?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
winebuild: Add a common helper to find an imported dll.
[wine.git] / dlls / xmllite / reader.c
blobd9ed7e851226f3bee680a06630bf7e3254b04462
1 /*
2 * IXmlReader implementation
3 *
4 * Copyright 2010, 2012-2013 Nikolay Sivov
5 *
6 * This library is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
10 *
11 * This library is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
15 *
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with this library; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
19 */
20
21 #define COBJMACROS
22
23 #include <stdio.h>
24 #include <stdarg.h>
25 #include "windef.h"
26 #include "winbase.h"
27 #include "initguid.h"
28 #include "objbase.h"
29 #include "xmllite.h"
30 #include "xmllite_private.h"
31
32 #include "wine/debug.h"
33 #include "wine/list.h"
34 #include "wine/unicode.h"
35
36 WINE_DEFAULT_DEBUG_CHANNEL(xmllite);
37
38 /* not defined in public headers */
39 DEFINE_GUID(IID_IXmlReaderInput, 0x0b3ccc9b, 0x9214, 0x428b, 0xa2, 0xae, 0xef, 0x3a, 0xa8, 0x71, 0xaf, 0xda);
40
41 typedef enum
42 {
43 XmlReadInState_Initial,
44 XmlReadInState_XmlDecl,
45 XmlReadInState_Misc_DTD,
46 XmlReadInState_DTD,
47 XmlReadInState_DTD_Misc,
48 XmlReadInState_Element,
49 XmlReadInState_Content,
50 XmlReadInState_MiscEnd, /* optional Misc at the end of a document */
51 XmlReadInState_Eof
52 } XmlReaderInternalState;
53
54 /* This state denotes where parsing was interrupted by input problem.
55 Reader resumes parsing using this information. */
56 typedef enum
57 {
58 XmlReadResumeState_Initial,
59 XmlReadResumeState_PITarget,
60 XmlReadResumeState_PIBody,
61 XmlReadResumeState_CDATA,
62 XmlReadResumeState_Comment,
63 XmlReadResumeState_STag,
64 XmlReadResumeState_CharData,
65 XmlReadResumeState_Whitespace
66 } XmlReaderResumeState;
67
68 /* saved pointer index to resume from particular input position */
69 typedef enum
70 {
71 XmlReadResume_Name, /* PITarget, name for NCName, prefix for QName */
72 XmlReadResume_Local, /* local for QName */
73 XmlReadResume_Body, /* PI body, comment text, CDATA text, CharData text */
74 XmlReadResume_Last
75 } XmlReaderResume;
76
77 typedef enum
78 {
79 StringValue_LocalName,
80 StringValue_Prefix,
81 StringValue_QualifiedName,
82 StringValue_Value,
83 StringValue_Last
84 } XmlReaderStringValue;
85
86 static const WCHAR utf16W[] = {'U','T','F','-','1','6',0};
87 static const WCHAR utf8W[] = {'U','T','F','-','8',0};
88
89 static const WCHAR dblquoteW[] = {'\"',0};
90 static const WCHAR quoteW[] = {'\'',0};
91 static const WCHAR ltW[] = {'<',0};
92 static const WCHAR gtW[] = {'>',0};
93 static const WCHAR commentW[] = {'<','!','-','-',0};
94 static const WCHAR piW[] = {'<','?',0};
95
96 static const char *debugstr_nodetype(XmlNodeType nodetype)
97 {
98 static const char * const type_names[] =
99 {
100 "None",
101 "Element",
102 "Attribute",
103 "Text",
104 "CDATA",
105 "",
106 "",
107 "ProcessingInstruction",
108 "Comment",
109 "",
110 "DocumentType",
111 "",
112 "",
113 "Whitespace",
114 "",
115 "EndElement",
116 "",
117 "XmlDeclaration"
118 };
119
120 if (nodetype > _XmlNodeType_Last)
121 return wine_dbg_sprintf("unknown type=%d", nodetype);
122
123 return type_names[nodetype];
124 }
125
126 static const char *debugstr_reader_prop(XmlReaderProperty prop)
127 {
128 static const char * const prop_names[] =
129 {
130 "MultiLanguage",
131 "ConformanceLevel",
132 "RandomAccess",
133 "XmlResolver",
134 "DtdProcessing",
135 "ReadState",
136 "MaxElementDepth",
137 "MaxEntityExpansion"
138 };
139
140 if (prop > _XmlReaderProperty_Last)
141 return wine_dbg_sprintf("unknown property=%d", prop);
142
143 return prop_names[prop];
144 }
145
146 struct xml_encoding_data
147 {
148 const WCHAR *name;
149 xml_encoding enc;
150 UINT cp;
151 };
152
153 static const struct xml_encoding_data xml_encoding_map[] = {
154 { utf16W, XmlEncoding_UTF16, ~0 },
155 { utf8W, XmlEncoding_UTF8, CP_UTF8 }
156 };
157
158 const WCHAR *get_encoding_name(xml_encoding encoding)
159 {
160 return xml_encoding_map[encoding].name;
161 }
162
163 xml_encoding get_encoding_from_codepage(UINT codepage)
164 {
165 int i;
166 for (i = 0; i < sizeof(xml_encoding_map)/sizeof(xml_encoding_map[0]); i++)
167 {
168 if (xml_encoding_map[i].cp == codepage) return xml_encoding_map[i].enc;
169 }
170 return XmlEncoding_Unknown;
171 }
172
173 typedef struct
174 {
175 char *data;
176 UINT cur;
177 unsigned int allocated;
178 unsigned int written;
179 } encoded_buffer;
180
181 typedef struct input_buffer input_buffer;
182
183 typedef struct
184 {
185 IXmlReaderInput IXmlReaderInput_iface;
186 LONG ref;
187 /* reference passed on IXmlReaderInput creation, is kept when input is created */
188 IUnknown *input;
189 IMalloc *imalloc;
190 xml_encoding encoding;
191 BOOL hint;
192 WCHAR *baseuri;
193 /* stream reference set after SetInput() call from reader,
194 stored as sequential stream, cause currently
195 optimizations possible with IStream aren't implemented */
196 ISequentialStream *stream;
197 input_buffer *buffer;
198 unsigned int pending : 1;
199 } xmlreaderinput;
200
201 static const struct IUnknownVtbl xmlreaderinputvtbl;
202
203 /* Structure to hold parsed string of specific length.
204
205 Reader stores node value as 'start' pointer, on request
206 a null-terminated version of it is allocated.
207
208 To init a strval variable use reader_init_strval(),
209 to set strval as a reader value use reader_set_strval().
210 */
211 typedef struct
212 {
213 WCHAR *str; /* allocated null-terminated string */
214 UINT len; /* length in WCHARs, altered after ReadValueChunk */
215 UINT start; /* input position where value starts */
216 } strval;
217
218 static WCHAR emptyW[] = {0};
219 static const strval strval_empty = { emptyW };
220
221 struct attribute
222 {
223 struct list entry;
224 strval localname;
225 strval value;
226 };
227
228 struct element
229 {
230 struct list entry;
231 strval qname;
232 strval localname;
233 };
234
235 typedef struct
236 {
237 IXmlReader IXmlReader_iface;
238 LONG ref;
239 xmlreaderinput *input;
240 IMalloc *imalloc;
241 XmlReadState state;
242 XmlReaderInternalState instate;
243 XmlReaderResumeState resumestate;
244 XmlNodeType nodetype;
245 DtdProcessing dtdmode;
246 IXmlResolver *resolver;
247 UINT line, pos; /* reader position in XML stream */
248 struct list attrs; /* attributes list for current node */
249 struct attribute *attr; /* current attribute */
250 UINT attr_count;
251 struct list elements;
252 strval strvalues[StringValue_Last];
253 UINT depth;
254 UINT max_depth;
255 BOOL empty_element;
256 UINT resume[XmlReadResume_Last]; /* offsets used to resume reader */
257 } xmlreader;
258
259 struct input_buffer
260 {
261 encoded_buffer utf16;
262 encoded_buffer encoded;
263 UINT code_page;
264 xmlreaderinput *input;
265 };
266
267 static inline xmlreader *impl_from_IXmlReader(IXmlReader *iface)
268 {
269 return CONTAINING_RECORD(iface, xmlreader, IXmlReader_iface);
270 }
271
272 static inline xmlreaderinput *impl_from_IXmlReaderInput(IXmlReaderInput *iface)
273 {
274 return CONTAINING_RECORD(iface, xmlreaderinput, IXmlReaderInput_iface);
275 }
276
277 /* reader memory allocation functions */
278 static inline void *reader_alloc(xmlreader *reader, size_t len)
279 {
280 return m_alloc(reader->imalloc, len);
281 }
282
283 static inline void reader_free(xmlreader *reader, void *mem)
284 {
285 m_free(reader->imalloc, mem);
286 }
287
288 /* Just return pointer from offset, no attempt to read more. */
289 static inline WCHAR *reader_get_ptr2(const xmlreader *reader, UINT offset)
290 {
291 encoded_buffer *buffer = &reader->input->buffer->utf16;
292 return (WCHAR*)buffer->data + offset;
293 }
294
295 static inline WCHAR *reader_get_strptr(const xmlreader *reader, const strval *v)
296 {
297 return v->str ? v->str : reader_get_ptr2(reader, v->start);
298 }
299
300 static HRESULT reader_strvaldup(xmlreader *reader, const strval *src, strval *dest)
301 {
302 *dest = *src;
303
304 if (src->str != strval_empty.str)
305 {
306 dest->str = reader_alloc(reader, (dest->len+1)*sizeof(WCHAR));
307 if (!dest->str) return E_OUTOFMEMORY;
308 memcpy(dest->str, reader_get_strptr(reader, src), dest->len*sizeof(WCHAR));
309 dest->str[dest->len] = 0;
310 dest->start = 0;
311 }
312
313 return S_OK;
314 }
315
316 /* reader input memory allocation functions */
317 static inline void *readerinput_alloc(xmlreaderinput *input, size_t len)
318 {
319 return m_alloc(input->imalloc, len);
320 }
321
322 static inline void *readerinput_realloc(xmlreaderinput *input, void *mem, size_t len)
323 {
324 return m_realloc(input->imalloc, mem, len);
325 }
326
327 static inline void readerinput_free(xmlreaderinput *input, void *mem)
328 {
329 m_free(input->imalloc, mem);
330 }
331
332 static inline WCHAR *readerinput_strdupW(xmlreaderinput *input, const WCHAR *str)
333 {
334 LPWSTR ret = NULL;
335
336 if(str) {
337 DWORD size;
338
339 size = (strlenW(str)+1)*sizeof(WCHAR);
340 ret = readerinput_alloc(input, size);
341 if (ret) memcpy(ret, str, size);
342 }
343
344 return ret;
345 }
346
347 static void reader_clear_attrs(xmlreader *reader)
348 {
349 struct attribute *attr, *attr2;
350 LIST_FOR_EACH_ENTRY_SAFE(attr, attr2, &reader->attrs, struct attribute, entry)
351 {
352 reader_free(reader, attr);
353 }
354 list_init(&reader->attrs);
355 reader->attr_count = 0;
356 reader->attr = NULL;
357 }
358
359 /* attribute data holds pointers to buffer data, so buffer shrink is not possible
360 while we are on a node with attributes */
361 static HRESULT reader_add_attr(xmlreader *reader, strval *localname, strval *value)
362 {
363 struct attribute *attr;
364
365 attr = reader_alloc(reader, sizeof(*attr));
366 if (!attr) return E_OUTOFMEMORY;
367
368 attr->localname = *localname;
369 attr->value = *value;
370 list_add_tail(&reader->attrs, &attr->entry);
371 reader->attr_count++;
372
373 return S_OK;
374 }
375
376 /* This one frees stored string value if needed */
377 static void reader_free_strvalued(xmlreader *reader, strval *v)
378 {
379 if (v->str != strval_empty.str)
380 {
381 reader_free(reader, v->str);
382 *v = strval_empty;
383 }
384 }
385
386 static inline void reader_init_strvalue(UINT start, UINT len, strval *v)
387 {
388 v->start = start;
389 v->len = len;
390 v->str = NULL;
391 }
392
393 static inline const char* debug_strval(const xmlreader *reader, const strval *v)
394 {
395 return debugstr_wn(reader_get_strptr(reader, v), v->len);
396 }
397
398 /* used to initialize from constant string */
399 static inline void reader_init_cstrvalue(WCHAR *str, UINT len, strval *v)
400 {
401 v->start = 0;
402 v->len = len;
403 v->str = str;
404 }
405
406 static void reader_free_strvalue(xmlreader *reader, XmlReaderStringValue type)
407 {
408 reader_free_strvalued(reader, &reader->strvalues[type]);
409 }
410
411 static void reader_free_strvalues(xmlreader *reader)
412 {
413 int type;
414 for (type = 0; type < StringValue_Last; type++)
415 reader_free_strvalue(reader, type);
416 }
417
418 /* This helper should only be used to test if strings are the same,
419 it doesn't try to sort. */
420 static inline int strval_eq(const xmlreader *reader, const strval *str1, const strval *str2)
421 {
422 if (str1->len != str2->len) return 0;
423 return !memcmp(reader_get_strptr(reader, str1), reader_get_strptr(reader, str2), str1->len*sizeof(WCHAR));
424 }
425
426 static void reader_clear_elements(xmlreader *reader)
427 {
428 struct element *elem, *elem2;
429 LIST_FOR_EACH_ENTRY_SAFE(elem, elem2, &reader->elements, struct element, entry)
430 {
431 reader_free_strvalued(reader, &elem->qname);
432 reader_free(reader, elem);
433 }
434 list_init(&reader->elements);
435 reader->empty_element = FALSE;
436 }
437
438 static HRESULT reader_inc_depth(xmlreader *reader)
439 {
440 if (++reader->depth > reader->max_depth) return SC_E_MAXELEMENTDEPTH;
441 return S_OK;
442 }
443
444 static void reader_dec_depth(xmlreader *reader)
445 {
446 if (reader->depth > 1) reader->depth--;
447 }
448
449 static HRESULT reader_push_element(xmlreader *reader, strval *qname, strval *localname)
450 {
451 struct element *elem;
452 HRESULT hr;
453
454 elem = reader_alloc(reader, sizeof(*elem));
