Do not use libxml to parse mozilla bookmarks. Use the galeon parser

2003-06-13  Marco Pesenti Gritti  <marco@it.gnome.org>

	* src/bookmarks/ephy-bookmarks-import.c:
	(gul_general_read_line_from_file), (gul_string_ascii_strcasestr),
	(ns_get_bookmark_item), (ns_parse_bookmark_item),
	(ephy_bookmarks_import_mozilla):

	Do not use libxml to parse mozilla bookmarks.
	Use the galeon parser instead, ugly but works.

2 files changed, 231 insertions, 49 deletions

diff --git a/ChangeLog b/ChangeLog
index 2443b5d87..4e6f99199 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,13 @@
+2003-06-13  Marco Pesenti Gritti  <marco@it.gnome.org>
+
+	* src/bookmarks/ephy-bookmarks-import.c:
+	(gul_general_read_line_from_file), (gul_string_ascii_strcasestr),
+	(ns_get_bookmark_item), (ns_parse_bookmark_item),
+	(ephy_bookmarks_import_mozilla):
+
+	Do not use libxml to parse mozilla bookmarks.
+	Use the galeon parser instead, ugly but works.
+
 2003-06-13  Piers Cornwell  <piersc@cogs.susx.ac.uk>
  
  	* data/glade/prefs-dialog.glade:
diff --git a/src/bookmarks/ephy-bookmarks-import.c b/src/bookmarks/ephy-bookmarks-import.c
index f574654ae..778186039 100644
--- a/src/bookmarks/ephy-bookmarks-import.c
+++ b/src/bookmarks/ephy-bookmarks-import.c
@@ -24,6 +24,19 @@
 #include "ephy-bookmarks-import.h"
 #include "ephy-debug.h"
 
+/**
+ * NSItemType: netscape bookmark item type
+ */
+typedef enum
+{
+	NS_SITE,
+	NS_NOTES,
+	NS_FOLDER,
+	NS_FOLDER_END,
+	NS_SEPARATOR,
+	NS_UNKNOWN
+} NSItemType;
+
 typedef struct _XbelInfo
 {
 	char *title;
@@ -71,42 +84,6 @@ ephy_bookmarks_import (EphyBookmarks *bookmarks,
 }
 
 static void
-mozilla_parse_bookmarks (EphyBookmarks *bookmarks,
-		         htmlNodePtr node,
-			 char **keyword)
-{
-	htmlNodePtr child = node;
-
-	while (child != NULL)
-	{
-		if (xmlStrEqual (child->name, "h3"))
-		{
-			*keyword = xmlNodeGetContent (child);
-		}
-		else if (xmlStrEqual (child->name, "a"))
-		{
-			xmlChar *title, *url;
-			EphyNode *bmk;
-
-			title = xmlNodeGetContent (child);
-			url = xmlGetProp (child, "href");
-			bmk = ephy_bookmarks_add (bookmarks,
-					          title,
-					          url);
-			set_folder (bookmarks, bmk, *keyword);
-			xmlFree (title);
-			xmlFree (url);
-		}
-
-		mozilla_parse_bookmarks (bookmarks,
-				         child->children,
-					 keyword);
-		child = child->next;
-	}
-}
-
-
-static void
 xbel_parse_single_bookmark (EphyBookmarks *bookmarks,
 			    xmlNodePtr node, XbelInfo *xbel)
 {
@@ -221,25 +198,220 @@ xbel_parse_bookmarks (EphyBookmarks *bookmarks,
 	}
 }
 
