaces) != GENX_SUCCESS ||
545 initPlist(w, &w->elements) != GENX_SUCCESS ||
546 initPlist(w, &w->attributes) != GENX_SUCCESS ||
547 initPlist(w, &w->prefixes) != GENX_SUCCESS ||
548 initPlist(w, &w->stack) != GENX_SUCCESS)
549 return NULL;
550
551 if ((w->status = initCollector(&w->arec.value)) != GENX_SUCCESS)
552 return NULL;
553
554 if ((w->empty = copy((constUtf8) "")) == NULL) {
555 w->status = GENX_ALLOC_FAILED;
556 return NULL;
557 }
558
559 w->xmlnsEquals = declareAttribute(w, NULL, (constUtf8) "xmlns", NULL, &w->status);
560 if (w->xmlnsEquals == NULL || w->status != GENX_SUCCESS)
561 return NULL;
562 w->defaultNsDeclared = False;
563
564 w->nextPrefix = 1;
565
566 genxSetCharProps(w->xmlChars);
567
568 w->etext[GENX_SUCCESS] = "Success";
569 w->etext[GENX_BAD_UTF8] = "Bad UTF8";
570 w->etext[GENX_NON_XML_CHARACTER] = "Non XML Character";
571 w->etext[GENX_BAD_NAME] = "Bad NAME";
572 w->etext[GENX_ALLOC_FAILED] = "Memory allocation failed";
573 w->etext[GENX_BAD_NAMESPACE_NAME] = "Bad namespace name";
574 w->etext[GENX_INTERNAL_ERROR] = "Internal error";
575 w->etext[GENX_DUPLICATE_PREFIX] = "Duplicate prefix";
576 w->etext[GENX_SEQUENCE_ERROR] = "Call out of sequence";
577 w->etext[GENX_NO_START_TAG] = "No Start-tag for EndElement call";
578 w->etext[GENX_IO_ERROR] = "I/O error";
579 w->etext[GENX_MISSING_VALUE] = "Missing attribute value";
580 w->etext[GENX_MALFORMED_COMMENT] = "Malformed comment body";
581 w->etext[GENX_MALFORMED_PI] = "?> in PI";
582 w->etext[GENX_XML_PI_TARGET] = "Target of PI matches [xX][mM][lL]";
583 w->etext[GENX_DUPLICATE_ATTRIBUTE] =
584 "Same attribute specified more than once";
585 w->etext[GENX_ATTRIBUTE_IN_DEFAULT_NAMESPACE] =
586 "Attribute cannot be in default namespace";
587 w->etext[GENX_DUPLICATE_NAMESPACE] =
588 "Declared namespace twice with different prefixes on one element.";
589 w->etext[GENX_BAD_DEFAULT_DECLARATION] =
590 "Declared a default namespace on an element which is in no namespace";
591
592 /* the xml: namespace is pre-wired */
593 xml = genxDeclareNamespace(w, (constUtf8) "http://www.w3.org/XML/1998/namespace",
594 (constUtf8) "xml", &w->status);
595 if (xml == NULL)
596 return NULL;
597 xml->declCount = 1;
598 xml->declaration = xml->defaultDecl;
599
600 return w;
601 }
602
603 /*
604 * get/set userData
605 */
606 // FUNGE_ATTR_FAST void genxSetUserData(genxWriter w, void * userData)
607 // {
608 // w->userData = userData;
609 // }
610 // FUNGE_ATTR_FAST void * genxGetUserData(genxWriter w)
611 // {
612 // return w->userData;
613 // }
614
615 /*
616 * get/set allocator
617 */
618 // FUNGE_ATTR_FAST void genxSetAlloc(genxWriter w, genxAlloc alloc)
619 // {
620 // w->alloc = alloc;
621 // }
622 // FUNGE_ATTR_FAST
623 // void genxSetDealloc(genxWriter w, genxDealloc dealloc)
624 // {
625 // w->dealloc = dealloc;
626 // }
627 // FUNGE_ATTR_FAST
628 // genxAlloc genxGetAlloc(genxWriter w)
629 // {
630 // return w->alloc;
631 // }
632 // FUNGE_ATTR_FAST
633 // genxDealloc genxGetDealloc(genxWriter w)
634 // {
635 // return w->dealloc;
636 // }
637
638 /*
639 * Clean up
640 */
641 FUNGE_ATTR_FAST void genxDispose(genxWriter w)
642 {
643 ssize_t i;
644 genxNamespace * nn = (genxNamespace *) w->namespaces.pointers;
645 genxElement * ee = (genxElement *) w->elements.pointers;
646 genxAttribute * aa = (genxAttribute *) w->attributes.pointers;
647 utf8 * pp = (utf8 *) w->prefixes.pointers;
648
649 for (i = 0; i < w->namespaces.count; i++) {
650 deallocate(nn[i]->name);
651 deallocate(nn[i]);
652 }
653
654 for (i = 0; i < w->elements.count; i++) {
655 deallocate(ee[i]->type);
656 deallocate(ee[i]);
657 }
658
659 for (i = 0; i < w->attributes.count; i++) {
660 deallocate(aa[i]->name);
661 deallocate(aa[i]->value.buf);
662 deallocate(aa[i]);
663 }
664
665 for (i = 0; i < w->prefixes.count; i++)
666 deallocate(pp[i]);
667
668 deallocate(w->namespaces.pointers);
669 deallocate(w->elements.pointers);
670 deallocate(w->attributes.pointers);
671 deallocate(w->prefixes.pointers);
672 deallocate(w->stack.pointers);
673
674 deallocate(w->arec.value.buf);
675
676 deallocate(w->empty);
677
678 /* how Oscar dealt with Igli */
679 deallocate(w);
680 }
681
682 /*******************************
683 * External utility routines
684 */
685
686 /*
687 * scan a buffer and report problems with UTF-8 encoding or non-XML characters
688 */
689 FUNGE_ATTR_FAST genxStatus genxCheckText(const genxWriter restrict w, constUtf8 s)
690 {
691 while (*s) {
692 int c = genxNextUnicodeChar(&s);
693 if (c == -1)
694 return GENX_BAD_UTF8;
695
696 if (!isXMLChar(w, c))
697 return GENX_NON_XML_CHARACTER;
698 }
699 return GENX_SUCCESS;
700 }
701
702 /*
703 * Purify some text
704 */
705 FUNGE_ATTR_FAST int genxScrubText(const genxWriter restrict w, constUtf8 in, utf8 restrict out)
706 {
707 int problems = 0;
708 constUtf8 last = in;
709
710 while (*in) {
711 int c = genxNextUnicodeChar(&in);
