Merge pull request #2726 from lbryio/add_video_transcoding

added video file validation and optimization when publishing streams (using `--validate_file` and `--optimize_file` arguments)

.gitignore

@@ -5,6 +5,7 @@
 /.tox
 /.coverage*
 /lbry-venv
+/venv
 
 lbry.egg-info
 __pycache__
@@ -14,4 +15,3 @@ _trial_temp/
 /tests/.coverage.*
 
 /lbry/wallet/bin
-/venv

.gitlab-ci.yml

@@ -46,6 +46,8 @@ test:datanetwork-integration:
 test:blockchain-integration:
   stage: test
   script:
+    - apt-get update
+    - apt-get install -y --no-install-recommends ffmpeg
     - pip install tox-travis
     - tox -e blockchain
 

lbry/conf.py

@@ -462,7 +462,20 @@ class BaseConfig:
             self.persisted.save()
 
 
-class CLIConfig(BaseConfig):
+class TranscodeConfig(BaseConfig):
+
+    ffmpeg_folder = String('The path to ffmpeg and ffprobe', '')
+    video_encoder = String('FFmpeg codec and parameters for the video encoding. '
+                           'Example: libaom-av1 -crf 25 -b:v 0 -strict experimental',
+                           'libx264 -crf 18 -vf "format=yuv420p"')
+    audio_encoder = String('FFmpeg codec and parameters for the audio encoding. '
+                           'Example: libopus -b:a 128k',
+                           'aac -b:a 192k')
+    volume_filter = String('FFmpeg filter for audio normalization.', '-af loudnorm')
+    volume_analysis_time = Integer('Maximum seconds into the file that we examine audio volume (0 to disable).', '240')
+
+
+class CLIConfig(TranscodeConfig):
 
     api = String('Host name and port for lbrynet daemon API.', 'localhost:5279', metavar='HOST:PORT')
 

lbry/extras/cli.py

@@ -3,6 +3,7 @@ import sys
 import shutil
 import signal
 import pathlib
+import platform
 import json
 import asyncio
 import argparse
@@ -262,6 +263,9 @@ def setup_logging(logger: logging.Logger, args: argparse.Namespace, conf: Config
 
 
 def run_daemon(args: argparse.Namespace, conf: Config):
+    if sys.version_info < (3, 8) and platform.system() == "Windows":
+        # TODO: remove after we move to requiring Python 3.8
+        asyncio.set_event_loop(asyncio.ProactorEventLoop())
     loop = asyncio.get_event_loop()
     if args.verbose is not None:
         loop.set_debug(True)

lbry/extras/daemon/daemon.py

@@ -44,6 +44,7 @@ from lbry.extras.daemon.componentmanager import ComponentManager
 from lbry.extras.daemon.json_response_encoder import JSONResponseEncoder
 from lbry.extras.daemon import comment_client
 from lbry.extras.daemon.undecorated import undecorated
+from lbry.file_analysis import VideoFileAnalyzer
 from lbry.schema.claim import Claim
 from lbry.schema.url import URL
 
@@ -296,6 +297,7 @@ class Daemon(metaclass=JSONRPCServerType):
 
     def __init__(self, conf: Config, component_manager: typing.Optional[ComponentManager] = None):
         self.conf = conf
+        self._video_file_analyzer = VideoFileAnalyzer(conf)
         self._node_id = None
         self._installation_id = None
         self.session_id = base58.b58encode(utils.generate_id()).decode()
@@ -851,6 +853,7 @@ class Daemon(metaclass=JSONRPCServerType):
         """
 
         connection_code = await self.get_connection_status()
+        ffmpeg_status = await self._video_file_analyzer.status()
 
         response = {
             'installation_id': self.installation_id,
@@ -861,6 +864,7 @@ class Daemon(metaclass=JSONRPCServerType):
                 'code': connection_code,
                 'message': CONNECTION_MESSAGES[connection_code],
             },
+            'ffmpeg_status': ffmpeg_status
         }
         for component in self.component_manager.components:
             status = await component.get_status()
@@ -2808,6 +2812,7 @@ class Daemon(metaclass=JSONRPCServerType):
 
