Move dimming callback to ModestWindow
[modest] / src / modest-formatter.c
index 8eb4c75..c496ea9 100644 (file)
 #include <tny-simple-list.h>
 #include <tny-gtk-text-buffer-stream.h>
 #include <tny-camel-mem-stream.h>
+#include <tny-camel-html-to-text-stream.h>
 #include "modest-formatter.h"
 #include "modest-text-utils.h"
 #include "modest-tny-platform-factory.h"
 #include <modest-runtime.h>
 
+#define LINE_WRAP 78
+#define MAX_BODY_LINES 1024
+#define MAX_BODY_LENGTH 1024*128
+
 typedef struct _ModestFormatterPrivate ModestFormatterPrivate;
 struct _ModestFormatterPrivate {
        gchar *content_type;
@@ -63,23 +68,191 @@ static gchar*  modest_formatter_wrapper_inline (ModestFormatter *self, const gch
 
 static TnyMimePart *find_body_parent (TnyMimePart *part);
 
+static guint
+count_end_tag_lines (const gchar *haystack, const gchar *needle)
+{
+       gchar *tmp;
+       guint lines = 0;
+
+       tmp = g_strstr_len (haystack, g_utf8_strlen (haystack, -1), ">\n");
+       while (tmp && (tmp <= needle)) {
+               lines++;
+               tmp += 2;
+               tmp = g_strstr_len (tmp, g_utf8_strlen (tmp, -1), ">\n");
+       }
+
+       return lines;
+}
+
 static gchar *
 extract_text (ModestFormatter *self, TnyMimePart *body)
 {
+       TnyStream *mp_stream;
        TnyStream *stream;
+       TnyStream *input_stream;
        GtkTextBuffer *buf;
        GtkTextIter start, end;
-       gchar *text, *converted_text;
+       gchar *text;
        ModestFormatterPrivate *priv;
+       gint total, lines, total_lines, line_chars;
+       gboolean is_html;
 