712 if (c == -1) {
713 problems++;
714 last = in;
715 continue;
716 }
717
718 if (!isXMLChar(w, c)) {
719 problems++;
720 last = in;
721 continue;
722 }
723
724 while (last < in)
725 *out++ = *last++;
726 }
727 *out = 0;
728 return problems;
729 }
730
731 /*
732 * check one character
733 */
734 FUNGE_ATTR_FAST int genxCharClass(const genxWriter restrict w, int c)
735 {
736 int ret = 0;
737
738 if (isXMLChar(w, c))
739 ret |= GENX_XML_CHAR;
740 if (isNameChar(w, c))
741 ret |= GENX_NAMECHAR;
742 if (isLetter(w, c))
743 ret |= GENX_LETTER;
744 return ret;
745 }
746
747 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
748 static genxStatus checkNCName(genxWriter w, constUtf8 name)
749 {
750 int c;
751
752 if (name == NULL || *name == 0)
753 return GENX_BAD_NAME;
754
755 c = genxNextUnicodeChar(&name);
756 if (!isLetter(w, c) && c != ':' && c != '_')
757 return GENX_BAD_NAME;
758
759 while (*name) {
760 c = genxNextUnicodeChar(&name);
761 if (c == -1)
762 return GENX_BAD_UTF8;
763 if (!isNameChar(w, c))
764 return GENX_BAD_NAME;
765 }
766 return GENX_SUCCESS;
767 }
768
769 FUNGE_ATTR_FAST const char * genxGetErrorMessage(const genxWriter restrict w, genxStatus status)
770 {
771 return w->etext[status];
772 }
773 FUNGE_ATTR_FAST const char * genxLastErrorMessage(const genxWriter restrict w)
774 {
775 return w->etext[w->status];
776 }
777
778 /*******************************
779 * Declarations: namespace/element/attribute
780 */
781
782 /*
783 * DeclareNamespace - by far the most complex routine in Genx
784 */
785 FUNGE_ATTR_FAST
786 genxNamespace genxDeclareNamespace(genxWriter w, constUtf8 uri,
787 constUtf8 defaultPref,
788 genxStatus * statusP)
789 {
790 genxNamespace ns;
791 genxAttribute defaultDecl;
792 unsigned char newPrefix[100];
793
794 if (uri == NULL || uri[0] == 0) {
795 w->status = GENX_BAD_NAMESPACE_NAME;
796 goto busted;
797 }
798
799 if ((w->status = genxCheckText(w, uri)) != GENX_SUCCESS)
800 goto busted;
801
802 /* if a prefix is provided, it has to be an NCname */
803 if (defaultPref != NULL && defaultPref[0] != 0 &&
804 (w->status = checkNCName(w, defaultPref)) != GENX_SUCCESS)
805 goto busted;
806
807 /* previously declared? */
808 if ((ns = findNamespace(&w->namespaces, uri))) {
809 /* just a lookup, really */
810 if ((defaultPref == NULL) ||
811 (defaultPref[0] == 0 && ns->defaultDecl == w->xmlnsEquals) ||
812 (strcmp((const char *) ns->defaultDecl->name + STRLEN_XMLNS_COLON,
813 (const char *) defaultPref) == 0)) {
814 w->status = *statusP = GENX_SUCCESS;
815 return ns;
816 }
817 }
818
819 /* wasn't already declared */
820 else {
821
822 /* make a default prefix if none provided */
823 if (defaultPref == NULL) {
824 snprintf((char *) newPrefix, sizeof(newPrefix), "g%d", w->nextPrefix++);
825 defaultPref = newPrefix;
826 }
827
828 ns = (genxNamespace) allocate(sizeof(struct genxNamespace_rec));
829 if (ns == NULL) {
830 w->status = GENX_ALLOC_FAILED;
831 goto busted;
832 }
833 ns->writer = w;
834 ns->baroque = False;
835
836 if ((ns->name = copy(uri)) == NULL) {
837 w->status = GENX_ALLOC_FAILED;
838 goto busted;
839 }
840
841 if ((w->status = listAppend(&w->namespaces, ns)) != GENX_SUCCESS)
842 goto busted;
843 ns->defaultDecl = ns->declaration = NULL;
844 ns->declCount = 0;
845 }
846
847 if (defaultPref[0] == 0) {
848 if (w->defaultNsDeclared) {
849 w->status = GENX_DUPLICATE_PREFIX;
850 goto busted;
851 }
852 defaultDecl = w->xmlnsEquals;
853 w->defaultNsDeclared = True;
854 } else {
855 /* this catches dupes too */
856 if ((defaultPref = storePrefix(w, defaultPref, False)) == NULL)
857 goto busted;
858
859 defaultDecl = declareAttribute(w, NULL, defaultPref, ns->name, statusP);
860 if (defaultDecl == NULL || *statusP != GENX_SUCCESS) {
861 w->status = *statusP;
862 return NULL;
863 }
864 }
865
866 if (ns->defaultDecl != NULL && defaultDecl != ns->defaultDecl)
867 ns->baroque = True;
868 ns->defaultDecl = defaultDecl;
869
870 *statusP = GENX_SUCCESS;
871 return ns;
872
873 busted:
874 *statusP = w->status;
875 return NULL;
876 }
877
878 /**
879 * get namespace prefix
880 */
881 FUNGE_ATTR_FAST utf8 genxGetNamespacePrefix(genxNamespace ns)
882 {
883 if (ns->declaration == NULL)
884 return NULL;
885
886 if (ns->declaration == ns->writer->xmlnsEquals)
887 return ns->writer->empty;
888
889 return ns->declaration->name + STRLEN_XMLNS_COLON;
890 }
891
892 /**
893 * DeclareElement - see genx.h for details
894 */
895 FUNGE_ATTR_FAST
896 genxElement genxDeclareElement(genxWriter w,
897 genxNamespace ns, constUtf8 type,
898 genxStatus * statusP)
899 {
900 genxElement old;
901 genxElement el;
902
903 if ((w->status = checkNCName(w, type)) != GENX_SUCCESS) {
904 *statusP = w->status;
905 return NULL;
906 }
907
908 /* already declared? */
909 old = findElement(&w->elements, (ns == NULL) ? NULL : ns->name, type);
910 if (old)
911 return old;
912
