[geary] Stricter regex for finding HTML tags

[Robert Schroll <rschroll src gnome org>]

commit 3d14719aa0208e6c552146f0467005577deffb76
Author: Robert Schroll <rschroll gmail com>
Date:   Tue Feb 3 00:19:12 2015 -0500

    Stricter regex for finding HTML tags
    
    This function is moved from the client to the engine, since it is
    somewhat specialized.
    
    https://bugzilla.gnome.org/show_bug.cgi?id=743898

 src/client/accounts/add-edit-page.vala   |    2 +-
 src/client/composer/composer-widget.vala |    4 ++--
 src/client/util/util-webkit.vala         |   17 +++++++++++++++++
 src/engine/util/util-html.vala           |   15 ---------------
 4 files changed, 20 insertions(+), 18 deletions(-)
---
diff --git a/src/client/accounts/add-edit-page.vala b/src/client/accounts/add-edit-page.vala
index a78d5df..650098a 100644
--- a/src/client/accounts/add-edit-page.vala
+++ b/src/client/accounts/add-edit-page.vala
@@ -581,7 +581,7 @@ public class AddEditPage : Gtk.Box {
     
     private void on_signature_stack_changed() {
         if (signature_stack.visible_child_name == "preview_window")
-            preview_webview.load_html_string(Geary.HTML.smart_escape(email_signature, true), "");
+            preview_webview.load_html_string(smart_escape(email_signature, true), "");
     }
 
     private uint16 get_default_smtp_port() {
diff --git a/src/client/composer/composer-widget.vala b/src/client/composer/composer-widget.vala
index 459557e..9626b2f 100644
--- a/src/client/composer/composer-widget.vala
+++ b/src/client/composer/composer-widget.vala
@@ -993,7 +993,7 @@ public class ComposerWidget : Gtk.EventBox {
                     set_cursor();
                     return;
                 }
-                signature = Geary.HTML.smart_escape(signature, false);
+                signature = smart_escape(signature, false);
             } catch (Error error) {
                 debug("Error reading signature file %s: %s", signature_file.get_path(), error.message);
                 set_cursor();
@@ -1005,7 +1005,7 @@ public class ComposerWidget : Gtk.EventBox {
                 set_cursor();
                 return;
             }
-            signature = Geary.HTML.smart_escape(signature, true);
+            signature = smart_escape(signature, true);
         }
         
         if (body_html == null)
diff --git a/src/client/util/util-webkit.vala b/src/client/util/util-webkit.vala
index d870388..fd06773 100644
--- a/src/client/util/util-webkit.vala
+++ b/src/client/util/util-webkit.vala
@@ -467,3 +467,20 @@ public bool dissasemble_data_uri(string uri, out Geary.Memory.Buffer? buffer) {
     return true;
 }
 
+// Escape reserved HTML entities if the string does not have HTML tags.  If there are no tags,
+// or if preserve_whitespace_in_html is true, wrap the string a div to preserve whitespace.
+public string smart_escape(string? text, bool preserve_whitespace_in_html) {
+    if (text == null)
+        return text;
+    
+    string res = text;
+    if (!Regex.match_simple("<([A-Z]*)(?: [^>]*)?>.*</(\\1)>|<[A-Z]*(?: [^>]*)?/>", res,
+        RegexCompileFlags.CASELESS)) {
+        res = Geary.HTML.escape_markup(res);
+        preserve_whitespace_in_html = true;
+    }
+    if (preserve_whitespace_in_html)
+        res = @"<div style='white-space: pre;'>$res</div>";
+    return res;
+}
+
diff --git a/src/engine/util/util-html.vala b/src/engine/util/util-html.vala
index 561c9d9..ce34ec5 100644
--- a/src/engine/util/util-html.vala
+++ b/src/engine/util/util-html.vala
@@ -88,21 +88,6 @@ public string preserve_whitespace(string? text) {
     return output;
 }
 
-public string smart_escape(string? text, bool preserve_whitespace_in_html) {
-    if (text == null)
-        return text;
-    
-    string res = text;
-    if (!Regex.match_simple("<([A-Z]*)[^>]*>.*</(\\1)>|<[^>]*/>", res,
-        RegexCompileFlags.CASELESS)) {
-        res = escape_markup(res);
-        preserve_whitespace_in_html = true;
-    }
-    if (preserve_whitespace_in_html)
-        res = @"<div style='white-space: pre;'>$res</div>";
-    return res;
-}
-
 // Removes any text between < and >.  Additionally, if input terminates in the middle of a tag, 
 // the tag will be removed.
 // If the HTML is invalid, the original string will be returned.