455 if (!elem) return E_OUTOFMEMORY;
456
457 hr = reader_strvaldup(reader, qname, &elem->qname);
458 if (FAILED(hr)) {
459 reader_free(reader, elem);
460 return hr;
461 }
462
463 hr = reader_strvaldup(reader, localname, &elem->localname);
464 if (FAILED(hr))
465 {
466 reader_free_strvalued(reader, &elem->qname);
467 reader_free(reader, elem);
468 return hr;
469 }
470
471 if (!list_empty(&reader->elements))
472 {
473 hr = reader_inc_depth(reader);
474 if (FAILED(hr)) {
475 reader_free(reader, elem);
476 return hr;
477 }
478 }
479
480 list_add_head(&reader->elements, &elem->entry);
481 reader->empty_element = FALSE;
482 return hr;
483 }
484
485 static void reader_pop_element(xmlreader *reader)
486 {
487 struct element *elem = LIST_ENTRY(list_head(&reader->elements), struct element, entry);
488
489 if (elem)
490 {
491 list_remove(&elem->entry);
492 reader_free_strvalued(reader, &elem->qname);
493 reader_free_strvalued(reader, &elem->localname);
494 reader_free(reader, elem);
495 reader_dec_depth(reader);
496 }
497 }
498
499 /* Always make a copy, cause strings are supposed to be null terminated. Null pointer for 'value'
500 means node value is to be determined. */
501 static void reader_set_strvalue(xmlreader *reader, XmlReaderStringValue type, const strval *value)
502 {
503 strval *v = &reader->strvalues[type];
504
505 reader_free_strvalue(reader, type);
506 if (!value)
507 {
508 v->str = NULL;
509 v->start = 0;
510 v->len = 0;
511 return;
512 }
513
514 if (value->str == strval_empty.str)
515 *v = *value;
516 else
517 {
518 if (type == StringValue_Value)
519 {
520 /* defer allocation for value string */
521 v->str = NULL;
522 v->start = value->start;
523 v->len = value->len;
524 }
525 else
526 {
527 v->str = reader_alloc(reader, (value->len + 1)*sizeof(WCHAR));
528 memcpy(v->str, reader_get_strptr(reader, value), value->len*sizeof(WCHAR));
529 v->str[value->len] = 0;
530 v->len = value->len;
531 }
532 }
533 }
534
535 static inline int is_reader_pending(xmlreader *reader)
536 {
537 return reader->input->pending;
538 }
539
540 static HRESULT init_encoded_buffer(xmlreaderinput *input, encoded_buffer *buffer)
541 {
542 const int initial_len = 0x2000;
543 buffer->data = readerinput_alloc(input, initial_len);
544 if (!buffer->data) return E_OUTOFMEMORY;
545
546 memset(buffer->data, 0, 4);
547 buffer->cur = 0;
548 buffer->allocated = initial_len;
549 buffer->written = 0;
550
551 return S_OK;
552 }
553
554 static void free_encoded_buffer(xmlreaderinput *input, encoded_buffer *buffer)
555 {
556 readerinput_free(input, buffer->data);
557 }
558
559 HRESULT get_code_page(xml_encoding encoding, UINT *cp)
560 {
561 if (encoding == XmlEncoding_Unknown)
562 {
563 FIXME("unsupported encoding %d\n", encoding);
564 return E_NOTIMPL;
565 }
566
567 *cp = xml_encoding_map[encoding].cp;
568
569 return S_OK;
570 }
571
572 xml_encoding parse_encoding_name(const WCHAR *name, int len)
573 {
574 int min, max, n, c;
575
576 if (!name) return XmlEncoding_Unknown;
577
578 min = 0;
579 max = sizeof(xml_encoding_map)/sizeof(struct xml_encoding_data) - 1;
580
581 while (min <= max)
582 {
583 n = (min+max)/2;
584
585 if (len != -1)
586 c = strncmpiW(xml_encoding_map[n].name, name, len);
587 else
588 c = strcmpiW(xml_encoding_map[n].name, name);
589 if (!c)
590 return xml_encoding_map[n].enc;
591
592 if (c > 0)
593 max = n-1;
594 else
595 min = n+1;
596 }
597
598 return XmlEncoding_Unknown;
599 }
600
601 static HRESULT alloc_input_buffer(xmlreaderinput *input)
602 {
603 input_buffer *buffer;
604 HRESULT hr;
605
606 input->buffer = NULL;
607
608 buffer = readerinput_alloc(input, sizeof(*buffer));
609 if (!buffer) return E_OUTOFMEMORY;
610
611 buffer->input = input;
612 buffer->code_page = ~0; /* code page is unknown at this point */
613 hr = init_encoded_buffer(input, &buffer->utf16);
614 if (hr != S_OK) {
615 readerinput_free(input, buffer);
616 return hr;
617 }
618
619 hr = init_encoded_buffer(input, &buffer->encoded);
620 if (hr != S_OK) {
621 free_encoded_buffer(input, &buffer->utf16);
622 readerinput_free(input, buffer);
623 return hr;
624 }
625
626 input->buffer = buffer;
627 return S_OK;
628 }
629
630 static void free_input_buffer(input_buffer *buffer)
631 {
632 free_encoded_buffer(buffer->input, &buffer->encoded);
633 free_encoded_buffer(buffer->input, &buffer->utf16);
634 readerinput_free(buffer->input, buffer);
635 }
636
637 static void readerinput_release_stream(xmlreaderinput *readerinput)
638 {
639 if (readerinput->stream) {
640 ISequentialStream_Release(readerinput->stream);
641 readerinput->stream = NULL;
642 }
643 }
644
645 /* Queries already stored interface for IStream/ISequentialStream.
646 Interface supplied on creation will be overwritten */
647 static inline HRESULT readerinput_query_for_stream(xmlreaderinput *readerinput)
648 {
649 HRESULT hr;
650
651 readerinput_release_stream(readerinput);
652 hr = IUnknown_QueryInterface(readerinput->input, &IID_IStream, (void**)&readerinput->stream);
653 if (hr != S_OK)
654 hr = IUnknown_QueryInterface(readerinput->input, &IID_ISequentialStream, (void**)&readerinput->stream);
655
656 return hr;
657 }
658
659 /* reads a chunk to raw buffer */
660 static HRESULT readerinput_growraw(xmlreaderinput *readerinput)
661 {
662 encoded_buffer *buffer = &readerinput->buffer->encoded;
663 /* to make sure aligned length won't exceed allocated length */
664 ULONG len = buffer->allocated - buffer->written - 4;
665 ULONG read;
666 HRESULT hr;
667
668 /* always try to get aligned to 4 bytes, so the only case we can get partially read characters is
669 variable width encodings like UTF-8 */
670 len = (len + 3) & ~3;
671 /* try to use allocated space or grow */
672 if (buffer->allocated - buffer->written < len)
673 {
674 buffer->allocated *= 2;
675 buffer->data = readerinput_realloc(readerinput, buffer->data, buffer->allocated);
676 len = buffer->allocated - buffer->written;
677 }
678
679 read = 0;
680 hr = ISequentialStream_Read(readerinput->stream, buffer->data + buffer->written, len, &read);
681 TRACE("written=%d, alloc=%d, requested=%d, read=%d, ret=0x%08x\n", buffer->written, buffer->allocated, len, read, hr);
682 readerinput->pending = hr == E_PENDING;
683 if (FAILED(hr)) return hr;
684 buffer->written += read;
685
686 return hr;
687 }
688
689 /* grows UTF-16 buffer so it has at least 'length' WCHAR chars free on return */
690 static void readerinput_grow(xmlreaderinput *readerinput, int length)
691 {
692 encoded_buffer *buffer = &readerinput->buffer->utf16;
693
694 length *= sizeof(WCHAR);
695 /* grow if needed, plus 4 bytes to be sure null terminator will fit in */
696 if (buffer->allocated < buffer->written + length + 4)
697 {
698 int grown_size = max(2*buffer->allocated, buffer->allocated + length);
699 buffer->data = readerinput_realloc(readerinput, buffer->data, grown_size);
700 buffer->allocated = grown_size;
701 }
702 }
703
704 static inline BOOL readerinput_is_utf8(xmlreaderinput *readerinput)
705 {
706 static const char startA[] = {'<','?'};
707 static const char commentA[] = {'<','!'};
708 encoded_buffer *buffer = &readerinput->buffer->encoded;
709 unsigned char *ptr = (unsigned char*)buffer->data;
710
711 return !memcmp(buffer->data, startA, sizeof(startA)) ||
712 !memcmp(buffer->data, commentA, sizeof(commentA)) ||
713 /* test start byte */
714 (ptr[0] == '<' &&
715 (
716 (ptr[1] && (ptr[1] <= 0x7f)) ||
717 (buffer->data[1] >> 5) == 0x6 || /* 2 bytes */
718 (buffer->data[1] >> 4) == 0xe || /* 3 bytes */
719 (buffer->data[1] >> 3) == 0x1e) /* 4 bytes */
720 );
721 }
722
723 static HRESULT readerinput_detectencoding(xmlreaderinput *readerinput, xml_encoding *enc)
724 {
725 encoded_buffer *buffer = &readerinput->buffer->encoded;
726 static const WCHAR startW[] = {'<','?'};
727 static const WCHAR commentW[] = {'<','!'};
728 static const char utf8bom[] = {0xef,0xbb,0xbf};
729 static const char utf16lebom[] = {0xff,0xfe};
730
731 *enc = XmlEncoding_Unknown;
732
733 if (buffer->written <= 3)
734 {
735 HRESULT hr = readerinput_growraw(readerinput);
736 if (FAILED(hr)) return hr;
737 if (buffer->written <= 3) return MX_E_INPUTEND;
738 }
739
740 /* try start symbols if we have enough data to do that, input buffer should contain
741 first chunk already */
742 if (readerinput_is_utf8(readerinput))
743 *enc = XmlEncoding_UTF8;
744 else if (!memcmp(buffer->data, startW, sizeof(startW)) ||
745 !memcmp(buffer->data, commentW, sizeof(commentW)))
746 *enc = XmlEncoding_UTF16;
747 /* try with BOM now */
748 else if (!memcmp(buffer->data, utf8bom, sizeof(utf8bom)))
749 {
750 buffer->cur += sizeof(utf8bom);
751 *enc = XmlEncoding_UTF8;
752 }
753 else if (!memcmp(buffer->data, utf16lebom, sizeof(utf16lebom)))
754 {
755 buffer->cur += sizeof(utf16lebom);
756 *enc = XmlEncoding_UTF16;
757 }
758
759 return S_OK;
760 }
761
762 static int readerinput_get_utf8_convlen(xmlreaderinput *readerinput)
763 {
764 encoded_buffer *buffer = &readerinput->buffer->encoded;
765 int len = buffer->written;
766
767 /* complete single byte char */
768 if (!(buffer->data[len-1] & 0x80)) return len;
769
770 /* find start byte of multibyte char */
771 while (--len && !(buffer->data[len] & 0xc0))
772 ;
773
774 return len;
775 }
776
777 /* Returns byte length of complete char sequence for buffer code page,
778 it's relative to current buffer position which is currently used for BOM handling
779 only. */
780 static int readerinput_get_convlen(xmlreaderinput *readerinput)
781 {
782 encoded_buffer *buffer = &readerinput->buffer->encoded;
783 int len;
784
785 if (readerinput->buffer->code_page == CP_UTF8)
786 len = readerinput_get_utf8_convlen(readerinput);
787 else
788 len = buffer->written;
789
790 TRACE("%d\n", len - buffer->cur);
791 return len - buffer->cur;
792 }
793
794 /* It's possible that raw buffer has some leftovers from last conversion - some char
795 sequence that doesn't represent a full code point. Length argument should be calculated with
796 readerinput_get_convlen(), if it's -1 it will be calculated here. */
797 static void readerinput_shrinkraw(xmlreaderinput *readerinput, int len)
798 {
799 encoded_buffer *buffer = &readerinput->buffer->encoded;
800
801 if (len == -1)
802 len = readerinput_get_convlen(readerinput);
803
804 memmove(buffer->data, buffer->data + buffer->cur + (buffer->written - len), len);
805 /* everything below cur is lost too */
806 buffer->written -= len + buffer->cur;
807 /* after this point we don't need cur offset really,
808 it's used only to mark where actual data begins when first chunk is read */
809 buffer->cur = 0;
810 }
811
812 /* note that raw buffer content is kept */
813 static void readerinput_switchencoding(xmlreaderinput *readerinput, xml_encoding enc)
814 {
815 encoded_buffer *src = &readerinput->buffer->encoded;
816 encoded_buffer *dest = &readerinput->buffer->utf16;
817 int len, dest_len;
818 HRESULT hr;
819 WCHAR *ptr;
820 UINT cp;
821
822 hr = get_code_page(enc, &cp);
823 if (FAILED(hr)) return;
824
825 readerinput->buffer->code_page = cp;
826 len = readerinput_get_convlen(readerinput);
827
828 TRACE("switching to cp %d\n", cp);
829
830 /* just copy in this case */
831 if (enc == XmlEncoding_UTF16)
832 {
833 readerinput_grow(readerinput, len);
834 memcpy(dest->data, src->data + src->cur, len);
835 dest->written += len*sizeof(WCHAR);
836 return;
837 }
838
839 dest_len = MultiByteToWideChar(cp, 0, src->data + src->cur, len, NULL, 0);
840 readerinput_grow(readerinput, dest_len);
841 ptr = (WCHAR*)dest->data;
842 MultiByteToWideChar(cp, 0, src->data + src->cur, len, ptr, dest_len);
843 ptr[dest_len] = 0;
844 dest->written += dest_len*sizeof(WCHAR);
845 }
846
847 /* shrinks parsed data a buffer begins with */
848 static void reader_shrink(xmlreader *reader)
849 {
850 encoded_buffer *buffer = &reader->input->buffer->utf16;
851
852 /* avoid to move too often using threshold shrink length */
853 if (buffer->cur*sizeof(WCHAR) > buffer->written / 2)
854 {
855 buffer->written -= buffer->cur*sizeof(WCHAR);
856 memmove(buffer->data, (WCHAR*)buffer->data + buffer->cur, buffer->written);
857 buffer->cur = 0;
858 *(WCHAR*)&buffer->data[buffer->written] = 0;
859 }
860 }
861
862 /* This is a normal way for reader to get new data converted from raw buffer to utf16 buffer.
