Searching in epub documents

I added the capability to search through documents. The search is quite buggy, and we'll be taking care of each bug one at a time.

7 files changed, 271 insertions, 132 deletions

diff --git a/backend/epub/epub-document.c b/backend/epub/epub-document.c
index e9443c09..3a574151 100644
--- a/backend/epub/epub-document.c
+++ b/backend/epub/epub-document.c
@@ -29,14 +29,13 @@
 #include "ev-document-misc.h"
 #include <libxml/parser.h>
 #include <libxml/xmlmemory.h>
-
+#include <libxml/HTMLparser.h>
 #include <config.h>
 
 #include <glib/gi18n.h>
 #include <glib/gstdio.h>
 
 #include <gtk/gtk.h>
-#include <stdio.h>
 
 /*For strcasestr(),strstr()*/
 #include <string.h>
@@ -130,47 +129,72 @@ epub_document_thumbnails_get_thumbnail (EvDocumentThumbnails *document,
 }
 
 static gboolean
+in_tag(const char* found)
+{
+    const char* bracket = found ;
+
+    /* Since the dump started with the body tag, the '<' will be the first
+     * character in the haystack.
+     */
+    while (*bracket != '<') {
+        bracket--;
+        if (*bracket == '>') {
+            /*We encounted a close brace before an open*/
+            return FALSE ;
+        }
+    }
+
+    return TRUE;
+}
+
+static int 
+get_substr_count(const char * haystack,const char *needle,gboolean case_sensitive)
+{
+    const char* tmp = haystack ;
+    char* (*string_compare_function)(const char*,const char*);
+    int count=0;
+    if (case_sensitive) {
+        string_compare_function = strstr ;
+    }
+    else {
+        string_compare_function = strcasestr;
+    }
+
+    while ((tmp=string_compare_function(tmp,needle))) {
+        if (!in_tag(tmp)) {
+            count++;
+        }
+        tmp = tmp + strlen(needle);
+    }
+
+    return count;
+}
+
+static guint
 epub_document_check_hits(EvDocumentFind *document_find,
                          EvPage         *page,
                          const gchar    *text,
                          gboolean        case_sensitive)
 {
 	gchar *filepath = g_filename_from_uri((gchar*)page->backend_page,NULL,NULL);
-	FILE *fp = fopen(filepath,"r");
-	GString *buffer; 
-	gchar *found ;
-	
-	while (!feof(fp)) {
-		gchar c;
-		gint pos=0;
-		buffer = g_string_sized_new (1024);
-		
-		while ((c = fgetc(fp)) != '\n' && !feof(fp)) {
-			g_string_insert_c(buffer,pos++,c);
-		}
+	htmlDocPtr htmldoc =  xmlParseFile(filepath); 
+	htmlNodePtr htmltag = xmlDocGetRootElement(htmldoc);
+	int count=0;
+	htmlNodePtr bodytag = htmltag->xmlChildrenNode;
 
-		g_string_insert_c(buffer,pos,'\0');
-		
-		if (case_sensitive) {
-			if ((found = strstr(buffer->str,text)) != NULL) {
-				g_string_free(buffer,TRUE);
-				fclose(fp);
-				return TRUE;
-			}
-		}
-		else {
-			
-			if ( (found = strcasestr(buffer->str,text)) != NULL) {
-				g_string_free(buffer,TRUE);
-				fclose(fp);
-				return TRUE;
-			}
-		}
-		g_string_free(buffer,TRUE);
+	while ( xmlStrcmp(bodytag->name,(xmlChar*)"body") ) {
+		bodytag = bodytag->next;
 	}
 	
-	fclose(fp);
-	return FALSE;
+	xmlBufferPtr bodybuffer = xmlBufferCreate();
+	xmlNodeDump(bodybuffer,htmldoc,bodytag,0,1);
+
+	count = get_substr_count((char*)bodybuffer->content,text,case_sensitive);
+	
+	xmlBufferFree(bodybuffer);
+	xmlFreeDoc(htmldoc);
+
+	return count;
 }
 
 static gboolean
diff --git a/libdocument/ev-document-find.c b/libdocument/ev-document-find.c
index c22b913d..92dfffe5 100644
--- a/libdocument/ev-document-find.c
+++ b/libdocument/ev-document-find.c
@@ -40,7 +40,7 @@ ev_document_find_find_text (EvDocumentFind *document_find,
 	return iface->find_text (document_find, page, text, case_sensitive);
 }
 
