xml_parser.c

一个用于智能手机的多媒体库适合S60 WinCE的跨平台开发库

		parser->text_start = parser->text_end = 0;
		return;
	}

	offset = 0;
	while (offset<parser->text_end) {
		c = parser->buffer[parser->text_end-2-offset];
		if (c=='\r') offset++;
		else if (c==' ') offset++;
		else if (c=='\n') {
			parser->line++;
			offset++;
		} else {
			break;
		}
	}
	parser->text_end-=offset;
	assert(parser->text_start < parser->text_end);

	c = parser->buffer[parser->text_end-1];
	parser->buffer[parser->text_end-1] = 0;
	text = parser->buffer + parser->text_start-1;

	/*solve XML built-in entities*/
	if (strchr(text, '&') && strchr(text, ';')) {
		char *xml_text = xml_translate_xml_string(text);
		if (xml_text) {
			parser->sax_text_content(parser->sax_cbck, xml_text, (parser->sax_state==SAX_STATE_CDATA) ? 1 : 0);
			free(xml_text);
		}
	} else {
		parser->sax_text_content(parser->sax_cbck, text, (parser->sax_state==SAX_STATE_CDATA) ? 1 : 0);
	}
	parser->buffer[parser->text_end-1] = c;
	parser->text_start = parser->text_end = 0;
}

static void xml_sax_store_text(GF_SAXParser *parser, u32 txt_len)
{
	if (!txt_len) return;

	if (!parser->text_start) {
		parser->text_start = parser->current_pos + 1;
		parser->text_end = parser->text_start + txt_len;
		parser->current_pos += txt_len;
		assert(parser->current_pos <= parser->line_size);
		return;
	}
	/*contiguous text*/
	if (parser->text_end && (parser->text_end-1 == parser->current_pos)) {
		parser->text_end += txt_len;
		parser->current_pos += txt_len;
		assert(parser->current_pos <= parser->line_size);
		return;
	}
	/*need to flush*/
	xml_sax_flush_text(parser);

	parser->text_start = parser->current_pos + 1;
	parser->text_end = parser->text_start + txt_len;
	parser->current_pos += txt_len;
	assert(parser->current_pos <= parser->line_size);
}

static char *xml_get_current_text(GF_SAXParser *parser)
{
	char *text, c;
	if (!parser->text_start) return NULL;

	c = parser->buffer[parser->text_end-1];
	parser->buffer[parser->text_end-1] = 0;
	text = strdup(parser->buffer + parser->text_start-1);
	parser->buffer[parser->text_end-1] = c;
	parser->text_start = parser->text_end = 0;
	return text;
}

static void xml_sax_skip_doctype(GF_SAXParser *parser)
{
	while (parser->current_pos < parser->line_size) {
		if (parser->buffer[parser->current_pos]=='>') {
			parser->sax_state = SAX_STATE_ELEMENT;
			parser->current_pos++;
			xml_sax_swap(parser);
			return;
		}
		parser->current_pos++;
	}
}

static void xml_sax_skip_xml_proc(GF_SAXParser *parser)
{
	while (parser->current_pos + 1 < parser->line_size) {
		if ((parser->buffer[parser->current_pos]=='?') && (parser->buffer[parser->current_pos+1]=='>')) {
			parser->sax_state = SAX_STATE_ELEMENT;
			parser->current_pos++;
			xml_sax_swap(parser);
			return;
		}
		parser->current_pos++;
	}
}


static void xml_sax_parse_entity(GF_SAXParser *parser)
{
	char szName[1024];
	u32 i = 0;
	XML_Entity *ent = (XML_Entity *)gf_list_last(parser->entities);
	char *skip_chars = " \t\n\r";
	i=0;
	if (ent && ent->value) ent = NULL;
	if (ent) skip_chars = NULL;

