[Refact] Refactoring code & Add docstring

- Docstring을 이용하여 api에 대한 설명을 작성.
- 코드 리팩토링 진행
issue #91

serving/backend/app/api/face.py

@@ -28,7 +28,6 @@ class People(BaseModel):
     people_img: Dict[str, bytes]
     file_name: str
     message: Optional[str]
-    # created_at : datetime = Field(default_factory=datetime.now)
 
 
 class ItemValue(BaseModel):
@@ -39,38 +38,29 @@ class ItemValue(BaseModel):
 # get으로 바꾸어 사진 보여주기
 @router.get("/show-people", description="face clustering으로 추출한 인물의 사진을 보여줍니다.")
 def show_people(id: UUID):
-    # 사진 넣어주기
+    """face clustering으로 추출한 인물의 사진을 보여주는 api
+
+    Args:
+        id (UUID): clustering 결과에 접근하기 위한, 식별자
+
+    Returns:
+        Response(dict):
+            id (str) : 영상을 구분할 수 있는 구분자.
+            people_img (dict) : 인물의 대표이미지를 gcs에 올린 후, gcs에 접근할 수 있는 url을 제공
+    """
+
     people_img = {}
     result_path = os.path.join(FILE_DIR, str(id), 'result', 'result.npy')
     result_data = np.load(result_path, allow_pickle=True).item()
 
-    # people_img = {}
-    # result_path = os.path.join(FILE_DIR, str(id), 'result')
-    # dir_list = os.listdir(result_path)
-    # people_list = [dir for dir in dir_list if dir.startswith('person')]
-
     for person in result_data.keys():
-        # 현재는 첫 번째 이미지를 가져옴. 이후에 다른 이미지를 가져오는 알고리즘이 있다면 사용하기
-        # img_file = os.listdir(os.path.join(str(result_path), str(person)))[0]   # first_image
-        # img_path = os.path.join(str(result_path), str(person), str(img_file))
         img_path = result_data[person]['repr_img_path']
 
         blob_dir = os.path.join(str(id), 'people', person)
         blob = bucket.blob(blob_dir)
         blob.upload_from_filename(img_path)
         people_img[person] = blob_dir
 
-    # saving wav file to GCS for STT
-    server_path = os.path.join(FILE_DIR, str(id))
-    wav_path = os.path.join(server_path, 'original_audio')
-    original_file = [ori_file for ori_file in os.listdir(server_path) if ori_file.startswith('original.')]
-    if original_file:
-        video_path = os.path.join(server_path, original_file[0])
-        os.system('ffmpeg -i {} -acodec pcm_s16le -ar 16000 {}.wav'.format(video_path, wav_path))
-    blob = bucket.blob(os.path.join(str(id), 'original_audio.wav'))
-    blob.upload_from_filename(wav_path + '.wav')
-
-    # print(people_img)
     return {"id": id, "people_img": people_img}
 
 
@@ -101,14 +91,5 @@ def get_timeline_face(info: dict):
 
     save_path = os.path.join(FILE_DIR, info['id'], 'face_timelines.npy')
     np.save(save_path, timelines)
-    # timelines = {}
-
-    # for face in info['face']:
-    #     image_file = os.listdir(os.path.join(result_path, 'result', face))[0]
-
-    #     image = os.path.join(result_path, 'result', face, image_file)
 
-    #     timeline = FaceRecognition(video, [image])
-    #     timelines[face] = timeline
-    # FE에서 선택한 사람을 받아 face recognition 진행 예정
     return {"id" : info['id']}

serving/backend/app/api/video.py

@@ -14,13 +14,6 @@
 bucket_name = 'snowman-bucket'
 bucket = storage_client.bucket(bucket_name)
 
-
-# # google cloud storage
-# from google.cloud import storage
-# storage_client = storage.Client()
-# bucket_name = 'snowman-bucket'
-# bucket = storage_client.batch(bucket_name)
-
 from ml.face_functions import FaceClustering
 
 
@@ -38,8 +31,18 @@ class Video(BaseModel):
 
 @router.post("/upload-video", description="비디오를 업로드합니다.")
 def create_video_file(file: UploadFile = File(...)):
+    """Upload video to server, and execute Face Clustering
+
+    Args:
+        file (UploadFile, optional): formData come from GCP Server. Defaults to File(...).
+
+    Returns:
+        Video (BaseModel):
+            id (UUID): 비디오를 구분할 수 있는 구분자
+            file_name (str): 파일의 이름
+            created_at (datetime): 파일이 업로드된 시간.
+    """
     new_video = Video(file_name=file.filename)
-    # video_contents = await file.read()
     os.makedirs(os.path.join(FILE_DIR, str(new_video.id)))
     id_path = os.path.join(FILE_DIR, str(new_video.id))
     server_path = os.path.join(id_path, ('original' + os.path.splitext(file.filename)[1]))
@@ -53,13 +56,24 @@ def create_video_file(file: UploadFile = File(...)):
 
 class YTVideo(BaseModel):
     id: UUID = Field(default_factory=uuid4)
-    video: Optional[str]
     file_name: str
     created_at : datetime = Field(default_factory=datetime.now)
 
 
 @router.post("/upload-video-youtube", description="유튜브 URL을 이용하여 비디오를 업로드합니다.")
 def create_video_file_from_youtube(info: dict):
+    """유튜브 URL을 이용하여 비디오를 업로드하는 api
+
+    Args:
+        info (dict): 
+            url (str): youtube영상을 업로드 하기 위한 url을 담고 있음
+
+    Returns:
+        YTVideo (BaseModel): 
+            id (UUID): 비디오를 구분할 수 있는 구분자
+            file_name (str): 파일의 이름
+            created_at (datetime) : 파일이 업로드 된 시간
+    """
     yt_video = YouTube(info['url'])
     new_video = YTVideo(file_name=yt_video.title + '.mp4')
 

serving/backend/app/ml/face_functions.py

@@ -17,8 +17,6 @@
 
 ########## Face Clustering ############
 def FaceClustering(video_path: str = "", save_dir:str = ""):
-    # Load config
-    # cfg = load_json('./ml/config.json')
 
