From 6879c02a332824ff63ce0b56de36996d2f0d48cf Mon Sep 17 00:00:00 2001
From: James Bursa <james@netsurf-browser.org>
Date: Sun, 6 Jun 2004 19:39:17 +0000
Subject: [project @ 2004-06-06 19:39:17 by bursa] Add and improve error
 handling of save_complete. Save HTML as UTF-8 to prevent encoding issues.

svn path=/import/netsurf/; revision=931
---
 riscos/save_complete.c | 522 +++++++++++++++++++++++++++++--------------------
 1 file changed, 306 insertions(+), 216 deletions(-)

(limited to 'riscos/save_complete.c')

diff --git a/riscos/save_complete.c b/riscos/save_complete.c
index bdf8682db..4ad1668b6 100644
--- a/riscos/save_complete.c
+++ b/riscos/save_complete.c
@@ -12,6 +12,7 @@
 
 #include <assert.h>
 #include <ctype.h>
+#include <errno.h>
 #include <string.h>
 #include <sys/types.h>
 #include <regex.h>
@@ -33,24 +34,24 @@ regex_t save_complete_import_re;
 
 /** An entry in save_complete_list. */
 struct save_complete_entry {
-        char *url;   /**< Fully qualified URL, as per url_join output */
-        int ptr;    /**< Pointer to object's location in memory */
-        struct save_complete_entry *next; /**< Next entry in list */
+	struct content *content;
+	struct save_complete_entry *next; /**< Next entry in list */
 };
 
 /** List of urls seen and saved so far. */
-static struct save_complete_entry *save_complete_list;
+static struct save_complete_entry *save_complete_list = 0;
 
-static void save_complete_html(struct content *c, const char *path,
-                bool index);
-static void save_imported_sheets(struct content *c, const char *path);
+static bool save_complete_html(struct content *c, const char *path,
+		bool index);
+static bool save_imported_sheets(struct content *c, const char *path);
 static char * rewrite_stylesheet_urls(const char *source, unsigned int size,
 		int *osize, const char *base);
-static int rewrite_document_urls(xmlDoc *doc, const char *base);
-static int rewrite_urls(xmlNode *n, const char *base);
-static void rewrite_url(xmlNode *n, const char *attr, const char *base);
-static void save_complete_add_url(const char *url, int id);
-static int save_complete_find_url(const char *url);
+static bool rewrite_document_urls(xmlDoc *doc, const char *base);
+static bool rewrite_urls(xmlNode *n, const char *base);
+static bool rewrite_url(xmlNode *n, const char *attr, const char *base);
+static bool save_complete_list_add(struct content *content);
+static struct content * save_complete_list_find(const char *url);
+static bool save_complete_list_check(struct content *content);
 
 
 /**
@@ -58,64 +59,89 @@ static int save_complete_find_url(const char *url);
  *
  * \param  c     CONTENT_HTML to save
  * \param  path  directory to save to (must exist)
+ * \return  true on success, false on error and error reported
  */
 
-void save_complete(struct content *c, const char *path)
+bool save_complete(struct content *c, const char *path)
 {
-      	save_complete_list = 0;
+	bool result;
 
-        save_complete_html(c, path, true);
+	result = save_complete_html(c, path, true);
 
-       	/* free save_complete_list */
+	/* free save_complete_list */
 	while (save_complete_list) {
 		struct save_complete_entry *next = save_complete_list->next;
-		free(save_complete_list->url);
 		free(save_complete_list);
 		save_complete_list = next;
 	}
+
+	return result;
 }
 
+
 /**
  * Save an HTML page with all dependencies, recursing through imported pages.
  *
- * \param  c     CONTENT_HTML to save
- * \param  path  directory to save to (must exist)
+ * \param  c      CONTENT_HTML to save
+ * \param  path   directory to save to (must exist)
+ * \param  index  true to save as "index"
+ * \return  true on success, false on error and error reported
  */
 