-gboolean
+guint
 ev_document_find_check_for_hits(EvDocumentFind *document_find,
                                 EvPage         *page,
                                 const gchar    *text,
diff --git a/libdocument/ev-document-find.h b/libdocument/ev-document-find.h
index 7fb5d22c..2b6ba68b 100644
--- a/libdocument/ev-document-find.h
+++ b/libdocument/ev-document-find.h
@@ -53,7 +53,7 @@ struct _EvDocumentFindInterface
 				    const gchar    *text,
 				    gboolean        case_sensitive);
 
-	gboolean (* check_for_hits)(EvDocumentFind *document_find,
+	guint (* check_for_hits)   (EvDocumentFind *document_find,
 				    EvPage         *page,
 				    const gchar    *text,
 				    gboolean        case_sensitive);
@@ -65,7 +65,7 @@ GList *ev_document_find_find_text (EvDocumentFind *document_find,
 				   const gchar    *text,
 				   gboolean        case_sensitive);
 
-gboolean ev_document_find_check_for_hits(EvDocumentFind *document_find,
+guint ev_document_find_check_for_hits   (EvDocumentFind *document_find,
                                          EvPage         *page,
                                          const gchar    *text,
                                          gboolean        case_sensitive);
diff --git a/libview/ev-jobs.c b/libview/ev-jobs.c
index 155990e9..3a7dfd27 100644
--- a/libview/ev-jobs.c
+++ b/libview/ev-jobs.c
@@ -1379,7 +1379,7 @@ ev_job_find_run (EvJob *job)
 	ev_page = ev_document_get_page (job->document, job_find->current_page);
 
 	if (job->document->iswebdocument) {
-		job_find->has_results = ev_document_find_check_for_hits(find, ev_page, job_find->text,
+		job_find->results[job_find->current_page] = ev_document_find_check_for_hits(find, ev_page, job_find->text,
 		                                                        job_find->case_sensitive);
 	}else {
 		matches = ev_document_find_find_text (find, ev_page, job_find->text,
@@ -1393,10 +1393,14 @@ ev_job_find_run (EvJob *job)
 	if (!job_find->has_results && !job->document->iswebdocument) {
 		job_find->has_results = (matches != NULL);
 	}
-	
+	else if (!job_find->has_results && job->document->iswebdocument){
+		job_find->has_results = (job_find->results[job_find->current_page] > 0);
+	}
+
 	if (job->document->iswebdocument == FALSE) {
 		job_find->pages[job_find->current_page] = matches;
 	}
+
 	g_signal_emit (job_find, job_find_signals[FIND_UPDATED], 0, job_find->current_page);
 		       
 	job_find->current_page = (job_find->current_page + 1) % job_find->n_pages;
diff --git a/libview/ev-web-view.c b/libview/ev-web-view.c
index 152e2542..0756b0f0 100644
--- a/libview/ev-web-view.c
+++ b/libview/ev-web-view.c
@@ -37,13 +37,19 @@
 #define EV_IS_WEB_VIEW_CLASS(klass) (G_TYPE_CHECK_CLASS_TYPE ((klass), EV_TYPE_WEB_VIEW))
 #define EV_WEB_VIEW_GET_CLASS(obj)  (G_TYPE_INSTANCE_GET_CLASS ((obj), EV_TYPE_WEB_VIEW, EvWebViewClass))
 
+ typedef enum {
+ 	EV_WEB_VIEW_FIND_NEXT,
+ 	EV_WEB_VIEW_FIND_PREV
+ } EvWebViewFindDirection;
+
 typedef struct _SearchParams {
 	gboolean case_sensitive;
-	guint    page_current;
 	gboolean search_jump;
 	gchar*   search_string;
-	guint    on_result;
+	gint     on_result;
 	guint    n_results;
+	guint   *results;
+	EvWebViewFindDirection direction;
 }SearchParams;
 
 struct _EvWebView
@@ -65,20 +71,21 @@ struct _EvWebViewClass
 
 /*** Callbacks ***/
 static void       ev_web_view_change_page                   (EvWebView          *webview,
-						          							  gint                new_page);
+						          							 gint                new_page);
 
 static void       ev_web_view_page_changed_cb               (EvDocumentModel     *model,
 														      gint                 old_page,
 															  gint                 new_page,
 														      EvWebView           *webview);
 /*** GObject ***/
-static void       ev_web_view_dispose                           (GObject             *object);
+static void       ev_web_view_dispose                       (GObject             *object);
 