         Usage:
             publish (<name> | --name=<name>) [--bid=<bid>] [--file_path=<file_path>]
+                    [--validate_file] [--optimize_file]
                     [--fee_currency=<fee_currency>] [--fee_amount=<fee_amount>] [--fee_address=<fee_address>]
                     [--title=<title>] [--description=<description>] [--author=<author>]
                     [--tags=<tags>...] [--languages=<languages>...] [--locations=<locations>...]
@@ -2823,6 +2828,11 @@ class Daemon(metaclass=JSONRPCServerType):
             --name=<name>                  : (str) name of the content (can only consist of a-z A-Z 0-9 and -(dash))
             --bid=<bid>                    : (decimal) amount to back the claim
             --file_path=<file_path>        : (str) path to file to be associated with name.
+            --validate_file                : (bool) validate that the video container and encodings match
+                                             common web browser support or that optimization succeeds if specified.
+                                             FFmpeg is required
+            --optimize_file                : (bool) transcode the video & audio if necessary to ensure
+                                             common web browser support. FFmpeg is required
             --fee_currency=<fee_currency>  : (string) specify fee currency
             --fee_amount=<fee_amount>      : (decimal) content download fee
             --fee_address=<fee_address>    : (str) address where to send fee payments, will use
@@ -2994,12 +3004,13 @@ class Daemon(metaclass=JSONRPCServerType):
             self, name, bid, file_path, allow_duplicate_name=False,
             channel_id=None, channel_name=None, channel_account_id=None,
             account_id=None, wallet_id=None, claim_address=None, funding_account_ids=None,
-            preview=False, blocking=False, **kwargs):
+            preview=False, blocking=False, validate_file=False, optimize_file=False, **kwargs):
         """
         Make a new stream claim and announce the associated file to lbrynet.
 
         Usage:
             stream_create (<name> | --name=<name>) (<bid> | --bid=<bid>) (<file_path> | --file_path=<file_path>)
+                    [--validate_file] [--optimize_file]
                     [--allow_duplicate_name=<allow_duplicate_name>]
                     [--fee_currency=<fee_currency>] [--fee_amount=<fee_amount>] [--fee_address=<fee_address>]
                     [--title=<title>] [--description=<description>] [--author=<author>]
@@ -3016,6 +3027,11 @@ class Daemon(metaclass=JSONRPCServerType):
             --name=<name>                  : (str) name of the content (can only consist of a-z A-Z 0-9 and -(dash))
             --bid=<bid>                    : (decimal) amount to back the claim
             --file_path=<file_path>        : (str) path to file to be associated with name.
+            --validate_file                : (bool) validate that the video container and encodings match
+                                             common web browser support or that optimization succeeds if specified.
+                                             FFmpeg is required
+            --optimize_file                : (bool) transcode the video & audio if necessary to ensure
+                                             common web browser support. FFmpeg is required
         --allow_duplicate_name=<allow_duplicate_name> : (bool) create new claim even if one already exists with
                                               given name. default: false.
             --fee_currency=<fee_currency>  : (string) specify fee currency
@@ -3107,6 +3123,8 @@ 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)
+
         claim = Claim()
         claim.stream.update(file_path=file_path, sd_hash='0' * 96, **kwargs)
         tx = await Transaction.claim_create(

