[rygel] media-export: Deduce DVD title from file name

[Jens Georg <jensgeorg src gnome org>]

commit e790df5b0da584c096f4d45ca5364a91d613fd12
Author: Jens Georg <mail jensge org>
Date:   Sat Feb 13 17:10:20 2016 +0100

    media-export: Deduce DVD title from file name
    
    Do not use information from volume label, it is apparently more often
    broken than helpful.
    
    Signed-off-by: Jens Georg <mail jensge org>

 .../rygel-media-export-dvd-container.vala          |    4 +-
 .../rygel-media-export-item-factory.vala           |  122 +++++++++++++++++---
 2 files changed, 108 insertions(+), 18 deletions(-)
---
diff --git a/src/plugins/media-export/rygel-media-export-dvd-container.vala 
b/src/plugins/media-export/rygel-media-export-dvd-container.vala
index af1c78a..6c05896 100644
--- a/src/plugins/media-export/rygel-media-export-dvd-container.vala
+++ b/src/plugins/media-export/rygel-media-export-dvd-container.vala
@@ -55,9 +55,7 @@ internal class Rygel.MediaExport.DVDContainer : SimpleContainer, UpdatableObject
         var child_count = 0;
 
         while (it != null) {
-            if (it->name == "title") {
-                this.title = it->children->content;
-            } else if (it->name == "track") {
+            if (it->name == "track") {
                 child_count++;
             }
 
diff --git a/src/plugins/media-export/rygel-media-export-item-factory.vala 
b/src/plugins/media-export/rygel-media-export-item-factory.vala
index f04a83b..2520203 100644
--- a/src/plugins/media-export/rygel-media-export-item-factory.vala
+++ b/src/plugins/media-export/rygel-media-export-item-factory.vala
@@ -38,6 +38,27 @@ namespace Rygel.MediaExport.ItemFactory {
         MISMATCH
     }
 
+    private const string INVALID_CHARS = "()[]<>{}! #$^&*+=|\\/\"'?~";
+    private const string CONVERT_CHARS = "\t_\\.";
+    private const string BLOCK_PATTERN = "%s[^%s]*%s";
+    private const string[] BLOCKS = { "()", "{}", "[]", "<>" };
+    private const string[] BLACKLIST = {
+        "720p", "1080p", "x264", "ws", "proper", "real.repack", "repack",
+        "hdtv", "pdtv", "notv", "dsr", "DVDRip", "divx", "xvid"
+    };
+
+    private const string[] VIDEO_SUFFIXES = {
+        "webm", "mkv", "flv", "ogv", "ogg", "avi", "mov", "wmv", "mp4",
+        "m4v", "mpeg", "mpg", "iso"
+    };
+
+    private static Regex char_remove_regex;
+    private static Regex char_convert_regex;
+    private static Regex space_compress_regex;
+    private static Regex[] block_regexes;
+    private static Regex[] blacklist_regexes;
+    private static Regex[] video_suffix_regexes;
+
     private static bool check_variant_type (Variant v,
                                             string typestring) throws Error {
         if (!v.is_of_type (new VariantType (typestring))) {
@@ -173,6 +194,12 @@ namespace Rygel.MediaExport.ItemFactory {
             case DVDContainer.UPNP_CLASS:
                 object = new DVDContainer ("dvd:" + id, parent, "", file.get_path ());
                 object.add_uri (file.get_uri ());
+
+                if (file_info != null) {
+                    apply_file_info (object, file_info);
+                    object.title =  strip_invalid_entities (object.title);
+                }
+
                 return object;
             default:
                 return null;
@@ -184,14 +211,14 @@ namespace Rygel.MediaExport.ItemFactory {
             apply_dlna_profile (item, dlna_profile);
         }
 
-        if (file_info != null) {
-            apply_file_info (item, file_info);
-        }
-
         if (info != null) {
             apply_info (item, info);
         }
 