-static void       ev_web_view_finalize						 (GObject             *object);
+static void       ev_web_view_finalize						(GObject             *object);
 static void       ev_web_view_class_init                    (EvWebViewClass      *klass);
 static void       ev_web_view_init                          (EvWebView           *webview);
 
 G_DEFINE_TYPE (EvWebView, ev_web_view, WEBKIT_TYPE_WEB_VIEW)
+
 static void
 web_view_update_range_and_current_page (EvWebView *webview)
 {
@@ -106,12 +113,18 @@ ev_web_view_dispose (GObject *object)
 	if (webview->model) {
 		g_object_unref(webview->model);
 		webview->model = NULL;
-	};
+	}
+	
 	if (webview->hlink) {
 		g_free(webview->hlink);
 		webview->hlink = NULL;
 	}
 
+	if (webview->search) {
+		g_free(webview->search);
+		webview->search = NULL;
+	}
+
 	G_OBJECT_CLASS (ev_web_view_parent_class)->dispose (object);
 }
 
@@ -132,7 +145,10 @@ ev_web_view_init (EvWebView *webview)
 	webview->current_page = 0;
 
 	webview->search = g_new0(SearchParams, 1);
-	
+	webview->search->search_string = NULL;
+	webview->search->on_result = -1 ;
+	webview->search->n_results = 0;
+	webview->search->results = NULL;
 	webview->search->search_jump = TRUE ;
 
 	webview->fullscreen = FALSE;
@@ -400,112 +416,205 @@ ev_web_view_handle_link(EvWebView *webview,EvLink *link)
 }
 
 /* Searching */
-void 
-ev_web_view_find_next(EvWebView *webview)
-{
-	/*First search for the next item on the current page*/
-	 webkit_web_view_search_text (WEBKIT_WEB_VIEW(webview),
-                                  webview->search->search_string,
-                                  webview->search->case_sensitive,
-                                  TRUE,
-                                  FALSE);
-}
 
-void 
-ev_web_view_find_previous(EvWebView *webview)
+static void
+jump_to_find_result_on_page(EvWebView *webview,EvWebViewFindDirection direction)
 {
+	gboolean forward,wrap;
+	
+	if (direction == EV_WEB_VIEW_FIND_NEXT) {
+		forward = TRUE;
+		wrap = FALSE;
+	}
+	else {
+		forward = FALSE;
+		wrap = FALSE;
+	}
+
 	webkit_web_view_search_text (WEBKIT_WEB_VIEW(webview),
-                          		webview->search->search_string,
-                          		webview->search->case_sensitive,
-                          		FALSE,
-                          		TRUE);
+	                             webview->search->search_string,
+	                             webview->search->case_sensitive,
+	                             forward,
+	                             wrap);
 }
 
-void
-ev_web_view_find_set_highlight_search(EvWebView *webview, gboolean visible)
+static void
+jump_to_find_result(EvWebView *webview,
+                    GParamSpec *pspec,
+                    gpointer    data)
 {
-	webkit_web_view_set_highlight_text_matches(WEBKIT_WEB_VIEW(webview),visible);
+	gint n_results;
+	gboolean forward ;
+	gboolean wrap ;
+	
+	if (webkit_web_view_get_load_status(WEBKIT_WEB_VIEW(webview)) != WEBKIT_LOAD_FINISHED) {
+		return;
+	}
+
+	if (!webview->search->search_string) {
+		return;
+	}
+
+	n_results = webkit_web_view_mark_text_matches (WEBKIT_WEB_VIEW(webview),
+	                                               webview->search->search_string,
+	                                               webview->search->case_sensitive,
+	                                               0);
+
+	ev_web_view_find_set_highlight_search(webview,TRUE);
+
+	if (webview->search->direction == EV_WEB_VIEW_FIND_NEXT) {
+		forward = TRUE ;
+		wrap = FALSE;
+	}
+	else {
+		forward = FALSE;
+		wrap = TRUE ;
+	}
+
+	if (n_results > 0 && webview->search->on_result < n_results) {
+		webkit_web_view_search_text (WEBKIT_WEB_VIEW(webview),
+			                         webview->search->search_string,
+			                         webview->search->case_sensitive,
+			                         forward,
+			                         wrap);
+	    
+		webview->search->search_jump = FALSE;
+	}
 }
 