-void save_complete_html(struct content *c, const char *path, bool index)
+bool save_complete_html(struct content *c, const char *path, bool index)
 {
 	char spath[256];
 	unsigned int i;
-	htmlParserCtxtPtr toSave;
+	htmlParserCtxtPtr parser;
+	os_error *error;
 
 	if (c->type != CONTENT_HTML)
-		return;
+		return false;
+
+	if (save_complete_list_check(c))
+		return true;
 
-        /* save stylesheets, ignoring the base sheet */
-        for (i = 1; i != c->data.html.stylesheet_count; i++) {
+	/* save stylesheets, ignoring the base sheet */
+	for (i = 1; i != c->data.html.stylesheet_count; i++) {
 		struct content *css = c->data.html.stylesheet_content[i];
 		char *source;
 		int source_len;
 
-                if (!css)
-                        continue;
+		if (!css)
+			continue;
+		if (save_complete_list_check(css))
+			continue;
 
-		save_complete_add_url(css->url, (int) css);
+		if (!save_complete_list_add(css)) {
+			warn_user("NoMemory", 0);
+			return false;
+		}
 
-                save_imported_sheets(css, path);
+		if (!save_imported_sheets(css, path))
+			return false;
 
-                if (i == 1) continue; /* don't save <style> elements */
+		if (i == 1)
+			continue; /* don't save <style> elements */
 
 		snprintf(spath, sizeof spath, "%s.%x", path,
 				(unsigned int) css);
 		source = rewrite_stylesheet_urls(css->source_data,
 				css->source_size, &source_len, css->url);
-		if (source) {
-			xosfile_save_stamped(spath, 0xf79, source,
+		if (!source) {
+			warn_user("NoMemory", 0);
+			return false;
+		}
+
+		error = xosfile_save_stamped(spath, 0xf79, source,
 					source + source_len);
-			free(source);
+		free(source);
+		if (error) {
+			LOG(("xosfile_save_stamped: 0x%x: %s",
+					error->errnum, error->errmess));
+			warn_user("SaveError", error->errmess);
+			return false;
 		}
-        }
+	}
 
 	/* save objects */
 	for (i = 0; i != c->data.html.object_count; i++) {
@@ -124,79 +150,134 @@ void save_complete_html(struct content *c, const char *path, bool index)
 		/* skip difficult content types */
 		if (!obj || obj->type >= CONTENT_OTHER || !obj->source_data)
 			continue;
+		if (save_complete_list_check(obj))
+			continue;
 
-		save_complete_add_url(obj->url, (int) obj);
-
-                if (obj->type == CONTENT_HTML) {
-                        save_complete_html(obj, path, false);
-                }
-                else {
-		        snprintf(spath, sizeof spath, "%s.%x", path,
-			        (unsigned int) obj);
-		        xosfile_save_stamped(spath,
-			        	ro_content_filetype(obj),
-				        obj->source_data,
-				        obj->source_data + obj->source_size);
+		if (!save_complete_list_add(obj)) {
+			warn_user("NoMemory", 0);
+			return false;
+		}
+
+		if (obj->type == CONTENT_HTML) {
+			if (!save_complete_html(obj, path, false))
+				return false;
+			continue;
+		}
+
+		snprintf(spath, sizeof spath, "%s.%x", path,
+				(unsigned int) obj);
+		error = xosfile_save_stamped(spath,
+				ro_content_filetype(obj),
+				obj->source_data,
+				obj->source_data + obj->source_size);
+		if (error) {
+			LOG(("xosfile_save_stamped: 0x%x: %s",
+					error->errnum, error->errmess));
+			warn_user("SaveError", error->errmess);
+			return false;
 		}
 	}
 
 	/* make a copy of the document tree */
-	toSave = htmlCreateMemoryParserCtxt(c->source_data, c->source_size);
-	htmlParseDocument(toSave);
+	parser = htmlCreateMemoryParserCtxt(c->source_data, c->source_size);
+	if (!parser) {
+		warn_user("NoMemory", 0);
+		return false;
+	}
+	/** \todo set parser charset */
+	htmlParseDocument(parser);
 
