From b96ca28eb480ae17ee97b3dd2697b1509931e0a9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Simon=20K=C3=BCnzel?= <simonk@fsmpi.rwth-aachen.de>
Date: Thu, 24 Apr 2025 02:38:13 +0200
Subject: [PATCH] Add file_format to medium metadata
---
api/api_specification.json | 9 ++++++
api/api_specification.md | 14 +++++++--
api/api_specification_template.md | 7 ++++-
api/config/db_test_data.sql | 22 +++++++-------
api/migration.sql | 30 +++++++++++++++++++
.../media_process/basic_targets.py | 8 ++++-
.../src/videoag_common/objects/medium.py | 6 ++++
.../src/videoag_common/test/object_data.py | 10 +++++++
.../jobs/media_process_scheduler/job.py | 7 +++++
9 files changed, 98 insertions(+), 15 deletions(-)
diff --git a/api/api_specification.json b/api/api_specification.json
index 5790d4b..90509bf 100644
--- a/api/api_specification.json
+++ b/api/api_specification.json
@@ -3740,6 +3740,15 @@
"medium_metadata": {
"fields": {
"": {
+ "file_format": {
+ "config_directly_modifiable": false,
+ "id": "file_format",
+ "notes": "",
+ "object_variant": null,
+ "only_mod": false,
+ "optional": false,
+ "type": "string"
+ },
"file_size": {
"config_directly_modifiable": false,
"id": "file_size",
diff --git a/api/api_specification.md b/api/api_specification.md
index 28e8aeb..eaa2b9f 100644
--- a/api/api_specification.md
+++ b/api/api_specification.md
@@ -1,4 +1,4 @@
-# Specification of the Web API for the Video-AG Website (v0.82).
+# Specification of the Web API for the Video-AG Website (v0.84).
## Introduction
@@ -2710,7 +2710,12 @@ Additionally, the following objects may appear as the type of some field:
</thead>
<tbody>
<tr>
- <td rowspan="4">any</td>
+ <td rowspan="5">any</td>
+ <td>file_format</td>
+ <td>string</td>
+ <td></td>
+ </tr>
+ <tr>
<td>file_size</td>
<td>int</td>
<td></td>
@@ -3213,6 +3218,11 @@ Possible `error_code`:
## Changelog
+### v0.84
+
+* Updated `medium_metadata`
+ * Added field `file_format`
+
### v0.83
* Updated `GET /lecture/{lecture_id}/media_process_overview`
diff --git a/api/api_specification_template.md b/api/api_specification_template.md
index 36b13ef..21874e0 100644
--- a/api/api_specification_template.md
+++ b/api/api_specification_template.md
@@ -1,4 +1,4 @@
-# Specification of the Web API for the Video-AG Website (v0.82).
+# Specification of the Web API for the Video-AG Website (v0.84).
## Introduction
@@ -139,6 +139,11 @@ Possible `error_code`:
## Changelog
+### v0.84
+
+* Updated `medium_metadata`
+ * Added field `file_format`
+
### v0.83
* Updated `GET /lecture/{lecture_id}/media_process_overview`
diff --git a/api/config/db_test_data.sql b/api/config/db_test_data.sql
index c3c942c..1407fb2 100644
--- a/api/config/db_test_data.sql
+++ b/api/config/db_test_data.sql
@@ -90,17 +90,17 @@ SELECT setval('sorter_file_id_seq', 1000);
-- TODO figure out correct/realistic file sizes
-INSERT INTO medium_metadata (id,file_id,"type",file_size,duration_sec,vertical_resolution,horizontal_resolution,video_frame_rate_numerator,video_frame_rate_denominator,audio_sample_rate,audio_channel_count) VALUES
- (6,1,'thumbnail'::medium_metadata_type,42,NULL,640,640,NULL,NULL,NULL,NULL),
- (8,2,'thumbnail'::medium_metadata_type,42,NULL,640,640,NULL,NULL,NULL,NULL),
- (17,3,'thumbnail'::medium_metadata_type,42,NULL,640,640,NULL,NULL,NULL,NULL),
- (20,4,'thumbnail'::medium_metadata_type,42,NULL,640,640,NULL,NULL,NULL,NULL),
- (15,5,'plain_video'::medium_metadata_type,42,5431,1080,1920,25,1,44000,2),
- (5,6,'plain_video'::medium_metadata_type,42,5243,720,1280,25,1,44000,2),
- (7,7,'plain_video'::medium_metadata_type,42,5420,720,1280,25,1,44000,2),
- (19,8,'plain_video'::medium_metadata_type,42,5001,720,1280,25,1,44000,2),
- (16,9,'plain_video'::medium_metadata_type,42,5431,720,1280,25,1,44000,2),
- (25,10,'plain_video'::medium_metadata_type,42,5431,480,854,25,1,44000,2);
+INSERT INTO medium_metadata (id,file_id,"type",file_size,file_format,duration_sec,vertical_resolution,horizontal_resolution,video_frame_rate_numerator,video_frame_rate_denominator,audio_sample_rate,audio_channel_count) VALUES