-typedef struct _FindCBStruct {
-	EvJobFind *job;
-	gint page;
-}FindCBStruct;
+static gint
+ev_web_view_find_get_n_results (EvWebView *webview, gint page)
+{
+	return webview->search->results[page];
+}
 
+/**
+ * jump_to_find_page
+ * @webview: #EvWebView instance
+ * @direction: Direction to look
+ * @shift: Shift from current page
+ *
+ * Jumps to the first page that has occurences of searched word.
+ * Uses a direction where to look and a shift from current page. 
+ *
+ */
 static void
-find_page_change_cb(WebKitWebView  *webview,
-                    WebKitWebFrame *webframe,
-                    FindCBStruct   *findcbs)
-{
-	findcbs->job->results[findcbs->page] = webkit_web_view_mark_text_matches(WEBKIT_WEB_VIEW(webview),
-	                                                                         findcbs->job->text,
-	                                                                         findcbs->job->case_sensitive,
-	                                                                         0);
-	ev_web_view_find_set_highlight_search(EV_WEB_VIEW(webview), TRUE);
-	
-	webkit_web_view_search_text (WEBKIT_WEB_VIEW(webview),
-	                             findcbs->job->text,
-	                             findcbs->job->case_sensitive,
-                                 TRUE,
-                                 FALSE);
+jump_to_find_page (EvWebView *webview, EvWebViewFindDirection direction, gint shift)
+{
+	int n_pages, i;
+
+	n_pages = ev_document_get_n_pages (webview->document);
+
+	for (i = 0; i < n_pages; i++) {
+		int page;
+		
+		if (direction == EV_WEB_VIEW_FIND_NEXT)
+			page = webview->current_page + i;
+		else
+			page = webview->current_page - i;		
+		page += shift;
+		
+		if (page >= n_pages) {
+			page = page - n_pages;
+		} else if (page < 0) 
+			page = page + n_pages;
+
+		if (page == webview->current_page) 
+			jump_to_find_result_on_page(webview,EV_WEB_VIEW_FIND_NEXT);
+
+		if (ev_web_view_find_get_n_results (webview, page) > 0) {
+			webkit_web_view_unmark_text_matches (WEBKIT_WEB_VIEW(webview));
+			ev_document_model_set_page (webview->model, page);
+			webview->search->direction = direction;
+			break;
+		}
+	}
 }
+
 void
-ev_web_view_find_changed(EvWebView *webview, gint page_found_on,EvJobFind *job)
+ev_web_view_find_changed (EvWebView *webview, guint *results, gchar *text,gboolean case_sensitive)
 {
-	if (job->has_results == FALSE)
-		return;
+	webview->search->results = results;
+	webview->search->search_string = g_strdup(text);
+	webview->search->case_sensitive = case_sensitive;
 	
 	if (webview->search->search_jump == TRUE) {
+		jump_to_find_page (webview, EV_WEB_VIEW_FIND_NEXT, 0);
+	} else {
+		jump_to_find_result_on_page(webview,EV_WEB_VIEW_FIND_NEXT);
+	}
+}	
 
-		webview->search->on_result = 1;
-		webview->search->case_sensitive = job->case_sensitive;
-		webview->search->search_string = g_strdup(job->text);
-		webview->search->search_jump = FALSE;
-		
-		if (page_found_on != webview->current_page) {
-			ev_web_view_change_page(webview, page_found_on);
+void
+ev_web_view_find_next (EvWebView *webview)
+{
+	gint n_results;
 