lbry/file_analysis.py

@@ -0,0 +1,355 @@
+import asyncio
+import json
+import logging
+import os
+import pathlib
+import re
+import shlex
+import shutil
+
+from lbry.conf import TranscodeConfig
+
+log = logging.getLogger(__name__)
+
+
+class VideoFileAnalyzer:
+
+    def __init__(self, conf: TranscodeConfig):
+        self._conf = conf
+        self._available_encoders = ""
+        self._ffmpeg_installed = False
+        self._which = None
+
+    async def _execute(self, command, arguments):
+        args = shlex.split(arguments)
+        process = await asyncio.create_subprocess_exec(self._conf.ffmpeg_folder + command, *args,
+                                                       stdout=asyncio.subprocess.PIPE, stderr=asyncio.subprocess.PIPE)
+        stdout, stderr = await process.communicate()  # returns when the streams are closed
+        return stdout.decode() + stderr.decode(), process.returncode
+
+    async def _verify_executable(self, name):
+        try:
+            version, code = await self._execute(name, "-version")
+        except Exception as e:
+            log.warning("Unable to run %s, but it was requested. Message: %s", name, str(e))
+            code = -1
+            version = ""
+        if code != 0 or not version.startswith(name):
+            raise FileNotFoundError(f"Unable to locate or run {name}. Please install FFmpeg "
+                                    f"and ensure that it is callable via PATH or conf.ffmpeg_folder")
+        return version
+
+    async def _verify_ffmpeg_installed(self):
+        if self._ffmpeg_installed:
+            return
+        await self._verify_executable("ffprobe")
+        version = await self._verify_executable("ffmpeg")
+        self._which = shutil.which(f"{self._conf.ffmpeg_folder}ffmpeg")
+        self._ffmpeg_installed = True
+        log.debug("Using %s at %s", version.splitlines()[0].split(" Copyright")[0], self._which)
+
+    async def status(self, reset=False):
+        if reset:
+            self._available_encoders = ""
+            self._ffmpeg_installed = False
+            self._which = None
+
+        installed = True
+        try:
+            await self._verify_ffmpeg_installed()
+        except FileNotFoundError:
+            installed = False
+
+        return {
+            "available": installed,
+            "which": self._which,
+            "analyze_audio_volume": int(self._conf.volume_analysis_time) > 0
+        }
+
+    @staticmethod
+    def _verify_container(scan_data: json):
+        container = scan_data["format"]["format_name"]
+        log.debug("   Detected container is %s", container)
+        if not {"webm", "mp4", "3gp", "ogg"}.intersection(container.split(",")):
+            return "Container format is not in the approved list of WebM, MP4. " \
+                   f"Actual: {container} [{scan_data['format']['format_long_name']}]"
+        return ""
+
+    @staticmethod
+    def _verify_video_encoding(scan_data: json):
+        for stream in scan_data["streams"]:
+            if stream["codec_type"] != "video":
+                continue
+            codec = stream["codec_name"]
+            log.debug("   Detected video codec is %s, format is %s", codec, stream["pix_fmt"])
+            if not {"h264", "vp8", "vp9", "av1", "theora"}.intersection(codec.split(",")):
+                return "Video codec is not in the approved list of H264, VP8, VP9, AV1, Theora. " \
+                       f"Actual: {codec} [{stream['codec_long_name']}]"
+
+            if "h264" in codec.split(",") and stream["pix_fmt"] != "yuv420p":
+                return "Video codec is H264, but its pixel format does not match the approved yuv420p. " \
+                       f"Actual: {stream['pix_fmt']}"
+
+        return ""
+
+    @staticmethod
+    def _verify_bitrate(scan_data: json):
+        if "bit_rate" not in scan_data["format"]:
+            return ""
+
+        bit_rate = float(scan_data["format"]["bit_rate"])
+        log.debug("   Detected bitrate is %s Mbps", str(bit_rate / 1000000.0))
+        pixels = -1.0
+        for stream in scan_data["streams"]:
+            if stream["codec_type"] == "video":
+                pieces = stream["r_frame_rate"].split('/', 1)
+                frame_rate = float(pieces[0]) if len(pieces) == 1 \
+                    else float(pieces[0]) / float(pieces[1])
+                pixels = max(pixels, float(stream["height"]) * float(stream["width"]) * frame_rate)
+
+        if pixels > 0.0 and pixels / bit_rate < 3.0:
+            return "Bits per second is excessive for this data; this may impact web streaming performance. " \
+                   f"Actual: {str(bit_rate / 1000000.0)} Mbps"
+
+        return ""
+
+    async def _verify_fast_start(self, scan_data: json, video_file):
+        container = scan_data["format"]["format_name"]
+        if {"webm", "ogg"}.intersection(container.split(",")):
+            return ""
+
+        result, _ = await self._execute("ffprobe", f'-v debug "{video_file}"')
