/* * Copyright 2007 James Bursa * * This file is part of NetSurf, http://www.netsurf-browser.org/ * * NetSurf is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; version 2 of the License. * * NetSurf is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program. If not, see . */ /** \file * Content for text/html (implementation). */ #include #include #include #include #include #include #include #include "utils/config.h" #include "content/content.h" #include "content/fetch.h" #include "content/fetchcache.h" #include "desktop/browser.h" #include "desktop/gui.h" #include "desktop/options.h" #include "render/box.h" #include "render/font.h" #include "render/html.h" #include "render/imagemap.h" #include "render/layout.h" #include "utils/log.h" #include "utils/messages.h" #include "utils/talloc.h" #include "utils/url.h" #include "utils/utils.h" #define CHUNK 4096 static bool html_set_parser_encoding(struct content *c, const char *encoding); static const char *html_detect_encoding(const char **data, unsigned int *size); static void html_convert_css_callback(content_msg msg, struct content *css, intptr_t p1, intptr_t p2, union content_msg_data data); static bool html_meta_refresh(struct content *c, xmlNode *head); static bool html_head(struct content *c, xmlNode *head); static bool html_find_stylesheets(struct content *c, xmlNode *html, xmlNode *head); static bool html_find_inline_stylesheets(struct content *c, xmlNode *html); static bool html_process_style_element(struct content *c, xmlNode *style); static void html_object_callback(content_msg msg, struct content *object, intptr_t p1, intptr_t p2, union content_msg_data data); static void html_object_done(struct box *box, struct content *object, bool background); static void html_object_failed(struct box *box, struct content *content, bool background); static bool html_object_type_permitted(const content_type type, const content_type *permitted_types); static void html_object_refresh(void *p); static void html_destroy_frameset(struct content_html_frames *frameset); static void html_destroy_iframe(struct content_html_iframe *iframe); static void html_set_status(struct content *c, const char *extra); static void html_dump_frameset(struct content_html_frames *frame, unsigned int depth); static const char empty_document[] = "" "" "" "Empty document" "" "" "

Empty document

" "

The document sent by the server is empty.

" "" ""; /** * Create a CONTENT_HTML. * * The content_html_data structure is initialized and the HTML parser is * created. */ bool html_create(struct content *c, const char *params[]) { unsigned int i; struct content_html_data *html = &c->data.html; union content_msg_data msg_data; html->parser = 0; html->encoding_handler = 0; html->encoding = 0; html->getenc = true; html->base_url = c->url; html->base_target = NULL; html->layout = 0; html->background_colour = TRANSPARENT; html->stylesheet_count = 0; html->stylesheet_content = 0; html->style = 0; html->working_stylesheet = 0; html->object_count = 0; html->object = 0; html->forms = 0; html->imagemaps = 0; html->bw = 0; html->frameset = 0; html->iframe = 0; html->page = 0; html->index = 0; html->box = 0; for (i = 0; params[i]; i += 2) { if (strcasecmp(params[i], "charset") == 0) { html->encoding = talloc_strdup(c, params[i + 1]); if (!html->encoding) goto no_memory; html->encoding_source = ENCODING_SOURCE_HEADER; html->getenc = false; break; } } html->parser = htmlCreatePushParserCtxt(0, 0, "", 0, 0, XML_CHAR_ENCODING_NONE); if (!html->parser) goto no_memory; if (html->encoding) { /* an encoding was specified in the Content-Type header */ if (!html_set_parser_encoding(c, html->encoding)) return false; } return true; no_memory: msg_data.error = messages_get("NoMemory"); content_broadcast(c, CONTENT_MSG_ERROR, msg_data); return