 	/* rewrite all urls we know about */
-       	if (rewrite_document_urls(toSave->myDoc, c->data.html.base_url) == 0) {
-       	        xfree(spath);
-       	        xmlFreeDoc(toSave->myDoc);
-       	        htmlFreeParserCtxt(toSave);
-       	        return;
-       	}
+	if (!rewrite_document_urls(parser->myDoc, c->data.html.base_url)) {
+		xmlFreeDoc(parser->myDoc);
+		htmlFreeParserCtxt(parser);
+		warn_user("NoMemory", 0);
+		return false;
+	}
 
 	/* save the html file out last of all */
 	if (index)
-        	snprintf(spath, sizeof spath, "%s.index", path);
-        else
-                snprintf(spath, sizeof spath, "%s.%x", path, (unsigned int)c);
-	htmlSaveFile(spath, toSave->myDoc);
-	xosfile_set_type(spath, 0xfaf);
-
-        xmlFreeDoc(toSave->myDoc);
-        htmlFreeParserCtxt(toSave);
+		snprintf(spath, sizeof spath, "%s.index", path);
+	else
+		snprintf(spath, sizeof spath, "%s.%x", path, (unsigned int)c);
+
+	errno = 0;
+	if (htmlSaveFileFormat(spath, parser->myDoc, 0, 0) == -1) {
+		if (errno)
+			warn_user("SaveError", strerror(errno));
+		else
+			warn_user("SaveError", "htmlSaveFileFormat failed");
+		return false;
+	}
+
+	error = xosfile_set_type(spath, 0xfaf);
+	if (error) {
+		LOG(("xosfile_set_type: 0x%x: %s",
+				error->errnum, error->errmess));
+		warn_user("SaveError", error->errmess);
+		return false;
+	}
+
+	xmlFreeDoc(parser->myDoc);
+	htmlFreeParserCtxt(parser);
+
+	return true;
 }
 
+
 /**
- * Save all imported stylesheets
+ * Save stylesheets imported by a CONTENT_CSS.
  *
- * \param  c The content containing the stylesheet
- * \param path Path to save to
+ * \param  c     a CONTENT_CSS
+ * \param  path  path to save to
+ * \return  true on success, false on error and error reported
  */
-void save_imported_sheets(struct content *c, const char *path)
+
+bool save_imported_sheets(struct content *c, const char *path)
 {
 	char spath[256];
-        unsigned int j;
-        char *source;
-        int source_len;
+	unsigned int j;
+	char *source;
+	int source_len;
+	os_error *error;
 
-        for (j = 0; j != c->data.css.import_count; j++) {
+	for (j = 0; j != c->data.css.import_count; j++) {
 		struct content *css = c->data.css.import_content[j];
 
-                if (!css)
-                        continue;
+		if (!css)
+			continue;
+		if (save_complete_list_check(css))
+			continue;
 
-		save_complete_add_url(css->url, (int) css);
+		if (!save_complete_list_add(css)) {
+			warn_user("NoMemory", 0);
+			return false;
+		}
 
-                save_imported_sheets(css, path);
+		if (!save_imported_sheets(css, path))
+			return false;
 
 		snprintf(spath, sizeof spath, "%s.%x", path,
 				(unsigned int) css);
 		source = rewrite_stylesheet_urls(css->source_data,
 				css->source_size, &source_len, css->url);
-		if (source) {
-			xosfile_save_stamped(spath, 0xf79, source,
+		if (!source) {
+			warn_user("NoMemory", 0);
+			return false;
+		}
+
+		error = xosfile_save_stamped(spath, 0xf79, source,
 					source + source_len);
-			free(source);
+		free(source);
+		if (error) {
+			LOG(("xosfile_save_stamped: 0x%x: %s",
+					error->errnum, error->errmess));
+			warn_user("SaveError", error->errmess);
+			return false;
 		}
-        }
+	}
+
+	return true;
 }
 
