Merge pull request #32 from nattvara/update-capture-preview

Update capture preview job

jobs/__init__.py

@@ -142,15 +142,17 @@ def schedule_analysis_of_lecture(
 
     save_message_for_analysis(analysis, 'Analysis scheduled', 'Waiting for a worker to pick it up.')
 
-    next(get_metadata_queue()).enqueue(capture_preview.job, lecture.public_id, lecture.language, job_timeout=capture_preview.TIMEOUT)  # noqa: E501
-    next(get_metadata_queue()).enqueue(fetch_metadata.job, lecture.public_id, lecture.language, job_timeout=fetch_metadata.TIMEOUT)  # noqa: E501
-    job_1 = next(queue_download()).enqueue(download_lecture.job, lecture.public_id, lecture.language, job_timeout=download_lecture.TIMEOUT)  # noqa: E501
-    job_2 = next(queue_extract()).enqueue(extract_audio.job, lecture.public_id, lecture.language, job_timeout=extract_audio.TIMEOUT, depends_on=job_1)  # noqa: E501
-    job_3 = next(queue_transcribe()).enqueue(transcribe_audio.job, lecture.public_id, lecture.language, job_timeout=transcribe_audio.TIMEOUT, depends_on=job_2)  # noqa: E501
-    job_4 = next(queue_summarise()).enqueue(summarise_transcript.job, lecture.public_id, lecture.language, job_timeout=summarise_transcript.TIMEOUT, depends_on=job_3)  # noqa: E501
-    job_5 = next(queue_metadata()).enqueue(clean_lecture.job, lecture.public_id, lecture.language, depends_on=job_4)  # noqa: F841, E501
-    job_6 = next(queue_metadata()).enqueue(capture_preview.job, lecture.public_id, lecture.language, job_timeout=capture_preview.TIMEOUT, depends_on=job_4)  # noqa: E501
-    job_7 = next(queue_metadata()).enqueue(index_lecture.job, lecture.public_id, lecture.language, depends_on=job_6)  # noqa: F841, E501
+    # analysis sequence
+    download = next(queue_download()).enqueue(download_lecture.job, lecture.public_id, lecture.language, job_timeout=download_lecture.TIMEOUT)  # noqa: E501
+    extract = next(queue_extract()).enqueue(extract_audio.job, lecture.public_id, lecture.language, job_timeout=extract_audio.TIMEOUT, depends_on=download)  # noqa: E501
+    transcribe = next(queue_transcribe()).enqueue(transcribe_audio.job, lecture.public_id, lecture.language, job_timeout=transcribe_audio.TIMEOUT, depends_on=extract)  # noqa: E501
+    summarise = next(queue_summarise()).enqueue(summarise_transcript.job, lecture.public_id, lecture.language, job_timeout=summarise_transcript.TIMEOUT, depends_on=transcribe)  # noqa: E501
+
+    # other jobs that depend on different steps in the analysis
+    metadata = next(get_metadata_queue()).enqueue(fetch_metadata.job, lecture.public_id, lecture.language, job_timeout=fetch_metadata.TIMEOUT)  # noqa: F841, E501
+    preview = next(queue_metadata()).enqueue(capture_preview.job, lecture.public_id, lecture.language, job_timeout=capture_preview.TIMEOUT, depends_on=download)  # noqa: F841, E501
+    clean = next(queue_metadata()).enqueue(clean_lecture.job, lecture.public_id, lecture.language, depends_on=summarise)  # noqa: F841, E501
+    index = next(queue_metadata()).enqueue(index_lecture.job, lecture.public_id, lecture.language, depends_on=summarise)  # noqa: F841, E501
 
     return analysis
 

jobs/capture_preview.py

@@ -1,45 +1,25 @@
 from config.settings import settings
 from redis import Redis
 from rq import Queue
-import logging
 
 from db.crud import get_lecture_by_public_id_and_language
-from tools.video.img import save_photo_from_video
-from tools.web.camera import save_photo
 from db.models import Lecture
 import jobs.capture_preview
+import tools.video.img
 
 # 5min timeout
 TIMEOUT = 5 * 60
 
 
 def job(lecture_id: str, language: str):
-    logger = logging.getLogger('rq.worker')
-
     lecture = get_lecture_by_public_id_and_language(lecture_id, language)
-    if lecture.raw_content_link is not None:
-        logger.info('content has raw link, using mp4 file')
-        if lecture.mp4_filepath is None:
-            raise ValueError('mp4_filepath was not specified')
-
-        mp4 = lecture.mp4_filepath
-        save_photo_from_video(mp4, lecture.preview_filename())
-        save_photo_from_video(mp4, lecture.preview_small_filename(), True)
-
-        lecture.refresh()
-        lecture.img_preview = lecture.preview_filename()
-        lecture.img_preview_small = lecture.preview_small_filename()
-        lecture.save()
-        return
-
-    url = lecture.content_link()
-    logger.info(f'taking photo of {url}')
 