        buf = gtk_text_buffer_new (NULL);
        stream = TNY_STREAM (tny_gtk_text_buffer_stream_new (buf));
        tny_stream_reset (stream);
-       tny_mime_part_decode_to_stream (body, stream, NULL);
+       mp_stream = tny_mime_part_get_decoded_stream (body);
+
+       is_html = (g_strcmp0 (tny_mime_part_get_content_type (body), "text/html") == 0);
+       if (is_html) {
+               input_stream = tny_camel_html_to_text_stream_new (mp_stream);
+       } else {
+               input_stream = g_object_ref (mp_stream);
+       }
+
+       total = 0;
+       total_lines = 0;
+       line_chars = 0;
+       lines = 0;
+
+       /* For pure HTML emails tny_camel_html_to_text_stream inserts
+          a \n for every ">\n" found in the email including the HTML
+          headers (<html>, <head> ...). For that reason we need to
+          remove them from the resulting text as it is artificially
+          added by the stream */
+       if (is_html) {
+               const guint BUFFER_SIZE = 1024;
+               TnyStream *is;
+               gboolean look_for_end_tag, found;
+               gchar buffer [BUFFER_SIZE + 1];
+               gchar *needle;
+
+               is = g_object_ref (mp_stream);
+               look_for_end_tag = FALSE;
+               found = FALSE;
+
+               /* This algorithm does not work if the body tag is
+                  spread along 2 different stream reads. But there
+                  are not a lot of changes for this to happen as the
+                  buffer size is big enough in most situations. In
+                  the worst case, when it's not found we just accept
+                  the original translation with the extra "\n" */
+               while (!tny_stream_is_eos (is) && !found) {
+                       gint n_read;
+
+                       needle = NULL;
+                       memset (buffer, 0, BUFFER_SIZE);
+                       n_read = tny_stream_read (is, buffer, BUFFER_SIZE);
+
+                       if (G_UNLIKELY (n_read < 0))
+                               break;
+
+                       buffer[n_read] = '\0';
+
+                       /* If we found body,then look for the end of the tag */
+                       if (look_for_end_tag) {
+                               needle = strchr (buffer, '>');
+
+                               if (needle) {
+                                       found = TRUE;
+                                       lines += count_end_tag_lines (buffer, needle);
+                                       break;
+                               }
+                       } else {
+                               gchar *closing;
+
+                               /* Try to find the <body> tag. There
+                                  is no other HTML tag starting by
+                                  "bo", and we can detect more cases
+                                  were <body> tag falls into two
+                                  different stream reads */
+                               needle = g_strstr_len (buffer, n_read, "<bo");
+
+                               if (needle)
+                                       look_for_end_tag = TRUE;
+                               else
+                                       needle = &(buffer[n_read]);
+
+                               lines += count_end_tag_lines (buffer, needle);
+
+                               closing = strchr (needle, '>');
+                               if (closing) {
+                                       if (*(closing + 1) == '\n')
+                                               lines++;
+                                       found = TRUE;
+                                       break;
+                               }
+                       }
+               }
+               if (!found)
+                       lines = 0;
+               tny_stream_reset (is);
+
+               g_object_unref (is);
+       }
+
+       while (!tny_stream_is_eos (input_stream)) {
+               gchar buffer [128];
+               gchar *offset;
+               gint n_read;
+               gint next_read;
+
+               next_read = MIN (128, MAX_BODY_LENGTH - total);
+               if (next_read == 0)
+                       break;
+               n_read = tny_stream_read (input_stream, buffer, next_read);
+
+               if (G_UNLIKELY (n_read < 0))
+                       break;
+
+               offset = buffer;
+               while (offset < buffer + n_read) {
+
+                       if (*offset == '\n') {
+                               total_lines ++;
+                               line_chars = 0;
+                       } else {
+                               line_chars ++;
+                               if (line_chars >= LINE_WRAP) {
+                                       total_lines ++;
+                                       line_chars = 0;
+                               }
+                       }
+                       if (total_lines >= MAX_BODY_LINES)
+                               break;
+                       offset++;
+               }
+
+               if (offset - buffer > 0) {
+                       gint n_write = 0, to_write = 0;
+                       gchar *buffer_ptr;
+
+                       /* Discard lines artificially inserted by
+                          Camel when translating from HTML to text */
+                       buffer_ptr = buffer;
+                       if (lines) {
+                               int i;
+                               for (i=0; i < lines; i++) {
+                                       buffer_ptr = strchr (buffer_ptr, '\n');
+                                       buffer_ptr++;
+                               }
+                       }
+                       to_write = offset - buffer_ptr;
+                       n_write = tny_stream_write (stream, buffer_ptr, to_write);
+                       total += n_write;
+               } else if (n_read == -1) {
+                       break;
+               }
+
+               if (total_lines >= MAX_BODY_LINES)
+                       break;
+       }
+
        tny_stream_reset (stream);
 
        g_object_unref (G_OBJECT(stream));
-       
+       g_object_unref (G_OBJECT (mp_stream));
+       g_object_unref (G_OBJECT (input_stream));
+
        gtk_text_buffer_get_bounds (buf, &start, &end);
        text = gtk_text_buffer_get_text (buf, &start, &end, FALSE);
        g_object_unref (G_OBJECT(buf));
@@ -87,15 +260,6 @@ extract_text (ModestFormatter *self, TnyMimePart *body)
        /* Convert to desired content type if needed */
        priv = MODEST_FORMATTER_GET_PRIVATE (self);
 
-       if (strcmp (tny_mime_part_get_content_type (body), priv->content_type) == 0) {
-               if (!strcmp (priv->content_type, "text/html"))
-                       converted_text = modest_text_utils_convert_to_html  (text);
-               else
-                       converted_text = g_strdup (text);
-
-               g_free (text);
-               text = converted_text;
-       }
        return text;
 }
 