 
@@ -244,7 +325,7 @@ void save_complete_init(void)
  * @param  size    size of source
  * @param  osize   updated with the size of the result
  * @param  base    url of stylesheet
- * @return  converted source, or 0 on error
+ * @return  converted source, or 0 on out of memory
  */
 
 char * rewrite_stylesheet_urls(const char *source, unsigned int size,
@@ -256,7 +337,7 @@ char * rewrite_stylesheet_urls(const char *source, unsigned int size,
 	char buf[20];
 	unsigned int offset = 0;
 	int url_len = 0;
-	int id;
+	struct content *content;
 	int m;
 	unsigned int i;
 	unsigned int imports = 0;
@@ -276,10 +357,8 @@ char * rewrite_stylesheet_urls(const char *source, unsigned int size,
 	}
 
 	res = malloc(size + imports * 20);
-	if (!res) {
-		warn_user("NoMemory", 0);
+	if (!res)
 		return 0;
-	}
 	*osize = 0;
 
 	while (offset < size) {
@@ -317,26 +396,25 @@ char * rewrite_stylesheet_urls(const char *source, unsigned int size,
 
 		url2 = strndup(url, url_len);
 		if (!url2) {
-			warn_user("NoMemory", 0);
 			free(res);
 			return 0;
 		}
 		url = url_join(url2, base);
 		free(url2);
 		if (!url) {
-			warn_user("NoMemory", 0);
 			free(res);
 			return 0;
-                }
+		}
 
 		/* copy data before match */
 		memcpy(res + *osize, source + offset, match[0].rm_so);
 		*osize += match[0].rm_so;
 
-		id = save_complete_find_url(url);
-		if (id) {
+		content = save_complete_list_find(url);
+		if (content) {
 			/* replace import */
-			sprintf(buf, "@import '%x'", id);
+			snprintf(buf, sizeof buf, "@import '%x'",
+					(unsigned int) content);
 			memcpy(res + *osize, buf, strlen(buf));
 			*osize += strlen(buf);
 		} else {
@@ -361,116 +439,110 @@ char * rewrite_stylesheet_urls(const char *source, unsigned int size,
 
 
 /**
- * Rewrite URLs in a HTML document to be relative
+ * Rewrite URLs in a HTML document to be relative.
  *
- * @param doc The root of the document tree
- * @return 0 on error. >0 otherwise
+ * \param  doc   root of the document tree
  * \param  base  base url of document
+ * \return  true on success, false on out of memory
  */
 
-int rewrite_document_urls(xmlDoc *doc, const char *base)
+bool rewrite_document_urls(xmlDoc *doc, const char *base)
 {
-        xmlNode *html;
+	xmlNode *node;
 
-        /* find the html element */
-        for (html = doc->children;
-             html!=0 && html->type != XML_ELEMENT_NODE;
-             html = html->next)
-                ;
-        if (html == 0 || strcmp((const char*)html->name, "html") != 0) {
-                return 0;
-        }
+	for (node = doc->children; node; node = node->next)
+		if (node->type == XML_ELEMENT_NODE)
+			if (!rewrite_urls(node, base))
+				return false;
 
-	rewrite_urls(html, base);
-
-        return 1;
+	return true;
 }
 
 
 /**
  * Traverse tree, rewriting URLs as we go.
  *
+ * \param  n     xmlNode of type XML_ELEMENT_NODE to rewrite
  * \param  base  base url of document
- * @param n The root of the tree
- * @return 0 on error. >0 otherwise
+ * \return  true on success, false on out of memory
+ *
+ * URLs in the tree rooted at element n are rewritten.
  */
 