863 It won't attempt to shrink but will grow destination buffer if needed */
864 static HRESULT reader_more(xmlreader *reader)
865 {
866 xmlreaderinput *readerinput = reader->input;
867 encoded_buffer *src = &readerinput->buffer->encoded;
868 encoded_buffer *dest = &readerinput->buffer->utf16;
869 UINT cp = readerinput->buffer->code_page;
870 int len, dest_len;
871 HRESULT hr;
872 WCHAR *ptr;
873
874 /* get some raw data from stream first */
875 hr = readerinput_growraw(readerinput);
876 len = readerinput_get_convlen(readerinput);
877
878 /* just copy for UTF-16 case */
879 if (cp == ~0)
880 {
881 readerinput_grow(readerinput, len);
882 memcpy(dest->data + dest->written, src->data + src->cur, len);
883 dest->written += len*sizeof(WCHAR);
884 return hr;
885 }
886
887 dest_len = MultiByteToWideChar(cp, 0, src->data + src->cur, len, NULL, 0);
888 readerinput_grow(readerinput, dest_len);
889 ptr = (WCHAR*)(dest->data + dest->written);
890 MultiByteToWideChar(cp, 0, src->data + src->cur, len, ptr, dest_len);
891 ptr[dest_len] = 0;
892 dest->written += dest_len*sizeof(WCHAR);
893 /* get rid of processed data */
894 readerinput_shrinkraw(readerinput, len);
895
896 return hr;
897 }
898
899 static inline UINT reader_get_cur(xmlreader *reader)
900 {
901 return reader->input->buffer->utf16.cur;
902 }
903
904 static inline WCHAR *reader_get_ptr(xmlreader *reader)
905 {
906 encoded_buffer *buffer = &reader->input->buffer->utf16;
907 WCHAR *ptr = (WCHAR*)buffer->data + buffer->cur;
908 if (!*ptr) reader_more(reader);
909 return (WCHAR*)buffer->data + buffer->cur;
910 }
911
912 static int reader_cmp(xmlreader *reader, const WCHAR *str)
913 {
914 int i=0;
915 const WCHAR *ptr = reader_get_ptr(reader);
916 while (str[i])
917 {
918 if (!ptr[i])
919 {
920 reader_more(reader);
921 ptr = reader_get_ptr(reader);
922 }
923 if (str[i] != ptr[i])
924 return ptr[i] - str[i];
925 i++;
926 }
927 return 0;
928 }
929
930 /* moves cursor n WCHARs forward */
931 static void reader_skipn(xmlreader *reader, int n)
932 {
933 encoded_buffer *buffer = &reader->input->buffer->utf16;
934 const WCHAR *ptr = reader_get_ptr(reader);
935
936 while (*ptr++ && n--)
937 {
938 buffer->cur++;
939 reader->pos++;
940 }
941 }
942
943 static inline BOOL is_wchar_space(WCHAR ch)
944 {
945 return ch == ' ' || ch == '\t' || ch == '\r' || ch == '\n';
946 }
947
948 /* [3] S ::= (#x20 | #x9 | #xD | #xA)+ */
949 static int reader_skipspaces(xmlreader *reader)
950 {
951 encoded_buffer *buffer = &reader->input->buffer->utf16;
952 const WCHAR *ptr = reader_get_ptr(reader);
953 UINT start = reader_get_cur(reader);
954
955 while (is_wchar_space(*ptr))
956 {
957 if (*ptr == '\r')
958 reader->pos = 0;
959 else if (*ptr == '\n')
960 {
961 reader->line++;
962 reader->pos = 0;
963 }
964 else
965 reader->pos++;
966
967 buffer->cur++;
968 ptr = reader_get_ptr(reader);
969 }
970
971 return reader_get_cur(reader) - start;
972 }
973
974 /* [26] VersionNum ::= '1.' [0-9]+ */
975 static HRESULT reader_parse_versionnum(xmlreader *reader, strval *val)
976 {
977 static const WCHAR onedotW[] = {'1','.',0};
978 WCHAR *ptr, *ptr2;
979 UINT start;
980
981 if (reader_cmp(reader, onedotW)) return WC_E_XMLDECL;
982
983 start = reader_get_cur(reader);
984 /* skip "1." */
985 reader_skipn(reader, 2);
986
987 ptr2 = ptr = reader_get_ptr(reader);
988 while (*ptr >= '0' && *ptr <= '9')
989 {
990 reader_skipn(reader, 1);
991 ptr = reader_get_ptr(reader);
992 }
993
994 if (ptr2 == ptr) return WC_E_DIGIT;
995 reader_init_strvalue(start, reader_get_cur(reader)-start, val);
996 TRACE("version=%s\n", debug_strval(reader, val));
997 return S_OK;
998 }
999
1000 /* [25] Eq ::= S? '=' S? */
1001 static HRESULT reader_parse_eq(xmlreader *reader)
1002 {
1003 static const WCHAR eqW[] = {'=',0};
1004 reader_skipspaces(reader);
1005 if (reader_cmp(reader, eqW)) return WC_E_EQUAL;
1006 /* skip '=' */
1007 reader_skipn(reader, 1);
1008 reader_skipspaces(reader);
1009 return S_OK;
1010 }
1011
1012 /* [24] VersionInfo ::= S 'version' Eq ("'" VersionNum "'" | '"' VersionNum '"') */
1013 static HRESULT reader_parse_versioninfo(xmlreader *reader)
1014 {
1015 static const WCHAR versionW[] = {'v','e','r','s','i','o','n',0};
1016 strval val, name;
1017 HRESULT hr;
1018
1019 if (!reader_skipspaces(reader)) return WC_E_WHITESPACE;
1020
1021 if (reader_cmp(reader, versionW)) return WC_E_XMLDECL;
1022 reader_init_strvalue(reader_get_cur(reader), 7, &name);
1023 /* skip 'version' */
1024 reader_skipn(reader, 7);
1025
1026 hr = reader_parse_eq(reader);
1027 if (FAILED(hr)) return hr;
1028
1029 if (reader_cmp(reader, quoteW) && reader_cmp(reader, dblquoteW))
1030 return WC_E_QUOTE;
1031 /* skip "'"|'"' */
1032 reader_skipn(reader, 1);
1033
1034 hr = reader_parse_versionnum(reader, &val);
1035 if (FAILED(hr)) return hr;
1036
1037 if (reader_cmp(reader, quoteW) && reader_cmp(reader, dblquoteW))
1038 return WC_E_QUOTE;
1039
1040 /* skip "'"|'"' */
1041 reader_skipn(reader, 1);
1042
1043 return reader_add_attr(reader, &name, &val);
1044 }
1045
1046 /* ([A-Za-z0-9._] | '-') */
1047 static inline BOOL is_wchar_encname(WCHAR ch)
1048 {
1049 return ((ch >= 'A' && ch <= 'Z') ||
1050 (ch >= 'a' && ch <= 'z') ||
1051 (ch >= '0' && ch <= '9') ||
1052 (ch == '.') || (ch == '_') ||
1053 (ch == '-'));
1054 }
1055
1056 /* [81] EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')* */
1057 static HRESULT reader_parse_encname(xmlreader *reader, strval *val)
1058 {
1059 WCHAR *start = reader_get_ptr(reader), *ptr;
1060 xml_encoding enc;
1061 int len;
1062
1063 if ((*start < 'A' || *start > 'Z') && (*start < 'a' || *start > 'z'))
1064 return WC_E_ENCNAME;
1065
1066 val->start = reader_get_cur(reader);
1067
1068 ptr = start;
1069 while (is_wchar_encname(*++ptr))
1070 ;
1071
1072 len = ptr - start;
1073 enc = parse_encoding_name(start, len);
1074 TRACE("encoding name %s\n", debugstr_wn(start, len));
1075 val->str = start;
1076 val->len = len;
1077
1078 if (enc == XmlEncoding_Unknown)
1079 return WC_E_ENCNAME;
1080
1081 /* skip encoding name */
1082 reader_skipn(reader, len);
1083 return S_OK;
1084 }
1085
1086 /* [80] EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | "'" EncName "'" ) */
1087 static HRESULT reader_parse_encdecl(xmlreader *reader)
1088 {
1089 static const WCHAR encodingW[] = {'e','n','c','o','d','i','n','g',0};
1090 strval name, val;
1091 HRESULT hr;
1092
1093 if (!reader_skipspaces(reader)) return S_FALSE;
1094
1095 if (reader_cmp(reader, encodingW)) return S_FALSE;
1096 name.str = reader_get_ptr(reader);
1097 name.start = reader_get_cur(reader);
1098 name.len = 8;
1099 /* skip 'encoding' */
1100 reader_skipn(reader, 8);
1101
1102 hr = reader_parse_eq(reader);
1103 if (FAILED(hr)) return hr;
1104
1105 if (reader_cmp(reader, quoteW) && reader_cmp(reader, dblquoteW))
1106 return WC_E_QUOTE;
1107 /* skip "'"|'"' */
1108 reader_skipn(reader, 1);
1109
1110 hr = reader_parse_encname(reader, &val);
1111 if (FAILED(hr)) return hr;
1112
1113 if (reader_cmp(reader, quoteW) && reader_cmp(reader, dblquoteW))
1114 return WC_E_QUOTE;
1115
1116 /* skip "'"|'"' */
1117 reader_skipn(reader, 1);
1118
1119 return reader_add_attr(reader, &name, &val);
1120 }
1121
1122 /* [32] SDDecl ::= S 'standalone' Eq (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no') '"')) */
1123 static HRESULT reader_parse_sddecl(xmlreader *reader)
1124 {
1125 static const WCHAR standaloneW[] = {'s','t','a','n','d','a','l','o','n','e',0};
1126 static const WCHAR yesW[] = {'y','e','s',0};
1127 static const WCHAR noW[] = {'n','o',0};
1128 strval name, val;
1129 UINT start;
1130 HRESULT hr;
1131
1132 if (!reader_skipspaces(reader)) return S_FALSE;
1133
1134 if (reader_cmp(reader, standaloneW)) return S_FALSE;
1135 reader_init_strvalue(reader_get_cur(reader), 10, &name);
1136 /* skip 'standalone' */
1137 reader_skipn(reader, 10);
1138
1139 hr = reader_parse_eq(reader);
1140 if (FAILED(hr)) return hr;
1141
1142 if (reader_cmp(reader, quoteW) && reader_cmp(reader, dblquoteW))
1143 return WC_E_QUOTE;
1144 /* skip "'"|'"' */
1145 reader_skipn(reader, 1);
1146
1147 if (reader_cmp(reader, yesW) && reader_cmp(reader, noW))
1148 return WC_E_XMLDECL;
1149
1150 start = reader_get_cur(reader);
1151 /* skip 'yes'|'no' */
1152 reader_skipn(reader, reader_cmp(reader, yesW) ? 2 : 3);
1153 reader_init_strvalue(start, reader_get_cur(reader)-start, &val);
1154 TRACE("standalone=%s\n", debug_strval(reader, &val));
1155
1156 if (reader_cmp(reader, quoteW) && reader_cmp(reader, dblquoteW))
1157 return WC_E_QUOTE;
1158 /* skip "'"|'"' */
1159 reader_skipn(reader, 1);
1160
1161 return reader_add_attr(reader, &name, &val);
1162 }
1163
1164 /* [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>' */
1165 static HRESULT reader_parse_xmldecl(xmlreader *reader)
1166 {
1167 static const WCHAR xmldeclW[] = {'<','?','x','m','l',' ',0};
1168 static const WCHAR declcloseW[] = {'?','>',0};
