[project @ 2004-03-24 23:51:08 by jmb]

Text Export support.

svn path=/import/netsurf/; revision=665

10 files changed, 177 insertions, 4 deletions

diff --git a/!NetSurf/Resources/en/Messages b/!NetSurf/Resources/en/Messages
index 4cb9aee78..6fa0198ce 100644
--- a/!NetSurf/Resources/en/Messages
+++ b/!NetSurf/Resources/en/Messages
@@ -13,6 +13,7 @@ Save:Save
 Export:Export
 ExportAs:Export as
 Draw:Draw
+Text:Text
 Print:Print...
 Selection:Selection
 Copy:Copy to clipboard
@@ -77,3 +78,4 @@ SelectMenu:Select
 
 SaveSource:Source
 SaveDraw:Webpage
+SaveText:Webpage
diff --git a/!NetSurf/Resources/fr/Messages b/!NetSurf/Resources/fr/Messages
index 275a4af23..e3f32f4d9 100644
--- a/!NetSurf/Resources/fr/Messages
+++ b/!NetSurf/Resources/fr/Messages
@@ -13,6 +13,7 @@ Save:Sauver
 Export:Exporter
 ExportAs:Exporter sous
 Draw:Draw
+Text:Text
 Print:Imprimer...
 Selection:Sélection
 Copy:Copier vers le presse papier
@@ -77,3 +78,4 @@ SelectMenu:Sélection
 
 SaveSource:Source
 SaveDraw:PageWeb
+SaveText:PageWeb
diff --git a/makefile b/makefile
index 25c6b4cff..f5e319e3a 100644
--- a/makefile
+++ b/makefile
@@ -19,7 +19,7 @@ OBJECTS = $(OBJECTS_COMMON) \
 	draw.o gif.o jpeg.o plugin.o png.o sprite.o \
 	about.o filetype.o font.o uri.o url_protocol.o history.o \
 	version.o thumbnail.o \
-	save.o save_complete.o save_draw.o schedule.o
+	save.o save_complete.o save_draw.o save_text.o schedule.o
 OBJECTS_DEBUG = $(OBJECTS_COMMON) \
 	netsurfd.o \
 	options.o filetyped.o fontd.o
diff --git a/riscos/gui.h b/riscos/gui.h
index a0f7c6a19..24f96c21c 100644
--- a/riscos/gui.h
+++ b/riscos/gui.h
@@ -31,7 +31,7 @@ extern bool gui_redraw_debug;
 extern gui_window *current_gui;
 
 typedef enum { GUI_BROWSER_WINDOW, GUI_DOWNLOAD_WINDOW } gui_window_type;
-typedef enum { GUI_SAVE_SOURCE, GUI_SAVE_DRAW } gui_save_type;
+typedef enum { GUI_SAVE_SOURCE, GUI_SAVE_DRAW, GUI_SAVE_TEXT } gui_save_type;
 extern gui_save_type gui_current_save_type;
 typedef enum { GUI_DRAG_SELECTION, GUI_DRAG_DOWNLOAD_SAVE,
 		GUI_DRAG_SAVE } gui_drag_type;
diff --git a/riscos/menus.c b/riscos/menus.c
index dda598b89..0cb8cb288 100644
--- a/riscos/menus.c
+++ b/riscos/menus.c
@@ -48,10 +48,11 @@ wimp_menu *iconbar_menu = (wimp_menu *) & (wimp_MENU(4)) {
 int iconbar_menu_height = 4 * 44;
 
 /* browser window menu structure - based on Style Guide */
-static wimp_MENU(1) export_menu = {
+static wimp_MENU(2) export_menu = {
   { "ExportAs" }, 7,2,7,0, 200, 44, 0,
   {
-    { wimp_MENU_LAST | wimp_MENU_GIVE_WARNING, wimp_NO_SUB_MENU, DEFAULT_FLAGS, { "Draw" } }
+    { wimp_MENU_GIVE_WARNING, wimp_NO_SUB_MENU, DEFAULT_FLAGS, { "Draw" } },
+    { wimp_MENU_LAST | wimp_MENU_GIVE_WARNING, wimp_NO_SUB_MENU, DEFAULT_FLAGS, { "Text" } }
   }
 };
 static wimp_menu *browser_export_menu = (wimp_menu *) &export_menu;
@@ -125,6 +126,7 @@ void ro_gui_menus_init(void)
 	iconbar_menu->entries[0].sub_menu = (wimp_menu *) dialog_info;
 	browser_page_menu->entries[1].sub_menu = (wimp_menu *) dialog_saveas;
 	browser_export_menu->entries[0].sub_menu = (wimp_menu *) dialog_saveas;
+	browser_export_menu->entries[1].sub_menu = (wimp_menu *) dialog_saveas;
 	browser_view_menu->entries[0].sub_menu = (wimp_menu *) dialog_zoom;
 }
 