-int rewrite_urls(xmlNode *n, const char *base)
+bool rewrite_urls(xmlNode *n, const char *base)
 {
-        xmlNode *this;
-
-        /**
-         * We only need to consider the following cases:
-         *
-         * Attribute:      Elements:
-         *
-         * 1)   data         <object>
-         * 2)   href         <a> <area> <link> <base>
-         * 3)   src          <script> <input> <frame> <iframe> <img>
-         * 4)   n/a          <style>
-         */
-        if (n->type == XML_ELEMENT_NODE) {
-                /* 1 */
-                if (strcmp(n->name, "object") == 0) {
-                      rewrite_url(n, "data", base);
-                }
-                /* 2 */
-                else if (strcmp(n->name, "a") == 0 ||
-                         strcmp(n->name, "area") == 0 ||
-                         strcmp(n->name, "link") == 0 ||
-                         strcmp(n->name, "base") == 0) {
-                      rewrite_url(n, "href", base);
-                }
-                /* 3 */
-                else if (strcmp(n->name, "frame") == 0 ||
-                         strcmp(n->name, "iframe") == 0 ||
-                         strcmp(n->name, "input") == 0 ||
-                         strcmp(n->name, "img") == 0 ||
-                         strcmp(n->name, "script") == 0) {
-                      rewrite_url(n, "src", base);
-                }
-                /* 4 */
-                else if (strcmp(n->name, "style") == 0) {
-                        unsigned int len;
-                        xmlChar *content;
-
-                        for (this = n->children; this != 0; this = this->next) {
-                                /* Get current content */
-                                content = xmlNodeGetContent(this);
-                                if (!content) continue;
-
-                                /* Rewrite @import rules */
-                                char *rewritten = rewrite_stylesheet_urls(
-                                                  content,
-		                                  strlen((char*)content),
-		                                  &len, base);
-                                if (!rewritten) {
-                                        xmlFree(content);
-                                        continue;
-                                }
-
-                                /* set new content */
-                                xmlNodeSetContentLen(this,
-                                             (const xmlChar*)rewritten,
-                                             len);
-
-                                /* free old content */
-                                xmlFree(content);
-                        }
-
-                        return 1;
-                }
-        }
-        else {
-                return 0;
-        }
-
-        /* now recurse */
-	for (this = n->children; this != 0; this = this->next) {
-                rewrite_urls(this, base);
+	xmlNode *child;
+
+	/**
+	 * We only need to consider the following cases:
+	 *
+	 * Attribute:      Elements:
+	 *
+	 * 1)   data         <object>
+	 * 2)   href         <a> <area> <link> <base>
+	 * 3)   src          <script> <input> <frame> <iframe> <img>
+	 * 4)   n/a          <style>
+	 */
+	/* 1 */
+	if (strcmp(n->name, "object") == 0) {
+		if (!rewrite_url(n, "data", base))
+			return false;
+	}
+	/* 2 */
+	else if (strcmp(n->name, "a") == 0 ||
+			strcmp(n->name, "area") == 0 ||
+			strcmp(n->name, "link") == 0 ||
+			strcmp(n->name, "base") == 0) {
+		if (!rewrite_url(n, "href", base))
+			return false;
+	}
+	/* 3 */
+	else if (strcmp(n->name, "frame") == 0 ||
+			strcmp(n->name, "iframe") == 0 ||
+			strcmp(n->name, "input") == 0 ||
+			strcmp(n->name, "img") == 0 ||
+			strcmp(n->name, "script") == 0) {
+		if (!rewrite_url(n, "src", base))
+			return false;
+	}
+	/* 4 */
+	else if (strcmp(n->name, "style") == 0) {
+		unsigned int len;
+		xmlChar *content;
+
+		for (child = n->children; child != 0; child = child->next) {
+			/* Get current content */
+			content = xmlNodeGetContent(child);
+			if (!content)
+				/* unfortunately we don't know if this is due to
+				 * memory exhaustion, or because there is no
+				 * content for this node */
+				continue;
+
+			/* Rewrite @import rules */
+			char *rewritten = rewrite_stylesheet_urls(
+					content,
+					strlen((char*)content),
+					&len, base);
+			xmlFree(content);
+			if (!rewritten)
+				return false;
+
+			/* set new content */
+			xmlNodeSetContentLen(child,
+					(const xmlChar*)rewritten,
+					len);
+		}
+
+		return true;
 	}
 