913 if ((el = (genxElement) allocate(sizeof(struct genxElement_rec))) == NULL) {
914 w->status = *statusP = GENX_ALLOC_FAILED;
915 return NULL;
916 }
917
918 el->writer = w;
919 el->ns = ns;
920 if ((el->type = copy(type)) == NULL) {
921 w->status = *statusP = GENX_ALLOC_FAILED;
922 return NULL;
923 }
924
925 if ((w->status = listAppend(&w->elements, el)) != GENX_SUCCESS) {
926 *statusP = w->status;
927 return NULL;
928 }
929
930 *statusP = GENX_SUCCESS;
931 return el;
932 }
933
934 /**
935 * C14n ordering for attributes:
936 * - first, namespace declarations by the prefix being declared
937 * - second, unprefixed attributes by attr name
938 * - third, prefixed attrs by ns uri then local part
939 */
940 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
941 static int orderAttributes(genxAttribute a1, genxAttribute a2)
942 {
943 if (a1->atype == a2->atype) {
944 if (a1->atype == ATTR_PREFIXED && a1->ns != a2->ns)
945 return strcmp((const char *) a1->ns->name, (const char *) a2->ns->name);
946 else
947 return strcmp((const char *) a1->name, (const char *) a2->name);
948 }
949
950 else if (a1->atype == ATTR_NSDECL)
951 return -1;
952
953 else if (a1->atype == ATTR_NAKED) {
954 if (a2->atype == ATTR_NSDECL)
955 return 1;
956 else
957 return -1;
958 }
959
960 else
961 return 1;
962 }
963
964 /*
965 * internal declare-attribute. This one allows colonized values for
966 * names, so that you can declare xmlns:-type attributes
967 */
968 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
969 static genxAttribute declareAttribute(genxWriter w, genxNamespace ns,
970 constUtf8 name, constUtf8 valuestr,
971 genxStatus * statusP)
972 {
973 ssize_t high, low;
974 genxAttribute * aa = (genxAttribute *) w->attributes.pointers;
975 genxAttribute a;
976
977 w->arec.ns = ns;
978 w->arec.name = (utf8) name;
979
980 if (ns)
981 w->arec.atype = ATTR_PREFIXED;
982 else if (strncmp((const char *) name, "xmlns", STRLEN_XMLNS_COLON - 1) == 0)
983 w->arec.atype = ATTR_NSDECL;
984 else
985 w->arec.atype = ATTR_NAKED;
986
987 if (ns && (ns->defaultDecl == w->xmlnsEquals)) {
988 w->status = GENX_ATTRIBUTE_IN_DEFAULT_NAMESPACE;
989 goto busted;
990 }
991
992 /* attribute list has to be kept sorted per c14n rules */
993 high = w->attributes.count; low = -1;
994 while (high - low > 1) {
995 ssize_t probe = (high + low) / 2;
996 if (orderAttributes(&w->arec, aa[probe]) < 0)
997 high = probe;
998 else
999 low = probe;
1000 }
1001
1002 /* if it was already there */
1003 if (low != -1 && orderAttributes(&w->arec, aa[low]) == 0)
1004 return aa[low];
1005
1006 /* not there, build it */
1007 a = (genxAttribute) allocate(sizeof(struct genxAttribute_rec));
1008 if (a == NULL) {
1009 w->status = GENX_ALLOC_FAILED;
1010 goto busted;
1011 }
1012
1013 a->writer = w;
1014 a->ns = ns;
1015 a->provided = False;
1016 a->atype = w->arec.atype;
1017
1018 if ((a->name = copy(name)) == NULL) {
1019 w->status = GENX_ALLOC_FAILED;
1020 goto busted;
1021 }
1022
1023 if ((w->status = initCollector(&a->value)) != GENX_SUCCESS)
1024 goto busted;
1025
1026 if (valuestr)
1027 if ((w->status = collectString(w, &a->value, valuestr)) != GENX_SUCCESS)
1028 goto busted;
1029
1030 w->status = listInsert(&w->attributes, a, high);
1031 if (w->status != GENX_SUCCESS)
1032 goto busted;
1033
1034 *statusP = GENX_SUCCESS;
1035 return a;
1036
1037 busted:
1038 *statusP = w->status;
1039 return NULL;
1040 }
1041
1042 /*
1043 * genxDeclareAttribute - see genx.h for details
1044 */
1045 FUNGE_ATTR_FAST
1046 genxAttribute genxDeclareAttribute(genxWriter w,
1047 genxNamespace ns, constUtf8 name,
1048 genxStatus * statusP)
1049 {
1050 if ((w->status = checkNCName(w, name)) != GENX_SUCCESS) {
1051 *statusP = w->status;
1052 return NULL;
1053 }
1054
1055 return declareAttribute(w, ns, name, NULL, statusP);
1056 }
1057
1058 /*******************************
1059 * I/O
1060 */
1061 FUNGE_ATTR_FAST
1062 static genxStatus sendx(genxWriter w, constUtf8 s)
1063 {
1064 if (w->sender)
1065 return (*w->sender->send)(w->userData, s);
1066 else {
1067 if (fputs((const char *) s, w->file) == -1)
1068 return GENX_IO_ERROR;
1069 else
1070 return GENX_SUCCESS;
1071 }
1072 }
1073
1074 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
1075 static genxStatus sendxBounded(genxWriter w, constUtf8 start, constUtf8 end)
1076 {
1077 if (w->sender)
1078 return (*w->sender->sendBounded)(w->userData, start, end);
1079 else
1080 if (fwrite(start, 1, end - start, w->file) != (size_t)(end - start))
1081 return GENX_IO_ERROR;
1082 else
1083 return GENX_SUCCESS;
1084 }
1085
1086 #define SendCheck(w,s) if ((w->status=sendx(w,(constUtf8)s))!=GENX_SUCCESS) return w->status;
1087
1088 /*******************************
1089 * XML writing routines. The semantics of the externally-facing ones are
1090 * written up in genx.h. Commentary here is implementation notes and
1091 * for internal routines.