+        iterator = re.finditer(r"\s+seeks:(\d+)\s+", result)
+        for match in iterator:
+            if int(match.group(1)) != 0:
+                return "Video stream descriptors are not at the start of the file (the faststart flag was not used)."
+        return ""
+
+    @staticmethod
+    def _verify_audio_encoding(scan_data: json):
+        for stream in scan_data["streams"]:
+            if stream["codec_type"] != "audio":
+                continue
+            codec = stream["codec_name"]
+            log.debug("   Detected audio codec is %s", codec)
+            if not {"aac", "mp3", "flac", "vorbis", "opus"}.intersection(codec.split(",")):
+                return "Audio codec is not in the approved list of AAC, FLAC, MP3, Vorbis, and Opus. " \
+                       f"Actual: {codec} [{stream['codec_long_name']}]"
+
+        return ""
+
+    async def _verify_audio_volume(self, seconds, video_file):
+        try:
+            validate_volume = int(seconds) > 0
+        except ValueError:
+            validate_volume = False
+
+        if not validate_volume:
+            return ""
+
+        result, _ = await self._execute("ffmpeg", f'-i "{video_file}" -t {seconds} '
+                                        f'-af volumedetect -vn -sn -dn -f null "{os.devnull}"')
+        try:
+            mean_volume = float(re.search(r"mean_volume:\s+([-+]?\d*\.\d+|\d+)", result).group(1))
+            max_volume = float(re.search(r"max_volume:\s+([-+]?\d*\.\d+|\d+)", result).group(1))
+        except Exception as e:
+            log.debug("   Failure in volume analysis. Message: %s", str(e))
+            return ""
+
+        if max_volume < -5.0 and mean_volume < -22.0:
+            return "Audio is at least five dB lower than prime. " \
+                   f"Actual max: {max_volume}, mean: {mean_volume}"
+
+        log.debug("   Detected audio volume has mean, max of %f, %f dB", mean_volume, max_volume)
+
+        return ""
+
+    @staticmethod
+    def _compute_crf(scan_data):
+        height = 240.0
+        for stream in scan_data["streams"]:
+            if stream["codec_type"] == "video":
+                height = max(height, float(stream["height"]))
+
+        # https://developers.google.com/media/vp9/settings/vod/
+        return int(-0.011 * height + 40)
+
+    async def _get_video_encoder(self, scan_data):
+        # use what the user said if it's there:
+        # if it's not there, use h264 if we can because it's way faster than the others
+        # if we don't have h264 use vp9; it's fairly compatible even though it's slow
+
+        if not self._available_encoders:
+            self._available_encoders, _ = await self._execute("ffmpeg", "-encoders -v quiet")
+
+        encoder = self._conf.video_encoder.split(" ", 1)[0]
+        if re.search(fr"^\s*V..... {encoder} ", self._available_encoders, re.MULTILINE):
+            return self._conf.video_encoder
+
+        if re.search(r"^\s*V..... libx264 ", self._available_encoders, re.MULTILINE):
+            if encoder:
+                log.warning("   Using libx264 since the requested encoder was unavailable. Requested: %s", encoder)
+            return 'libx264 -crf 19 -vf "format=yuv420p"'
+
+        if not encoder:
+            encoder = "libx264"
+
+        if re.search(r"^\s*V..... libvpx-vp9 ", self._available_encoders, re.MULTILINE):
+            log.warning("   Using libvpx-vp9 since the requested encoder was unavailable. Requested: %s", encoder)
+            crf = self._compute_crf(scan_data)
+            return f"libvpx-vp9 -crf {crf} -b:v 0"
+
+        if re.search(r"^\s*V..... libtheora", self._available_encoders, re.MULTILINE):
+            log.warning("   Using libtheora since the requested encoder was unavailable. Requested: %s", encoder)
+            return "libtheora -q:v 7"
+
+        raise Exception(f"The video encoder is not available. Requested: {encoder}")
+
+    async def _get_audio_encoder(self, extension):
+        # if the video encoding is theora or av1/vp8/vp9 use opus (or fallback to vorbis)
+        # or we don't have a video encoding but we have an ogg or webm container use opus
+        # if we need to use opus/vorbis see if the conf file has it else use our own params
+        # else use the user-set value if it exists
+        # else use aac
+
+        wants_opus = extension != "mp4"
+        if not self._available_encoders:
+            self._available_encoders, _ = await self._execute("ffmpeg", "-encoders -v quiet")
+
+        encoder = self._conf.audio_encoder.split(" ", 1)[0]
+        if wants_opus and 'opus' in encoder:
+            return self._conf.audio_encoder
+
+        if wants_opus and re.search(r"^\s*A..... libopus ", self._available_encoders, re.MULTILINE):
+            return "libopus -b:a 160k"
+
+        if wants_opus and 'vorbis' in encoder:
+            return self._conf.audio_encoder
+
+        if wants_opus and re.search(r"^\s*A..... libvorbis ", self._available_encoders, re.MULTILINE):
+            return "libvorbis -q:a 6"