1169 HRESULT hr;
1170
1171 /* check if we have "<?xml " */
1172 if (reader_cmp(reader, xmldeclW)) return S_FALSE;
1173
1174 reader_skipn(reader, 5);
1175 hr = reader_parse_versioninfo(reader);
1176 if (FAILED(hr))
1177 return hr;
1178
1179 hr = reader_parse_encdecl(reader);
1180 if (FAILED(hr))
1181 return hr;
1182
1183 hr = reader_parse_sddecl(reader);
1184 if (FAILED(hr))
1185 return hr;
1186
1187 reader_skipspaces(reader);
1188 if (reader_cmp(reader, declcloseW)) return WC_E_XMLDECL;
1189 reader_skipn(reader, 2);
1190
1191 reader_inc_depth(reader);
1192 reader->nodetype = XmlNodeType_XmlDeclaration;
1193 reader_set_strvalue(reader, StringValue_LocalName, &strval_empty);
1194 reader_set_strvalue(reader, StringValue_QualifiedName, &strval_empty);
1195 reader_set_strvalue(reader, StringValue_Value, &strval_empty);
1196
1197 return S_OK;
1198 }
1199
1200 /* [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' */
1201 static HRESULT reader_parse_comment(xmlreader *reader)
1202 {
1203 WCHAR *ptr;
1204 UINT start;
1205
1206 if (reader->resumestate == XmlReadResumeState_Comment)
1207 {
1208 start = reader->resume[XmlReadResume_Body];
1209 ptr = reader_get_ptr(reader);
1210 }
1211 else
1212 {
1213 /* skip '<!--' */
1214 reader_skipn(reader, 4);
1215 reader_shrink(reader);
1216 ptr = reader_get_ptr(reader);
1217 start = reader_get_cur(reader);
1218 reader->nodetype = XmlNodeType_Comment;
1219 reader->resume[XmlReadResume_Body] = start;
1220 reader->resumestate = XmlReadResumeState_Comment;
1221 reader_set_strvalue(reader, StringValue_LocalName, NULL);
1222 reader_set_strvalue(reader, StringValue_QualifiedName, NULL);
1223 reader_set_strvalue(reader, StringValue_Value, NULL);
1224 }
1225
1226 /* will exit when there's no more data, it won't attempt to
1227 read more from stream */
1228 while (*ptr)
1229 {
1230 if (ptr[0] == '-')
1231 {
1232 if (ptr[1] == '-')
1233 {
1234 if (ptr[2] == '>')
1235 {
1236 strval value;
1237
1238 reader_init_strvalue(start, reader_get_cur(reader)-start, &value);
1239 TRACE("%s\n", debug_strval(reader, &value));
1240
1241 /* skip rest of markup '->' */
1242 reader_skipn(reader, 3);
1243
1244 reader_set_strvalue(reader, StringValue_LocalName, &strval_empty);
1245 reader_set_strvalue(reader, StringValue_QualifiedName, &strval_empty);
1246 reader_set_strvalue(reader, StringValue_Value, &value);
1247 reader->resume[XmlReadResume_Body] = 0;
1248 reader->resumestate = XmlReadResumeState_Initial;
1249 return S_OK;
1250 }
1251 else
1252 return WC_E_COMMENT;
1253 }
1254 }
1255
1256 reader_skipn(reader, 1);
1257 ptr++;
1258 }
1259
1260 return S_OK;
1261 }
1262
1263 /* [2] Char ::= #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD] | [#x10000-#x10FFFF] */
1264 static inline BOOL is_char(WCHAR ch)
1265 {
1266 return (ch == '\t') || (ch == '\r') || (ch == '\n') ||
1267 (ch >= 0x20 && ch <= 0xd7ff) ||
1268 (ch >= 0xd800 && ch <= 0xdbff) || /* high surrogate */
1269 (ch >= 0xdc00 && ch <= 0xdfff) || /* low surrogate */
1270 (ch >= 0xe000 && ch <= 0xfffd);
1271 }
1272
1273 /* [13] PubidChar ::= #x20 | #xD | #xA | [a-zA-Z0-9] | [-'()+,./:=?;!*#@$_%] */
1274 static inline BOOL is_pubchar(WCHAR ch)
1275 {
1276 return (ch == ' ') ||
1277 (ch >= 'a' && ch <= 'z') ||
1278 (ch >= 'A' && ch <= 'Z') ||
1279 (ch >= '0' && ch <= '9') ||
1280 (ch >= '-' && ch <= ';') || /* '()*+,-./:; */
1281 (ch == '=') || (ch == '?') ||
1282 (ch == '@') || (ch == '!') ||
1283 (ch >= '#' && ch <= '%') || /* #$% */
1284 (ch == '_') || (ch == '\r') || (ch == '\n');
1285 }
1286
1287 static inline BOOL is_namestartchar(WCHAR ch)
1288 {
1289 return (ch == ':') || (ch >= 'A' && ch <= 'Z') ||
1290 (ch == '_') || (ch >= 'a' && ch <= 'z') ||
1291 (ch >= 0xc0 && ch <= 0xd6) ||
1292 (ch >= 0xd8 && ch <= 0xf6) ||
1293 (ch >= 0xf8 && ch <= 0x2ff) ||
1294 (ch >= 0x370 && ch <= 0x37d) ||
1295 (ch >= 0x37f && ch <= 0x1fff) ||
1296 (ch >= 0x200c && ch <= 0x200d) ||
1297 (ch >= 0x2070 && ch <= 0x218f) ||
1298 (ch >= 0x2c00 && ch <= 0x2fef) ||
1299 (ch >= 0x3001 && ch <= 0xd7ff) ||
1300 (ch >= 0xd800 && ch <= 0xdbff) || /* high surrogate */
1301 (ch >= 0xdc00 && ch <= 0xdfff) || /* low surrogate */
1302 (ch >= 0xf900 && ch <= 0xfdcf) ||
1303 (ch >= 0xfdf0 && ch <= 0xfffd);
1304 }
1305
1306 /* [4 NS] NCName ::= Name - (Char* ':' Char*) */
1307 static inline BOOL is_ncnamechar(WCHAR ch)
1308 {
1309 return (ch >= 'A' && ch <= 'Z') ||
1310 (ch == '_') || (ch >= 'a' && ch <= 'z') ||
1311 (ch == '-') || (ch == '.') ||
1312 (ch >= '0' && ch <= '9') ||
1313 (ch == 0xb7) ||
1314 (ch >= 0xc0 && ch <= 0xd6) ||
1315 (ch >= 0xd8 && ch <= 0xf6) ||
1316 (ch >= 0xf8 && ch <= 0x2ff) ||
1317 (ch >= 0x300 && ch <= 0x36f) ||
1318 (ch >= 0x370 && ch <= 0x37d) ||
1319 (ch >= 0x37f && ch <= 0x1fff) ||
1320 (ch >= 0x200c && ch <= 0x200d) ||
1321 (ch >= 0x203f && ch <= 0x2040) ||
1322 (ch >= 0x2070 && ch <= 0x218f) ||
1323 (ch >= 0x2c00 && ch <= 0x2fef) ||
1324 (ch >= 0x3001 && ch <= 0xd7ff) ||
1325 (ch >= 0xd800 && ch <= 0xdbff) || /* high surrogate */
1326 (ch >= 0xdc00 && ch <= 0xdfff) || /* low surrogate */
1327 (ch >= 0xf900 && ch <= 0xfdcf) ||
1328 (ch >= 0xfdf0 && ch <= 0xfffd);
1329 }
1330
1331 static inline BOOL is_namechar(WCHAR ch)
1332 {
1333 return (ch == ':') || is_ncnamechar(ch);
1334 }
1335
1336 static XmlNodeType reader_get_nodetype(const xmlreader *reader)
1337 {
1338 /* When we're on attribute always return attribute type, container node type is kept.
1339 Note that container is not necessarily an element, and attribute doesn't mean it's
1340 an attribute in XML spec terms. */
1341 return reader->attr ? XmlNodeType_Attribute : reader->nodetype;
1342 }
1343
1344 /* [4] NameStartChar ::= ":" | [A-Z] | "_" | [a-z] | [#xC0-#xD6] | [#xD8-#xF6] | [#xF8-#x2FF] | [#x370-#x37D] |
1345 [#x37F-#x1FFF] | [#x200C-#x200D] | [#x2070-#x218F] | [#x2C00-#x2FEF] | [#x3001-#xD7FF] |
1346 [#xF900-#xFDCF] | [#xFDF0-#xFFFD] | [#x10000-#xEFFFF]
1347 [4a] NameChar ::= NameStartChar | "-" | "." | [0-9] | #xB7 | [#x0300-#x036F] | [#x203F-#x2040]
1348 [5] Name ::= NameStartChar (NameChar)* */
1349 static HRESULT reader_parse_name(xmlreader *reader, strval *name)
1350 {
1351 WCHAR *ptr;
1352 UINT start;
1353
1354 if (reader->resume[XmlReadResume_Name])
1355 {
1356 start = reader->resume[XmlReadResume_Name];
1357 ptr = reader_get_ptr(reader);
1358 }
1359 else
1360 {
1361 ptr = reader_get_ptr(reader);
1362 start = reader_get_cur(reader);
1363 if (!is_namestartchar(*ptr)) return WC_E_NAMECHARACTER;
1364 }
1365
1366 while (is_namechar(*ptr))
1367 {
1368 reader_skipn(reader, 1);
1369 ptr = reader_get_ptr(reader);
1370 }
1371
1372 if (is_reader_pending(reader))
1373 {
1374 reader->resume[XmlReadResume_Name] = start;
1375 return E_PENDING;
1376 }
1377 else
1378 reader->resume[XmlReadResume_Name] = 0;
1379
1380 reader_init_strvalue(start, reader_get_cur(reader)-start, name);
1381 TRACE("name %s:%d\n", debug_strval(reader, name), name->len);
1382
1383 return S_OK;
1384 }
1385
1386 /* [17] PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l')) */
1387 static HRESULT reader_parse_pitarget(xmlreader *reader, strval *target)
1388 {
1389 static const WCHAR xmlW[] = {'x','m','l'};
1390 static const strval xmlval = { (WCHAR*)xmlW, 3 };
1391 strval name;
1392 WCHAR *ptr;
1393 HRESULT hr;
1394 UINT i;
1395
1396 hr = reader_parse_name(reader, &name);
1397 if (FAILED(hr)) return is_reader_pending(reader) ? E_PENDING : WC_E_PI;
1398
1399 /* now that we got name check for illegal content */
1400 if (strval_eq(reader, &name, &xmlval))
1401 return WC_E_LEADINGXML;
1402
1403 /* PITarget can't be a qualified name */
1404 ptr = reader_get_strptr(reader, &name);
1405 for (i = 0; i < name.len; i++)
1406 if (ptr[i] == ':')
1407 return i ? NC_E_NAMECOLON : WC_E_PI;
1408
1409 TRACE("pitarget %s:%d\n", debug_strval(reader, &name), name.len);
1410 *target = name;
1411 return S_OK;
1412 }
1413
1414 /* [16] PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>' */
1415 static HRESULT reader_parse_pi(xmlreader *reader)
1416 {
1417 strval target;
1418 WCHAR *ptr;
1419 UINT start;
1420 HRESULT hr;
1421
1422 switch (reader->resumestate)
1423 {
1424 case XmlReadResumeState_Initial:
1425 /* skip '<?' */
1426 reader_skipn(reader, 2);
1427 reader_shrink(reader);
1428 reader->resumestate = XmlReadResumeState_PITarget;
1429 case XmlReadResumeState_PITarget:
1430 hr = reader_parse_pitarget(reader, &target);
1431 if (FAILED(hr)) return hr;
1432 reader_set_strvalue(reader, StringValue_LocalName, &target);
1433 reader_set_strvalue(reader, StringValue_QualifiedName, &target);
1434 reader_set_strvalue(reader, StringValue_Value, &strval_empty);
1435 reader->resumestate = XmlReadResumeState_PIBody;
1436 reader->resume[XmlReadResume_Body] = reader_get_cur(reader);
1437 default:
1438 ;
1439 }
1440
1441 start = reader->resume[XmlReadResume_Body];
1442 ptr = reader_get_ptr(reader);
1443 while (*ptr)
1444 {
1445 if (ptr[0] == '?')