1092 */
1093
1094 /**
1095 * Start a document
1096 */
1097 FUNGE_ATTR_FAST genxStatus genxStartDocFile(genxWriter w, FILE * file)
1098 {
1099 if (w->sequence != SEQUENCE_NO_DOC)
1100 return w->status = GENX_SEQUENCE_ERROR;
1101
1102 w->sequence = SEQUENCE_PRE_DOC;
1103 w->file = file;
1104 w->sender = NULL;
1105 return GENX_SUCCESS;
1106 }
1107
1108 FUNGE_ATTR_FAST genxStatus genxStartDocSender(genxWriter w, genxSender * sender)
1109 {
1110 if (w->sequence != SEQUENCE_NO_DOC)
1111 return w->status = GENX_SEQUENCE_ERROR;
1112
1113 w->sequence = SEQUENCE_PRE_DOC;
1114 w->file = NULL;
1115 w->sender = sender;
1116 return GENX_SUCCESS;
1117 }
1118
1119 /**
1120 * Write out the attributes we've been gathering up for an element. We save
1121 * them until we've gathered them all so they can be writen in canonical
1122 * order.
1123 * Also, we end the start-tag.
1124 * The trick here is that we keep the attribute list properly sorted as
1125 * we build it, then as each attribute is added, we fill in its value and
1126 * mark the fact that it's been added, in the "provided" field.
1127 */
1128 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
1129 static genxStatus writeStartTag(genxWriter w)
1130 {
1131 ssize_t i;
1132 genxAttribute * aa = (genxAttribute *) w->attributes.pointers;
1133 genxElement e = w->nowStarting;
1134
1135 /*
1136 * make sure the right namespace decls are in effect;
1137 * if they are these might create an error, so ignore it
1138 */
1139 if (e->ns)
1140 addNamespace(e->ns, NULL);
1141 else
1142 unsetDefaultNamespace(w);
1143 w->status = GENX_SUCCESS;
1144
1145 SendCheck(w, "<");
1146 if (e->ns && (e->ns->declaration != w->xmlnsEquals)) {
1147 SendCheck(w, e->ns->declaration->name + STRLEN_XMLNS_COLON);
1148 SendCheck(w, ":");
1149 }
1150 SendCheck(w, e->type);
1151
1152 for (i = 0; i < w->attributes.count; i++) {
1153 if (aa[i]->provided) {
1154 if (aa[i]->ns && aa[i]->ns->baroque &&
1155 aa[i]->ns->declaration == w->xmlnsEquals)
1156 return w->status = GENX_ATTRIBUTE_IN_DEFAULT_NAMESPACE;
1157
1158 SendCheck(w, " ");
1159
1160 if (aa[i]->ns) {
1161 SendCheck(w, aa[i]->ns->declaration->name + STRLEN_XMLNS_COLON)
1162 SendCheck(w, ":");
1163 }
1164 SendCheck(w, aa[i]->name);
1165 SendCheck(w, "=\"");
1166 SendCheck(w, aa[i]->value.buf);
1167 SendCheck(w, "\"");
1168 }
1169 }
1170 SendCheck(w, ">");
1171 return GENX_SUCCESS;
1172 }
1173
1174 /*
1175 * internal clear-er; no sequence checking
1176 */
1177 FUNGE_ATTR_FAST
1178 static genxStatus unsetDefaultNamespace(genxWriter w)
1179 {
1180 ssize_t i;
1181 Boolean found = False;
1182
1183 /* don't put it in if not needed */
1184 i = w->stack.count - 1;
1185 while (found == False && i > 0) {
1186 while (w->stack.pointers[i] != NULL) {
1187 genxAttribute decl = (genxAttribute) w->stack.pointers[i--];
1188 genxNamespace ns = (genxNamespace) w->stack.pointers[i--];
1189
1190 /* if already unset */
1191 if (ns == NULL)
1192 return w->status = GENX_SUCCESS;
1193
1194 /*
1195 * the default namespace was declared. This namespace now
1196 * becomes baroque
1197 */
1198 if (decl == w->xmlnsEquals) {
1199 ns->baroque = True;
1200 found = True;
1201 break;
1202 }
1203 }
1204 i -= 2;
1205 }
1206
1207 if (!found)
1208 return GENX_SUCCESS;
1209
1210 /*
1211 * push a signal on the stack
1212 */
1213 if ((w->status = listAppend(&w->stack, NULL)) != GENX_SUCCESS)
1214 return w->status;
1215 w->status = listAppend(&w->stack, w->xmlnsEquals);
1216 if (w->status != GENX_SUCCESS)
1217 return w->status;
1218
1219 /* add the xmlns= attribute, it must be the first one */
1220 return addAttribute(w->xmlnsEquals, w->empty);
1221 }
1222
1223 /*
1224 * clear the default namespace declaration
1225 */
1226 FUNGE_ATTR_FAST genxStatus genxUnsetDefaultNamespace(genxWriter w)
1227 {
1228
1229 /* can only do this while in start-tag mode */
1230 if (w->sequence != SEQUENCE_START_TAG)
1231 return w->status = GENX_SEQUENCE_ERROR;
1232
1233 return unsetDefaultNamespace(w);
1234 }
1235
1236 FUNGE_ATTR_FAST genxStatus genxStartElement(genxElement e)
1237 {
1238 genxWriter w = e->writer;
1239 ssize_t i;
1240
1241 switch (w->sequence) {
1242 case SEQUENCE_NO_DOC:
1243 case SEQUENCE_POST_DOC:
1244 return w->status = GENX_SEQUENCE_ERROR;
1245 case SEQUENCE_START_TAG:
1246 case SEQUENCE_ATTRIBUTES:
1247 if ((w->status = writeStartTag(w)) != GENX_SUCCESS)
1248 return w->status;
1249 break;
1250 case SEQUENCE_PRE_DOC:
1251 case SEQUENCE_CONTENT:
1252 break;
1253 }
1254
1255 w->sequence = SEQUENCE_START_TAG;
1256
1257 /* clear provided attributes */
1258 for (i = 0; i < w->attributes.count; i++)
1259 ((genxAttribute) w->attributes.pointers[i])->provided = 0;
1260
1261 /*
1262 * push the stack. We push a NULL after a pointer to this element
1263 * because the stack will also contain pointers to the namespace
1264 * attributes that got declared here, so we can keep track of what's
1265 * in effect. I.e. a single stack entry consists logically of a pointer
1266 * to an element object, a NULL, then zero or more pairs of pointers to
1267 * namespace objects/declarations
1268 */
1269 if ((w->status = listAppend(&w->stack, e)) != GENX_SUCCESS)
1270 return w->status;