+
+        if re.search(fr"^\s*A..... {encoder} ", self._available_encoders, re.MULTILINE):
+            return self._conf.audio_encoder
+
+        if re.search(r"^\s*A..... aac ", self._available_encoders, re.MULTILINE):
+            return "aac -b:a 192k"
+
+        raise Exception(f"The audio encoder is not available. Requested: {encoder or 'aac'}")
+
+    async def _get_volume_filter(self):
+        return self._conf.volume_filter if self._conf.volume_filter else "-af loudnorm"
+
+    @staticmethod
+    def _get_best_container_extension(scan_data, video_encoder):
+        # the container is chosen by the video format
+        # if we are theora-encoded, we want ogg
+        # if we are vp8/vp9/av1 we want webm
+        # use mp4 for anything else
+
+        if not video_encoder:  # not re-encoding video
+            for stream in scan_data["streams"]:
+                if stream["codec_type"] != "video":
+                    continue
+                codec = stream["codec_name"].split(",")
+                if "theora" in codec:
+                    return "ogg"
+                if {"vp8", "vp9", "av1"}.intersection(codec):
+                    return "webm"
+
+        if "theora" in video_encoder:
+            return "ogg"
+        elif re.search(r"vp[89x]|av1", video_encoder.split(" ", 1)[0]):
+            return "webm"
+        return "mp4"
+
+    async def _get_scan_data(self, validate, file_path):
+        result, _ = await self._execute("ffprobe",
+                                        f'-v quiet -print_format json -show_format -show_streams "{file_path}"')
+        try:
+            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
+
+        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
+
+        return scan_data
+
+    async def verify_or_repair(self, validate, repair, file_path):
+        if not validate and not repair:
+            return file_path
+
+        await self._verify_ffmpeg_installed()
+        scan_data = await self._get_scan_data(validate, file_path)
+
+        fast_start_msg = await self._verify_fast_start(scan_data, file_path)
+        log.debug("Analyzing %s:", file_path)
+        log.debug("   Detected faststart is %s", "false" if fast_start_msg else "true")
+        container_msg = self._verify_container(scan_data)
+        bitrate_msg = self._verify_bitrate(scan_data)
+        video_msg = self._verify_video_encoding(scan_data)
+        audio_msg = self._verify_audio_encoding(scan_data)
+        volume_msg = await self._verify_audio_volume(self._conf.volume_analysis_time, file_path)
+        messages = [container_msg, bitrate_msg, fast_start_msg, video_msg, audio_msg, volume_msg]
+
+        if not any(messages):
+            return file_path
+
+        if not repair:
+            errors = ["Streamability verification failed:"]
+            errors.extend(filter(None, messages))
+            raise Exception("\n   ".join(errors))
+
+        # the plan for transcoding:
+        # we have to re-encode the video if it is in a nonstandard format
+        # we also re-encode if we are h264 but not yuv420p (both errors caught in video_msg)
+        # we also re-encode if our bitrate is too high
+
+        try:
+            transcode_command = [f'-i "{file_path}" -y -c:s copy -c:d copy -c:v']
+
+            video_encoder = ""
+            if video_msg or bitrate_msg:
+                video_encoder = await self._get_video_encoder(scan_data)
+                transcode_command.append(video_encoder)
+            else:
+                transcode_command.append("copy")
+
+            transcode_command.append("-movflags +faststart -c:a")
+            path = pathlib.Path(file_path)
+            extension = self._get_best_container_extension(scan_data, video_encoder)
+
+            if audio_msg or volume_msg:
+                audio_encoder = await self._get_audio_encoder(extension)
+                transcode_command.append(audio_encoder)
+                if volume_msg:
+                    volume_filter = await self._get_volume_filter()
+                    transcode_command.append(volume_filter)
+            else:
+                transcode_command.append("copy")
+
+            # TODO: put it in a temp folder and delete it after we upload?
+            output = path.parent / f"{path.stem}_fixed.{extension}"
+            transcode_command.append(f'"{output}"')
+
+            ffmpeg_command = " ".join(transcode_command)
+            log.info("Proceeding on transcode via: ffmpeg %s", ffmpeg_command)
+            result, code = await self._execute("ffmpeg", ffmpeg_command)
+            if code != 0:
+                raise Exception(f"Failure to complete the transcode command. Output: {result}")
+        except Exception as e:
+            if validate:
+                raise
+            log.info("Unable to transcode %s . Message: %s", file_path, str(e))
+            # TODO: delete partial output file here if it exists?
+            return file_path
+
+        return str(output)