@@ -302,6 +304,14 @@ void ro_gui_menu_warning(wimp_message_menu_warning *warning)
 					ICON_SAVE_PATH,
 					messages_get("SaveDraw"));
 			break;
+                case 1: /* Export as -> Text */
+			gui_current_save_type = GUI_SAVE_TEXT;
+			ro_gui_set_icon_string(dialog_saveas,
+					ICON_SAVE_ICON, "file_fff");
+			ro_gui_set_icon_string(dialog_saveas,
+					ICON_SAVE_PATH,
+					messages_get("SaveText"));
+			break;
 
 		case -1:
 		default: /* Save */
diff --git a/riscos/save.c b/riscos/save.c
index 33a5351ba..c3773bc71 100644
--- a/riscos/save.c
+++ b/riscos/save.c
@@ -15,6 +15,7 @@
 #include "oslib/wimp.h"
 #include "netsurf/riscos/gui.h"
 #include "netsurf/riscos/save_draw.h"
+#include "netsurf/riscos/save_text.h"
 #include "netsurf/utils/log.h"
 #include "netsurf/utils/messages.h"
 #include "netsurf/utils/utils.h"
@@ -124,6 +125,12 @@ void ro_gui_save_datasave_ack(wimp_message *message)
 				return;
 			save_as_draw(c, path);
 			break;
+
+		case GUI_SAVE_TEXT:
+			if (!c)
+				return;
+			save_as_text(c, path);
+			break;
 	}
 