1271 if ((w->status = listAppend(&w->stack, NULL)) != GENX_SUCCESS)
1272 return w->status;
1273
1274 w->nowStarting = e;
1275
1276 return GENX_SUCCESS;
1277 }
1278
1279 /*
1280 * internal namespace adder; no sequence checking
1281 */
1282 FUNGE_ATTR_FAST
1283 static genxStatus addNamespace(genxNamespace ns, constUtf8 prefix)
1284 {
1285 genxWriter w = ns->writer;
1286 genxAttribute decl;
1287 ssize_t i;
1288 genxElement e;
1289
1290 /*
1291 * first, we'll find the declaring attribute
1292 */
1293 if (prefix == NULL)
1294 decl = ns->defaultDecl;
1295 else {
1296 if (prefix[0] == 0)
1297 decl = w->xmlnsEquals;
1298 else {
1299 if ((prefix = storePrefix(w, prefix, True)) == NULL)
1300 return w->status;
1301 decl = declareAttribute(w, NULL, prefix, ns->name, &w->status);
1302 if (decl == NULL || w->status != GENX_SUCCESS)
1303 return w->status;
1304 }
1305 }
1306
1307 if (decl != ns->defaultDecl)
1308 ns->baroque = True;
1309
1310 /*
1311 * avoid doing anything if this namespace is already declared. If
1312 * they've shown good taste we can do this cheaply
1313 */
1314 if (!ns->baroque) {
1315 if (ns->declCount > 0)
1316 return w->status = GENX_SUCCESS;
1317 } else {
1318
1319 /*
1320 * First, we'll run all the way up the stack to see if there is
1321 * another declaration for this namespace/prefix in scope, in which
1322 * case it's a no-op; or, if there's another declaration for this
1323 * prefix on another namespace, in which case we have to over-ride
1324 */
1325 i = w->stack.count - 1;
1326 while (i > 0) {
1327 while (w->stack.pointers[i] != NULL) {
1328 genxAttribute otherDecl = (genxAttribute) w->stack.pointers[i--];
1329 genxNamespace otherNs = (genxNamespace) w->stack.pointers[i--];
1330
1331 if (ns == otherNs) {
1332 if (decl == otherDecl)
1333 return w->status = GENX_SUCCESS;
1334 else {
1335 i = 0;
1336 break;
1337 }
1338 } else {
1339 /* different namespace, same prefix? */
1340 if (decl == otherDecl) {
1341 i = 0;
1342 break;
1343 }
1344 }
1345 }
1346 i -= 2;
1347 }
1348 }
1349
1350 /*
1351 * If this namespace is already declared on
1352 * this element (with different prefix/decl) which is an error.
1353 */
1354 i = w->stack.count - 1;
1355 while (w->stack.pointers[i] != NULL) {
1356 genxNamespace otherNs;
1357 i--; /* don't need declaration */
1358 otherNs = (genxNamespace) w->stack.pointers[i--];
1359
1360 if (ns == otherNs)
1361 return w->status = GENX_DUPLICATE_NAMESPACE;
1362 }
1363
1364 /* move pointer from NULL to element */
1365 --i;
1366
1367 /*
1368 * It's also an error if this is a default-namespace declaration and the
1369 * element is in no namespace.
1370 */
1371 e = (genxElement) w->stack.pointers[i];
1372 if (e->ns == NULL && decl == w->xmlnsEquals)
1373 return w->status = GENX_BAD_DEFAULT_DECLARATION;
1374
1375 if ((w->status = listAppend(&w->stack, ns)) != GENX_SUCCESS)
1376 return w->status;
1377 if ((w->status = listAppend(&w->stack, decl)) != GENX_SUCCESS)
1378 return w->status;
1379
1380 ns->declaration = decl;
1381 ns->declCount++;
1382 return addAttribute(decl, ns->name);
1383 }
1384
1385 /*
1386 * Add a namespace declaration
1387 */
1388 FUNGE_ATTR_FAST genxStatus genxAddNamespace(genxNamespace ns, utf8 prefix)
1389 {
1390 if (ns->writer->sequence != SEQUENCE_START_TAG)
1391 return ns->writer->status = GENX_SEQUENCE_ERROR;
1392
1393 return addNamespace(ns, prefix);
1394 }
1395
1396 /*
1397 * Private attribute-adding code
1398 * most of the work here is normalizing the value, which is the same
1399 * as regular normalization except for " is replaced by "&quot;"
1400 */
1401 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
1402 static genxStatus addAttribute(genxAttribute a, constUtf8 valuestr)
1403 {
1404 constUtf8 lastv = (constUtf8) valuestr;
1405 genxWriter w = a->writer;
1406
1407 /* if valuestr not provided, this is an xmlns with a pre-cooked value */
1408 if (valuestr) {
1409 startCollect(&a->value);
1410 while (*valuestr) {
1411 int c = genxNextUnicodeChar(&valuestr);
1412
1413 if (c == -1)
1414 return w->status = GENX_BAD_UTF8;
1415
1416 if (!isXMLChar(w, c))
1417 return w->status = GENX_NON_XML_CHARACTER;
1418
1419 switch (c) {
1420 case 9:
1421 collectPiece(w, &a->value, "&#x9;", 5);
1422 break;
1423 case 0xa:
1424 collectPiece(w, &a->value, "&#xA;", 5);
1425 break;
1426 case 0xd:
1427 collectPiece(w, &a->value, "&#xD;", 5);
1428 break;
1429 case '"':
1430 collectPiece(w, &a->value, "&quot;", 6);
1431 break;
1432 case '<':
1433 collectPiece(w, &a->value, "&lt;", 4);
1434 break;
1435 case '&':
1436 collectPiece(w, &a->value, "&amp;", 5);
1437 break;
1438 /*
1439 case '>':
1440 collectPiece(w, &a->value, "&gt;", 4);
1441 break;
1442 */
1443 default:
1444 collectPiece(w, &a->value, (const char *) lastv, valuestr - lastv);
1445 break;
1446 }
1447 lastv = (constUtf8) valuestr;
1448 }
1449 endCollect(&a->value);
1450 }
1451
1452 /* now add the namespace attribute; might fail if it's bee hand-declared */
1453 if (a->ns)
1454 addNamespace(a->ns, NULL);
1455
1456 if (valuestr && a->provided)
1457 return w->status = GENX_DUPLICATE_ATTRIBUTE;
1458 a->provided = 1;
1459
1460 return GENX_SUCCESS;
1461 }
1462
1463 /*
1464 * public attribute adder.
