diff options
author | Daniel Silverstone <dsilvers@digital-scurf.org> | 2012-07-15 18:28:06 +0100 |
---|---|---|
committer | Daniel Silverstone <dsilvers@digital-scurf.org> | 2012-07-15 18:28:06 +0100 |
commit | 57bee28b15d3f8b878e3ebfe8aaa7aeb0fd7addf (patch) | |
tree | 56d43f0cfe4f7911293a77e453989d543ad25e79 | |
parent | cb0225a6d309cd951b595eb5c1b488f21a0f17be (diff) | |
parent | 2f4be485272446109e4451a19dc29ba6728f0039 (diff) | |
download | libdom-57bee28b15d3f8b878e3ebfe8aaa7aeb0fd7addf.tar.gz libdom-57bee28b15d3f8b878e3ebfe8aaa7aeb0fd7addf.tar.bz2 |
Merge expat support. Test suite now passes with expat.
-rw-r--r-- | bindings/xml/Makefile | 2 | ||||
-rw-r--r-- | bindings/xml/expat_xmlparser.c | 312 |
2 files changed, 259 insertions, 55 deletions
diff --git a/bindings/xml/Makefile b/bindings/xml/Makefile index e75e362..ea7653c 100644 --- a/bindings/xml/Makefile +++ b/bindings/xml/Makefile @@ -26,7 +26,7 @@ ifeq ($(WITH_EXPAT_BINDING),yes) endif ifeq ($(DO_XML_INSTALL),yes) - DIR_INSTALL_ITEMS := /include/dom/bindings/libxml:xmlerror.h;xmlparser.h + DIR_INSTALL_ITEMS := /include/dom/bindings/xml:xmlerror.h;xmlparser.h endif include $(NSBUILD)/Makefile.subdir diff --git a/bindings/xml/expat_xmlparser.c b/bindings/xml/expat_xmlparser.c index ce649d7..1077bc9 100644 --- a/bindings/xml/expat_xmlparser.c +++ b/bindings/xml/expat_xmlparser.c @@ -29,6 +29,7 @@ struct dom_xml_parser { bool complete; /**< Indicate stream completion */ struct dom_document *doc; /**< DOM Document we're building */ struct dom_node *current; /**< DOM node we're currently building */ + bool is_cdata; /**< If the character data is cdata or text */ }; /* Binding functions */ @@ -42,9 +43,7 @@ expat_xmlparser_start_element_handler(void *_parser, dom_exception err; dom_element *elem, *ins_elem; dom_string *tag_name; - - fprintf(stderr, "<%s>\n", name); - + err = dom_string_create((const uint8_t *)name, strlen(name), &tag_name); @@ -53,7 +52,7 @@ expat_xmlparser_start_element_handler(void *_parser, "No memory for tag name"); return; } - + err = dom_document_create_element(parser->doc, tag_name, &elem); if (err != DOM_NO_ERR) { dom_string_unref(tag_name); @@ -61,15 +60,15 @@ expat_xmlparser_start_element_handler(void *_parser, "Failed to create element '%s'", name); return; } - + dom_string_unref(tag_name); - + /* Add attributes to the element */ - + while (*atts) { dom_string *key, *value; - err = dom_string_create((const uint8_t *)(*atts), + err = dom_string_create((const uint8_t *)(*atts), strlen(*atts), &key); if (err != DOM_NO_ERR) { parser->msg(DOM_MSG_CRITICAL, parser->mctx, @@ -88,7 +87,7 @@ expat_xmlparser_start_element_handler(void *_parser, return; } atts++; - + err = dom_element_set_attribute(elem, key, value); dom_string_unref(key); dom_string_unref(value); @@ -99,7 +98,7 @@ expat_xmlparser_start_element_handler(void *_parser, return; } } - + err = dom_node_append_child(parser->current, elem, &ins_elem); if (err != DOM_NO_ERR) { dom_node_unref(elem); @@ -107,9 +106,9 @@ expat_xmlparser_start_element_handler(void *_parser, "No memory for appending child node"); return; } - + dom_node_unref(ins_elem); - + dom_node_unref(parser->current); parser->current = (struct dom_node *)elem; /* Steal initial ref */ } @@ -121,24 +120,38 @@ expat_xmlparser_end_element_handler(void *_parser, dom_xml_parser *parser = _parser; dom_exception err; dom_node *parent; - + UNUSED(name); - - fprintf(stderr, "</%s>\n", name); - + err = dom_node_get_parent_node(parser->current, &parent); - + if (err != DOM_NO_ERR) { parser->msg(DOM_MSG_CRITICAL, parser->mctx, "Unable to find a parent while closing element."); return; } - + dom_node_unref(parser->current); parser->current = parent; /* Takes the ref given by get_parent_node */ } static void +expat_xmlparser_start_cdata_handler(void *_parser) +{ + dom_xml_parser *parser = _parser; + + parser->is_cdata = true; +} + +static void +expat_xmlparser_end_cdata_handler(void *_parser) +{ + dom_xml_parser *parser = _parser; + + parser->is_cdata = false; +} + +static void expat_xmlparser_cdata_handler(void *_parser, const XML_Char *s, int len) @@ -146,29 +159,66 @@ expat_xmlparser_cdata_handler(void *_parser, dom_xml_parser *parser = _parser; dom_string *data; dom_exception err; - struct dom_cdata_section *cdata, *ins_cdata; - + struct dom_node *cdata, *ins_cdata, *lastchild = NULL; + dom_node_type ntype = 0; + err = dom_string_create((const uint8_t *)s, len, &data); if (err != DOM_NO_ERR) { parser->msg(DOM_MSG_CRITICAL, parser->mctx, "No memory for cdata section contents"); return; } - - err = dom_document_create_cdata_section(parser->doc, data, &cdata); + + err = dom_node_get_last_child(parser->current, &lastchild); + + if (err == DOM_NO_ERR && lastchild != NULL) { + err = dom_node_get_node_type(lastchild, &ntype); + } + if (err != DOM_NO_ERR) { dom_string_unref(data); + if (lastchild != NULL) + dom_node_unref(lastchild); parser->msg(DOM_MSG_CRITICAL, parser->mctx, "No memory for cdata section"); return; } - + + if (ntype == DOM_TEXT_NODE && parser->is_cdata == false) { + /* We can append this text instead */ + err = dom_characterdata_append_data( + (dom_characterdata *)lastchild, data); + dom_string_unref(data); + if (lastchild != NULL) + dom_node_unref(lastchild); + if (err != DOM_NO_ERR) { + parser->msg(DOM_MSG_CRITICAL, parser->mctx, + "No memory for cdata section"); + } + return; + } + + if (lastchild != NULL) + dom_node_unref(lastchild); + + /* We can't append directly, so make a new node */ + err = parser->is_cdata ? + dom_document_create_cdata_section(parser->doc, data, + (dom_cdata_section **)&cdata) : + dom_document_create_text_node(parser->doc, data, + (dom_text **)&cdata); + if (err != DOM_NO_ERR) { + dom_string_unref(data); + parser->msg(DOM_MSG_CRITICAL, parser->mctx, + "No memory for cdata section"); + return; + } + /* No longer need data */ dom_string_unref(data); /* Append cdata section to parent */ - err = dom_node_append_child(parser->current, (struct dom_node *) cdata, - (struct dom_node **) (void *) &ins_cdata); + err = dom_node_append_child(parser->current, cdata, &ins_cdata); if (err != DOM_NO_ERR) { dom_node_unref((struct dom_node *) cdata); parser->msg(DOM_MSG_ERROR, parser->mctx, @@ -178,10 +228,149 @@ expat_xmlparser_cdata_handler(void *_parser, /* We're not interested in the inserted cdata section */ if (ins_cdata != NULL) - dom_node_unref((struct dom_node *) ins_cdata); + dom_node_unref(ins_cdata); /* No longer interested in cdata section */ - dom_node_unref((struct dom_node *) cdata); + dom_node_unref(cdata); +} + +static int +expat_xmlparser_external_entity_ref_handler(XML_Parser parser, + const XML_Char *context, + const XML_Char *base, + const XML_Char *system_id, + const XML_Char *public_id) +{ + FILE *fh; + XML_Parser subparser; + unsigned char data[1024]; + size_t len; + enum XML_Status status; + + UNUSED(base); + UNUSED(public_id); + + if (system_id == NULL) + return XML_STATUS_OK; + + fh = fopen(system_id, "r"); + + if (fh == NULL) + return XML_STATUS_OK; + + subparser = XML_ExternalEntityParserCreate(parser, + context, + NULL); + + if (subparser == NULL) { + fclose(fh); + return XML_STATUS_OK; + } + + /* Parse the file bit by bit */ + while ((len = fread(data, 1, 1024, fh)) > 0) { + status = XML_Parse(subparser, (const char *)data, len, 0); + if (status != XML_STATUS_OK) { + XML_ParserFree(subparser); + fclose(fh); + return XML_STATUS_OK; + } + } + XML_Parse(subparser, "", 0, 1); + XML_ParserFree(subparser); + fclose(fh); + return XML_STATUS_OK; +} + +static void +expat_xmlparser_comment_handler(void *_parser, + const XML_Char *_comment) +{ + dom_xml_parser *parser = _parser; + struct dom_comment *comment, *ins_comment = NULL; + dom_string *data; + dom_exception err; + + /* Create DOM string data for comment */ + err = dom_string_create((const uint8_t *)_comment, + strlen((const char *) _comment), &data); + if (err != DOM_NO_ERR) { + parser->msg(DOM_MSG_CRITICAL, parser->mctx, + "No memory for comment data"); + return; + } + + /* Create comment */ + err = dom_document_create_comment(parser->doc, data, &comment); + if (err != DOM_NO_ERR) { + dom_string_unref(data); + parser->msg(DOM_MSG_CRITICAL, parser->mctx, + "No memory for comment node"); + return; + } + + /* No longer need data */ + dom_string_unref(data); + + /* Append comment to parent */ + err = dom_node_append_child(parser->current, (struct dom_node *) comment, + (struct dom_node **) (void *) &ins_comment); + if (err != DOM_NO_ERR) { + dom_node_unref((struct dom_node *) comment); + parser->msg(DOM_MSG_CRITICAL, parser->mctx, + "Failed attaching comment node"); + return; + } + + /* We're not interested in the inserted comment */ + if (ins_comment != NULL) + dom_node_unref((struct dom_node *) ins_comment); + + /* No longer interested in comment */ + dom_node_unref((struct dom_node *) comment); + +} + +static void +expat_xmlparser_start_doctype_decl_handler(void *_parser, + const XML_Char *doctype_name, + const XML_Char *system_id, + const XML_Char *public_id, + int has_internal_subset) +{ + dom_xml_parser *parser = _parser; + struct dom_document_type *doctype, *ins_doctype = NULL; + dom_exception err; + + UNUSED(has_internal_subset); + + err = dom_implementation_create_document_type( + doctype_name, system_id ? system_id : "", + public_id ? public_id : "", + &doctype); + + if (err != DOM_NO_ERR) { + parser->msg(DOM_MSG_CRITICAL, parser->mctx, + "Failed to create document type"); + return; + } + + /* Add doctype to document */ + err = dom_node_append_child(parser->doc, (struct dom_node *) doctype, + (struct dom_node **) (void *) &ins_doctype); + if (err != DOM_NO_ERR) { + dom_node_unref((struct dom_node *) doctype); + parser->msg(DOM_MSG_CRITICAL, parser->mctx, + "Failed attaching doctype"); + return; + } + + /* Not interested in inserted node */ + if (ins_doctype != NULL) + dom_node_unref((struct dom_node *) ins_doctype); + + /* No longer interested in doctype */ + dom_node_unref((struct dom_node *) doctype); } static void @@ -190,8 +379,8 @@ expat_xmlparser_unknown_data_handler(void *_parser, int len) { UNUSED(_parser); - - fprintf(stderr, "!!! %.