@@ -145,7 +309,7 @@ modest_formatter_do (ModestFormatter *self, TnyMimePart *body, TnyHeader *header
        priv = MODEST_FORMATTER_GET_PRIVATE (self);
        construct_from_text (TNY_MIME_PART (body_part), (const gchar*) txt, priv->content_type);
        g_object_unref (body_part);
-       
+
        /* Clean */
        g_free (body_text);
        g_free (txt);
@@ -177,6 +341,7 @@ modest_formatter_attach (ModestFormatter *self, TnyMsg *msg, TnyHeader *header)
        TnyMsg *new_msg = NULL;
        TnyMimePart *body_part = NULL;
        ModestFormatterPrivate *priv;
+       gchar *txt;
 
        /* Build new part */
        new_msg     = modest_formatter_create_message (self, TRUE, TRUE, FALSE);
@@ -184,7 +349,10 @@ modest_formatter_attach (ModestFormatter *self, TnyMsg *msg, TnyHeader *header)
 
        /* Create the two parts */
        priv = MODEST_FORMATTER_GET_PRIVATE (self);
-       construct_from_text (body_part, "", priv->content_type);
+       txt = modest_text_utils_cite ("", priv->content_type, priv->signature,
+                                     NULL, tny_header_get_date_sent (header));
+       construct_from_text (body_part, txt, priv->content_type);
+       g_free (txt);
        g_object_unref (body_part);
 
        if (msg) {
@@ -216,6 +384,7 @@ modest_formatter_instance_init (GTypeInstance *instance, gpointer g_class)
        ModestFormatterPrivate *priv = MODEST_FORMATTER_GET_PRIVATE (self);
 
        priv->content_type = NULL;
+       priv->signature = NULL;
 }
 
 static void
@@ -323,7 +492,7 @@ modest_formatter_wrapper_quote (ModestFormatter *self, const gchar *text, TnyHea
        GList *node = NULL;
        gchar *result = NULL;
        gchar *from;
-       
+
        /* First we need a GList of attachments filenames */
        for (node = attachments; node != NULL; node = g_list_next (node)) {
                TnyMimePart *part = (TnyMimePart *) node->data;
@@ -338,12 +507,13 @@ modest_formatter_wrapper_quote (ModestFormatter *self, const gchar *text, TnyHea
                        g_object_unref (header);
                } else {
                        filename = g_strdup (tny_mime_part_get_filename (part));
-                       if ((filename == NULL)||(filename[0] == '\0'))
+                       if ((filename == NULL)||(filename[0] == '\0')) {
+                               g_free (filename);
                                filename = g_strdup ("");
+                       }
                }
-               filenames = g_list_append (filenames, filename);
+               filenames = g_list_prepend (filenames, filename);
        }
-       filenames = g_list_reverse (filenames);
 
        /* TODO: get 80 from the configuration */
        from = tny_header_dup_from (header);
@@ -414,7 +584,7 @@ find_body_parent (TnyMimePart *part)
        msg_content_type = tny_mime_part_get_content_type (part);
 
        if ((msg_content_type != NULL) &&
-           (!g_strcasecmp (msg_content_type, "multipart/alternative")))
+           (!g_ascii_strcasecmp (msg_content_type, "multipart/alternative")))
                return g_object_ref (part);
        else if ((msg_content_type != NULL) &&
                 (g_str_has_prefix (msg_content_type, "multipart/"))) {
@@ -427,10 +597,10 @@ find_body_parent (TnyMimePart *part)
 
                while (!tny_iterator_is_done (iter)) {
                        TnyMimePart *part = TNY_MIME_PART (tny_iterator_get_current (iter));
-                       if (part && !g_strcasecmp(tny_mime_part_get_content_type (part), "multipart/alternative")) {
+                       if (part && !g_ascii_strcasecmp(tny_mime_part_get_content_type (part), "multipart/alternative")) {
                                alternative_part = part;
                                break;
-                       } else if (part && !g_strcasecmp (tny_mime_part_get_content_type (part), "multipart/related")) {
+                       } else if (part && !g_ascii_strcasecmp (tny_mime_part_get_content_type (part), "multipart/related")) {
                                related_part = part;
                                break;
                        }