+        if (file_info != null) {
+            apply_file_info (item, file_info);
+        }
+
         if (audio_info != null) {
             apply_audio_info (item, audio_info);
         }
@@ -306,7 +333,7 @@ namespace Rygel.MediaExport.ItemFactory {
         item.mime_type = it.next_value ().dup_string ();
     }
 
-    private static void apply_file_info (MediaFileItem item, Variant v)
+    private static void apply_file_info (MediaObject object, Variant v)
                                          throws Error {
         ItemFactory.check_variant_type (v, "(sstt)");
 
@@ -314,20 +341,85 @@ namespace Rygel.MediaExport.ItemFactory {
 
         Variant display_name;
         display_name = it.next_value ();
-        if (item.title == null || item.title == "") {
-            item.title = display_name.dup_string ();
+        if (object.title == null || object.title == "") {
+            object.title = display_name.dup_string ();
+        }
+
+        if (object is MediaFileItem) {
+            var item = object as MediaFileItem;
+
+            var mime = it.next_value ();
+            if (item.mime_type == null) {
+                item.mime_type = mime.dup_string ();
+            }
+
+            item.modified = (int64) it.next_value ().get_uint64 ();
+            if (item.date == null) {
+                TimeVal tv = { (long) item.modified, 0 };
+                item.date = tv.to_iso8601 ();
+            }
+            item.size = (int64) it.next_value ().get_uint64 ();
         }
+    }
 
-        var mime = it.next_value ();
-        if (item.mime_type == null) {
-            item.mime_type = mime.dup_string ();
+    private string strip_invalid_entities (string original) {
+        if (char_remove_regex == null) {
+            try {
+                var regex_string = Regex.escape_string (INVALID_CHARS);
+                char_remove_regex = new Regex ("[%s]".printf (regex_string));
+                regex_string = Regex.escape_string (CONVERT_CHARS);
+                char_convert_regex = new Regex ("[%s]".printf (regex_string));
+                space_compress_regex = new Regex ("\\s+");
+                block_regexes = new Regex[0];
+
+                foreach (var block in BLOCKS) {
+                    var block_re = BLOCK_PATTERN.printf (
+                                      Regex.escape_string ("%C".printf (block[0])),
+                                      Regex.escape_string ("%C".printf (block[1])),
+                                      Regex.escape_string ("%C".printf (block[1])));
+                    block_regexes += new Regex (block_re);
+                }
+
+                foreach (var blacklist in BLACKLIST) {
+                    blacklist_regexes += new Regex (Regex.escape_string
+                                                    (blacklist));
+                }
+
+                foreach (var suffix in VIDEO_SUFFIXES) {
+                    video_suffix_regexes += new Regex (Regex.escape_string
+                                                        (suffix));
+                }
+            } catch (RegexError error) {
+                assert_not_reached ();
+            }
         }
 
-        item.modified = (int64) it.next_value ().get_uint64 ();
-        if (item.date == null) {
-            TimeVal tv = { (long) item.modified, 0 };
-            item.date = tv.to_iso8601 ();
+        string p;
+
+        p = original;
+
+        try {
+            foreach (var re in blacklist_regexes) {
+                p = re.replace_literal (p, -1, 0, "");
+            }
+
+            foreach (var re in video_suffix_regexes) {
+                p = re.replace_literal (p, -1, 0, "");
+            }
+
+            foreach (var re in block_regexes) {
+                p = re.replace_literal (p, -1, 0, "");
+            }
+
+            p = char_remove_regex.replace_literal (p, -1, 0, "");
+            p = char_convert_regex.replace_literal (p, -1, 0, " ");
+            p = space_compress_regex.replace_literal (p, -1, 0, " ");
+
+            p._strip ();
+
+            return p;
+        } catch (RegexError error) {
+            assert_not_reached ();
         }
-        item.size = (int64) it.next_value ().get_uint64 ();
     }
 }