1446 {
1447 if (ptr[1] == '>')
1448 {
1449 UINT cur = reader_get_cur(reader);
1450 strval value;
1451
1452 /* strip all leading whitespace chars */
1453 while (start < cur)
1454 {
1455 ptr = reader_get_ptr2(reader, start);
1456 if (!is_wchar_space(*ptr)) break;
1457 start++;
1458 }
1459
1460 reader_init_strvalue(start, cur-start, &value);
1461
1462 /* skip '?>' */
1463 reader_skipn(reader, 2);
1464 TRACE("%s\n", debug_strval(reader, &value));
1465 reader->nodetype = XmlNodeType_ProcessingInstruction;
1466 reader->resumestate = XmlReadResumeState_Initial;
1467 reader->resume[XmlReadResume_Body] = 0;
1468 reader_set_strvalue(reader, StringValue_Value, &value);
1469 return S_OK;
1470 }
1471 }
1472
1473 reader_skipn(reader, 1);
1474 ptr = reader_get_ptr(reader);
1475 }
1476
1477 return S_OK;
1478 }
1479
1480 /* This one is used to parse significant whitespace nodes, like in Misc production */
1481 static HRESULT reader_parse_whitespace(xmlreader *reader)
1482 {
1483 switch (reader->resumestate)
1484 {
1485 case XmlReadResumeState_Initial:
1486 reader_shrink(reader);
1487 reader->resumestate = XmlReadResumeState_Whitespace;
1488 reader->resume[XmlReadResume_Body] = reader_get_cur(reader);
1489 reader->nodetype = XmlNodeType_Whitespace;
1490 reader_set_strvalue(reader, StringValue_LocalName, &strval_empty);
1491 reader_set_strvalue(reader, StringValue_QualifiedName, &strval_empty);
1492 reader_set_strvalue(reader, StringValue_Value, &strval_empty);
1493 /* fallthrough */
1494 case XmlReadResumeState_Whitespace:
1495 {
1496 strval value;
1497 UINT start;
1498
1499 reader_skipspaces(reader);
1500 if (is_reader_pending(reader)) return S_OK;
1501
1502 start = reader->resume[XmlReadResume_Body];
1503 reader_init_strvalue(start, reader_get_cur(reader)-start, &value);
1504 reader_set_strvalue(reader, StringValue_Value, &value);
1505 TRACE("%s\n", debug_strval(reader, &value));
1506 reader->resumestate = XmlReadResumeState_Initial;
1507 }
1508 default:
1509 ;
1510 }
1511
1512 return S_OK;
1513 }
1514
1515 /* [27] Misc ::= Comment | PI | S */
1516 static HRESULT reader_parse_misc(xmlreader *reader)
1517 {
1518 HRESULT hr = S_FALSE;
1519
1520 if (reader->resumestate != XmlReadResumeState_Initial)
1521 {
1522 hr = reader_more(reader);
1523 if (FAILED(hr)) return hr;
1524
1525 /* finish current node */
1526 switch (reader->resumestate)
1527 {
1528 case XmlReadResumeState_PITarget:
1529 case XmlReadResumeState_PIBody:
1530 return reader_parse_pi(reader);
1531 case XmlReadResumeState_Comment:
1532 return reader_parse_comment(reader);
1533 case XmlReadResumeState_Whitespace:
1534 return reader_parse_whitespace(reader);
1535 default:
1536 ERR("unknown resume state %d\n", reader->resumestate);
1537 }
1538 }
1539
1540 while (1)
1541 {
1542 const WCHAR *cur = reader_get_ptr(reader);
1543
1544 if (is_wchar_space(*cur))
1545 hr = reader_parse_whitespace(reader);
1546 else if (!reader_cmp(reader, commentW))
1547 hr = reader_parse_comment(reader);
1548 else if (!reader_cmp(reader, piW))
1549 hr = reader_parse_pi(reader);
1550 else
1551 break;
1552
1553 if (hr != S_FALSE) return hr;
1554 }
1555
1556 return hr;
1557 }
1558
1559 /* [11] SystemLiteral ::= ('"' [^"]* '"') | ("'" [^']* "'") */
1560 static HRESULT reader_parse_sys_literal(xmlreader *reader, strval *literal)
1561 {
1562 WCHAR *cur = reader_get_ptr(reader), quote;
1563 UINT start;
1564
1565 if (*cur != '"' && *cur != '\'') return WC_E_QUOTE;
1566
1567 quote = *cur;
1568 reader_skipn(reader, 1);
1569
1570 cur = reader_get_ptr(reader);
1571 start = reader_get_cur(reader);
1572 while (is_char(*cur) && *cur != quote)
1573 {
1574 reader_skipn(reader, 1);
1575 cur = reader_get_ptr(reader);
1576 }
1577 reader_init_strvalue(start, reader_get_cur(reader)-start, literal);
1578 if (*cur == quote) reader_skipn(reader, 1);
1579
1580 TRACE("%s\n", debug_strval(reader, literal));
1581 return S_OK;
1582 }
1583
1584 /* [12] PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'"
1585 [13] PubidChar ::= #x20 | #xD | #xA | [a-zA-Z0-9] | [-'()+,./:=?;!*#@$_%] */
1586 static HRESULT reader_parse_pub_literal(xmlreader *reader, strval *literal)
1587 {
1588 WCHAR *cur = reader_get_ptr(reader), quote;
1589 UINT start;
1590
1591 if (*cur != '"' && *cur != '\'') return WC_E_QUOTE;
1592
1593 quote = *cur;
1594 reader_skipn(reader, 1);
1595
1596 start = reader_get_cur(reader);
1597 cur = reader_get_ptr(reader);
1598 while (is_pubchar(*cur) && *cur != quote)
1599 {
1600 reader_skipn(reader, 1);
1601 cur = reader_get_ptr(reader);
1602 }
1603
1604 reader_init_strvalue(start, reader_get_cur(reader)-start, literal);
1605 TRACE("%s\n", debug_strval(reader, literal));
1606 return S_OK;
1607 }
1608
1609 /* [75] ExternalID ::= 'SYSTEM' S SystemLiteral | 'PUBLIC' S PubidLiteral S SystemLiteral */
1610 static HRESULT reader_parse_externalid(xmlreader *reader)
1611 {
1612 static WCHAR systemW[] = {'S','Y','S','T','E','M',0};
1613 static WCHAR publicW[] = {'P','U','B','L','I','C',0};
1614 strval name;
1615 HRESULT hr;
1616 int cnt;
1617
1618 if (reader_cmp(reader, systemW))
1619 {
1620 if (reader_cmp(reader, publicW))
1621 return S_FALSE;
1622 else
1623 {
1624 strval pub;
1625
1626 /* public id */
1627 reader_skipn(reader, 6);
1628 cnt = reader_skipspaces(reader);
1629 if (!cnt) return WC_E_WHITESPACE;
1630
1631 hr = reader_parse_pub_literal(reader, &pub);
1632 if (FAILED(hr)) return hr;
1633
1634 reader_init_cstrvalue(publicW, strlenW(publicW), &name);
1635 return reader_add_attr(reader, &name, &pub);
1636 }
1637 }
1638 else
1639 {
1640 strval sys;
1641
1642 /* system id */
1643 reader_skipn(reader, 6);
1644 cnt = reader_skipspaces(reader);
1645 if (!cnt) return WC_E_WHITESPACE;
1646
1647 hr = reader_parse_sys_literal(reader, &sys);
1648 if (FAILED(hr)) return hr;
1649
1650 reader_init_cstrvalue(systemW, strlenW(systemW), &name);
1651 return reader_add_attr(reader, &name, &sys);
1652 }
1653
1654 return hr;
1655 }
1656
1657 /* [28] doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S? ('[' intSubset ']' S?)? '>' */
1658 static HRESULT reader_parse_dtd(xmlreader *reader)
1659 {
1660 static const WCHAR doctypeW[] = {'<','!','D','O','C','T','Y','P','E',0};
1661 strval name;
1662 WCHAR *cur;
1663 HRESULT hr;
1664
1665 /* check if we have "<!DOCTYPE" */
1666 if (reader_cmp(reader, doctypeW)) return S_FALSE;
1667 reader_shrink(reader);
1668
1669 /* DTD processing is not allowed by default */
1670 if (reader->dtdmode == DtdProcessing_Prohibit) return WC_E_DTDPROHIBITED;
1671
1672 reader_skipn(reader, 9);
1673 if (!reader_skipspaces(reader)) return WC_E_WHITESPACE;
1674
1675 /* name */
1676 hr = reader_parse_name(reader, &name);
1677 if (FAILED(hr)) return WC_E_DECLDOCTYPE;
1678
1679 reader_skipspaces(reader);
1680
1681 hr = reader_parse_externalid(reader);
1682 if (FAILED(hr)) return hr;
1683
1684 reader_skipspaces(reader);
1685
1686 cur = reader_get_ptr(reader);
1687 if (*cur != '>')
1688 {
1689 FIXME("internal subset parsing not implemented\n");
1690 return E_NOTIMPL;
1691 }
1692
1693 /* skip '>' */
1694 reader_skipn(reader, 1);
1695
1696 reader->nodetype = XmlNodeType_DocumentType;
1697 reader_set_strvalue(reader, StringValue_LocalName, &name);
1698 reader_set_strvalue(reader, StringValue_QualifiedName, &name);
1699
1700 return S_OK;
1701 }
1702
1703 /* [11 NS] LocalPart ::= NCName */
1704 static HRESULT reader_parse_local(xmlreader *reader, strval *local)
1705 {
1706 WCHAR *ptr;
1707 UINT start;
1708
1709 if (reader->resume[XmlReadResume_Local])
1710 {
1711 start = reader->resume[XmlReadResume_Local];
1712 ptr = reader_get_ptr(reader);
1713 }
1714 else
1715 {
1716 ptr = reader_get_ptr(reader);
1717 start = reader_get_cur(reader);
1718 }
1719
1720 while (is_ncnamechar(*ptr))
1721 {
1722 reader_skipn(reader, 1);
1723 ptr = reader_get_ptr(reader);
1724 }
1725
1726 if (is_reader_pending(reader))
1727 {
1728 reader->resume[XmlReadResume_Local] = start;
1729 return E_PENDING;
1730 }
1731 else
1732 reader->resume[XmlReadResume_Local] = 0;
1733
1734 reader_init_strvalue(start, reader_get_cur(reader)-start, local);
1735
1736 return S_OK;
1737 }
1738
1739 /* [7 NS] QName ::= PrefixedName | UnprefixedName
1740 [8 NS] PrefixedName ::= Prefix ':' LocalPart
1741 [9 NS] UnprefixedName ::= LocalPart
1742 [10 NS] Prefix ::= NCName */
1743 static HRESULT reader_parse_qname(xmlreader *reader, strval *prefix, strval *local, strval *qname)
1744 {
1745 WCHAR *ptr;
1746 UINT start;
1747 HRESULT hr;
1748
1749 if (reader->resume[XmlReadResume_Name])
1750 {
1751 start = reader->resume[XmlReadResume_Name];
1752 ptr = reader_get_ptr(reader);
1753 }
1754 else
1755 {
1756 ptr = reader_get_ptr(reader);
1757 start = reader_get_cur(reader);
1758 reader->resume[XmlReadResume_Name] = start;
1759 if (!is_ncnamechar(*ptr)) return NC_E_QNAMECHARACTER;
1760 }
1761
1762 if (reader->resume[XmlReadResume_Local])
1763 {
1764 hr = reader_parse_local(reader, local);
1765 if (FAILED(hr)) return hr;
1766
1767 reader_init_strvalue(reader->resume[XmlReadResume_Name],
1768 local->start - reader->resume[XmlReadResume_Name] - 1,
1769 prefix);
1770 }
1771 else
1772 {
1773 /* skip prefix part */
1774 while (is_ncnamechar(*ptr))
1775 {
1776 reader_skipn(reader, 1);
1777 ptr = reader_get_ptr(reader);
1778 }
1779
1780 if (is_reader_pending(reader)) return E_PENDING;
1781
1782 /* got a qualified name */
1783 if (*ptr == ':')
1784 {
1785 reader_init_strvalue(start, reader_get_cur(reader)-start, prefix);
1786
1787 /* skip ':' */
1788 reader_skipn(reader, 1);
1789 hr = reader_parse_local(reader, local);
1790 if (FAILED(hr)) return hr;
1791 }
1792 else
1793 {
1794 reader_init_strvalue(reader->resume[XmlReadResume_Name], reader_get_cur(reader)-reader->resume[XmlReadResume_Name], local);
1795 reader_init_strvalue(0, 0, prefix);
1796 }
1797 }
1798
1799 reader_init_strvalue(start, reader_get_cur(reader)-start, local);
1800
1801 if (prefix->len)
1802 TRACE("qname %s:%s\n", debug_strval(reader, prefix), debug_strval(reader, local));
1803 else
1804 TRACE("ncname %s\n", debug_strval(reader, local));
1805
1806 reader_init_strvalue(prefix->len ? prefix->start : local->start,
1807 /* count ':' too */
1808 (prefix->len ? prefix->len + 1 : 0) + local->len,
1809 qname);
1810
1811 reader->resume[XmlReadResume_Name] = 0;
1812 reader->resume[XmlReadResume_Local] = 0;
1813
1814 return S_OK;
1815 }
1816
1817 /* Applies normalization rules to a single char, used for attribute values.
1818
1819 Rules include 2 steps:
1820
1821 1) replacing \r\n with a single \n;
1822 2) replacing all whitespace chars with ' '.
1823
1824 */
1825 static void reader_normalize_space(xmlreader *reader, WCHAR *ptr)
1826 {
1827 encoded_buffer *buffer = &reader->input->buffer->utf16;
1828
1829 if (!is_wchar_space(*ptr)) return;
1830
1831 if (*ptr == '\r' && *(ptr+1) == '\n')
1832 {
1833 int len = buffer->written - ((char*)ptr - buffer->data) - 2*sizeof(WCHAR);
1834 memmove(ptr+1, ptr+2, len);
1835 }
1836 *ptr = ' ';
1837 }
1838
1839 static WCHAR get_predefined_entity(const xmlreader *reader, const strval *name)
1840 {
1841 static const WCHAR entltW[] = {'l','t'};
1842 static const WCHAR entgtW[] = {'g','t'};
1843 static const WCHAR entampW[] = {'a','m','p'};
1844 static const WCHAR entaposW[] = {'a','p','o','s'};
1845 static const WCHAR entquotW[] = {'q','u','o','t'};
1846 static const strval lt = { (WCHAR*)entltW, 2 };
1847 static const strval gt = { (WCHAR*)entgtW, 2 };
1848 static const strval amp = { (WCHAR*)entampW, 3 };
1849 static const strval apos = { (WCHAR*)entaposW, 4 };
1850 static const strval quot = { (WCHAR*)entquotW, 4 };
1851 WCHAR *str = reader_get_strptr(reader, name);