1465 * The only difference is that it doesn't allow a NULL value
1466 */
1467 FUNGE_ATTR_FAST genxStatus genxAddAttribute(genxAttribute a, constUtf8 valuestr)
1468 {
1469 if (a->writer->sequence != SEQUENCE_START_TAG &&
1470 a->writer->sequence != SEQUENCE_ATTRIBUTES)
1471 return a->writer->status = GENX_SEQUENCE_ERROR;
1472 a->writer->sequence = SEQUENCE_ATTRIBUTES;
1473
1474 if (valuestr == NULL)
1475 return a->writer->status = GENX_MISSING_VALUE;
1476
1477 return addAttribute(a, valuestr);
1478 }
1479
1480 FUNGE_ATTR_FAST genxStatus genxEndElement(genxWriter w)
1481 {
1482 genxElement e;
1483 ssize_t i;
1484
1485 switch (w->sequence) {
1486 case SEQUENCE_NO_DOC:
1487 case SEQUENCE_PRE_DOC:
1488 case SEQUENCE_POST_DOC:
1489 return w->status = GENX_SEQUENCE_ERROR;
1490 case SEQUENCE_START_TAG:
1491 case SEQUENCE_ATTRIBUTES:
1492 if ((w->status = writeStartTag(w)) != GENX_SUCCESS)
1493 return w->status;
1494 break;
1495 case SEQUENCE_CONTENT:
1496 break;
1497 }
1498
1499 /*
1500 * first peek into the stack to find the right namespace declaration
1501 * (if any) so we can properly prefix the end-tag. Have to do this
1502 * before unwinding the stack because that might reset some xmlns
1503 * prefixes to the context in the parent element
1504 */
1505 for (i = w->stack.count - 1; w->stack.pointers[i] != NULL; i -= 2)
1506 ;
1507 e = (genxElement) w->stack.pointers[--i];
1508
1509 SendCheck(w, "</");
1510 if (e->ns && e->ns->declaration != w->xmlnsEquals) {
1511 SendCheck(w, e->ns->declaration->name + STRLEN_XMLNS_COLON);
1512 SendCheck(w, ":");
1513 }
1514 SendCheck(w, e->type);
1515 SendCheck(w, ">");
1516
1517 /*
1518 * pop zero or more namespace declarations, then a null, then the
1519 * start-element declaration off the stack
1520 */
1521 w->stack.count--;
1522 while (w->stack.pointers[w->stack.count] != NULL) {
1523 genxNamespace ns = (genxNamespace) w->stack.pointers[--w->stack.count];
1524 w->stack.count--; /* don't need decl */
1525
1526 /* if not a fake unset-default namespace */
1527 if (ns) {
1528 /*
1529 * if they've stupidly jammed in their own namespace-prefix
1530 * declarations, we have to go looking to see if there's another
1531 * one in effect
1532 */
1533 if (ns->baroque) {
1534 i = w->stack.count;
1535 while (i > 0) {
1536 while (w->stack.pointers[i] != NULL) {
1537 genxAttribute otherDecl = (genxAttribute) w->stack.pointers[i--];
1538 genxNamespace otherNs = (genxNamespace) w->stack.pointers[i--];
1539
1540 if (otherNs == ns) {
1541 ns->declaration = otherDecl;
1542 i = 0;
1543 break;
1544 }
1545 }
1546
1547 /* skip NULL & element */
1548 i -= 2;
1549 }
1550 }
1551 ns->declCount--;
1552 if (ns->declCount == 0)
1553 ns->baroque = False;
1554 }
1555 }
1556
1557 /* pop the NULL */
1558 --w->stack.count;
1559 if (w->stack.count < 0)
1560 return w->status = GENX_NO_START_TAG;
1561
1562 if (w->stack.count == 0)
1563 w->sequence = SEQUENCE_POST_DOC;
1564 else
1565 w->sequence = SEQUENCE_CONTENT;
1566
1567 return GENX_SUCCESS;
1568 }
1569
1570 /*
1571 * Internal character-adder. It tries to keep the number of sendx()
1572 * calls down by looking at each character but only doing the output
1573 * when it has to escape something; ordinary text gets saved up in
1574 * chunks the start of which is indicated by *breaker.
