Merge branch '222-update-import' into 'develop'

Resolve "Add flag during import to replace already present tracks with new version"

Closes #222

See merge request funkwhale/funkwhale!264

api/funkwhale_api/music/factories.py

@@ -89,6 +89,7 @@ class ImportJobFactory(factory.django.DjangoModelFactory):
     batch = factory.SubFactory(ImportBatchFactory)
     source = factory.Faker("url")
     mbid = factory.Faker("uuid4")
+    replace_if_duplicate = factory.Faker("boolean")
 
     class Meta:
         model = "music.ImportJob"

api/funkwhale_api/music/migrations/0028_importjob_replace_if_duplicate.py

@@ -0,0 +1,18 @@
+# Generated by Django 2.0.6 on 2018-06-22 13:36
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('music', '0027_auto_20180515_1808'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='importjob',
+            name='replace_if_duplicate',
+            field=models.BooleanField(default=False),
+        ),
+    ]

api/funkwhale_api/music/models.py

@@ -567,6 +567,7 @@ class ImportBatch(models.Model):
 
 class ImportJob(models.Model):
     uuid = models.UUIDField(unique=True, db_index=True, default=uuid.uuid4)
+    replace_if_duplicate = models.BooleanField(default=False)
     batch = models.ForeignKey(
         ImportBatch, related_name="jobs", on_delete=models.CASCADE
     )

api/funkwhale_api/music/tasks.py

@@ -80,10 +80,11 @@ def import_track_from_remote(library_track):
     )[0]
 
 
-def _do_import(import_job, replace=False, use_acoustid=False):
+def _do_import(import_job, use_acoustid=False):
     logger.info("[Import Job %s] starting job", import_job.pk)
     from_file = bool(import_job.audio_file)
     mbid = import_job.mbid
+    replace = import_job.replace_if_duplicate
     acoustid_track_id = None
     duration = None
     track = None
@@ -135,8 +136,8 @@ def _do_import(import_job, replace=False, use_acoustid=False):
 
     track_file = None
     if replace:
-        logger.info("[Import Job %s] replacing existing audio file", import_job.pk)
-        track_file = track.files.first()
+        logger.info("[Import Job %s] deleting existing audio file", import_job.pk)
+        track.files.all().delete()
     elif track.files.count() > 0:
         logger.info(
             "[Import Job %s] skipping, we already have a file for this track",
@@ -163,7 +164,7 @@ def _do_import(import_job, replace=False, use_acoustid=False):
             # no downloading, we hotlink
             pass
     elif not import_job.audio_file and not import_job.source.startswith("file://"):
-        # not an implace import, and we have a source, so let's download it
+        # not an inplace import, and we have a source, so let's download it
         logger.info("[Import Job %s] downloading audio file from remote", import_job.pk)
         track_file.download_file()
     elif not import_job.audio_file and import_job.source.startswith("file://"):
@@ -243,14 +244,14 @@ def get_cover_from_fs(dir_path):
 @celery.require_instance(
     models.ImportJob.objects.filter(status__in=["pending", "errored"]), "import_job"
 )
-def import_job_run(self, import_job, replace=False, use_acoustid=False):
+def import_job_run(self, import_job, use_acoustid=False):
     def mark_errored(exc):
         logger.error("[Import Job %s] Error during import: %s", import_job.pk, str(exc))
         import_job.status = "errored"
         import_job.save(update_fields=["status"])
 
     try:
-        tf = _do_import(import_job, replace, use_acoustid=use_acoustid)
+        tf = _do_import(import_job, use_acoustid=use_acoustid)
         return tf.pk if tf else None
     except Exception as exc:
         if not settings.DEBUG:

api/funkwhale_api/providers/audiofile/management/commands/import_files.py

@@ -55,6 +55,17 @@ class Command(BaseCommand):
                 "import and not much disk space available."
             ),
         )
+        parser.add_argument(
+            "--replace",
+            action="store_true",
+            dest="replace",
+            default=False,
+            help=(
+                "Use this flag to replace duplicates (tracks with same "
+                "musicbrainz mbid, or same artist, album and title) on import "
+                "with their newest version."
+            ),
+        )
         parser.add_argument(
             "--noinput",
             "--no-input",
@@ -112,16 +123,23 @@ class Command(BaseCommand):
                     "No superuser available, please provide a --username"
                 )
 
-        filtered = self.filter_matching(matching)
+        if options["replace"]:
+            filtered = {"initial": matching, "skipped": [], "new": matching}
+            message = "- {} files to be replaced"
+            import_paths = matching
+        else:
+            filtered = self.filter_matching(matching)
+            message = "- {} files already found in database"
+            import_paths = filtered["new"]
+
         self.stdout.write("Import summary:")
         self.stdout.write(
             "- {} files found matching this pattern: {}".format(
                 len(matching), options["path"]
             )
         )
-        self.stdout.write(
-            "- {} files already found in database".format(len(filtered["skipped"]))
-        )
+        self.stdout.write(message.format(len(filtered["skipped"])))
+
         self.stdout.write("- {} new files".format(len(filtered["new"])))
 
         self.stdout.write(
@@ -141,12 +159,12 @@ class Command(BaseCommand):
             if input("".join(message)) != "yes":
                 raise CommandError("Import cancelled.")
 
-        batch, errors = self.do_import(filtered["new"], user=user, options=options)
+        batch, errors = self.do_import(import_paths, user=user, options=options)
         message = "Successfully imported {} tracks"
         if options["async"]:
             message = "Successfully launched import for {} tracks"
 
-        self.stdout.write(message.format(len(filtered["new"])))
+        self.stdout.write(message.format(len(import_paths)))
         if len(errors) > 0:
             self.stderr.write("{} tracks could not be imported:".format(len(errors)))
 
@@ -196,7 +214,9 @@ class Command(BaseCommand):
         return batch, errors
 
     def import_file(self, path, batch, import_handler, options):
-        job = batch.jobs.create(source="file://" + path)
+        job = batch.jobs.create(
+            source="file://" + path, replace_if_duplicate=options["replace"]
+        )
         if not options["in_place"]:
             name = os.path.basename(path)
             with open(path, "rb") as f: