From 19ce0ab24652c17e81d2f31847470991186c2f71 Mon Sep 17 00:00:00 2001
From: Brannon King <countprimes@gmail.com>
Date: Tue, 11 Feb 2020 12:23:19 -0700
Subject: [PATCH] ogg -> ogv, ignore files that aren't video

---
 lbry/extras/daemon/daemon.py                |  5 ++++-
 lbry/file_analysis.py                       | 21 ++++++++++-----------
 lbry/schema/mime_types.py                   |  2 ++
 scripts/check_video.py                      |  4 ++--
 tests/integration/other/test_transcoding.py |  4 ++--
 5 files changed, 20 insertions(+), 16 deletions(-)

diff --git a/lbry/extras/daemon/daemon.py b/lbry/extras/daemon/daemon.py
index 256a8d812..6552f5eb0 100644
--- a/lbry/extras/daemon/daemon.py
+++ b/lbry/extras/daemon/daemon.py
@@ -3142,7 +3142,10 @@ class Daemon(metaclass=JSONRPCServerType):
                     f"Use --allow-duplicate-name flag to override."
                 )
 
-        file_path = await self._video_file_analyzer.verify_or_repair(validate_file, optimize_file, file_path)
+        try:
+            file_path = await self._video_file_analyzer.verify_or_repair(validate_file, optimize_file, file_path)
+        except ValueError:
+            pass  # it's not a video file
 
         claim = Claim()
         claim.stream.update(file_path=file_path, sd_hash='0' * 96, **kwargs)
diff --git a/lbry/file_analysis.py b/lbry/file_analysis.py
index 59abafa1c..c2106a896 100644
--- a/lbry/file_analysis.py
+++ b/lbry/file_analysis.py
@@ -257,12 +257,12 @@ class VideoFileAnalyzer:
                     continue
                 codec = stream["codec_name"].split(",")
                 if "theora" in codec:
-                    return "ogg"
+                    return "ogv"
                 if {"vp8", "vp9", "av1"}.intersection(codec):
                     return "webm"
 
         if "theora" in video_encoder:
-            return "ogg"
+            return "ogv"
         elif re.search(r"vp[89x]|av1", video_encoder.split(" ", 1)[0]):
             return "webm"
         return "mp4"
@@ -274,16 +274,15 @@ class VideoFileAnalyzer:
             scan_data = json.loads(result)
         except Exception as e:
             log.debug("Failure in JSON parsing ffprobe results. Message: %s", str(e))
-            if validate:
-                raise Exception(f'Invalid video file: {file_path}')
-            log.info("Unable to optimize %s . FFmpeg output was unreadable.", file_path)
-            return
+            raise ValueError(f'Absent or unreadable video file: {file_path}')
 
-        if "format" not in scan_data:
-            if validate:
-                raise FileNotFoundError(f'Unexpected or absent video file contents at: {file_path}')
-            log.info("Unable to optimize %s . FFmpeg output is missing the format section.", file_path)
-            return
+        if "format" not in scan_data or "duration" not in scan_data["format"]:
+            log.debug("Format data is missing from ffprobe results for: %s", file_path)
+            raise ValueError(f'Media file does not appear to contain video content at: {file_path}')
+
+        if float(scan_data["format"]["duration"]) < 0.1:
+            log.debug("Media file appears to be an image: %s", file_path)
+            raise ValueError(f'Assuming image file at: {file_path}')
 
         return scan_data
 
diff --git a/lbry/schema/mime_types.py b/lbry/schema/mime_types.py
index f873db897..00505e762 100644
--- a/lbry/schema/mime_types.py
+++ b/lbry/schema/mime_types.py
@@ -148,6 +148,7 @@ types_map = {
     '.mobi': ('application/x-mobipocket-ebook', 'document'),
     '.oga': ('audio/ogg', 'audio'),
     '.ogv': ('video/ogg', 'video'),
+    '.ogg': ('video/ogg', 'video'),
     '.pct': ('image/pict', 'image'),
     '.pic': ('image/pict', 'image'),
     '.pict': ('image/pict', 'image'),
@@ -162,6 +163,7 @@ types_map = {
 
 
 def guess_media_type(path):
+    # should we be using "file --mime-type -b $filename" on linux?
     _, ext = os.path.splitext(path)
     extension = ext.strip().lower()
     if extension[1:]:
diff --git a/scripts/check_video.py b/scripts/check_video.py
index ee3a26899..61331142b 100755
--- a/scripts/check_video.py
+++ b/scripts/check_video.py
@@ -26,8 +26,8 @@ async def process_video(analyzer, video_file):
     try:
         await analyzer.verify_or_repair(True, False, video_file)
         print("No concerns. Ship it!")
-    except FileNotFoundError as e:
-        print(str(e))
+    except (FileNotFoundError, ValueError) as e:
+        print("Analysis failed.", str(e))
     except Exception as e:
         print(str(e))
         transcode = input("Would you like to make a repaired clone now? [y/N] ")
diff --git a/tests/integration/other/test_transcoding.py b/tests/integration/other/test_transcoding.py
index c832ec4ec..60cad7992 100644
--- a/tests/integration/other/test_transcoding.py
+++ b/tests/integration/other/test_transcoding.py
@@ -130,7 +130,7 @@ class TranscodeValidation(ClaimTestCase):
 
         scan_data = await self.analyzer._get_scan_data(True, self.video_file_ogg)
         extension = self.analyzer._get_best_container_extension(scan_data, "")
-        self.assertEqual(extension, "ogg")
+        self.assertEqual(extension, "ogv")
 
         scan_data = await self.analyzer._get_scan_data(True, self.video_file_webm)
         extension = self.analyzer._get_best_container_extension(scan_data, "")
@@ -143,7 +143,7 @@ class TranscodeValidation(ClaimTestCase):
         self.assertEqual("webm", extension)
 
         extension = self.analyzer._get_best_container_extension("", "libtheora")
-        self.assertEqual("ogg", extension)
+        self.assertEqual("ogv", extension)
 
     async def test_no_ffmpeg(self):
         self.conf.ffmpeg_folder = "I don't really exist/"