     # Initialize Face Extractor
     extractor = FaceExtractor(
@@ -62,8 +60,6 @@ def FinalTimeline(laugh_timeline : list, people_timeline : dict, id : str):
 
     shorts = []
     for target_person in iter(people_timeline.keys()):
-        print(target_person)
-        print(people_timeline[target_person])
         if people_timeline[target_person] == [[]]:
             continue
         final_timeline, total_length = make_final_timeline(laugh_timeline, people_timeline[target_person])

serving/backend/app/ml/final_shorts/final_timeline.py

@@ -33,7 +33,7 @@ def make_final_timeline(laughter_timeline,person_timeline,max_length=None):
         sorted_timeline = sorted(final,key=lambda x:-x[2])
         choose_index = [False for _ in range(len(sorted_timeline))]
         total_length = 0
-        for start,end,interest,ratio in sorted_timeline:
+        for start,end,interest,ratio, duration in sorted_timeline:
             length = end - start
             if total_length + length > max_length:
                 break

serving/backend/app/ml/final_shorts/make_shorts.py

@@ -17,24 +17,17 @@ def make_shorts(final_highlights, total_length, id, target_person):
     SHORTS_STORAGE_DIR = os.path.join(id, 'shorts')
     os.makedirs(SHORTS_DIR, exist_ok=True)
 
-
     VIDEO_DIR = os.path.join(FILE_DIR, id, "original.mp4")
-    print(VIDEO_DIR)
-
-    # CURRENT_DIR = os.getcwd()
 
     in_file = ffmpeg.input(VIDEO_DIR)
 
     target_person_shorts = []
     for idx, (start, end, interest, ratio, during) in enumerate(final_highlights):
-        print("Making Clips...")
         HIGHLIGHT_PATH = os.path.join(SHORTS_DIR, f"short_{target_person}_{idx}.mp4")
 
         # save to gcs
         HIGHLIGHT_STORAGE_DIR = os.path.join(SHORTS_STORAGE_DIR, f"short_{target_person}_{idx}.mp4")
         blob = bucket.blob(HIGHLIGHT_STORAGE_DIR)
-
-        trim_and_fade(VIDEO_DIR, start, end, HIGHLIGHT_PATH)
 
         vid = (
             in_file.video
@@ -59,14 +52,3 @@ def make_shorts(final_highlights, total_length, id, target_person):
         target_person_shorts.append([target_person, HIGHLIGHT_STORAGE_DIR, during, interest])
 
     return target_person_shorts
-
-
-def trim_and_fade(original_path, start, end, save_path):
-    print(f"start making {save_path}")
-    os.system(f'ffmpeg -ss {start} -i {original_path} -to {end} -filter_complex \
-         "fade=in:st={start}:d=1, fade=out:st={end-1}:d=1; \
-         afade=in:st={start}:d=1, afade=out:st={end-1}:d=1" \
-        -c:v copy -c:a copy {save_path}')
-    # os.system(f'ffmpeg -i {original_path} -vf "trim=start={start}:end={end}, fade=in:st={start}:d=1, fade=out:st={end-1}:d=1" \
-    #     -af "atrim=start={start}:end={end}, afade=in:st={start}:d=1, afade=out:st={end-1}:d=1"\
-    #      {save_path}')

serving/backend/app_laughter/api/laughter.py

@@ -33,6 +33,16 @@ class VideoTimeline(BaseModel):
 
 @router.post("/laughter-detection", description="laughter timeline을 추출하는 전과정을 수행합니다.")
 def laughter_detection(file: UploadFile = File(...)):
+    """Upload된 영상을 통하여 laughter timeline을 추출하는 전 과정을 수행합니다.
+
+    Args:
+        file (UploadFile, optional): formData로 보내진 video data. Defaults to File(...).
+
+    Returns:
+        VideoTimeline (BaseModel): 
+            id (UUID): 영상을 구분할 수 있는 구분자.
+            laugh (List[Tuple], optional): 전체 영상에 대하여 웃음에 해당하는 timeline.
+    """
     # download the uploaded video
     new_video = Video(file_name=file.filename)
     os.makedirs(os.path.join(FILE_DIR, str(new_video.id_laughter)))
@@ -56,6 +66,17 @@ def laughter_detection(file: UploadFile = File(...)):
 
 @router.post("/laughter-detection-youtube", description="유튜브 URL을 이용하여 laughter timeline을 추출하는 전과정을 수행합니다.")
 def laughter_detection_from_youtube(info: dict):
+    """유튜브 URL을 이용하여 laughter timeline을 추출하는 전 과정을 수행합니다.
+
+    Args:
+        info (dict):
+            url (str): 유튜브 영상 url
+
+    Returns:
+        VideoTimeline: 
+            id (UUID): 영상을 구분할 수 있는 구분자.
+            laugh (List[Tuple], optional): 전체 영상에 대하여 웃음에 해당하는 timeline.
+    """
     yt_video = YouTube(info['url'])
     new_video = Video(file_name=yt_video.title + '.mp4')
 

serving/backend/app_laughter/ml/laughter_detection.py

@@ -28,6 +28,5 @@ def LaughterDetection(video_path: str = "", wav_path: str = "", ml_path: str = "
 
     # calculate interest and make final timeline
     final_laughter_timeline = detector.calculate_interest(laughter_timeline)
-    print(final_laughter_timeline)
 
     return final_laughter_timeline