*s !!!\n", len, s); + UNUSED(s); + UNUSED(len); } /** * Create an XML parser instance @@ -210,29 +399,29 @@ dom_xml_parser_create(const char *enc, const char *int_enc, { dom_xml_parser *parser; dom_exception err; - + UNUSED(int_enc); - + parser = calloc(sizeof(*parser), 1); if (parser == NULL) { msg(DOM_MSG_CRITICAL, mctx, "No memory for parser"); return NULL; } - + parser->msg = msg; parser->mctx = mctx; - + parser->parser = XML_ParserCreateNS(enc, ':'); - + if (parser->parser == NULL) { free(parser); msg(DOM_MSG_CRITICAL, mctx, "No memory for parser"); return NULL; } - + parser->complete = false; parser->doc = NULL; - + err = dom_implementation_create_document( DOM_IMPLEMENTATION_XML, /* namespace */ NULL, @@ -240,32 +429,47 @@ dom_xml_parser_create(const char *enc, const char *int_enc, /* doctype */ NULL, NULL, &parser->doc); - + if (err != DOM_NO_ERR) { - parser->msg(DOM_MSG_CRITICAL, parser->mctx, + parser->msg(DOM_MSG_CRITICAL, parser->mctx, "Failed creating document"); XML_ParserFree(parser->parser); free(parser); return NULL; } - + XML_SetUserData(parser->parser, parser); - + XML_SetElementHandler(parser->parser, expat_xmlparser_start_element_handler, expat_xmlparser_end_element_handler); - + + XML_SetCdataSectionHandler(parser->parser, + expat_xmlparser_start_cdata_handler, + expat_xmlparser_end_cdata_handler); + XML_SetCharacterDataHandler(parser->parser, expat_xmlparser_cdata_handler); - - XML_SetParamEntityParsing(parser->parser, + + XML_SetParamEntityParsing(parser->parser, XML_PARAM_ENTITY_PARSING_ALWAYS); - - XML_SetDefaultHandler(parser->parser, + + XML_SetExternalEntityRefHandler(parser->parser, + expat_xmlparser_external_entity_ref_handler); + + XML_SetCommentHandler(parser->parser, + expat_xmlparser_comment_handler); + + XML_SetStartDoctypeDeclHandler(parser->parser, + expat_xmlparser_start_doctype_decl_handler); + + XML_SetDefaultHandlerExpand(parser->parser, expat_xmlparser_unknown_data_handler); - + parser->current = dom_node_ref(parser->doc); - + + parser->is_cdata = false; + return parser; } @@ -278,7 +482,7 @@ void dom_xml_parser_destroy(dom_xml_parser *parser) { XML_ParserFree(parser->parser); - + free(parser); } @@ -294,14 +498,14 @@ dom_xml_error dom_xml_parser_parse_chunk(dom_xml_parser *parser, uint8_t *data, size_t len) { enum XML_Status status; - + status = XML_Parse(parser->parser, (const char *)data, len, 0); if (status != XML_STATUS_OK) { parser->msg(DOM_MSG_ERROR, parser->mctx, "XML_Parse failed: %d", status); return DOM_XML_EXTERNAL_ERR | status; } - + return DOM_XML_OK; } @@ -313,20 +517,20 @@ dom_xml_parser_parse_chunk(dom_xml_parser *parser, uint8_t *data, size_t len) * * This will force any remaining data through the parser */ -dom_xml_error +dom_xml_error dom_xml_parser_completed(dom_xml_parser *parser) { enum XML_Status status; - + status = XML_Parse(parser->parser, "", 0, 1); if (status != XML_STATUS_OK) { parser->msg(DOM_MSG_ERROR, parser->mctx, "XML_Parse failed: %d", status); return DOM_XML_EXTERNAL_ERR | status; } - + parser->complete = true; - + return DOM_XML_OK; } |