scripts/check_video.py

@@ -0,0 +1,62 @@
+#!/usr/bin/env python3
+
+import asyncio
+import logging
+import platform
+import sys
+
+# noinspection PyUnresolvedReferences
+import lbry.wallet  # needed to make the following line work (it's a bug):
+from lbry.conf import TranscodeConfig
+from lbry.file_analysis import VideoFileAnalyzer
+
+
+def enable_logging():
+    root = logging.getLogger()
+    root.setLevel(logging.DEBUG)
+
+    handler = logging.StreamHandler(sys.stdout)
+    handler.setLevel(logging.DEBUG)
+    formatter = logging.Formatter('%(message)s')  # %(asctime)s - %(levelname)s -
+    handler.setFormatter(formatter)
+    root.addHandler(handler)
+
+
+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 Exception as e:
+        print(str(e))
+        transcode = input("Would you like to make a repaired clone now? [y/N] ")
+        if transcode == "y":
+            try:
+                new_video_file = await analyzer.verify_or_repair(True, True, video_file)
+                print("Successfully created ", new_video_file)
+            except Exception as e:
+                print("Unable to complete the transcode. Message: ", str(e))
+
+
+def main():
+    if len(sys.argv) < 2:
+        print("Usage: check_video.py <path to video file>", file=sys.stderr)
+        sys.exit(1)
+
+    enable_logging()
+
+    video_file = sys.argv[1]
+    conf = TranscodeConfig()
+    analyzer = VideoFileAnalyzer(conf)
+    if sys.version_info < (3, 8) and platform.system() == "Windows":
+        # TODO: remove after we move to requiring Python 3.8
+        asyncio.set_event_loop(asyncio.ProactorEventLoop())
+    try:
+        asyncio.run(process_video(analyzer, video_file))
+    except KeyboardInterrupt:
+        pass
+
+
+if __name__ == '__main__':
+    main()