-    save_photo(url, lecture)
-    logger.info(f'photo saved at {lecture.preview_filename()}')
+    if lecture.mp4_filepath is None:
+        raise ValueError('mp4_filepath was not specified')
 
-    save_photo(url, lecture, small=True)
-    logger.info(f'photo saved at {lecture.preview_small_filename()}')
+    mp4 = lecture.mp4_filepath
+    tools.video.img.save_photo_from_video(mp4, lecture.preview_filename())
+    tools.video.img.save_photo_from_video(mp4, lecture.preview_small_filename(), True)
 
     lecture.refresh()
     lecture.img_preview = lecture.preview_filename()

jobs/download_lecture.py

@@ -73,16 +73,6 @@ def job(lecture_id: str, language: str):
             download_mp4_from_youtube(lecture)
         elif lecture.source == Lecture.Source.KTH_RAW:
             download_mp4_from_m3u8(lecture.raw_content_link, lecture)
-
-            logger.info('rq-queueing preview capture job')
-            from jobs import get_metadata_queue
-            q = next(get_metadata_queue())
-            q.enqueue(
-                jobs.capture_preview.job,
-                lecture.public_id,
-                lecture.language,
-                job_timeout=jobs.capture_preview.TIMEOUT
-            )
         else:
             raise ValueError(f'unsupported lecture source {lecture.source}')
 

tests/conftest.py

@@ -171,3 +171,8 @@ def mp4_file():
 @pytest.fixture
 def mp3_file():
     return os.path.join(os.path.dirname(__file__), 'files') + '/example.mp3'
+
+
+@pytest.fixture
+def img_file():
+    return os.path.join(os.path.dirname(__file__), 'files') + '/example.png'

tests/feature/jobs/test_capture_preview.py

@@ -0,0 +1,27 @@
+import shutil
+
+from jobs import capture_preview
+from db.models import Lecture
+
+
+def test_download_of_kth_play_lecture_saves_mp4_file(mocker, mp4_file, img_file):
+    def save_photo_from_video(mp4_filepath: str, output_file: str, small=False):
+        shutil.copy(img_file, output_file)
+
+    func = mocker.patch('tools.video.img.save_photo_from_video', side_effect=save_photo_from_video)
+
+    lecture = Lecture(
+        public_id='some_id',
+        language=Lecture.Language.SWEDISH,
+        mp4_filepath=mp4_file,
+        approved=True,
+        source=Lecture.Source.KTH,
+    )
+    lecture.save()
+
+    capture_preview.job(lecture.public_id, lecture.language)
+    lecture.refresh()
+
+    assert lecture.img_preview == lecture.preview_filename()
+    assert lecture.img_preview_small == lecture.preview_small_filename()
+    assert func.call_count == 2

web-ui/src/components/preview.less

@@ -56,6 +56,10 @@
   color: black;
 }
 
+.source {
+  color: black;
+}
+
 .length {
   color: black;
 }

web-ui/src/components/preview.tsx

@@ -50,6 +50,8 @@ export function PreviewCompact(props: PreviewCompactProps) {
     icon = youtubeLogoSmall;
   } else if (lecture.source === 'kth') {
     icon = kthLogo;
+  } else if (lecture.source === 'kth_raw') {
+    icon = kthLogo;
   }
 
   return (
@@ -134,6 +136,15 @@ export default function Preview(props: PreviewProps) {
     flagIcon = enFlag;
   }
 
+  let icon = '';
+  if (lecture.source === 'youtube') {
+    icon = youtubeLogoSmall;
+  } else if (lecture.source === 'kth') {
+    icon = kthLogo;
+  } else if (lecture.source === 'kth_raw') {
+    icon = kthLogo;
+  }
+
   let dateString = '';
   if (lecture.date !== null) {
     dateString = new Date(lecture.date).toISOString().split('T')[0];
@@ -186,6 +197,18 @@ export default function Preview(props: PreviewProps) {
                           preview={false}
                         />
                       </Col>
+                      <Col>
+                        <h3 className={styles.source}>From</h3>
+                      </Col>
+                      <Col>
+                        <Image
+                          src={icon}
+                          height={20}
+                          width={20}
+                          className={styles.logo}
+                          preview={false}
+                        />
+                      </Col>
                     </Space>
                   </Row>
                 </div>