+ (6,1,'thumbnail'::medium_metadata_type,42,'jpg',NULL,640,640,NULL,NULL,NULL,NULL),
+ (8,2,'thumbnail'::medium_metadata_type,42,'jpg',NULL,640,640,NULL,NULL,NULL,NULL),
+ (17,3,'thumbnail'::medium_metadata_type,42,'jpg',NULL,640,640,NULL,NULL,NULL,NULL),
+ (20,4,'thumbnail'::medium_metadata_type,42,'jpg',NULL,640,640,NULL,NULL,NULL,NULL),
+ (15,5,'plain_video'::medium_metadata_type,42,'mp4',5431,1080,1920,25,1,44000,2),
+ (5,6,'plain_video'::medium_metadata_type,42,'mp4',5243,720,1280,25,1,44000,2),
+ (7,7,'plain_video'::medium_metadata_type,42,'mp4',5420,720,1280,25,1,44000,2),
+ (19,8,'plain_video'::medium_metadata_type,42,'mp4',5001,720,1280,25,1,44000,2),
+ (16,9,'plain_video'::medium_metadata_type,42,'mp4',5431,720,1280,25,1,44000,2),
+ (25,10,'plain_video'::medium_metadata_type,42,'mp4',5431,480,854,25,1,44000,2);
SELECT setval('medium_metadata_id_seq', 1000);
diff --git a/api/migration.sql b/api/migration.sql
index 8eacea3..8427bd1 100644
--- a/api/migration.sql
+++ b/api/migration.sql
@@ -245,6 +245,30 @@ SELECT id, deleted, visible, lecture_id, "time", text
FROM old_data.chapters
;
+CREATE OR REPLACE FUNCTION temp_get_file_extension(path text) RETURNS text AS $$
+ DECLARE
+ str text;
+ BEGIN
+ SELECT path INTO str;
+
+ -- delete path in front
+ SELECT regexp_replace(str, '.*/', '', 'g') INTO str;
+
+ IF (str NOT LIKE '%.%') THEN
+ RAISE EXCEPTION 'File has no extension: %', path;
+ END IF;
+
+ -- delete part before extension
+ SELECT regexp_replace(str, '.*\.', '', 'g') INTO str;
+
+ IF (str = '') THEN
+ RAISE EXCEPTION 'File has no extension: %', path;
+ END IF;
+
+ RETURN str;
+ END;
+$$ LANGUAGE plpgsql;
+
DO $$
DECLARE
lecture record;
@@ -343,6 +367,7 @@ DO $$
file_id,
type,
file_size,
+ file_format,
duration_sec,
audio_sample_rate,
audio_channel_count,
@@ -354,6 +379,7 @@ DO $$
medium_file_id,
'plain_video',
video.file_size,
+ temp_get_file_extension(video.path),
video.duration,
44000,
2,
@@ -444,6 +470,7 @@ DO $$
file_id,
type,
file_size,
+ file_format,
duration_sec,
audio_sample_rate,
audio_channel_count,
@@ -457,6 +484,7 @@ DO $$
-- Some of these values are all probably wrong, but we can't get them here. Can be updated later by
-- a script inspecting the actual files
0,
+ 'mp4',
video.duration,
44000,
2,
@@ -536,12 +564,14 @@ DO $$
file_id,
type,
file_size,
+ file_format,
vertical_resolution,
horizontal_resolution
) VALUES (
thumbnail_medium_file_id,
'thumbnail',
0,
+ 'jpg',
640,
640
) RETURNING id INTO thumbnail_medium_metadata_id;
diff --git a/common_py/src/videoag_common/media_process/basic_targets.py b/common_py/src/videoag_common/media_process/basic_targets.py
index 55252a8..fe103a5 100644
--- a/common_py/src/videoag_common/media_process/basic_targets.py
+++ b/common_py/src/videoag_common/media_process/basic_targets.py
@@ -1,9 +1,11 @@
import re
import math
+from pathlib import Path
from typing import TYPE_CHECKING
from videoag_common.database import *
from videoag_common.miscellaneous import *
+from ..ffmpeg import get_file_extension
if TYPE_CHECKING:
# Can't actually import due to circular dependency
@@ -60,6 +62,10 @@ class SourceFileTargetProducer(SingleOutputTargetProducer["SourceMedium"]):
assert isinstance(intermediate, SorterFile)
sorter_file: SorterFile = intermediate
+
+ if get_file_extension(Path(sorter_file.file_path)) is None:
+ raise MediaProcessException(f"Sorter File {sorter_file.file_path} (#{sorter_file.id}) does not have a file"
+ f" extension")
medium_file = output_files_by_id[self.output_id]
medium_file.file_path = sorter_file.file_path
@@ -92,7 +98,7 @@ class RescaleVideoTargetProducer(SingleInputTargetProducer, SingleOutputTargetPr
raise MediaProcessException(f"Input {self.input_id} is not of type {MediumMetadataType.PLAIN_VIDEO}")
output_file = output_files_by_id[self.output_id]
- output_file.file_path += ".mp4"
+ output_file.file_path += f".{input_medium.file_format}"
return "rescale_video", {