1575 * c is the character, next points to the UTF8 representing the next
1576 * lastsP indirectly points to the UTF8 representing the
1577 * character, breakerP* indirectly points to the last place genx
1578 * changed the UTF8, e.g. by escaping a '<'
1579 */
1580 FUNGE_ATTR_FAST FUNGE_ATTR_WARN_UNUSED
1581 static genxStatus addChar(genxWriter w, int c, constUtf8 next,
1582 constUtf8 * lastsP, constUtf8 * breakerP)
1583 {
1584 if (c == -1)
1585 return GENX_BAD_UTF8;
1586
1587 if (!isXMLChar(w, c))
1588 return GENX_NON_XML_CHARACTER;
1589
1590 switch (c) {
1591 case 0xd:
1592 if ((w->status = sendxBounded(w, *breakerP, *lastsP)) != GENX_SUCCESS)
1593 return w->status;
1594 *breakerP = next;
1595 sendx(w, (constUtf8) "&#xD;");
1596 break;
1597 case '<':
1598 if ((w->status = sendxBounded(w, *breakerP, *lastsP)) != GENX_SUCCESS)
1599 return w->status;
1600 *breakerP = next;
1601 sendx(w, (constUtf8) "&lt;");
1602 break;
1603 case '&':
1604 if ((w->status = sendxBounded(w, *breakerP, *lastsP)) != GENX_SUCCESS)
1605 return w->status;
1606 *breakerP = next;
1607 sendx(w, (constUtf8) "&amp;");
1608 break;
1609 case '>':
1610 if ((w->status = sendxBounded(w, *breakerP, *lastsP)) != GENX_SUCCESS)
1611 return w->status;
1612 *breakerP = next;
1613 sendx(w, (constUtf8) "&gt;");
1614 break;
1615 default:
1616 break;
1617 }
1618 *lastsP = next;
1619 return GENX_SUCCESS;
1620 }
1621
1622 FUNGE_ATTR_FAST genxStatus genxAddText(genxWriter w, constUtf8 start)
1623 {
1624 constUtf8 lasts = start;
1625 constUtf8 breaker = start;
1626
1627 if (w->sequence == SEQUENCE_START_TAG ||
1628 w->sequence == SEQUENCE_ATTRIBUTES) {
1629 if ((w->status = writeStartTag(w)) != GENX_SUCCESS)
1630 return w->status;
1631 w->sequence = SEQUENCE_CONTENT;
1632 }
1633
1634 if (w->sequence != SEQUENCE_CONTENT)
1635 return w->status = GENX_SEQUENCE_ERROR;
1636
1637 while (*start) {
1638 int c = genxNextUnicodeChar(&start);
1639
1640 w->status = addChar(w, c, start, &lasts, &breaker);
1641 if (w->status != GENX_SUCCESS)
1642 return w->status;
1643 }
1644 return sendxBounded(w, breaker, (constUtf8) start);
1645 }
1646
1647 FUNGE_ATTR_FAST genxStatus genxAddBoundedText(genxWriter w, constUtf8 start, constUtf8 end)
1648 {
1649 constUtf8 lasts = start;
1650 constUtf8 breaker = start;
1651
1652 if (w->sequence == SEQUENCE_START_TAG ||
1653 w->sequence == SEQUENCE_ATTRIBUTES) {
1654 if ((w->status = writeStartTag(w)) != GENX_SUCCESS)
1655 return w->status;
1656 w->sequence = SEQUENCE_CONTENT;
1657 }
1658
1659 if (w->sequence != SEQUENCE_CONTENT)
1660 return w->status = GENX_SEQUENCE_ERROR;
1661
1662 while (start < end) {
1663 int c = genxNextUnicodeChar(&start);
1664
1665 w->status = addChar(w, c, (constUtf8) start, &lasts, &breaker);
1666 if (w->status != GENX_SUCCESS)
1667 return w->status;
1668 }
1669 return sendxBounded(w, breaker, (constUtf8) start);
1670 }
1671
1672 FUNGE_ATTR_FAST genxStatus genxAddCountedText(genxWriter w, constUtf8 start, int byteCount)
1673 {
1674 constUtf8 end = (constUtf8)(start + byteCount);
1675
1676 return genxAddBoundedText(w, start, end);
1677 }
1678
1679 FUNGE_ATTR_FAST genxStatus genxAddCharacter(genxWriter w, int c)
1680 {
1681 unsigned char cUTF8[10];
1682 utf8 lasts, breaker, next;
1683
1684 if (w->sequence == SEQUENCE_START_TAG ||
1685 w->sequence == SEQUENCE_ATTRIBUTES) {
1686 if ((w->status = writeStartTag(w)) != GENX_SUCCESS)
1687 return w->status;
1688 w->sequence = SEQUENCE_CONTENT;
1689 }
1690
1691 if (w->sequence != SEQUENCE_CONTENT)
1692 return w->status = GENX_SEQUENCE_ERROR;
1693
1694 if (!isXMLChar(w, c))
1695 return w->status = GENX_NON_XML_CHARACTER;
1696
1697 /* make UTF8 representation of character */
1698 lasts = breaker = next = cUTF8;
1699
1700 if (c < 0x80)
1701 *next++ = c;
1702 else if (c < 0x800) {
1703 *next++ = 0xc0 | (c >> 6);
1704 *next++ = 0x80 | (c & 0x3f);
1705 } else if (c < 0x10000) {
1706 *next++ = 0xe0 | (c >> 12);
1707 *next++ = 0x80 | ((c & 0xfc0) >> 6);
1708 *next++ = 0x80 | (c & 0x3f);
1709 } else {
1710 *next++ = 0xf0 | (c >> 18);
1711 *next++ = 0x80 | ((c & 0x3f000) >> 12);
1712 *next++ = 0x80 | ((c & 0xfc0) >> 6);
1713 *next++ = 0x80 | (c & 0x3f);
1714 }
1715 *next = 0;
1716
1717 w->status =
1718 addChar(w, c, next, (constUtf8 *) & lasts, (constUtf8 *) & breaker);
1719 if (w->status != GENX_SUCCESS)
1720 return w->status;
1721
1722 return sendxBounded(w, breaker, next);
1723 }
1724
1725 FUNGE_ATTR_FAST genxStatus genxEndDocument(genxWriter w)
1726 {
1727 if (w->sequence != SEQUENCE_POST_DOC)
1728 return w->status = GENX_SEQUENCE_ERROR;
1729
1730 if (w->file)
1731 fflush(w->file);
1732 else
1733 if ((w->status = (*w->sender->flush)(w->userData)) != GENX_SUCCESS)