tests/integration/blockchain/test_transcoding.py

@@ -0,0 +1,151 @@
+import logging
+import pathlib
+import time
+
+from .test_claim_commands import ClaimTestCase
+from lbry.conf import TranscodeConfig
+from lbry.file_analysis import VideoFileAnalyzer
+
+log = logging.getLogger(__name__)
+
+
+class MeasureTime:
+    def __init__(self, text):
+        print(text, end="...", flush=True)
+
+    def __enter__(self):
+        self.start = time.perf_counter()
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        end = time.perf_counter()
+        print(f" done in {end - self.start:.6f}s", flush=True)
+
+
+class TranscodeValidation(ClaimTestCase):
+
+    def make_name(self, name, extension=""):
+        path = pathlib.Path(self.video_file_name)
+        return path.parent / f"{path.stem}_{name}{extension or path.suffix}"
+
+    async def asyncSetUp(self):
+        await super().asyncSetUp()
+        self.conf = TranscodeConfig()
+        self.conf.volume_analysis_time = 0  # disable it as the test file isn't very good here
+        self.analyzer = VideoFileAnalyzer(self.conf)
+        file_ogg = self.make_name("ogg", ".ogg")
+        self.video_file_ogg = str(file_ogg)
+        if not file_ogg.exists():
+            command = f'-i "{self.video_file_name}" -c:v libtheora -q:v 4 -c:a libvorbis -q:a 4 ' \
+                      f'-c:s copy -c:d copy "{file_ogg}"'
+            with MeasureTime(f"Creating {file_ogg.name}"):
+                output, code = await self.analyzer._execute("ffmpeg", command)
+                self.assertEqual(code, 0, output)
+
+        file_webm = self.make_name("webm", ".webm")
+        self.video_file_webm = str(file_webm)
+        if not file_webm.exists():
+            command = f'-i "{self.video_file_name}" -c:v libvpx-vp9 -crf 36 -b:v 0 -cpu-used 2 ' \
+                      f'-c:a libopus -b:a 128k -c:s copy -c:d copy "{file_webm}"'
+            with MeasureTime(f"Creating {file_webm.name}"):
+                output, code = await self.analyzer._execute("ffmpeg", command)
+                self.assertEqual(code, 0, output)
+
+    async def test_should_work(self):
+        new_file_name = await self.analyzer.verify_or_repair(True, False, self.video_file_name)
+        self.assertEqual(self.video_file_name, new_file_name)
+        new_file_name = await self.analyzer.verify_or_repair(True, False, self.video_file_ogg)
+        self.assertEqual(self.video_file_ogg, new_file_name)
+        new_file_name = await self.analyzer.verify_or_repair(True, False, self.video_file_webm)
+        self.assertEqual(self.video_file_webm, new_file_name)
+
+    async def test_volume(self):
+        self.conf.volume_analysis_time = 200
+        with self.assertRaisesRegex(Exception, "lower than prime"):
+            await self.analyzer.verify_or_repair(True, False, self.video_file_name)
+
+    async def test_container(self):
+        file_name = self.make_name("bad_container", ".avi")
+        if not file_name.exists():
+            command = f'-i "{self.video_file_name}" -c copy -map 0 "{file_name}"'
+            with MeasureTime(f"Creating {file_name.name}"):
+                output, code = await self.analyzer._execute("ffmpeg", command)
+                self.assertEqual(code, 0, output)
+
+        with self.assertRaisesRegex(Exception, "Container format is not in the approved list"):
+            await self.analyzer.verify_or_repair(True, False, file_name)
+
+        fixed_file = await self.analyzer.verify_or_repair(True, True, file_name)
+        pathlib.Path(fixed_file).unlink()
+
+    async def test_video_codec(self):
+        file_name = self.make_name("bad_video_codec_1")
+        if not file_name.exists():
+            command = f'-i "{self.video_file_name}" -c copy -map 0 -c:v libx265 -preset superfast "{file_name}"'
+            with MeasureTime(f"Creating {file_name.name}"):