1852
1853 switch (*str)
1854 {
1855 case 'l':
1856 if (strval_eq(reader, name, &lt)) return '<';
1857 break;
1858 case 'g':
1859 if (strval_eq(reader, name, &gt)) return '>';
1860 break;
1861 case 'a':
1862 if (strval_eq(reader, name, &amp))
1863 return '&';
1864 else if (strval_eq(reader, name, &apos))
1865 return '\'';
1866 break;
1867 case 'q':
1868 if (strval_eq(reader, name, &quot)) return '\"';
1869 break;
1870 default:
1871 ;
1872 }
1873
1874 return 0;
1875 }
1876
1877 /* [66] CharRef ::= '&#' [0-9]+ ';' | '&#x' [0-9a-fA-F]+ ';'
1878 [67] Reference ::= EntityRef | CharRef
1879 [68] EntityRef ::= '&' Name ';' */
1880 static HRESULT reader_parse_reference(xmlreader *reader)
1881 {
1882 encoded_buffer *buffer = &reader->input->buffer->utf16;
1883 WCHAR *start = reader_get_ptr(reader), *ptr;
1884 UINT cur = reader_get_cur(reader);
1885 WCHAR ch = 0;
1886 int len;
1887
1888 /* skip '&' */
1889 reader_skipn(reader, 1);
1890 ptr = reader_get_ptr(reader);
1891
1892 if (*ptr == '#')
1893 {
1894 reader_skipn(reader, 1);
1895 ptr = reader_get_ptr(reader);
1896
1897 /* hex char or decimal */
1898 if (*ptr == 'x')
1899 {
1900 reader_skipn(reader, 1);
1901 ptr = reader_get_ptr(reader);
1902
1903 while (*ptr != ';')
1904 {
1905 if ((*ptr >= '0' && *ptr <= '9'))
1906 ch = ch*16 + *ptr - '0';
1907 else if ((*ptr >= 'a' && *ptr <= 'f'))
1908 ch = ch*16 + *ptr - 'a' + 10;
1909 else if ((*ptr >= 'A' && *ptr <= 'F'))
1910 ch = ch*16 + *ptr - 'A' + 10;
1911 else
1912 return ch ? WC_E_SEMICOLON : WC_E_HEXDIGIT;
1913 reader_skipn(reader, 1);
1914 ptr = reader_get_ptr(reader);
1915 }
1916 }
1917 else
1918 {
1919 while (*ptr != ';')
1920 {
1921 if ((*ptr >= '0' && *ptr <= '9'))
1922 {
1923 ch = ch*10 + *ptr - '0';
1924 reader_skipn(reader, 1);
1925 ptr = reader_get_ptr(reader);
1926 }
1927 else
1928 return ch ? WC_E_SEMICOLON : WC_E_DIGIT;
1929 }
1930 }
1931
1932 if (!is_char(ch)) return WC_E_XMLCHARACTER;
1933
1934 /* normalize */
1935 if (is_wchar_space(ch)) ch = ' ';
1936
1937 len = buffer->written - ((char*)ptr - buffer->data) - sizeof(WCHAR);
1938 memmove(start+1, ptr+1, len);
1939 buffer->cur = cur + 1;
1940
1941 *start = ch;
1942 }
1943 else
1944 {
1945 strval name;
1946 HRESULT hr;
1947
1948 hr = reader_parse_name(reader, &name);
1949 if (FAILED(hr)) return hr;
1950
1951 ptr = reader_get_ptr(reader);
1952 if (*ptr != ';') return WC_E_SEMICOLON;
1953
1954 /* predefined entities resolve to a single character */
1955 ch = get_predefined_entity(reader, &name);
1956 if (ch)
1957 {
1958 len = buffer->written - ((char*)ptr - buffer->data) - sizeof(WCHAR);
1959 memmove(start+1, ptr+1, len);
1960 buffer->cur = cur + 1;
1961
1962 *start = ch;
1963 }
1964 else
1965 {
1966 FIXME("undeclared entity %s\n", debug_strval(reader, &name));
1967 return WC_E_UNDECLAREDENTITY;
1968 }
1969
1970 }
1971
1972 return S_OK;
1973 }
1974
1975 /* [10 NS] AttValue ::= '"' ([^<&"] | Reference)* '"' | "'" ([^<&'] | Reference)* "'" */
1976 static HRESULT reader_parse_attvalue(xmlreader *reader, strval *value)
1977 {
1978 WCHAR *ptr, quote;
1979 UINT start;
1980
1981 ptr = reader_get_ptr(reader);
1982
1983 /* skip opening quote */
1984 quote = *ptr;
1985 if (quote != '\"' && quote != '\'') return WC_E_QUOTE;
1986 reader_skipn(reader, 1);
1987
1988 ptr = reader_get_ptr(reader);
1989 start = reader_get_cur(reader);
1990 while (*ptr)
1991 {
1992 if (*ptr == '<') return WC_E_LESSTHAN;
1993
1994 if (*ptr == quote)
1995 {
1996 reader_init_strvalue(start, reader_get_cur(reader)-start, value);
1997 /* skip closing quote */
1998 reader_skipn(reader, 1);
1999 return S_OK;
2000 }
2001
2002 if (*ptr == '&')
2003 {
2004 HRESULT hr = reader_parse_reference(reader);
2005 if (FAILED(hr)) return hr;
2006 }
2007 else
2008 {
2009 reader_normalize_space(reader, ptr);
2010 reader_skipn(reader, 1);
2011 }
2012 ptr = reader_get_ptr(reader);
2013 }
2014
2015 return WC_E_QUOTE;
2016 }
2017
2018 /* [1 NS] NSAttName ::= PrefixedAttName | DefaultAttName
2019 [2 NS] PrefixedAttName ::= 'xmlns:' NCName
2020 [3 NS] DefaultAttName ::= 'xmlns'
2021 [15 NS] Attribute ::= NSAttName Eq AttValue | QName Eq AttValue */
2022 static HRESULT reader_parse_attribute(xmlreader *reader)
2023 {
2024 static const WCHAR xmlnsW[] = {'x','m','l','n','s',0};
2025 strval prefix, local, qname, xmlns, value;
2026 HRESULT hr;
2027
2028 hr = reader_parse_qname(reader, &prefix, &local, &qname);
2029 if (FAILED(hr)) return hr;
2030
2031 reader_init_cstrvalue((WCHAR*)xmlnsW, 5, &xmlns);
2032
2033 if (strval_eq(reader, &prefix, &xmlns))
2034 {
2035 FIXME("namespace definitions not supported\n");
2036 return E_NOTIMPL;
2037 }
2038
2039 if (strval_eq(reader, &qname, &xmlns))
2040 FIXME("default namespace definitions not supported\n");
2041
2042 hr = reader_parse_eq(reader);
2043 if (FAILED(hr)) return hr;
2044
2045 hr = reader_parse_attvalue(reader, &value);
2046 if (FAILED(hr)) return hr;
2047
2048 TRACE("%s=%s\n", debug_strval(reader, &local), debug_strval(reader, &value));
2049 return reader_add_attr(reader, &local, &value);
2050 }
2051
2052 /* [12 NS] STag ::= '<' QName (S Attribute)* S? '>'
2053 [14 NS] EmptyElemTag ::= '<' QName (S Attribute)* S? '/>' */
2054 static HRESULT reader_parse_stag(xmlreader *reader, strval *prefix, strval *local, strval *qname, int *empty)
2055 {
2056 HRESULT hr;
2057
2058 hr = reader_parse_qname(reader, prefix, local, qname);
2059 if (FAILED(hr)) return hr;
2060
2061 while (1)
2062 {
2063 static const WCHAR endW[] = {'/','>',0};
2064
2065 reader_skipspaces(reader);
2066
2067 /* empty element */
2068 if ((*empty = !reader_cmp(reader, endW)))
2069 {
2070 /* skip '/>' */
2071 reader_skipn(reader, 2);
2072 reader->empty_element = TRUE;
2073 return S_OK;
2074 }
2075
2076 /* got a start tag */
2077 if (!reader_cmp(reader, gtW))
2078 {
2079 /* skip '>' */
2080 reader_skipn(reader, 1);
2081 return reader_push_element(reader, qname, local);
2082 }
2083
2084 hr = reader_parse_attribute(reader);
2085 if (FAILED(hr)) return hr;
2086 }
2087
2088 return S_OK;
2089 }
2090
2091 /* [39] element ::= EmptyElemTag | STag content ETag */
2092 static HRESULT reader_parse_element(xmlreader *reader)
2093 {
2094 HRESULT hr;
2095
2096 switch (reader->resumestate)
2097 {
2098 case XmlReadResumeState_Initial:
2099 /* check if we are really on element */
2100 if (reader_cmp(reader, ltW)) return S_FALSE;
2101
2102 /* skip '<' */
2103 reader_skipn(reader, 1);
2104
2105 reader_shrink(reader);
2106 reader->resumestate = XmlReadResumeState_STag;
2107 case XmlReadResumeState_STag:
2108 {
2109 strval qname, prefix, local;
2110 int empty = 0;
2111
2112 /* this handles empty elements too */
2113 hr = reader_parse_stag(reader, &prefix, &local, &qname, &empty);
2114 if (FAILED(hr)) return hr;
2115
2116 /* FIXME: need to check for defined namespace to reject invalid prefix,
2117 currently reject all prefixes */
2118 if (prefix.len) return NC_E_UNDECLAREDPREFIX;
2119
2120 /* if we got empty element and stack is empty go straight to Misc */
2121 if (empty && list_empty(&reader->elements))
2122 reader->instate = XmlReadInState_MiscEnd;
2123 else
2124 reader->instate = XmlReadInState_Content;
2125
2126 reader->nodetype = XmlNodeType_Element;
2127 reader->resumestate = XmlReadResumeState_Initial;
2128 reader_set_strvalue(reader, StringValue_LocalName, &local);
2129 reader_set_strvalue(reader, StringValue_Prefix, &prefix);
2130 reader_set_strvalue(reader, StringValue_QualifiedName, &qname);
2131 break;
2132 }
2133 default:
2134 hr = E_FAIL;
2135 }
2136
2137 return hr;
2138 }
2139
2140 /* [13 NS] ETag ::= '</' QName S? '>' */
2141 static HRESULT reader_parse_endtag(xmlreader *reader)
2142 {
2143 strval prefix, local, qname;
2144 struct element *elem;
2145 HRESULT hr;
2146
2147 /* skip '</' */
2148 reader_skipn(reader, 2);
2149
2150 hr = reader_parse_qname(reader, &prefix, &local, &qname);
2151 if (FAILED(hr)) return hr;
2152
2153 reader_skipspaces(reader);
2154
2155 if (reader_cmp(reader, gtW)) return WC_E_GREATERTHAN;
2156
2157 /* skip '>' */
2158 reader_skipn(reader, 1);
2159
2160 /* Element stack should never be empty at this point, cause we shouldn't get to
2161 content parsing if it's empty. */
2162 elem = LIST_ENTRY(list_head(&reader->elements), struct element, entry);
2163 if (!strval_eq(reader, &elem->qname, &qname)) return WC_E_ELEMENTMATCH;
2164
2165 reader_pop_element(reader);
2166
2167 /* It was a root element, the rest is expected as Misc */
2168 if (list_empty(&reader->elements))
2169 reader->instate = XmlReadInState_MiscEnd;
2170
2171 reader->nodetype = XmlNodeType_EndElement;
2172 reader_set_strvalue(reader, StringValue_LocalName, &local);
2173 reader_set_strvalue(reader, StringValue_QualifiedName, &qname);
2174
2175 return S_OK;
2176 }
2177
2178 /* [18] CDSect ::= CDStart CData CDEnd
2179 [19] CDStart ::= '<![CDATA['
2180 [20] CData ::= (Char* - (Char* ']]>' Char*))
2181 [21] CDEnd ::= ']]>' */
2182 static HRESULT reader_parse_cdata(xmlreader *reader)
2183 {
2184 WCHAR *ptr;
2185 UINT start;
2186
2187 if (reader->resumestate == XmlReadResumeState_CDATA)
2188 {
2189 start = reader->resume[XmlReadResume_Body];
2190 ptr = reader_get_ptr(reader);
2191 }
2192 else
2193 {
2194 /* skip markup '<![CDATA[' */
2195 reader_skipn(reader, 9);
2196 reader_shrink(reader);
2197 ptr = reader_get_ptr(reader);
2198 start = reader_get_cur(reader);
2199 reader->nodetype = XmlNodeType_CDATA;
2200 reader->resume[XmlReadResume_Body] = start;
2201 reader->resumestate = XmlReadResumeState_CDATA;
2202 reader_set_strvalue(reader, StringValue_LocalName, NULL);
2203 reader_set_strvalue(reader, StringValue_QualifiedName, NULL);
2204 reader_set_strvalue(reader, StringValue_Value, NULL);
2205 }
2206
2207 while (*ptr)
2208 {
2209 if (*ptr == ']' && *(ptr+1) == ']' && *(ptr+2) == '>')
2210 {
2211 strval value;
2212
2213 reader_init_strvalue(start, reader_get_cur(reader)-start, &value);
2214
2215 /* skip ']]>' */
2216 reader_skipn(reader, 3);
2217 TRACE("%s\n", debug_strval(reader, &value));
2218
2219 reader_set_strvalue(reader, StringValue_LocalName, &strval_empty);
2220 reader_set_strvalue(reader, StringValue_QualifiedName, &strval_empty);
2221 reader_set_strvalue(reader, StringValue_Value, &value);
2222 reader->resume[XmlReadResume_Body] = 0;
2223 reader->resumestate = XmlReadResumeState_Initial;
2224 return S_OK;
2225 }
2226 else
2227 {
2228 /* Value normalization is not fully implemented, rules are:
2229
2230 - single '\r' -> '\n';
2231 - sequence '\r\n' -> '\n', in this case value length changes;
2232 */
2233 if (*ptr == '\r') *ptr = '\n';
2234 reader_skipn(reader, 1);
2235 ptr++;
2236 }
2237 }
2238
2239 return S_OK;
2240 }
2241
2242 /* [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*) */
2243 static HRESULT reader_parse_chardata(xmlreader *reader)
2244 {
2245 WCHAR *ptr;
2246 UINT start;
2247
2248 if (reader->resumestate == XmlReadResumeState_CharData)
2249 {
2250 start = reader->resume[XmlReadResume_Body];
2251 ptr = reader_get_ptr(reader);
2252 }
2253 else
2254 {
2255 reader_shrink(reader);
2256 ptr = reader_get_ptr(reader);
2257 start = reader_get_cur(reader);
2258 /* There's no text */
2259 if (!*ptr || *ptr == '<') return S_OK;
2260 reader->nodetype = is_wchar_space(*ptr) ? XmlNodeType_Whitespace : XmlNodeType_Text;
2261 reader->resume[XmlReadResume_Body] = start;
2262 reader->resumestate = XmlReadResumeState_CharData;
2263 reader_set_strvalue(reader, StringValue_LocalName, &strval_empty);
2264 reader_set_strvalue(reader, StringValue_QualifiedName, &strval_empty);
2265 reader_set_strvalue(reader, StringValue_Value, NULL);
2266 }
2267
2268 while (*ptr)
2269 {