-			FindCBStruct *findstruct = g_new0 (FindCBStruct, 1);
-			findstruct->job = job;
-			findstruct->page = page_found_on;
-			
-			g_signal_connect(WEBKIT_WEB_VIEW(webview),"document-load-finished",G_CALLBACK(find_page_change_cb),findstruct);
-		}
-		else {
-			job->results[webview->current_page] = webkit_web_view_mark_text_matches(WEBKIT_WEB_VIEW(webview),
-			                                                                        job->text,
-			                                                                        job->case_sensitive,
-			                                                                        0);
+	n_results = ev_web_view_find_get_n_results (webview, webview->current_page);
+	webview->search->on_result++;
 
-				ev_web_view_find_set_highlight_search(webview, TRUE);
-		}
+	if (webview->search->on_result >= n_results) {
+		webview->search->on_result = 0;
+		jump_to_find_page (webview, EV_WEB_VIEW_FIND_NEXT, 1);
+	} 
+	else {
+		jump_to_find_result_on_page (webview, EV_WEB_VIEW_FIND_NEXT);
 	}
 }
 
 void
-ev_web_view_find_search_changed(EvWebView *webview)
+ev_web_view_find_previous (EvWebView *webview)
 {
-	ev_web_view_find_set_highlight_search(webview,FALSE);
-	webkit_web_view_unmark_text_matches (WEBKIT_WEB_VIEW(webview));
+	webview->search->on_result--;
+
+	if (webview->search->on_result < 0) {
+		jump_to_find_page (webview, EV_WEB_VIEW_FIND_PREV, -1);
+		webview->search->on_result = MAX (0, ev_web_view_find_get_n_results (webview, webview->current_page) - 1);
+	} else {
+		jump_to_find_result_on_page (webview,EV_WEB_VIEW_FIND_PREV);
+	}
+}
+
+void
+ev_web_view_find_search_changed (EvWebView *webview,gboolean visible)
+{
+	/* search string has changed, focus on new search result */
+	if (visible) {
+		g_signal_connect(webview,
+		                 "notify::load-status",
+		                 G_CALLBACK(jump_to_find_result),
+		                 NULL);
+	}
+	else {
+		g_signal_handlers_disconnect_by_func(webview,
+											 jump_to_find_result,
+											 NULL);		
+	}
+	webkit_web_view_unmark_text_matches(WEBKIT_WEB_VIEW(webview));
+
 	webview->search->search_jump = TRUE;
+	
+	if (webview->search->search_string) {
+		g_free(webview->search->search_string);
+		webview->search->search_string = NULL;
+	}
 }
 
 void
-ev_web_view_find_cancel(EvWebView *webview)
+ev_web_view_find_set_highlight_search (EvWebView *webview, gboolean value)
 {
-	ev_web_view_find_set_highlight_search(webview,FALSE);
-	webkit_web_view_unmark_text_matches (WEBKIT_WEB_VIEW(webview));
+	webkit_web_view_set_highlight_text_matches (WEBKIT_WEB_VIEW(webview),value);
 }
 
 void
-ev_web_view_empty_search(EvWebView *webview)
-{
-	SearchParams *search = webview->search ;
-	search->case_sensitive = FALSE;
-	if (search->search_string) 
-		g_free(search->search_string);
-	search->search_string = NULL;
-	search->search_jump = TRUE ;
+ev_web_view_find_cancel (EvWebView *webview)
+{
+	g_signal_handlers_disconnect_by_func(webview,
+										 jump_to_find_result,
+										 NULL);
+
+	webkit_web_view_unmark_text_matches(WEBKIT_WEB_VIEW(webview));
+	ev_web_view_find_set_highlight_search(webview,FALSE);
 }
 
