[tracker-miners/wip/carlosg/album-dates: 918/918] tracker-extract-gstreamer: Make date/time detection smarter

[Sam Thursfield <sthursfield src gnome org>]

commit 5384635182ed8188c22cda15757287afefb95cac
Author: Carlos Garnacho <carlosg gnome org>
Date:   Sun Aug 5 15:00:48 2018 +0200

    tracker-extract-gstreamer: Make date/time detection smarter
    
    There's files that may have both GST_TAG_DATE_TIME and GST_TAG_DATE,
    in some cases both will be "the same" and the former will be just
    a more exact representation.
    
    However for other cases (notably albums ripped in m4a), it is more
    likely that the album release date is contained in GST_TAG_DATE,
    while the other tag is more likely to contain the ripping time, or
    something equally random.
    
    For those cases where both dates are present but different, actually
    prefer GST_TAG_DATE even though it's "less exact" in terms of
    resolution. This also is more in line with how totem and rhythmbox
    handle these fields.
    
    https://gitlab.gnome.org/GNOME/tracker-miners/issues/16
    
    Closes: #16

 src/tracker-extract/tracker-extract-gstreamer.c | 59 ++++++++++++++++++-------
 1 file changed, 44 insertions(+), 15 deletions(-)
---
diff --git a/src/tracker-extract/tracker-extract-gstreamer.c b/src/tracker-extract/tracker-extract-gstreamer.c
index 007940469..e52b983d0 100644
--- a/src/tracker-extract/tracker-extract-gstreamer.c
+++ b/src/tracker-extract/tracker-extract-gstreamer.c
@@ -226,31 +226,56 @@ extract_gst_date_time (gchar       *buf,
 {
        GstDateTime *date_time = NULL;
        GDate *date = NULL;
-       gboolean ret = FALSE;
+       gboolean ret = FALSE, use_datetime = FALSE, use_date = FALSE;
 
        buf[0] = '\0';
 
-       if (gst_tag_list_get_date_time (tag_list, tag_date_time, &date_time)) {
-               gboolean complete;
+       /* See https://gitlab.gnome.org/GNOME/tracker-miners/-/issues/16 for why
+        * this code looks at both GST_TAG_DATE_TIME and GST_TAG_DATE
+        */
+       gst_tag_list_get_date_time (tag_list, tag_date_time, &date_time);
+       gst_tag_list_get_date (tag_list, tag_date, &date);
+
+       if (date_time && date) {
+               if (!gst_date_time_has_year (date_time) ||
+                   !gst_date_time_has_month (date_time) ||
+                   !gst_date_time_has_day (date_time)) {
+                       use_date = TRUE;
+               } else if (gst_date_time_get_year (date_time) == g_date_get_year (date) &&
+                          gst_date_time_get_month (date_time) == g_date_get_month (date) &&
+                          gst_date_time_get_day (date_time) == g_date_get_day (date)) {
+                       /* Dates are equal, prefer datetime as it has more precission */
+                       use_datetime = TRUE;
+               } else {
+                       /* Dates are dissimilar, prefer date alone as it's
+                        * traditionally preferred for "release" dates
+                        */
+                       use_date = TRUE;
+               }
+       } else if (date_time) {
+               use_datetime = TRUE;
+       } else {
+               use_date = TRUE;
+       }
 
-               ret = gst_date_time_has_year (date_time);
-               complete = get_gst_date_time_to_buf (date_time, buf, size);
-               gst_date_time_unref (date_time);
+       if (date_time && use_datetime) {
+               ret = get_gst_date_time_to_buf (date_time, buf, size);
 
-               if (!complete) {
+               if (!ret) {
                        g_debug ("GstDateTime was not complete, parts of the date/time were missing (e.g. 
hours, minutes, seconds)");
+                       use_date = TRUE;
+
+                       /* Return success even if the DATE tag is empty, as we have at
+                        * least some information from the DATE_TIME tag. */
+                       ret = TRUE;
                }
-       } else if (gst_tag_list_get_date (tag_list, tag_date, &date)) {
-               if (date && g_date_valid (date)) {
-                       if (date->julian)
-                               ret = g_date_valid_julian (date->julian_days);
-                       if (date->dmy)
-                               ret = g_date_valid_dmy (date->day, date->month, date->year);
-               }
+       }
 
-               if (ret) {
+       if (date && use_date) {
+               if (g_date_valid (date)) {
                        /* GDate does not carry time zone information, assume UTC */
                        g_date_strftime (buf, size, "%Y-%m-%dT%H:%M:%SZ", date);
+                       ret = TRUE;
                }
        }
 
@@ -258,6 +283,10 @@ extract_gst_date_time (gchar       *buf,
                g_date_free (date);
        }
 
+       if (date_time) {
+               gst_date_time_unref (date_time);
+       }
+
        return ret;
 }