2270 /* CDATA closing sequence ']]>' is not allowed */
2271 if (ptr[0] == ']' && ptr[1] == ']' && ptr[2] == '>')
2272 return WC_E_CDSECTEND;
2273
2274 /* Found next markup part */
2275 if (ptr[0] == '<')
2276 {
2277 strval value;
2278
2279 reader_init_strvalue(start, reader_get_cur(reader)-start, &value);
2280 reader_set_strvalue(reader, StringValue_Value, &value);
2281 reader->resume[XmlReadResume_Body] = 0;
2282 reader->resumestate = XmlReadResumeState_Initial;
2283 return S_OK;
2284 }
2285
2286 reader_skipn(reader, 1);
2287
2288 /* this covers a case when text has leading whitespace chars */
2289 if (!is_wchar_space(*ptr)) reader->nodetype = XmlNodeType_Text;
2290 ptr++;
2291 }
2292
2293 return S_OK;
2294 }
2295
2296 /* [43] content ::= CharData? ((element | Reference | CDSect | PI | Comment) CharData?)* */
2297 static HRESULT reader_parse_content(xmlreader *reader)
2298 {
2299 static const WCHAR cdstartW[] = {'<','!','[','C','D','A','T','A','[',0};
2300 static const WCHAR etagW[] = {'<','/',0};
2301 static const WCHAR ampW[] = {'&',0};
2302
2303 if (reader->resumestate != XmlReadResumeState_Initial)
2304 {
2305 switch (reader->resumestate)
2306 {
2307 case XmlReadResumeState_CDATA:
2308 return reader_parse_cdata(reader);
2309 case XmlReadResumeState_Comment:
2310 return reader_parse_comment(reader);
2311 case XmlReadResumeState_PIBody:
2312 case XmlReadResumeState_PITarget:
2313 return reader_parse_pi(reader);
2314 case XmlReadResumeState_CharData:
2315 return reader_parse_chardata(reader);
2316 default:
2317 ERR("unknown resume state %d\n", reader->resumestate);
2318 }
2319 }
2320
2321 reader_shrink(reader);
2322
2323 /* handle end tag here, it indicates end of content as well */
2324 if (!reader_cmp(reader, etagW))
2325 return reader_parse_endtag(reader);
2326
2327 if (!reader_cmp(reader, commentW))
2328 return reader_parse_comment(reader);
2329
2330 if (!reader_cmp(reader, piW))
2331 return reader_parse_pi(reader);
2332
2333 if (!reader_cmp(reader, cdstartW))
2334 return reader_parse_cdata(reader);
2335
2336 if (!reader_cmp(reader, ampW))
2337 return reader_parse_reference(reader);
2338
2339 if (!reader_cmp(reader, ltW))
2340 return reader_parse_element(reader);
2341
2342 /* what's left must be CharData */
2343 return reader_parse_chardata(reader);
2344 }
2345
2346 static HRESULT reader_parse_nextnode(xmlreader *reader)
2347 {
2348 HRESULT hr;
2349
2350 if (!is_reader_pending(reader))
2351 reader_clear_attrs(reader);
2352
2353 while (1)
2354 {
2355 switch (reader->instate)
2356 {
2357 /* if it's a first call for a new input we need to detect stream encoding */
2358 case XmlReadInState_Initial:
2359 {
2360 xml_encoding enc;
2361
2362 hr = readerinput_growraw(reader->input);
2363 if (FAILED(hr)) return hr;
2364
2365 /* try to detect encoding by BOM or data and set input code page */
2366 hr = readerinput_detectencoding(reader->input, &enc);
2367 TRACE("detected encoding %s, 0x%08x\n", debugstr_w(xml_encoding_map[enc].name), hr);
2368 if (FAILED(hr)) return hr;
2369
2370 /* always switch first time cause we have to put something in */
2371 readerinput_switchencoding(reader->input, enc);
2372
2373 /* parse xml declaration */
2374 hr = reader_parse_xmldecl(reader);
2375 if (FAILED(hr)) return hr;
2376
2377 readerinput_shrinkraw(reader->input, -1);
2378 reader->instate = XmlReadInState_Misc_DTD;
2379 if (hr == S_OK) return hr;
2380 }
2381 break;
2382 case XmlReadInState_Misc_DTD:
2383 hr = reader_parse_misc(reader);
2384 if (FAILED(hr)) return hr;
2385
2386 if (hr == S_FALSE)
2387 reader->instate = XmlReadInState_DTD;
2388 else
2389 return hr;
2390 break;
2391 case XmlReadInState_DTD:
2392 hr = reader_parse_dtd(reader);
2393 if (FAILED(hr)) return hr;
2394
2395 if (hr == S_OK)
2396 {
2397 reader->instate = XmlReadInState_DTD_Misc;
2398 return hr;
2399 }
2400 else
2401 reader->instate = XmlReadInState_Element;
2402 break;
2403 case XmlReadInState_DTD_Misc:
2404 hr = reader_parse_misc(reader);
2405 if (FAILED(hr)) return hr;
2406
2407 if (hr == S_FALSE)
2408 reader->instate = XmlReadInState_Element;
2409 else
2410 return hr;
2411 break;
2412 case XmlReadInState_Element:
2413 return reader_parse_element(reader);
2414 case XmlReadInState_Content:
2415 return reader_parse_content(reader);
2416 case XmlReadInState_MiscEnd:
2417 hr = reader_parse_misc(reader);
2418 if (FAILED(hr)) return hr;
2419
2420 if (hr == S_FALSE)
2421 reader->instate = XmlReadInState_Eof;
2422 return hr;
2423 case XmlReadInState_Eof:
2424 return S_FALSE;
2425 default:
2426 FIXME("internal state %d not handled\n", reader->instate);
2427 return E_NOTIMPL;
2428 }
2429 }
2430
2431 return E_NOTIMPL;
2432 }
2433
2434 static HRESULT WINAPI xmlreader_QueryInterface(IXmlReader *iface, REFIID riid, void** ppvObject)
2435 {
2436 xmlreader *This = impl_from_IXmlReader(iface);
2437
2438 TRACE("(%p)->(%s %p)\n", This, debugstr_guid(riid), ppvObject);
2439
2440 if (IsEqualGUID(riid, &IID_IUnknown) ||
2441 IsEqualGUID(riid, &IID_IXmlReader))
2442 {
2443 *ppvObject = iface;
2444 }
2445 else
2446 {
2447 FIXME("interface %s not implemented\n", debugstr_guid(riid));
2448 *ppvObject = NULL;
2449 return E_NOINTERFACE;
2450 }
2451
2452 IXmlReader_AddRef(iface);
2453
2454 return S_OK;
2455 }
2456
2457 static ULONG WINAPI xmlreader_AddRef(IXmlReader *iface)
2458 {
2459 xmlreader *This = impl_from_IXmlReader(iface);
2460 ULONG ref = InterlockedIncrement(&This->ref);
2461 TRACE("(%p)->(%d)\n", This, ref);
2462 return ref;
2463 }
2464
2465 static ULONG WINAPI xmlreader_Release(IXmlReader *iface)
2466 {
2467 xmlreader *This = impl_from_IXmlReader(iface);
2468 LONG ref = InterlockedDecrement(&This->ref);
2469
2470 TRACE("(%p)->(%d)\n", This, ref);
2471
2472 if (ref == 0)
2473 {
2474 IMalloc *imalloc = This->imalloc;
2475 if (This->input) IUnknown_Release(&This->input->IXmlReaderInput_iface);
2476 if (This->resolver) IXmlResolver_Release(This->resolver);
2477 reader_clear_attrs(This);
2478 reader_clear_elements(This);
2479 reader_free_strvalues(This);
2480 reader_free(This, This);
2481 if (imalloc) IMalloc_Release(imalloc);
2482 }
2483
2484 return ref;
2485 }
2486
2487 static HRESULT WINAPI xmlreader_SetInput(IXmlReader* iface, IUnknown *input)
2488 {
2489 xmlreader *This = impl_from_IXmlReader(iface);
2490 IXmlReaderInput *readerinput;
2491 HRESULT hr;
2492
2493 TRACE("(%p)->(%p)\n", This, input);
2494
2495 if (This->input)
2496 {
2497 readerinput_release_stream(This->input);
2498 IUnknown_Release(&This->input->IXmlReaderInput_iface);
2499 This->input = NULL;
2500 }
2501
2502 This->line = This->pos = 0;
2503 reader_clear_elements(This);
2504 This->depth = 0;
2505 This->resumestate = XmlReadResumeState_Initial;
2506 memset(This->resume, 0, sizeof(This->resume));
2507
2508 /* just reset current input */
2509 if (!input)
2510 {
2511 This->state = XmlReadState_Initial;
2512 return S_OK;
2513 }
2514
2515 /* now try IXmlReaderInput, ISequentialStream, IStream */
2516 hr = IUnknown_QueryInterface(input, &IID_IXmlReaderInput, (void**)&readerinput);
2517 if (hr == S_OK)
2518 {
2519 if (readerinput->lpVtbl == &xmlreaderinputvtbl)
2520 This->input = impl_from_IXmlReaderInput(readerinput);
2521 else
2522 {
2523 ERR("got external IXmlReaderInput implementation: %p, vtbl=%p\n",
2524 readerinput, readerinput->lpVtbl);
2525 IUnknown_Release(readerinput);
2526 return E_FAIL;
2527
2528 }
2529 }
2530
2531 if (hr != S_OK || !readerinput)
2532 {
2533 /* create IXmlReaderInput basing on supplied interface */
2534 hr = CreateXmlReaderInputWithEncodingName(input,
2535 This->imalloc, NULL, FALSE, NULL, &readerinput);
2536 if (hr != S_OK) return hr;
2537 This->input = impl_from_IXmlReaderInput(readerinput);
2538 }
2539
2540 /* set stream for supplied IXmlReaderInput */
2541 hr = readerinput_query_for_stream(This->input);
2542 if (hr == S_OK)
2543 {
2544 This->state = XmlReadState_Initial;
2545 This->instate = XmlReadInState_Initial;
2546 }
2547
2548 return hr;
2549 }
2550
2551 static HRESULT WINAPI xmlreader_GetProperty(IXmlReader* iface, UINT property, LONG_PTR *value)
2552 {
2553 xmlreader *This = impl_from_IXmlReader(iface);
2554
2555 TRACE("(%p)->(%s %p)\n", This, debugstr_reader_prop(property), value);
2556
2557 if (!value) return E_INVALIDARG;
2558
2559 switch (property)
2560 {
2561 case XmlReaderProperty_XmlResolver:
2562 *value = (LONG_PTR)This->resolver;
2563 if (This->resolver)
2564 IXmlResolver_AddRef(This->resolver);
2565 break;
2566 case XmlReaderProperty_DtdProcessing:
2567 *value = This->dtdmode;
2568 break;
2569 case XmlReaderProperty_ReadState:
2570 *value = This->state;
2571 break;
2572 default:
2573 FIXME("Unimplemented property (%u)\n", property);
2574 return E_NOTIMPL;
2575 }
2576
2577 return S_OK;
2578 }
2579
2580 static HRESULT WINAPI xmlreader_SetProperty(IXmlReader* iface, UINT property, LONG_PTR value)
2581 {
2582 xmlreader *This = impl_from_IXmlReader(iface);
2583
2584 TRACE("(%p)->(%s 0x%lx)\n", This, debugstr_reader_prop(property), value);
2585
2586 switch (property)
2587 {
2588 case XmlReaderProperty_XmlResolver:
2589 if (This->resolver)
2590 IXmlResolver_Release(This->resolver);
2591 This->resolver = (IXmlResolver*)value;
2592 if (This->resolver)
2593 IXmlResolver_AddRef(This->resolver);
2594 break;
2595 case XmlReaderProperty_DtdProcessing:
2596 if (value < 0 || value > _DtdProcessing_Last) return E_INVALIDARG;
2597 This->dtdmode = value;
2598 break;
2599 default:
2600 FIXME("Unimplemented property (%u)\n", property);
2601 return E_NOTIMPL;
2602 }
2603
2604 return S_OK;
2605 }
2606
2607 static HRESULT WINAPI xmlreader_Read(IXmlReader* iface, XmlNodeType *nodetype)
2608 {
2609 xmlreader *This = impl_from_IXmlReader(iface);
2610 XmlNodeType oldtype = This->nodetype;
2611 HRESULT hr;
2612
2613 TRACE("(%p)->(%p)\n", This, nodetype);
2614
2615 if (This->state == XmlReadState_Closed) return S_FALSE;
2616
2617 hr = reader_parse_nextnode(This);
2618 if (oldtype == XmlNodeType_None && This->nodetype != oldtype)
2619 This->state = XmlReadState_Interactive;
2620 if (hr == S_OK)
2621 {
2622 TRACE("node type %s\n", debugstr_nodetype(This->nodetype));
2623 *nodetype = This->nodetype;
2624 }
2625
2626 return hr;
2627 }
2628
2629 static HRESULT WINAPI xmlreader_GetNodeType(IXmlReader* iface, XmlNodeType *node_type)
2630 {
2631 xmlreader *This = impl_from_IXmlReader(iface);
2632 TRACE("(%p)->(%p)\n", This, node_type);
2633
2634 *node_type = reader_get_nodetype(This);
2635 return This->state == XmlReadState_Closed ? S_FALSE : S_OK;
2636 }
2637
2638 static HRESULT WINAPI xmlreader_MoveToFirstAttribute(IXmlReader* iface)
2639 {
2640 xmlreader *This = impl_from_IXmlReader(iface);
2641
2642 TRACE("(%p)\n", This);
2643
2644 if (!This->attr_count) return S_FALSE;
2645 This->attr = LIST_ENTRY(list_head(&This->attrs), struct attribute, entry);
2646 reader_set_strvalue(This, StringValue_LocalName, &This->attr->localname);
2647 reader_set_strvalue(This, StringValue_Value, &This->attr->value);
2648
2649 return S_OK;
2650 }
2651
2652 static HRESULT WINAPI xmlreader_MoveToNextAttribute(IXmlReader* iface)
2653 {
2654 xmlreader *This = impl_from_IXmlReader(iface);
2655 const struct list *next;
2656
2657 TRACE("(%p)\n", This);
2658
2659 if (!This->attr_count) return S_FALSE;
2660
2661 if (!This->attr)
2662 return IXmlReader_MoveToFirstAttribute(iface);
2663
2664 next = list_next(&This->attrs, &This->attr->entry);
2665 if (next)
2666 {
2667 This->attr = LIST_ENTRY(next, struct attribute, entry);
2668 reader_set_strvalue(This, StringValue_LocalName, &This->attr->localname);
2669 reader_set_strvalue(This, StringValue_Value, &This->attr->value);
2670 }
2671
2672 return next ? S_OK : S_FALSE;
2673 }
2674
2675 static HRESULT WINAPI xmlreader_MoveToAttributeByName(IXmlReader* iface,
2676 LPCWSTR local_name,
2677 LPCWSTR namespaceUri)
2678 {
2679 FIXME("(%p %p %p): stub\n", iface, local_name, namespaceUri);
2680 return E_NOTIMPL;
2681 }
2682
2683 static HRESULT WINAPI xmlreader_MoveToElement(IXmlReader* iface)
2684 {
2685 xmlreader *This = impl_from_IXmlReader(iface);
2686 struct element *elem;
2687
2688 TRACE("(%p)\n", This);
2689
2690 if (!This->attr_count) return S_FALSE;
2691 This->attr = NULL;
2692
2693 /* FIXME: support other node types with 'attributes' like DTD */
2694 elem = LIST_ENTRY(list_head(&This->elements), struct element, entry);
2695 if (elem)
2696 {
2697 reader_set_strvalue(This, StringValue_QualifiedName, &elem->qname);
2698 reader_set_strvalue(This, StringValue_LocalName, &elem->localname);
2699 }
2700
2701 return S_OK;
2702 }
2703
2704 static HRESULT WINAPI xmlreader_GetQualifiedName(IXmlReader* iface, LPCWSTR *name, UINT *len)
2705 {
2706 xmlreader *This = impl_from_IXmlReader(iface);
2707
2708 TRACE("(%p)->(%p %p)\n", This, name, len);
2709 *name = This->strvalues[StringValue_QualifiedName].str;
2710 if (len) *len = This->strvalues[StringValue_QualifiedName].len;
2711 return S_OK;
2712 }
2713
2714 static HRESULT WINAPI xmlreader_GetNamespaceUri(IXmlReader* iface,
2715 LPCWSTR *namespaceUri,
2716 UINT *namespaceUri_length)
2717 {
2718 FIXME("(%p %p %p): stub\n", iface, namespaceUri, namespaceUri_length);
2719 return E_NOTIMPL;
2720 }
2721
2722 static HRESULT WINAPI xmlreader_GetLocalName(IXmlReader* iface, LPCWSTR *name, UINT *len)
2723 {
2724 xmlreader *This = impl_from_IXmlReader(iface);
2725
2726 TRACE("(%p)->(%p %p)\n", This, name, len);
2727 *name = This->strvalues[StringValue_LocalName].str;
2728 if (len) *len = This->strvalues[StringValue_LocalName].len;
2729 return S_OK;
2730 }
2731
2732 static HRESULT WINAPI xmlreader_GetPrefix(IXmlReader* iface, LPCWSTR *prefix, UINT *len)
2733 {
2734 xmlreader *This = impl_from_IXmlReader(iface);
2735
2736 TRACE("(%p)->(%p %p)\n", This, prefix, len);
2737 *prefix = This->strvalues[StringValue_Prefix].str;
2738 if (len) *len = This->strvalues[StringValue_Prefix].len;
2739 return S_OK;
2740 }
2741
2742 static HRESULT WINAPI xmlreader_GetValue(IXmlReader* iface, const WCHAR **value, UINT *len)
2743 {
2744 xmlreader *reader = impl_from_IXmlReader(iface);
2745 strval *val = &reader->strvalues[StringValue_Value];
2746
2747 TRACE("(%p)->(%p %p)\n", reader, value, len);
2748
2749 *value = NULL;
2750
2751 if ((reader->nodetype == XmlNodeType_Comment && !val->str) || is_reader_pending(reader))
2752 {
2753 XmlNodeType type;
2754 HRESULT hr;
2755
2756 hr = IXmlReader_Read(iface, &type);
2757 if (FAILED(hr)) return hr;
2758
2759 /* return if still pending, partially read values are not reported */
2760 if (is_reader_pending(reader)) return E_PENDING;
2761 }
2762
2763 if (!val->str)
2764 {
2765 WCHAR *ptr = reader_alloc(reader, (val->len+1)*sizeof(WCHAR));
2766 if (!ptr) return E_OUTOFMEMORY;
2767 memcpy(ptr, reader_get_strptr(reader, val), val->len*sizeof(WCHAR));
2768 ptr[val->len] = 0;
2769 val->str = ptr;
2770 }
2771
2772 *value = val->str;
2773 if (len) *len = val->len;
2774 return S_OK;
2775 }
2776
2777 static HRESULT WINAPI xmlreader_ReadValueChunk(IXmlReader* iface, WCHAR *buffer, UINT chunk_size, UINT *read)
2778 {
2779 xmlreader *reader = impl_from_IXmlReader(iface);
2780 strval *val = &reader->strvalues[StringValue_Value];
2781 UINT len;
2782
2783 TRACE("(%p)->(%p %u %p)\n", reader, buffer, chunk_size, read);
2784
2785 /* Value is already allocated, chunked reads are not possible. */
2786 if (val->str) return S_FALSE;
2787
2788 if (val->len)
2789 {
2790 len = min(chunk_size, val->len);
2791 memcpy(buffer, reader_get_ptr2(reader, val->start), len);
2792 val->start += len;
2793 val->len -= len;
2794 if (read) *read = len;
2795 }
2796
2797 return S_OK;
2798 }
2799
2800 static HRESULT WINAPI xmlreader_GetBaseUri(IXmlReader* iface,
2801 LPCWSTR *baseUri,
2802 UINT *baseUri_length)
2803 {
2804 FIXME("(%p %p %p): stub\n", iface, baseUri, baseUri_length);
2805 return E_NOTIMPL;
2806 }
2807
2808 static BOOL WINAPI xmlreader_IsDefault(IXmlReader* iface)
2809 {
2810 FIXME("(%p): stub\n", iface);
2811 return FALSE;
2812 }
2813
2814 static BOOL WINAPI xmlreader_IsEmptyElement(IXmlReader* iface)
2815 {
2816 xmlreader *This = impl_from_IXmlReader(iface);
2817 TRACE("(%p)\n", This);
2818 /* Empty elements are not placed in stack, it's stored as a global reader flag that makes sense
2819 when current node is start tag of an element */
2820 return (reader_get_nodetype(This) == XmlNodeType_Element) ? This->empty_element : FALSE;
2821 }
2822
2823 static HRESULT WINAPI xmlreader_GetLineNumber(IXmlReader* iface, UINT *lineNumber)
2824 {
2825 xmlreader *This = impl_from_IXmlReader(iface);
2826
2827 TRACE("(%p %p)\n", This, lineNumber);
2828
2829 if (!lineNumber) return E_INVALIDARG;
2830
2831 *lineNumber = This->line;
2832
2833 return S_OK;
2834 }
2835
2836 static HRESULT WINAPI xmlreader_GetLinePosition(IXmlReader* iface, UINT *linePosition)
2837 {
2838 xmlreader *This = impl_from_IXmlReader(iface);
2839
2840 TRACE("(%p %p)\n", This, linePosition);
2841
2842 if (!linePosition) return E_INVALIDARG;
2843
2844 *linePosition = This->pos;
2845
2846 return S_OK;
2847 }
2848
2849 static HRESULT WINAPI xmlreader_GetAttributeCount(IXmlReader* iface, UINT *count)
2850 {
2851 xmlreader *This = impl_from_IXmlReader(iface);
2852
2853 TRACE("(%p)->(%p)\n", This, count);
2854
2855 if (!count) return E_INVALIDARG;
2856
2857 *count = This->attr_count;
2858 return S_OK;
2859 }
2860
2861 static HRESULT WINAPI xmlreader_GetDepth(IXmlReader* iface, UINT *depth)
2862 {
2863 xmlreader *This = impl_from_IXmlReader(iface);
2864 TRACE("(%p)->(%p)\n", This, depth);
2865 *depth = This->depth;
2866 return S_OK;
2867 }
2868
2869 static BOOL WINAPI xmlreader_IsEOF(IXmlReader* iface)
2870 {
2871 FIXME("(%p): stub\n", iface);
2872 return FALSE;
2873 }
2874
2875 static const struct IXmlReaderVtbl xmlreader_vtbl =
2876 {
2877 xmlreader_QueryInterface,
2878 xmlreader_AddRef,
2879 xmlreader_Release,
2880 xmlreader_SetInput,
2881 xmlreader_GetProperty,
2882 xmlreader_SetProperty,
2883 xmlreader_Read,
2884 xmlreader_GetNodeType,
2885 xmlreader_MoveToFirstAttribute,
2886 xmlreader_MoveToNextAttribute,
2887 xmlreader_MoveToAttributeByName,
2888 xmlreader_MoveToElement,
2889 xmlreader_GetQualifiedName,
2890 xmlreader_GetNamespaceUri,
2891 xmlreader_GetLocalName,
2892 xmlreader_GetPrefix,
2893 xmlreader_GetValue,
2894 xmlreader_ReadValueChunk,
2895 xmlreader_GetBaseUri,
2896 xmlreader_IsDefault,
2897 xmlreader_IsEmptyElement,
2898 xmlreader_GetLineNumber,
2899 xmlreader_GetLinePosition,
2900 xmlreader_GetAttributeCount,
2901 xmlreader_GetDepth,
2902 xmlreader_IsEOF
2903 };
2904
2905 /** IXmlReaderInput **/
2906 static HRESULT WINAPI xmlreaderinput_QueryInterface(IXmlReaderInput *iface, REFIID riid, void** ppvObject)
2907 {
2908 xmlreaderinput *This = impl_from_IXmlReaderInput(iface);
2909
2910 TRACE("(%p)->(%s %p)\n", This, debugstr_guid(riid), ppvObject);
2911
2912 if (IsEqualGUID(riid, &IID_IXmlReaderInput) ||
2913 IsEqualGUID(riid, &IID_IUnknown))
2914 {
2915 *ppvObject = iface;
2916 }
2917 else
2918 {
2919 WARN("interface %s not implemented\n", debugstr_guid(riid));
2920 *ppvObject = NULL;
2921 return E_NOINTERFACE;
2922 }
2923
2924 IUnknown_AddRef(iface);
2925
2926 return S_OK;
2927 }
2928
2929 static ULONG WINAPI xmlreaderinput_AddRef(IXmlReaderInput *iface)
2930 {
2931 xmlreaderinput *This = impl_from_IXmlReaderInput(iface);
2932 ULONG ref = InterlockedIncrement(&This->ref);
2933 TRACE("(%p)->(%d)\n", This, ref);
2934 return ref;
2935 }
2936
2937 static ULONG WINAPI xmlreaderinput_Release(IXmlReaderInput *iface)
2938 {
2939 xmlreaderinput *This = impl_from_IXmlReaderInput(iface);
2940 LONG ref = InterlockedDecrement(&This->ref);
2941
2942 TRACE("(%p)->(%d)\n", This, ref);
2943
2944 if (ref == 0)
2945 {
2946 IMalloc *imalloc = This->imalloc;
2947 if (This->input) IUnknown_Release(This->input);
2948 if (This->stream) ISequentialStream_Release(This->stream);
2949 if (This->buffer) free_input_buffer(This->buffer);
2950 readerinput_free(This, This->baseuri);
2951 readerinput_free(This, This);
2952 if (imalloc) IMalloc_Release(imalloc);
2953 }
2954
2955 return ref;
2956 }
2957
2958 static const struct IUnknownVtbl xmlreaderinputvtbl =
2959 {
2960 xmlreaderinput_QueryInterface,
2961 xmlreaderinput_AddRef,
2962 xmlreaderinput_Release
2963 };
2964
2965 HRESULT WINAPI CreateXmlReader(REFIID riid, void **obj, IMalloc *imalloc)
2966 {
2967 xmlreader *reader;
2968 int i;
2969
2970 TRACE("(%s, %p, %p)\n", wine_dbgstr_guid(riid), obj, imalloc);
2971
2972 if (!IsEqualGUID(riid, &IID_IXmlReader))
2973 {
2974 ERR("Unexpected IID requested -> (%s)\n", wine_dbgstr_guid(riid));
2975 return E_FAIL;
2976 }
2977
2978 if (imalloc)
2979 reader = IMalloc_Alloc(imalloc, sizeof(*reader));
2980 else
2981 reader = heap_alloc(sizeof(*reader));
2982 if(!reader) return E_OUTOFMEMORY;
2983
2984 reader->IXmlReader_iface.lpVtbl = &xmlreader_vtbl;
2985 reader->ref = 1;
2986 reader->input = NULL;
2987 reader->state = XmlReadState_Closed;
2988 reader->instate = XmlReadInState_Initial;
2989 reader->resumestate = XmlReadResumeState_Initial;
2990 reader->dtdmode = DtdProcessing_Prohibit;
2991 reader->resolver = NULL;
2992 reader->line = reader->pos = 0;
2993 reader->imalloc = imalloc;
2994 if (imalloc) IMalloc_AddRef(imalloc);
2995 reader->nodetype = XmlNodeType_None;
2996 list_init(&reader->attrs);
2997 reader->attr_count = 0;
2998 reader->attr = NULL;
2999 list_init(&reader->elements);
3000 reader->depth = 0;
3001 reader->max_depth = 256;
3002 reader->empty_element = FALSE;
3003 memset(reader->resume, 0, sizeof(reader->resume));
3004
3005 for (i = 0; i < StringValue_Last; i++)
3006 reader->strvalues[i] = strval_empty;
3007
3008 *obj = &reader->IXmlReader_iface;
3009
3010 TRACE("returning iface %p\n", *obj);
3011
3012 return S_OK;
3013 }
3014
3015 HRESULT WINAPI CreateXmlReaderInputWithEncodingName(IUnknown *stream,
3016 IMalloc *imalloc,
3017 LPCWSTR encoding,
3018 BOOL hint,
3019 LPCWSTR base_uri,
3020 IXmlReaderInput **ppInput)
3021 {
3022 xmlreaderinput *readerinput;
3023 HRESULT hr;
3024
3025 TRACE("%p %p %s %d %s %p\n", stream, imalloc, wine_dbgstr_w(encoding),
3026 hint, wine_dbgstr_w(base_uri), ppInput);
3027
3028 if (!stream || !ppInput) return E_INVALIDARG;
3029
3030 if (imalloc)
3031 readerinput = IMalloc_Alloc(imalloc, sizeof(*readerinput));
3032 else
3033 readerinput = heap_alloc(sizeof(*readerinput));
3034 if(!readerinput) return E_OUTOFMEMORY;
3035
3036 readerinput->IXmlReaderInput_iface.lpVtbl = &xmlreaderinputvtbl;
3037 readerinput->ref = 1;
3038 readerinput->imalloc = imalloc;
3039 readerinput->stream = NULL;
3040 if (imalloc) IMalloc_AddRef(imalloc);
3041 readerinput->encoding = parse_encoding_name(encoding, -1);
3042 readerinput->hint = hint;
3043 readerinput->baseuri = readerinput_strdupW(readerinput, base_uri);
3044 readerinput->pending = 0;
3045
3046 hr = alloc_input_buffer(readerinput);
3047 if (hr != S_OK)
3048 {
3049 readerinput_free(readerinput, readerinput->baseuri);
3050 readerinput_free(readerinput, readerinput);
3051 if (imalloc) IMalloc_Release(imalloc);
3052 return hr;
3053 }
3054 IUnknown_QueryInterface(stream, &IID_IUnknown, (void**)&readerinput->input);
3055
3056 *ppInput = &readerinput->IXmlReaderInput_iface;
3057
3058 TRACE("returning iface %p\n", *ppInput);
3059
3060 return S_OK;
3061 }
Windows API implementation