+static gchar *
+gul_general_read_line_from_file (FILE *f)
+{
+	gchar *line = g_strdup ("");
+	gchar *t;
+	gchar *buf = g_new0 (gchar, 256);
+	while ( ! ( strchr (buf, '\n') || feof (f) ) ) {
+		fgets(buf, 256, f);
+		t = line;
+		line = g_strconcat (line, buf, NULL);
+		g_free (t);
+	}
+	g_free (buf);
+	return line;
+}
+
+static const gchar *
+gul_string_ascii_strcasestr (const gchar *a, const gchar *b)
+{
+	gchar *down_a;
+	gchar *down_b;
+	gchar *ptr;
+
+	/* copy and lower case the strings */
+	down_a = g_strdup (a);
+	down_b = g_strdup (b);
+	g_ascii_strdown (down_a, -1);
+	g_ascii_strdown (down_b, -1);
+
+	/* compare */
+	ptr = strstr (down_a, down_b);
+
+	/* free allocated strings */
+	g_free (down_a);
+	g_free (down_b);
+
+	/* return result of comparison */
+	return ptr == NULL ? NULL : (a + (ptr - down_a));
+}
+
+/**
+ * Parses a line of a mozilla/netscape bookmark file. File must be open.
+ */
+/* this has been tested fairly well */
+static NSItemType
+ns_get_bookmark_item (FILE *f, GString *name, GString *url)
+{
+	char *line = NULL;
+	char *found;
+
+	line = gul_general_read_line_from_file (f);
+
+	if ((found = (char *) gul_string_ascii_strcasestr (line, "<A HREF=")))
+	{  /* declare site? */
+		g_string_assign (url, found+9);  /* url=URL+ ADD_DATE ... */
+		g_string_truncate (url, strstr(url->str, "\"")-url->str);
+		found = (char *) strstr (found+9+url->len, "\">");
+		if (!found)
+		{
+			g_free (line);
+			return NS_UNKNOWN;
+		}
+		g_string_assign (name, found+2);
+		g_string_truncate (name, gul_string_ascii_strcasestr (name->str,
+						       "</A>")-name->str);
+		g_free (line);
+		return NS_SITE;
+	}
+	else if ((found = (char *) gul_string_ascii_strcasestr (line, "<DT><H3")))
+	{ /* declare folder? */
+		found = (char *) strstr(found+7, ">");
+		if (!found) return NS_UNKNOWN;
+		g_string_assign (name, found+1);
+		g_string_truncate (name, gul_string_ascii_strcasestr (name->str,
+				   "</H3>") - name->str);
+		g_free (line);
+		return NS_FOLDER;
+	}
+	else if ((found = (char *) gul_string_ascii_strcasestr (line, "</DL>")))
+	{     /* end folder? */
+		g_free (line);
+		return NS_FOLDER_END;
+	}
+
+	g_free (line);
+	return NS_UNKNOWN;
+}
+
+/**
+ * This function replaces some weird elements
+ * like &amp; &le;, etc..
+ * More info : http://www.w3.org/TR/html4/charset.html#h-5.3.2
+ * NOTE : We don't support &#D or &#xH.
+ * Patch courtesy of Almer S. Tigelaar <almer1@dds.nl>
+ */
+static char *
+ns_parse_bookmark_item (GString *string)
+{
+	char *iterator, *temp;
+	int cnt = 0;
+	GString *result = g_string_new (NULL);
+
+	g_return_val_if_fail (string != NULL, NULL);
+	g_return_val_if_fail (string->str != NULL, NULL);
+
+	iterator = string->str;
+
+	for (cnt = 0, iterator = string->str;
+	     cnt <= (int)(strlen (string->str));
+	     cnt++, iterator++) {
+		if (*iterator == '&') {
+			int jump = 0;
+			int i;
+
+			if (g_ascii_strncasecmp (iterator, "&amp;", 5) == 0)
+			{
+				g_string_append_c (result, '&');
+				jump = 5;
+			}
+			else if (g_ascii_strncasecmp (iterator, "&lt;", 4) == 0)
+			{
+				g_string_append_c (result, '<');
+				jump = 4;
+			}
+			else if (g_ascii_strncasecmp (iterator, "&gt;", 4) == 0)
+			{
+				g_string_append_c (result, '>');
+				jump = 4;
+			}
+			else if (g_ascii_strncasecmp (iterator, "&quot;", 6) == 0)
+			{
+				g_string_append_c (result, '\"');
+				jump = 6;
+			}
+			else
+			{
+				/* It must be some numeric thing now */
+
+				iterator++;
+
+				if (iterator && *iterator == '#') {
+					int val;
+					char *num, *tmp;
+
+					iterator++;
+
+					val = atoi (iterator);
+
+					tmp = g_strdup_printf ("%d", val);
+					jump = strlen (tmp);
+					g_free (tmp);
+
+					num = g_strdup_printf ("%c", (char) val);
+					g_string_append (result, num);
+					g_free (num);
+				}
+			}
+
+			for (i = jump - 1; i > 0; i--)
+			{
+				iterator++;
+				if (iterator == NULL)
+					break;
+			}
+		}
+		else
+		{
+			g_string_append_c (result, *iterator);
+		}
+	}
+	temp = result->str;
+	g_string_free (result, FALSE);
+	return temp;
+}
+
 gboolean
 ephy_bookmarks_import_mozilla (EphyBookmarks *bookmarks,
 			       const char *filename)
 {
-	htmlDocPtr doc;
-	htmlNodePtr child;
-	char *keyword = NULL;
-
-	if (g_file_test (filename, G_FILE_TEST_EXISTS) == FALSE)
+	FILE *bf;  /* bookmark file */
+	GString *name = g_string_new (NULL);
+	gchar *parsedname;
+	GString *url = g_string_new (NULL);
+	char *current_folder = NULL;
+	EphyNode *bmk;
+
+	if (!(bf = fopen (filename, "r"))) {
+		g_warning ("Failed to open file: %s\n", filename);
 		return FALSE;
+	}
 
-	doc = htmlParseFile (filename, "UTF-8");
-	g_assert (doc != NULL);
-
-	child = doc->children;
-	mozilla_parse_bookmarks (bookmarks, child, &keyword);
-
-	g_free (keyword);
-        xmlFreeDoc (doc);
+	while (!feof (bf)) {
+		NSItemType t;
+		t = ns_get_bookmark_item (bf, name, url);
+		switch (t)
+		{
+		case NS_FOLDER:
+			g_free (current_folder);
+			current_folder = g_strdup (name->str);
+			break;
+		case NS_SITE:
+			parsedname = ns_parse_bookmark_item (name);
+			bmk = ephy_bookmarks_add (bookmarks,
+					          parsedname,
+					          url->str);
+			set_folder (bookmarks, bmk, current_folder);
+			break;
+		default:
+			break;
+		}
+	}
+	fclose (bf);
+	g_string_free (name, TRUE);
+	g_string_free (url, TRUE);
 
 	ephy_bookmarks_save (bookmarks);