	while (parser->current_pos+i < parser->line_size) {
		u8 c = parser->buffer[parser->current_pos+i];
		if (skip_chars && strchr(skip_chars, c)) {
			if (c=='\n') parser->line++;
			parser->current_pos++;
			continue;
		}
		if (!ent && (c=='%')) {
			parser->current_pos+=i+1;
			parser->sax_state = SAX_STATE_SKIP_DOCTYPE;
			return;
		}
		else if (!ent && ((c=='\"') || (c=='\'')) ) {
			szName[i] = 0;
			GF_SAFEALLOC(ent, XML_Entity);
			ent->name = strdup(szName);
			ent->sep = c;
			parser->current_pos += 1+i;
			assert(parser->current_pos < parser->line_size);
			xml_sax_swap(parser);
			i=0;
			gf_list_add(parser->entities, ent);
			skip_chars = NULL;
		} else if (ent && c==ent->sep) {
			xml_sax_store_text(parser, i);

			ent->value = xml_get_current_text(parser);
			if (!ent->value) ent->value = strdup("");

			parser->current_pos += 1;
			assert(parser->current_pos < parser->line_size);
			xml_sax_swap(parser);
			parser->sax_state = SAX_STATE_SKIP_DOCTYPE;
			return;
		} else if (!ent) {
			szName[i] = c;
			i++;
		} else {
			i++;
		}
	}
	xml_sax_store_text(parser, i);
}

static void xml_sax_cdata(GF_SAXParser *parser)
{
	char *cd_end = strstr(parser->buffer + parser->current_pos, "]]>");
	if (!cd_end) {
		xml_sax_store_text(parser, parser->line_size - parser->current_pos);
	} else {
		u32 size = cd_end - (parser->buffer + parser->current_pos);
		xml_sax_store_text(parser, size);
		xml_sax_flush_text(parser);
		parser->current_pos += 3;
		assert(parser->current_pos < parser->line_size);
		parser->sax_state = SAX_STATE_TEXT_CONTENT;
	}
}

static Bool xml_sax_parse_comments(GF_SAXParser *parser)
{
	char *end = strstr(parser->buffer + parser->current_pos, "-->");
	if (!end) {
		if (parser->line_size>3)
			parser->current_pos = parser->line_size-3;
		xml_sax_swap(parser);
		return 0;
	}

	parser->current_pos += 3 + (u32) (end - (parser->buffer + parser->current_pos) );
	assert(parser->current_pos <= parser->line_size);
	parser->sax_state = SAX_STATE_TEXT_CONTENT;
	parser->text_start = parser->text_end = 0;
	xml_sax_swap(parser);
	return 1;
}



static GF_Err xml_sax_parse(GF_SAXParser *parser, Bool force_parse)
{
	u32 i = 0;
	Bool is_text, is_end;
	u8 c;
	char *elt, sep;

	is_text = 0;
	while (parser->current_pos<parser->line_size) {
		if (!force_parse && parser->suspended) goto exit;

restart:
		is_text = 0;
		switch (parser->sax_state) {
		/*load an XML element*/
		case SAX_STATE_TEXT_CONTENT:
			is_text = 1;
		case SAX_STATE_ELEMENT:
			elt = NULL;
			i=0;
			while ((c = parser->buffer[parser->current_pos+i]) !='<') {
				if ((parser->init_state==2) && (c ==']')) {
					parser->sax_state = SAX_STATE_ATT_NAME;
					parser->current_pos+=i+1;
					goto restart;
				}
				i++;
				if (!is_text && (c=='\n')) parser->line++;
				if (parser->current_pos+i==parser->line_size) goto exit;
			}
			if (is_text && i) {
				xml_sax_store_text(parser, i);
				is_text = 0;
				parser->sax_state = SAX_STATE_ELEMENT;
			} else if (i) {
				parser->current_pos += i;
				assert(parser->current_pos < parser->line_size);
			}
			is_end = 0;
			i = 0;
			while (1) {
				char c = parser->buffer[parser->current_pos+1+i];
				if (!c) {
					i = 0;
					goto exit;
				}
				if ((c=='\t') || (c=='\r') || (c==' ') ) {
					if (i) break;
					else parser->current_pos++;
				}
				else if (c=='\n') {
					parser->line++;
					if (i) break;
					else parser->current_pos++;
				}
				else if (c=='>') break;
				else if (c=='=') break;
				else if (c=='/') {
					is_end = !i ? 1 : 2;
					i++;
				} else {
					i++;
				}
//				if ((c=='[') && (parser->buffer[parser->elt_name_start-1 + i-2]=='A') ) break;
				if (parser->current_pos+1+i==parser->line_size) {
					i=0;
					goto exit;
				}
			}
			if (i) {
				parser->elt_name_start = parser->current_pos+1 + 1;
				if (is_end==1) parser->elt_name_start ++;
				if (is_end==2) parser->elt_name_end = parser->current_pos+1+i;
				else parser->elt_name_end = parser->current_pos+1+i + 1;
			} 
			if (is_end) {
				xml_sax_flush_text(parser);
				parser->elt_end_pos = parser->file_pos + parser->current_pos + i;
				if (is_end==2) {
					parser->sax_state = SAX_STATE_ELEMENT;
					xml_sax_node_start(parser);
					xml_sax_node_end(parser, 0);
				} else {
					parser->elt_end_pos += parser->elt_name_end - parser->elt_name_start;
					xml_sax_node_end(parser, 1);
				}
				if (parser->sax_state == SAX_STATE_SYNTAX_ERROR) break;
				parser->current_pos+=2+i;
				parser->sax_state = SAX_STATE_TEXT_CONTENT;
				break;
			}
			sep = parser->buffer[parser->elt_name_end-1];
			parser->buffer[parser->elt_name_end-1] = 0;
			elt = parser->buffer + parser->elt_name_start-1;