false; } /** * Process data for CONTENT_HTML. * * The data is parsed in chunks of size CHUNK, multitasking in between. */ bool html_process_data(struct content *c, char *data, unsigned int size) { unsigned long x; if (c->data.html.getenc) { /* No encoding was specified in the Content-Type header. * Attempt to detect if the encoding is not 8-bit. If the * encoding is 8-bit, leave the parser unchanged, so that it * searches for a . */ const char *encoding; encoding = html_detect_encoding((const char **) &data, &size); if (encoding) { if (!html_set_parser_encoding(c, encoding)) return false; c->data.html.encoding = talloc_strdup(c, encoding); if (!c->data.html.encoding) return false; c->data.html.encoding_source = ENCODING_SOURCE_DETECTED; } c->data.html.getenc = false; /* The data we received may have solely consisted of a BOM. * If so, it will have been stripped by html_detect_encoding. * Therefore, we'll have nothing to do in that case. */ if (size == 0) return true; } for (x = 0; x + CHUNK <= size; x += CHUNK) { htmlParseChunk(c->data.html.parser, data + x, CHUNK, 0); gui_multitask(); } htmlParseChunk(c->data.html.parser, data + x, (int) (size - x), 0); if (!c->data.html.encoding && c->data.html.parser->input->encoding) { /* The encoding was not in headers or detected, * and the parser found a . */ /* However, if that encoding is non-ASCII-compatible, * ignore it, as it can't possibly be correct */ if (strncasecmp((const char *) c->data.html.parser-> input->encoding, "UTF-16", 6) == 0 || /* UTF-16(LE|BE)? */ strncasecmp((const char *) c->data.html.parser-> input->encoding, "UTF-32", 6) == 0) { /* UTF-32(LE|BE)? */ c->data.html.encoding = talloc_strdup(c, "ISO-8859-1"); c->data.html.encoding_source = ENCODING_SOURCE_DETECTED; } else { c->data.html.encoding = talloc_strdup(c, (const char *) c->data.html.parser-> input->encoding); c->data.html.encoding_source = ENCODING_SOURCE_META; } if (!c->data.html.encoding) { union content_msg_data msg_data; msg_data.error = messages_get("NoMemory"); content_broadcast(c, CONTENT_MSG_ERROR, msg_data); return false; } /* have the encoding; don't attempt to detect it */ c->data.html.getenc = false; /* now, we must reset the parser such that it reparses * using the correct charset, and then reparse any document * source we've got. we achieve this by recreating the * parser in its entirety as this is simpler than resetting * the existing one and ensuring it's still set up correctly. */ if (c->data.html.parser->myDoc) xmlFreeDoc(c->data.html.parser->myDoc); htmlFreeParserCtxt(c->data.html.parser); c->data.html.parser = htmlCreatePushParserCtxt(0, 0, "", 0, 0, XML_CHAR_ENCODING_NONE); if (!c->data.html.parser) { union content_msg_data msg_data; msg_data.error = messages_get("NoMemory"); content_broadcast(c, CONTENT_MSG_ERROR, msg_data); return false; } if (!html_set_parser_encoding(c, c->data.html.encoding)) return false; /* and reparse received document source - the recursion * is safe as we've just set c->data.html.encoding so * we'll never get back in here. */ if (!html_process_data(c, c->source_data, c->source_size)) return false; } return true; } /** * Set the HTML parser character encoding. * * \param c content of type CONTENT_HTML * \param encoding name of encoding * \return true on success, false on error and error reported */ bool html_set_parser_encoding(struct content *c, const char *encoding) { struct content_html_data *html = &c->data.html; xmlError *error; char error_message[500]; union content_msg_data msg_data; html->encoding_handler = xmlFindCharEncodingHandler(encoding); if (!html->encoding_handler) { /* either out of memory, or no handler available */ /* assume no handler available, which is not a fatal error */ LOG(("no encoding handler for \"%s\"", encoding)); /* \todo warn user and ask them to install iconv? */ return true; } xmlCtxtResetLastError(html->parser); if (xmlSwitchToEncoding(html->parser, html->encoding_handler)) { error = xmlCtxtGetLastError(html->parser); snprintf(error_message, sizeof error_message, "%s xmlSwitchToEncoding(): %s", messages_get("MiscError"), error ? error->message : "failed"); msg_data.error = error_message; content_broadcast(c, CONTENT_MSG_ERROR, msg_data); return false; } /* Dirty hack to get around libxml oddness: * 1) When creating a push parser context, the input flow's encoding * string is not set (whether an encoding is specified or not) * 2) When switching encoding (as above), the input flow's encoding * string is never changed * 3) When handling a meta charset, the input flow's encoding string * is checked to determine if an encoding has already been set. * If it has been set, then the meta charset is ignored. * * The upshot of this is that, if we don't explicitly set the input * flow's encoding string here, any meta charset in the document * will override our setting, which is incorrect behaviour. * * Ideally, this would be fixed in libxml, but that requires rather * more knowledge than I currently have of what libxml is doing. */ if (!html->parser->input->encoding) html->parser->input->encoding = xmlStrdup((const xmlChar *) encoding); /* Ensure noone else attempts to reset the encoding */ html->getenc = false; return true; } /** * Attempt to detect the encoding of some HTML data. * * \param data Pointer to HTML source data * \param size Pointer to length of data * \return a constant string giving the encoding, or 0 if the encoding * appears to be some 8-bit encoding * * If a BOM is encountered, *data and *size will be modified to skip over it */ const char *html_detect_encoding(const char **data, unsigned int *size) { const unsigned char *d = (const unsigned char *) *data; /* this detection assumes that the first two characters are <= 0xff */ if (*size < 4) return 0; if (d[0] == 0x00 && d[1] == 0x00 && d[2] == 0xfe && d[3] == 0xff) { /* BOM 00 00 fe ff */ *data += 4; *size -= 4; return "UTF-32BE"; } else if (d[0] == 0xff && d[1] == 0xfe && d[2] == 0x00 && d[3] == 0x00) { /* BOM ff fe 00 00 */ *data += 4; *size -= 4; return "UTF-32LE"; } else if (d[0] == 0x00 && d[1] != 0x00 && d[2] == 0x00 && d[3] != 0x00) /* 00 xx 00 xx */ return "UTF-16BE"; else if (d[0] != 0x00 && d[1] == 0x00 && d[2] != 0x00 && d[3] == 0x00) /* xx 00 xx 00 */ return "UTF-16LE"; else if (d[0] == 0x00 && d[1] == 0x00 && d[2] == 0x00 && d[3] != 0x00) /* 00 00 00 xx */ return "ISO-10646-UCS-4"; else if (d[0] != 0x00 && d[1] == 0x00 && d[2] == 0x00 && d[3] == 0x00) /* xx 00 00 00 */ return "ISO-10646-UCS-4"; else if (d[0] == 0xfe && d[1] == 0xff) { /* BOM fe ff */ *data += 2; *size -= 2; return "UTF-16BE"; } else if (d[0] == 0xff && d[1] == 0xfe) { /* BOM ff fe */ *data += 2; *size -= 2; return "UTF-16LE"; } else if (d[0] == 0xef && d[1] == 0xbb && d[2] == 0xbf) { /* BOM ef bb bf */ *data += 3; *size -= 3; return "UTF-8"; } return 0; } /** * Convert a CONTENT_HTML for display. * * The following steps are carried out in order: * * - parsing to an XML tree is completed * - stylesheets are fetched * - the XML tree is converted to a box tree and object fetches are started * - the box tree is laid out * * On exit, the content status will be either CONTENT_STATUS_DONE if the * document is completely loaded or CONTENT_STATUS_READY if objects are still * being fetched. */ bool html_convert(struct content *c, int width, int height) { xmlDoc *document; xmlNode *html, *head; union content_msg_data msg_data; unsigned int time_before, time_taken; /* finish parsing */ if (c->source_size == 0) htmlParseChunk(c->data.html.parser, empty_document, sizeof empty_document, 