+                output, code = await self.analyzer._execute("ffmpeg", command)
+                self.assertEqual(code, 0, output)
+
+        with self.assertRaisesRegex(Exception, "Video codec is not in the approved list"):
+            await self.analyzer.verify_or_repair(True, False, file_name)
+        with self.assertRaisesRegex(Exception, "faststart flag was not used"):
+            await self.analyzer.verify_or_repair(True, False, file_name)
+
+        fixed_file = await self.analyzer.verify_or_repair(True, True, file_name)
+        pathlib.Path(fixed_file).unlink()
+
+    async def test_video_format(self):
+        file_name = self.make_name("bad_video_format_1")
+        if not file_name.exists():
+            command = f'-i "{self.video_file_name}" -c copy -map 0 -c:v libx264 ' \
+                      f'-vf format=yuv444p "{file_name}"'
+            with MeasureTime(f"Creating {file_name.name}"):
+                output, code = await self.analyzer._execute("ffmpeg", command)
+                self.assertEqual(code, 0, output)
+
+        with self.assertRaisesRegex(Exception, "pixel format does not match the approved"):
+            await self.analyzer.verify_or_repair(True, False, file_name)
+
+        fixed_file = await self.analyzer.verify_or_repair(True, True, file_name)
+        pathlib.Path(fixed_file).unlink()
+
+    async def test_audio_codec(self):
+        file_name = self.make_name("bad_audio_codec_1", ".mkv")
+        if not file_name.exists():
+            command = f'-i "{self.video_file_name}" -c copy -map 0 -c:a pcm_s16le "{file_name}"'
+            with MeasureTime(f"Creating {file_name.name}"):
+                output, code = await self.analyzer._execute("ffmpeg", command)
+                self.assertEqual(code, 0, output)
+
+        with self.assertRaisesRegex(Exception, "Audio codec is not in the approved list"):
+            await self.analyzer.verify_or_repair(True, False, file_name)
+
+        fixed_file = await self.analyzer.verify_or_repair(True, True, file_name)
+        pathlib.Path(fixed_file).unlink()
+
+    async def test_extension_choice(self):
+
+        scan_data = await self.analyzer._get_scan_data(True, self.video_file_name)
+        extension = self.analyzer._get_best_container_extension(scan_data, "")
+        self.assertEqual(extension, pathlib.Path(self.video_file_name).suffix[1:])
+
+        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")
+
+        scan_data = await self.analyzer._get_scan_data(True, self.video_file_webm)
+        extension = self.analyzer._get_best_container_extension(scan_data, "")
+        self.assertEqual(extension, "webm")
+
+        extension = self.analyzer._get_best_container_extension("", "libx264 -crf 23")
+        self.assertEqual("mp4", extension)
+
+        extension = self.analyzer._get_best_container_extension("", "libvpx-vp9 -crf 23")
+        self.assertEqual("webm", extension)
+
+        extension = self.analyzer._get_best_container_extension("", "libtheora")
+        self.assertEqual("ogg", extension)
+
+    async def test_no_ffmpeg(self):
+        self.conf.ffmpeg_folder = "I don't really exist/"
+        with self.assertRaisesRegex(Exception, "Unable to locate"):
+            await self.analyzer.verify_or_repair(True, False, self.video_file_name)

tests/integration/blockchain/test_wallet_commands.py

@@ -199,7 +199,8 @@ class WalletEncryptionAndSynchronization(CommandTestCase):
         await self.confirm_tx(sendtxid, self.daemon2.ledger)
 
     def assertWalletEncrypted(self, wallet_path, encrypted):
-        wallet = json.load(open(wallet_path))
+        with open(wallet_path) as opened:
+            wallet = json.load(opened)
             self.assertEqual(wallet['accounts'][0]['private_key'][1:4] != 'prv', encrypted)
 
     async def test_sync(self):