[gnumeric] Clipboard: Skip MS HTML clipboard headers.

[Morten Welinder <mortenw src gnome org>]

commit 5f5041601b8d7760de18864c876a004f96b5871b
Author: Morten Welinder <terra gnome org>
Date:   Fri Jun 25 09:29:33 2010 -0400

    Clipboard: Skip MS HTML clipboard headers.

 ChangeLog           |    5 +++++
 NEWS                |    1 +
 src/gui-clipboard.c |   28 ++++++++++++++++++++++++++--
 3 files changed, 32 insertions(+), 2 deletions(-)
---
diff --git a/ChangeLog b/ChangeLog
index d59f599..b6e1c0b 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,8 @@
+2010-06-25  Morten Welinder  <terra gnome org>
+
+	* src/gui-clipboard.c (table_content_received): Skip headers for
+	MS HTML Format.
+
 2010-06-25  Andreas J. Guelzow <aguelzow pyrshep ca>
 
 	* src/Makefile.am (undo.[ch]): added
diff --git a/NEWS b/NEWS
index 896b9f2..4b84365 100644
--- a/NEWS
+++ b/NEWS
@@ -13,6 +13,7 @@ Morten:
 	* Improve function syntax tooltips.
 	* Fix excessive memory use problem with xlsx.  [#622365]
 	* Make BackSpace work as Delete on pane.  [#622599]
+	* Improve handling of MS-style HTML in clipboard.
 
 --------------------------------------------------------------------------
 Gnumeric 1.10.6
diff --git a/src/gui-clipboard.c b/src/gui-clipboard.c
index e7c5b22..b57dd73 100644
--- a/src/gui-clipboard.c
+++ b/src/gui-clipboard.c
@@ -415,9 +415,33 @@ table_content_received (GtkClipboard *clipboard, GtkSelectionData *sel,
 						 sel->length);
 	} else if (sel->target == gdk_atom_intern (HTML_ATOM_NAME_UNIX, FALSE) ||
 		   sel->target == gdk_atom_intern (HTML_ATOM_NAME_WINDOWS, FALSE)) {
+		char *data = sel->data;
+		size_t length = sel->length;
+
+		if (sel->target == gdk_atom_intern (HTML_ATOM_NAME_WINDOWS, FALSE)) {
+			/* See bug 143084 */
+			size_t i = 0;
+			while (i + 9 < length) {
+				if (memcmp (data + i, "<!DOCTYPE", 9) == 0) {
+					if (debug_clipboard ())
+						g_printerr ("Skipping %d bytes of headers.\n", i);
+					data += i;
+					length -= i;
+					break;
+				}
+				while (i < length) {
+					if (data[i] == '\n' || data[i] == '\r') {
+						while (i < length && g_ascii_isspace (data[i]))
+							i++;
+						break;
+					}
+					i++;
+				}
+			}
+		}
+
 		content = table_cellregion_read (wbc, "Gnumeric_html:html",
-						 pt, sel->data,
-						 sel->length);
+						 pt, data, length);
 	} else if ((sel->target == gdk_atom_intern ( BIFF8_ATOM_NAME, FALSE)) ||
 		   (sel->target == gdk_atom_intern ( BIFF8_ATOM_NAME_CITRIX, FALSE)) ||
 		   (sel->target == gdk_atom_intern ( BIFF5_ATOM_NAME, FALSE)) ||