"input_file": input_medium.file.file_path,
diff --git a/common_py/src/videoag_common/objects/medium.py b/common_py/src/videoag_common/objects/medium.py
index c8de5dc..9fbabc0 100644
--- a/common_py/src/videoag_common/objects/medium.py
+++ b/common_py/src/videoag_common/objects/medium.py
@@ -313,6 +313,12 @@ class MediumMetadata(ApiObject, Base):
include_in_data=True
)
)
+ file_format: Mapped[str] = api_mapped(
+ mapped_column(Text(collation=STRING_COLLATION), nullable=False),
+ ApiStringField(
+ include_in_data=True
+ )
+ )
file: Mapped["MediumFile"] = relationship(
primaryjoin=lambda: MediumMetadata.file_id == MediumFile.id,
diff --git a/common_py/src/videoag_common/test/object_data.py b/common_py/src/videoag_common/test/object_data.py
index 1d76807..75ac520 100644
--- a/common_py/src/videoag_common/test/object_data.py
+++ b/common_py/src/videoag_common/test/object_data.py
@@ -183,6 +183,7 @@ TEST_DATA_MEDIUM_METADATA_5 = \
"id": 5,
"type": "plain_video",
"file_size": 42,
+ "file_format": "mp4",
"vertical_resolution": 720,
"horizontal_resolution": 1280,
"video_frame_rate_numerator": 25,
@@ -196,6 +197,7 @@ TEST_DATA_MEDIUM_METADATA_6 = \
"id": 6,
"type": "thumbnail",
"file_size": 42,
+ "file_format": "jpg",
"vertical_resolution": 640,
"horizontal_resolution": 640,
}
@@ -206,6 +208,7 @@ TEST_DATA_MEDIUM_METADATA_7 = \
"id": 7,
"type": "plain_video",
"file_size": 42,
+ "file_format": "mp4",
"vertical_resolution": 720,
"horizontal_resolution": 1280,
"video_frame_rate_numerator": 25,
@@ -219,6 +222,7 @@ TEST_DATA_MEDIUM_METADATA_8 = \
"id": 8,
"type": "thumbnail",
"file_size": 42,
+ "file_format": "jpg",
"vertical_resolution": 640,
"horizontal_resolution": 640,
}
@@ -229,6 +233,7 @@ TEST_DATA_MEDIUM_METADATA_15 = \
"id": 15,
"type": "plain_video",
"file_size": 42,
+ "file_format": "mp4",
"vertical_resolution": 1080,
"horizontal_resolution": 1920,
"video_frame_rate_numerator": 25,
@@ -242,6 +247,7 @@ TEST_DATA_MEDIUM_METADATA_16 = \
"id": 16,
"type": "plain_video",
"file_size": 42,
+ "file_format": "mp4",
"vertical_resolution": 720,
"horizontal_resolution": 1280,
"video_frame_rate_numerator": 25,
@@ -255,6 +261,7 @@ TEST_DATA_MEDIUM_METADATA_25 = \
"id": 25,
"type": "plain_video",
"file_size": 42,
+ "file_format": "mp4",
"vertical_resolution": 480,
"horizontal_resolution": 854,
"video_frame_rate_numerator": 25,
@@ -268,6 +275,7 @@ TEST_DATA_MEDIUM_METADATA_17 = \
"id": 17,
"type": "thumbnail",
"file_size": 42,
+ "file_format": "jpg",
"vertical_resolution": 640,
"horizontal_resolution": 640,
}
@@ -278,6 +286,7 @@ TEST_DATA_MEDIUM_METADATA_19 = \
"id": 19,
"type": "plain_video",
"file_size": 42,
+ "file_format": "mp4",
"vertical_resolution": 720,
"horizontal_resolution": 1280,
"video_frame_rate_numerator": 25,
@@ -291,6 +300,7 @@ TEST_DATA_MEDIUM_METADATA_20 = \
"id": 20,
"type": "thumbnail",
"file_size": 42,
+ "file_format": "jpg",
"vertical_resolution": 640,
"horizontal_resolution": 640,
}
diff --git a/job_controller/jobs/media_process_scheduler/job.py b/job_controller/jobs/media_process_scheduler/job.py
index a7fa1f8..012fc46 100644
--- a/job_controller/jobs/media_process_scheduler/job.py
+++ b/job_controller/jobs/media_process_scheduler/job.py
@@ -218,6 +218,13 @@ class ProcessScheduler:
file.medium_metadata = res
assert isinstance(file.medium_metadata, MediumMetadata)
file.medium_metadata.file_size = probe_context.file_size()
+ if file.medium_metadata.file_format is None:
+ extension = get_file_extension(Path(file.file_path))
+ if extension is None:
+ raise ValueError(f"Got no file format from try_create_for_file for {medium_type} and file"
+ f" {file.file_path} does not have an extension. Target Producers need to make sure"
+ f" that their medium files have an extension.")
+ file.medium_metadata.file_format = extension
self._session.flush() # Ensure metadata id present
logger.info(f"Assigned metadata {file.medium_metadata.id} of type {medium_type} to medium file {file.file_path}"
f" ({file.process_target_id}, {file.id})")
--
GitLab