 	wimp_create_menu(wimp_CLOSE_MENU, 0, 0);
diff --git a/riscos/save_text.c b/riscos/save_text.c
new file mode 100644
index 000000000..c9a3ed2b8
--- /dev/null
+++ b/riscos/save_text.c
@@ -0,0 +1,130 @@
+/*
+ * This file is part of NetSurf, http://netsurf.sourceforge.net/
+ * Licensed under the GNU General Public License,
+ *                http://www.opensource.org/licenses/gpl-license
+ * Copyright 2004 John M Bell <jmb202@ecs.soton.ac.uk>
+ */
+
+#include <stdbool.h>
+#include <string.h>
+
+#include "libxml/HTMLtree.h"
+
+#include "oslib/osfile.h"
+
+#include "netsurf/utils/config.h"
+#include "netsurf/content/content.h"
+#include "netsurf/riscos/save_text.h"
+#include "netsurf/utils/log.h"
+#include "netsurf/utils/utils.h"
+
+#ifdef WITH_TEXT_EXPORT
+
+static void extract_text(xmlDoc *doc);
+static void extract_text_from_tree(xmlNode *n);
+
+static char *buffer = 0;
+static int output_size = 0;
+
+void save_as_text(struct content *c, char *path) {
+
+        htmlParserCtxtPtr toSave;
+
+	if (c->type != CONTENT_HTML) {
+		return;
+	}
+
+	/* allocate a buffer the same size as the source
+	 * the output is guaranteed to be less than this
+	 */
+	buffer = xcalloc(c->source_size, sizeof(char));
+
+	toSave = htmlCreateMemoryParserCtxt(c->source_data, c->source_size);
+	htmlParseDocument(toSave);
+
+	extract_text(toSave->myDoc);
+
+	if (output_size > 0) {
+        	xosfile_save_stamped(path, 0xfff, (byte*)buffer,
+	                             (byte*)buffer+output_size);
+	}
+
+	xmlFreeDoc(toSave->myDoc);
+	htmlFreeParserCtxt(toSave);
+	xfree(buffer);
+}
+
+void extract_text(xmlDoc *doc)
+{
+        xmlNode *html;
+
+        /* 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;
+        }
+
+        extract_text_from_tree(html);
+}
+
+void extract_text_from_tree(xmlNode *n)
+{
+        xmlNode *this;
+        char *text;
+        int len = 0;
+        int need_nl = 0;
+
+        if (n->type == XML_ELEMENT_NODE) {
+                if (strcmp(n->name, "dl") == 0 ||
+                    strcmp(n->name, "h1") == 0 ||
+                    strcmp(n->name, "h2") == 0 ||
+                    strcmp(n->name, "h3") == 0 ||
+                    strcmp(n->name, "ol") == 0 ||
+                    strcmp(n->name, "title") == 0 ||
+                    strcmp(n->name, "ul") == 0) {
+                        need_nl = 2;
+                }
+                else if (strcmp(n->name, "applet") == 0 ||
+                         strcmp(n->name, "br") == 0 ||
+                         strcmp(n->name, "div") == 0 ||
+                         strcmp(n->name, "dt") == 0 ||
+                         strcmp(n->name, "h4") == 0 ||
+                         strcmp(n->name, "h5") == 0 ||
+                         strcmp(n->name, "h6") == 0 ||
+                         strcmp(n->name, "li") == 0 ||
+                         strcmp(n->name, "object") == 0 ||
+                         strcmp(n->name, "p") == 0 ||
+                         strcmp(n->name, "tr") == 0) {
+                        need_nl = 1;
+                }
+                /* do nothing, we just recurse through these nodes */
+        }
+        else if (n->type == XML_TEXT_NODE) {
+                text = squash_tolat1(n->content);
+                len = strlen(text);
+                strcat(buffer, text);
+                output_size += len;
+                xfree(text);
+                return;
+        }
+        else {
+                return;
+        }
+
+        /* now recurse */
+	for (this = n->children; this != 0; this = this->next) {
+                extract_text_from_tree(this);
+	}
+
+	if (need_nl) {
+	        for (len = 0; len != need_nl; len++) {
+	                strcat(buffer, "\n");
+                        output_size += 1;
+                }
+	}
+}
+
+#endif
diff --git a/riscos/save_text.h b/riscos/save_text.h
new file mode 100644
index 000000000..2ed39d97f
--- /dev/null
+++ b/riscos/save_text.h
@@ -0,0 +1,15 @@
+/*
+ * This file is part of NetSurf, http://netsurf.sourceforge.net/
+ * Licensed under the GNU General Public License,
+ *                http://www.opensource.org/licenses/gpl-license
+ * Copyright 2004 John M Bell <jmb202@ecs.soton.ac.uk>
+ */
+
+#ifndef _NETSURF_RISCOS_SAVE_TEXT_H_
+#define _NETSURF_RISCOS_SAVE_TEXT_H_
+
+struct content;
+
+void save_as_text(struct content *c, char *path);
+
+#endif
diff --git a/riscos/window.c b/riscos/window.c
index a10114d6c..270fcd52b 100644
--- a/riscos/window.c
+++ b/riscos/window.c
@@ -24,6 +24,7 @@
 #include "netsurf/riscos/options.h"
 #include "netsurf/riscos/save_complete.h"
 #include "netsurf/riscos/save_draw.h"
+#include "netsurf/riscos/save_text.h"
 #include "netsurf/riscos/theme.h"
 #include "netsurf/riscos/thumbnail.h"
 #include "netsurf/utils/log.h"
@@ -785,6 +786,11 @@ bool ro_gui_window_keypress(gui_window *g, int key, bool toolbar)
 #endif
 			);
 			return true;
+#ifdef WITH_TEXT_EXPORT
+                case wimp_KEY_CONTROL + wimp_KEY_F3:
+/*                        save_as_text(g->data.browser.bw->current_content);*/
+                        return true;
+#endif
 #ifdef WITH_SAVE_COMPLETE
                 case wimp_KEY_SHIFT + wimp_KEY_F3:
                         save_complete(g->data.browser.bw->current_content);
diff --git a/utils/config.h b/utils/config.h
index 3520df52b..1211b606f 100644
--- a/utils/config.h
+++ b/utils/config.h
@@ -48,6 +48,7 @@
 /* Export modules */
 #define WITH_SAVE_COMPLETE
 #define WITH_DRAW_EXPORT
+#define WITH_TEXT_EXPORT
 
 #endif