-        return 1;
+	/* now recurse */
+	for (child = n->children; child; child = child->next)
+		if (!rewrite_urls(child, base))
+			return false;
+
+	return true;
 }
 
 
@@ -480,80 +552,98 @@ int rewrite_urls(xmlNode *n, const char *base)
  * \param  n     The node to modify
  * \param  attr  The html attribute to modify
  * \param  base  base url of document
+ * \return  true on success, false on out of memory
  */
 
-void rewrite_url(xmlNode *n,const char *attr, const char *base)
+bool rewrite_url(xmlNode *n, const char *attr, const char *base)
 {
-        char *url, *data;
-        char rel[256];
-	int id;
+	char *url, *data;
+	char rel[20];
+	struct content *content;
 
-        data = xmlGetProp(n, (const xmlChar*)attr);
+	if (!xmlHasProp(n, (const xmlChar *) attr))
+		return true;
 
-        if (!data) return;
+	data = xmlGetProp(n, (const xmlChar *) attr);
+	if (!data)
+		return false;
 
-        url = url_join(data, base);
-        if (!url) {
-                xmlFree(data);
-                return;
-        }
+	url = url_join(data, base);
+	xmlFree(data);
+	if (!url)
+		return false;
 
-	id = save_complete_find_url(url);
-	if (id) {
+	content = save_complete_list_find(url);
+	if (content) {
 		/* found a match */
-		snprintf(rel, sizeof rel, "%x", id);
-		xmlSetProp(n, (const xmlChar *) attr, (xmlChar *) rel);
+		free(url);
+		snprintf(rel, sizeof rel, "%x", (unsigned int) content);
+		if (!xmlSetProp(n, (const xmlChar *) attr, (xmlChar *) rel))
+			return false;
 	} else {
 		/* no match found */
-		xmlSetProp(n, (const xmlChar *) attr, (xmlChar *) url);
+		if (!xmlSetProp(n, (const xmlChar *) attr, (xmlChar *) url)) {
+			free(url);
+			return false;
+		}
+		free(url);
 	}
 
-        free(url);
-        xmlFree(data);
+	return true;
 }
 
 
 /**
- * Add a url to the save_complete_list.
+ * Add a content to the save_complete_list.
  *
- * \param  url  url to add (copied)
- * \param  id   id to use for url
+ * \param  content  content to add
+ * \return  true on success, false on out of memory
  */
 
-void save_complete_add_url(const char *url, int id)
+bool save_complete_list_add(struct content *content)
 {
 	struct save_complete_entry *entry;
 	entry = malloc(sizeof (*entry));
 	if (!entry)
-		return;
-	entry->url = strdup(url);
-	if (!url) {
-		free(entry);
-		return;
-	}
-	entry->ptr = id;
+		return false;
+	entry->content = content;
 	entry->next = save_complete_list;
 	save_complete_list = entry;
+	return true;
 }
 
 
 /**
  * Look up a url in the save_complete_list.
  *
- * \param  url   url to find
- * \param  len   length of url
- * \return  id to use for url, or 0 if not present
+ * \param  url  url to find
+ * \return  content if found, 0 otherwise
  */
 
-int save_complete_find_url(const char *url)
+struct content * save_complete_list_find(const char *url)
 {
 	struct save_complete_entry *entry;
 	for (entry = save_complete_list; entry; entry = entry->next)
-		if (strcmp(url, entry->url) == 0)
-			break;
-	if (entry)
-		return entry->ptr;
+		if (strcmp(url, entry->content->url) == 0)
+			return entry->content;
 	return 0;
 }
 
+
+/**
+ * Look up a content in the save_complete_list.
+ *
+ * \param  content  pointer to content
+ * \return  true if the content is in the save_complete_list
+ */
+
+bool save_complete_list_check(struct content *content)
+{
+	struct save_complete_entry *entry;
+	for (entry = save_complete_list; entry; entry = entry->next)
+		if (entry->content == content)
+			return true;
+	return false;
+}
+
 #endif
-- 
cgit v1.2.3