			parser->sax_state = SAX_STATE_ATT_NAME;
			assert(parser->elt_start_pos <= parser->file_pos + parser->current_pos);
			parser->elt_start_pos = parser->file_pos + parser->current_pos;

			if (!strncmp(elt, "!--", 3)) { 
				parser->sax_state = SAX_STATE_COMMENT;
				if (i>3) parser->current_pos -= (i-3);
			}
			else if (!strcmp(elt, "?xml")) parser->init_state = 1;
			else if (!strcmp(elt, "!DOCTYPE")) parser->init_state = 2;
			else if (!strcmp(elt, "!ENTITY")) parser->sax_state = SAX_STATE_ENTITY;
			else if (!strcmp(elt, "!ATTLIST") || !strcmp(elt, "!ELEMENT")) parser->sax_state = SAX_STATE_SKIP_DOCTYPE;
			else if (!strcmp(elt, "![CDATA[")) 
				parser->sax_state = SAX_STATE_CDATA;
			else if (elt[0]=='?') parser->sax_state = SAX_STATE_XML_PROC;
			/*node found*/
			else {
				xml_sax_flush_text(parser);
				if (parser->init_state) {
					parser->init_state = 0;
					/*that's a bit ugly: since we solve entities when appending text, we need to 
					reparse the current buffer*/
					if (gf_list_count(parser->entities)) {
						char *orig_buf;
						GF_Err e;
						parser->buffer[parser->elt_name_end-1] = sep;
						orig_buf = strdup(parser->buffer + parser->current_pos);
						parser->current_pos = 0;
						parser->line_size = 0;
						parser->elt_start_pos = 0;
						parser->sax_state = SAX_STATE_TEXT_CONTENT;
						e = gf_xml_sax_parse_intern(parser, orig_buf);
						free(orig_buf);
						return e;
					}
				}
			}
			parser->current_pos+=1+i;
			parser->buffer[parser->elt_name_end-1] = sep;
			break;
		case SAX_STATE_COMMENT:
			if (!xml_sax_parse_comments(parser)) {
				xml_sax_swap(parser);
				return GF_OK;
			}
			break;
		case SAX_STATE_ATT_NAME:
		case SAX_STATE_ATT_VALUE:
			if (xml_sax_parse_attribute(parser)) 
				goto exit;
			break;
		case SAX_STATE_ENTITY:
			xml_sax_parse_entity(parser);
			break;
		case SAX_STATE_SKIP_DOCTYPE:
			xml_sax_skip_doctype(parser);
			break;
		case SAX_STATE_XML_PROC:
			xml_sax_skip_xml_proc(parser);
			break;
		case SAX_STATE_CDATA:
			xml_sax_cdata(parser);
			break;
		case SAX_STATE_SYNTAX_ERROR:
			return GF_CORRUPTED_DATA;
		case SAX_STATE_DONE:
			return GF_EOS;
		}
	}
exit:
#if 0
	if (is_text) {
		if (i) xml_sax_store_text(parser, i);
		/*DON'T FLUSH TEXT YET, wait for next '<' to do so otherwise we may corrupt xml base entities (&apos;, ...)*/
	}
#endif
	xml_sax_swap(parser);
	return GF_OK;
}