0); htmlParseChunk(c->data.html.parser, "", 0, 1); document = c->data.html.parser->myDoc; /*xmlDebugDumpDocument(stderr, c->data.html.parser->myDoc);*/ htmlFreeParserCtxt(c->data.html.parser); c->data.html.parser = 0; if (!document) { LOG(("Parsing failed")); msg_data.error = messages_get("ParsingFail"); content_broadcast(c, CONTENT_MSG_ERROR, msg_data); return false; } /* locate html and head elements */ html = xmlDocGetRootElement(document); if (html == 0 || strcmp((const char *) html->name, "html") != 0) { LOG(("html element not found")); xmlFreeDoc(document); msg_data.error = messages_get("ParsingFail"); content_broadcast(c, CONTENT_MSG_ERROR, msg_data); return false; } for (head = html->children; head != 0 && head->type != XML_ELEMENT_NODE; head = head->next) ; if (head && strcmp((const char *) head->name, "head") != 0) { head = 0; LOG(("head element not found")); } if (head) { if (!html_head(c, head)) { msg_data.error = messages_get("NoMemory"); content_broadcast(c, CONTENT_MSG_ERROR, msg_data); return false; } /* handle meta refresh */ if (!html_meta_refresh(c, head)) return false; } /* get stylesheets */ if (!html_find_stylesheets(c, html, head)) return false; /* convert xml tree to box tree */ LOG(("XML to box")); content_set_status(c, messages_get("Processing")); content_broadcast(c, CONTENT_MSG_STATUS, msg_data); if (!xml_to_box(html, c)) { msg_data.error = messages_get("NoMemory"); content_broadcast(c, CONTENT_MSG_ERROR, msg_data); return false; } /*box_dump(c->data.html.layout->children, 0);*/ /*if (c->data.html.frameset) html_dump_frameset(c->data.html.frameset, 0);*/ /* extract image maps - can't do this sensibly in xml_to_box */ if (!imagemap_extract(html, c)) { LOG(("imagemap extraction failed")); msg_data.error = messages_get("NoMemory"); content_broadcast(c, CONTENT_MSG_ERROR, msg_data); return false; } /*imagemap_dump(c);*/ /* XML tree not required past this point */ xmlFreeDoc(document); /* layout the box tree */ html_set_status(c, messages_get("Formatting")); content_broadcast(c, CONTENT_MSG_STATUS, msg_data); LOG(("Layout document")); time_before = wallclock(); html_reformat(c, width, height); time_taken = wallclock() - time_before; LOG(("Layout took %dcs", time_taken)); c->reformat_time = wallclock() + ((time_taken < option_min_reflow_period ? option_min_reflow_period : time_taken * 1.25)); LOG(("Scheduling relayout no sooner than %dcs", c->reformat_time - wallclock())); /*box_dump(c->data.html.layout->children, 0);*/ if (c->active == 0) c->status = CONTENT_STATUS_DONE; else c->status = CONTENT_STATUS_READY; html_set_status(c, ""); return true; } /** * Process elements in . * * \param c content structure * \param head xml node of head element * \return true on success, false on memory exhaustion * * The title and base href are extracted if present. */ bool html_head(struct content *c, xmlNode *head) { xmlNode *node; xmlChar *s; c->title = 0; for (node = head->children; node != 0; node = node->next) { if (node->type != XML_ELEMENT_NODE) continue; LOG(("Node: %s", node->name)); if (!c->title && strcmp((const char *) node->name, "title") == 0) { xmlChar *title = xmlNodeGetContent(node); if (!title) return false; char *title2 = squash_whitespace((const char *) title); xmlFree(title); if (!title2) return false; c->title = talloc_strdup(c, title2); free(title2); if (!c->title) return false; } else if (strcmp((const char *) node->name, "base") == 0) { char *href = (char *) xmlGetProp(node, (const xmlChar *) "href"); if (href) { char *url; url_func_result res; res = url_normalize(href, &url); if (res == URL_FUNC_OK) { c->data.html.base_url = talloc_strdup(c, url); free(url); } xmlFree(href); } /* don't use the central values to ease freeing later on */ if ((s = xmlGetProp(node, (const xmlChar *) "target"))) { if ((!strcasecmp((const char *) s, "_blank")) || (!strcasecmp((const char *) s, "_top")) || (!strcasecmp((const char *) s, "_parent")) || (!strcasecmp((const char *) s, "_self")) || ('a' <= s[0] && s[0] <= 'z') || ('A' <= s[0] && s[0] <= 'Z')) { /* [6.16] */ c->data.html.base_target = talloc_strdup(c, (const char *) s); if (!c->data.html.base_target) { xmlFree(s); return false; } } xmlFree(s); } } } return true; } /** * Search for meta refresh * * http://wp.netscape.com/assist/net_sites/pushpull.html * * \param c content structure * \param head xml node of head element * \return true on success, false otherwise (error reported) */ bool html_meta_refresh(struct content *c, xmlNode *head) { xmlNode *n; xmlChar *equiv, *content; union content_msg_data msg_data; char *url, *end, *refresh; url_func_result res; for (n = head == 0 ? 0 : head->children; n; n = n->next) { if (n->type != XML_ELEMENT_NODE) continue; /* Recurse into noscript elements */ if (strcmp((const char *) n->name, "noscript") == 0) { if (!html_meta_refresh(c, n)) { /* Some error occurred */ return false; } else if (c->refresh) { /* Meta refresh found - stop */ return true; } } if (strcmp((const char *) n->name, "meta")) { continue; } equiv = xmlGetProp(n, (const xmlChar *) "http-equiv"); if (!equiv) continue; if (strcasecmp((const char *) equiv, "refresh")) { xmlFree(equiv); continue; } xmlFree(equiv); content = xmlGetProp(n, (const xmlChar *) "content"); if (!content) continue; end = (char *) content + strlen((const char *) content); msg_data.delay = (int)strtol((char *) content, &url, 10); /* a very small delay and self-referencing URL can cause a loop * that grinds machines to a halt. To prevent this we set a * minimum refresh delay of 1s. */ if (msg_data.delay < 1) msg_data.delay = 1; if (url == end) { /* Just delay specified, so refresh current page */ xmlFree(content); c->refresh = talloc_strdup(c, c->url); if (!c->refresh) { msg_data.error = messages_get("NoMemory"); content_broadcast(c, CONTENT_MSG_ERROR, msg_data); return false; } content_broadcast(c, CONTENT_MSG_REFRESH, msg_data); break; } for ( ; url <= end - 4; url++) { if (!strncasecmp(url, "url=", 4)) { url += 4; break; } } /* various sites contain junk meta refresh URL components, * so attempt to deal with this by stripping likely garbage * from the beginning and end of URLs */ while (url < end) { if (isspace(*url) || *url == '\'' || *url == '"') url++; else break; } while (end > url) { if (isspace(end[-1]) || end[-1] == '\'' || end[-1] == '"') *--end = '\0'; else break; } if (url < end) { res = url_join(url, c->data.html.base_url, &refresh); xmlFree(content); if (res == URL_FUNC_NOMEM) { msg_data.error = messages_get("NoMemory"); content_broadcast(c, CONTENT_MSG_ERROR, msg_data); return false; } else if (res == URL_FUNC_FAILED) { /* This isn't fatal so carry on looking */ continue; } c->refresh = talloc_strdup(c, refresh); free(refresh); if (!c->refresh) { msg_data.error = messages_get("NoMemory"); content_broadcast(c, CONTENT_MSG_ERROR, msg_data); return false; } content_broadcast(c, CONTENT_MSG_REFRESH, msg_data); break; } xmlFree(content); } return true; } /** * Process inline stylesheets and fetch linked stylesheets. * * \param c content structure * \param head xml node of html element * \param head xml node of head element, or 0 if none * \return true on success, false if an error occurred */ bool html_find_stylesheets(struct content *c, xmlNode *html, xmlNode *head) { xmlNode *node; char *rel, *type, *media, *href, *url; unsigned int i = STYLESHEET_START; unsigned int last_active = 0; union content_msg_data msg_data; url_func_result res; struct content **stylesheet_content; /* stylesheet 0 is the base style sheet, * stylesheet 1 is the adblocking stylesheet, * stylesheet 2 is any