1734 return w->status;
1735
1736 w->sequence = SEQUENCE_NO_DOC;
1737 return GENX_SUCCESS;
1738 }
1739
1740 FUNGE_ATTR_FAST genxStatus genxComment(genxWriter w, constUtf8 text)
1741 {
1742 int i;
1743
1744 if (w->sequence == SEQUENCE_NO_DOC)
1745 return w->status = GENX_SEQUENCE_ERROR;
1746
1747 if ((w->status = genxCheckText(w, text)) != GENX_SUCCESS)
1748 return w->status;
1749
1750 /* no leading '-', no trailing '-', no '--' */
1751 if (text[0] == '-')
1752 return w->status = GENX_MALFORMED_COMMENT;
1753 for (i = 0; text[i]; i++)
1754 if (text[i] == '-' && (text[i + 1] == '-' || text[i + 1] == 0))
1755 return w->status = GENX_MALFORMED_COMMENT;
1756
1757 if (w->sequence == SEQUENCE_START_TAG ||
1758 w->sequence == SEQUENCE_ATTRIBUTES) {
1759 if ((w->status = writeStartTag(w)) != GENX_SUCCESS)
1760 return w->status;
1761 w->sequence = SEQUENCE_CONTENT;
1762 }
1763
1764 else if (w->sequence == SEQUENCE_POST_DOC)
1765 if ((w->status = sendx(w, (constUtf8) "\n")) != GENX_SUCCESS)
1766 return w->status;
1767
1768 if ((w->status = sendx(w, (constUtf8) "<!--")) != GENX_SUCCESS)
1769 return w->status;
1770 if ((w->status = sendx(w, (constUtf8) text)) != GENX_SUCCESS)
1771 return w->status;
1772 if ((w->status = sendx(w, (constUtf8) "-->")) != GENX_SUCCESS)
1773 return w->status;
1774
1775 if (w->sequence == SEQUENCE_PRE_DOC)
1776 if ((w->status = sendx(w, (constUtf8) "\n")) != GENX_SUCCESS)
1777 return w->status;
1778
1779 return GENX_SUCCESS;
1780 }
1781
1782 FUNGE_ATTR_FAST genxStatus genxPI(genxWriter w, constUtf8 target, constUtf8 text)
1783 {
1784 int i;
1785
1786 if (w->sequence == SEQUENCE_NO_DOC)
1787 return w->status = GENX_SEQUENCE_ERROR;
1788
1789 if ((w->status = genxCheckText(w, target)) != GENX_SUCCESS)
1790 return w->status;
1791 if ((w->status = checkNCName(w, target)) != GENX_SUCCESS)
1792 return w->status;
1793 if ((strlen((const char *) target) >= 3) &&
1794 (target[0] == 'x' || target[0] == 'X') &&
1795 (target[1] == 'm' || target[1] == 'M') &&
1796 (target[2] == 'l' || target[2] == 'L') &&
1797 (target[3] == 0))
1798 return w->status = GENX_XML_PI_TARGET;
1799
1800 if ((w->status = genxCheckText(w, text)) != GENX_SUCCESS)
1801 return w->status;
1802
1803 /* no ?> within */
1804 for (i = 1; text[i]; i++)
1805 if (text[i] == '>' && text[i - 1] == '?')
1806 return w->status = GENX_MALFORMED_PI;
1807
1808 if (w->sequence == SEQUENCE_START_TAG ||
1809 w->sequence == SEQUENCE_ATTRIBUTES) {
1810 if ((w->status = writeStartTag(w)) != GENX_SUCCESS)
1811 return w->status;
1812 w->sequence = SEQUENCE_CONTENT;
1813 }
1814
1815 else if (w->sequence == SEQUENCE_POST_DOC)
1816 if ((w->status = sendx(w, (constUtf8) "\n")) != GENX_SUCCESS)
1817 return w->status;
1818
1819 if ((w->status = sendx(w, (constUtf8) "<?")) != GENX_SUCCESS)
1820 return w->status;
1821 if ((w->status = sendx(w, target)) != GENX_SUCCESS)
1822 return w->status;
1823 if (text[0]) {
1824 if ((w->status = sendx(w, (constUtf8) " ")) != GENX_SUCCESS)
1825 return w->status;
1826 if ((w->status = sendx(w, text)) != GENX_SUCCESS)
1827 return w->status;
1828 }
1829 if ((w->status = sendx(w, (constUtf8) "?>")) != GENX_SUCCESS)
1830 return w->status;
1831
1832 if (w->sequence == SEQUENCE_PRE_DOC)
1833 if ((w->status = sendx(w, (constUtf8) "\n")) != GENX_SUCCESS)
1834 return w->status;
1835
1836 return GENX_SUCCESS;
1837 }
1838
1839 /*******************************
1840 * Literal versions of the writing routines
1841 */
1842 FUNGE_ATTR_FAST
1843 genxStatus genxStartElementLiteral(genxWriter w,
1844 constUtf8 xmlns, constUtf8 type)
1845 {
1846 genxNamespace ns = NULL;
1847 genxElement e;
1848
1849 if (xmlns) {
1850 ns = genxDeclareNamespace(w, xmlns, NULL, &w->status);
1851 if (ns == NULL || w->status != GENX_SUCCESS)
1852 return w->status;
1853 }
1854 e = genxDeclareElement(w, ns, type, &w->status);
1855 if (e == NULL || w->status != GENX_SUCCESS)
1856 return w->status;
1857
1858 return genxStartElement(e);
1859 }
1860
1861 FUNGE_ATTR_FAST
1862 genxStatus genxAddAttributeLiteral(genxWriter w, constUtf8 xmlns,
1863 constUtf8 name, constUtf8 value)
1864 {
1865 genxNamespace ns = NULL;
1866 genxAttribute a;
1867
1868 if (xmlns) {
1869 ns = genxDeclareNamespace(w, xmlns, NULL, &w->status);
1870 if (ns == NULL && w->status != GENX_SUCCESS)
1871 return w->status;
1872 }
1873
1874 a = genxDeclareAttribute(w, ns, name, &w->status);
1875 if (a == NULL || w->status != GENX_SUCCESS)
1876 return w->status;
1877
1878 return genxAddAttribute(a, value);
1879 }
1880
1881 /*
1882 * return version
1883 */
1884 FUNGE_ATTR_FAST const char * genxGetVersion(void)
1885 {
1886 return GENX_VERSION;
1887 }
1888
1889 #endif /* !defined(CFUN_NO_FLOATS) && !defined(CFUN_NO_TURT) */