 /* Selection */
diff --git a/libview/ev-web-view.h b/libview/ev-web-view.h
index aa2d5492..751e7381 100644
--- a/libview/ev-web-view.h
+++ b/libview/ev-web-view.h
@@ -63,11 +63,15 @@ void       ev_web_view_handle_link (EvWebView *webview, EvLink* link);
 /* Searching */
 void     ev_web_view_find_next                 (EvWebView *webview);
 void     ev_web_view_find_previous             (EvWebView *webview);
-void     ev_web_view_find_changed              (EvWebView *webview, gint page_found_on,EvJobFind *job);
-void     ev_web_view_find_search_changed       (EvWebView *webview);
+
+void     ev_web_view_find_changed              (EvWebView *webview,
+                                                guint *results,
+                                                gchar *text, 
+                                                gboolean case_sensitive);
+
+void     ev_web_view_find_search_changed       (EvWebView *webview,gboolean visible);
 void     ev_web_view_find_cancel               (EvWebView *webview);
 void     ev_web_view_find_set_highlight_search (EvWebView *webview,gboolean visible);
-void     ev_web_view_empty_search              (EvWebView *webview);
 
 /* Selection */
 gboolean	ev_web_view_get_has_selection   (EvWebView *webview);
diff --git a/shell/ev-window.c b/shell/ev-window.c
index 43963fec..dfb3d7f2 100644
--- a/shell/ev-window.c
+++ b/shell/ev-window.c
@@ -3668,7 +3668,7 @@ static void
 ev_window_cmd_scroll_forward (GtkAction *action, EvWindow *window)
 {
 	/*If the webview is occupying the window*/
-	if ( window->priv->document->iswebdocument == FALSE) return ;
+	if ( window->priv->document->iswebdocument == TRUE) return ;
 	
 	ev_view_scroll (EV_VIEW (window->priv->view), GTK_SCROLL_PAGE_FORWARD, FALSE);
 }
@@ -3677,7 +3677,7 @@ static void
 ev_window_cmd_scroll_backward (GtkAction *action, EvWindow *window)
 {
 	/*If the webview is occupying the window*/
-	if ( window->priv->document->iswebdocument == FALSE ) return ;
+	if ( window->priv->document->iswebdocument == TRUE ) return ;
 	
 	ev_view_scroll (EV_VIEW (window->priv->view), GTK_SCROLL_PAGE_BACKWARD, FALSE);
 }
@@ -5069,7 +5069,7 @@ ev_window_find_job_updated_cb (EvJobFind *job,
 {
 	ev_window_update_actions (ev_window);
 	if (ev_window->priv->document->iswebdocument == TRUE ) {
-		ev_web_view_find_changed(EV_WEB_VIEW(ev_window->priv->webview), page,job);
+		ev_web_view_find_changed(EV_WEB_VIEW(ev_window->priv->webview), job->results,job->text, job->case_sensitive);
 	}
 	else {
 		ev_view_find_changed (EV_VIEW (ev_window->priv->view),
@@ -5150,7 +5150,7 @@ find_bar_search_changed_cb (EggFindBar *find_bar,
 	search_string = egg_find_bar_get_search_string (find_bar);
 
 	if (ev_window->priv->document->iswebdocument) {
-		ev_web_view_find_search_changed(EV_WEB_VIEW(ev_window->priv->webview));
+		ev_web_view_find_search_changed(EV_WEB_VIEW(ev_window->priv->webview),TRUE);
 	} else {
 		ev_view_find_search_changed (EV_VIEW (ev_window->priv->view));
 	}
@@ -5174,9 +5174,7 @@ find_bar_search_changed_cb (EggFindBar *find_bar,
 		ev_window_update_actions (ev_window);
 		egg_find_bar_set_status_text (EGG_FIND_BAR (ev_window->priv->find_bar),
 					      NULL);
-		if (ev_window->priv->document->iswebdocument == TRUE) {
-			ev_web_view_empty_search(EV_WEB_VIEW(ev_window->priv->webview));
-		} else {
+		if (ev_window->priv->document->iswebdocument == FALSE) {
 			gtk_widget_queue_draw (GTK_WIDGET (ev_window->priv->view));
 		}
 	}
@@ -5199,7 +5197,7 @@ find_bar_visibility_changed_cb (EggFindBar *find_bar,
 		}
 		else {
 			ev_web_view_find_set_highlight_search(EV_WEB_VIEW(ev_window->priv->webview),visible);
-			ev_web_view_find_search_changed(EV_WEB_VIEW(ev_window->priv->webview));
+			ev_web_view_find_search_changed(EV_WEB_VIEW(ev_window->priv->webview),visible);
 		}
 
 		ev_window_update_actions (ev_window);