static GF_Err xml_sax_append_string(GF_SAXParser *parser, char *string)
{
	u32 size = parser->line_size;
	u32 nl_size = strlen(string);
	
	if (parser->alloc_size < size+nl_size+1) {
		parser->buffer = realloc(parser->buffer, sizeof(char) * (size+nl_size+1) );
		if (!parser->buffer ) return GF_OUT_OF_MEM;
		parser->alloc_size = size+nl_size+1;
	}
	memcpy(parser->buffer+size, string, sizeof(char)*nl_size);
	parser->buffer[size+nl_size] = 0;
	parser->line_size = size+nl_size;
	return GF_OK;
}

static GF_Err gf_xml_sax_parse_intern(GF_SAXParser *parser, char *current)
{
	u32 i, count;
	/*solve entities*/
	count = gf_list_count(parser->entities);
	while (count) {
		char *entityEnd, szName[200];
		XML_Entity *ent;
		char *entityStart = strstr(current, "&");

		if (parser->in_entity) {
			entityEnd = strstr(current, ";");
			if (!entityEnd) return xml_sax_append_string(parser, current);
			entityStart = strrchr(parser->buffer, '&');

			strcpy(szName, entityStart+1);
			entityStart[0] = 0;
			entityEnd[0] = 0;
			strcat(szName, current);
			entityEnd[0] = ';';
			parser->in_entity = 0;
			current = entityEnd+1;
		} else {
			if (!entityStart) break;
			entityEnd = strstr(entityStart, ";");

			entityStart[0] = 0;
			xml_sax_append_string(parser, current);
			xml_sax_parse(parser, 1);
			entityStart[0] = '&';

			if (!entityEnd) {
				parser->in_entity = 1;
				/*store entity start*/
				return xml_sax_append_string(parser, entityStart);
			}
			strncpy(szName, entityStart+1, entityEnd - entityStart - 1);
			szName[entityEnd - entityStart - 1] = 0;
			current = entityEnd + 1;
		}

		for (i=0; i<count; i++) {
			ent = (XML_Entity *)gf_list_get(parser->entities, i);
			if (!strcmp(ent->name, szName)) {
				u32 line_num = parser->line;
				xml_sax_append_string(parser, ent->value);
				xml_sax_parse(parser, 1);
				parser->line = line_num;
				break;
			}
		}
	}
	xml_sax_append_string(parser, current);
	return xml_sax_parse(parser, 0);
}

GF_EXPORT
GF_Err gf_xml_sax_parse(GF_SAXParser *parser, void *string)
{
	GF_Err e;
	char *current;
	char *utf_conv = NULL;
	
	if (parser->unicode_type < 0) return GF_BAD_PARAM;

	if (parser->unicode_type>1) {
		const u16 *sptr = (const u16 *)string;
		u32 len = 2*gf_utf8_wcslen(sptr);
		utf_conv = (char *)malloc(sizeof(char)*(len+1));
		len = gf_utf8_wcstombs(utf_conv, len, &sptr);
		if (len==(u32) -1) {
			parser->sax_state = SAX_STATE_SYNTAX_ERROR;
			free(utf_conv);
			return GF_CORRUPTED_DATA;
		}
		utf_conv[len] = 0;
		current = utf_conv;
	} else {
		current = (char *)string;
	}

	e = gf_xml_sax_parse_intern(parser, current);
	if (utf_conv) free(utf_conv);
	return e;
}


GF_EXPORT
GF_Err gf_xml_sax_init(GF_SAXParser *parser, unsigned char *BOM)
{
	u32 offset;
	if (!BOM) parser->unicode_type = 0;
	if (parser->unicode_type >= 0) return gf_xml_sax_parse(parser, BOM);