MAJOR CHANGE: drop srt files, using only srt strings (wip); added moviepy font and color dropdowns + loading spinner (wip)

main.py

@@ -6,13 +6,22 @@ from utils.zip_response import zip_response
 from utils.read_html import read_html
 
 from fastapi import FastAPI, UploadFile, HTTPException, Form, Depends
+from fastapi.staticfiles import StaticFiles
 from fastapi.responses import HTMLResponse, Response
 from fastapi.security import HTTPBasic
 from pydantic import BaseModel, field_validator
 
+## THIS IS A BREAKING CHANGE. SRT FILE INPUT DEPRECATED. WIP.
+## TODO: add word level highlighting option
+## TODO: separate transcriber from subtitler logic + allow for interactive validation of trancription in-browser
+## TODO: add video preview component
+## TODO: improve loading spinner.
+
 app = FastAPI()
 security = HTTPBasic()
 api_configs_file = os.path.abspath("api_config.yml")
+static_dir = os.path.join(os.path.dirname(__file__), 'static')
+app.mount("/static", StaticFiles(directory=static_dir), name="static")
 
 class MP4Video(BaseModel):
     video_file: UploadFile
@@ -30,25 +39,6 @@ class MP4Video(BaseModel):
             raise HTTPException(status_code=500, detail='Invalid video file type. Please upload an MP4 file.')
         return v
 
-class SRTFile(BaseModel):
-    srt_file: Optional[UploadFile] = None
-    
-    @property
-    def filename(self):
-        return self.srt_file.filename
-    @property
-    def file(self):
-        return self.srt_file.file
-    @property
-    def size(self):
-        return self.srt_file.size
-
-    @field_validator('srt_file')
-    def validate_srt_file(cls, v):
-        if v.size > 0 and not v.filename.endswith('.srt'):
-            raise HTTPException(status_code=422, detail='Invalid subtitle file type. Please upload an SRT file.')
-        return v
-
 @app.get("/")
 async def root():
     html_content = f"""
@@ -73,7 +63,6 @@ async def get_temp_dir():
 
 @app.post("/process_video/")
 async def process_video_api(video_file: MP4Video = Depends(),
-                            srt_file: SRTFile = Depends(),
                             task: Optional[str] = Form("transcribe"),
                             model_version: Optional[str] = Form("deepdml/faster-whisper-large-v3-turbo-ct2"),
                             max_words_per_line: Optional[int] = Form(6),
@@ -93,25 +82,18 @@ async def process_video_api(video_file: MP4Video = Depends(),
             finally:
                 video_file.file.close()
             logging.info("Copying SRT UploadFile to the temp_input_path")
-            if srt_file.size > 0:
-                with open(os.path.join(temp_dir, f"{video_file.filename.split('.')[0]}.srt"), 'w+b') as temp_srt_file:
-                    try:
-                        shutil.copyfileobj(srt_file.file, temp_srt_file)
-                    finally:
-                        srt_file.file.close()
-                logging.info("Processing the video...")
-                output_path, _ = process_video(temp_file.name, temp_srt_file.name, task, model_version, max_words_per_line, fontsize, font, bg_color, text_color, caption_mode)
-                logging.info("Zipping response...")
-                with open(os.path.join(temp_dir, f"{video_file.filename.split('.')[0]}.zip"), 'w+b') as temp_zip_file:
-                    zip_file = zip_response(temp_zip_file.name, [output_path, srt_path])
-                return Response(content = zip_file)
-            with open(os.path.join(temp_dir, f"{video_file.filename.split('.')[0]}.srt"), 'w+b') as temp_srt_file:
+            with open(os.path.join(temp_dir, f"{video_file.filename.split('.')[0]}.srt"), 'w+') as temp_srt_file:
                 logging.info("Processing the video...")
-                output_path, srt_path = process_video(temp_file.name, None, task, model_version, max_words_per_line, fontsize, font, bg_color, text_color, caption_mode, api_configs_file)
-                logging.info("Zipping response...")
-                with open(os.path.join(temp_dir, f"{video_file.filename.split('.')[0]}.zip"), 'w+b') as temp_zip_file:
-                    zip_file = zip_response(temp_zip_file.name, [output_path, srt_path])
-                return Response(content = zip_file)  
+                output_path, srt_string = process_video(temp_file.name, task, model_version, max_words_per_line, fontsize, font, bg_color, text_color, caption_mode, api_configs_file)
+                temp_srt_file.write(srt_string)    
+            logging.info("Zipping response...")
+            with open(os.path.join(temp_dir, f"{video_file.filename.split('.')[0]}.zip"), 'w+b') as temp_zip_file:
+                zip_file = zip_response(temp_zip_file.name, [output_path, temp_srt_file.name])
+                return Response(
+                    content = zip_file,
+                    media_type="application/zip",
+                    headers={"Content-Disposition": f"attachment; filename={video_file.filename.split('.')[0]}.zip"}
+                    )
     except Exception as e:
         raise HTTPException(status_code=500, detail=str(e))
     

static/colors.txt

@@ -0,0 +1,678 @@
+AliceBlue
+AntiqueWhite
+AntiqueWhite1
+AntiqueWhite2
+AntiqueWhite3
+AntiqueWhite4
+aqua
+aquamarine
+aquamarine1
+aquamarine2
+aquamarine3
+aquamarine4
+azure
+azure1
+azure2
+azure3
+azure4
+beige
+bisque
+bisque1
+bisque2
+bisque3
+bisque4
+black
+BlanchedAlmond
+blue
+blue1
+blue2
+blue3
+blue4
+BlueViolet
+brown
+brown1
+brown2
+brown3
+brown4
+burlywood
+burlywood1
+burlywood2
+burlywood3
+burlywood4
+CadetBlue
+CadetBlue1
+CadetBlue2
+CadetBlue3
+CadetBlue4
+chartreuse
+chartreuse1
+chartreuse2
+chartreuse3
+chartreuse4
+chocolate
+chocolate1
+chocolate2
+chocolate3
+chocolate4
+coral
+coral1
+coral2
+coral3
+coral4
+CornflowerBlue
+cornsilk
+cornsilk1
+cornsilk2
+cornsilk3
+cornsilk4
+crimson
+cyan
+cyan1
+cyan2
+cyan3
+cyan4
+DarkBlue
+DarkCyan
+DarkGoldenrod
+DarkGoldenrod1
+DarkGoldenrod2
+DarkGoldenrod3
+DarkGoldenrod4
+DarkGray
+DarkGreen
+DarkGrey
+DarkKhaki
+DarkMagenta
+DarkOliveGreen
+DarkOliveGreen1
+DarkOliveGreen2
+DarkOliveGreen3
+DarkOliveGreen4
+DarkOrange
+DarkOrange1
+DarkOrange2
+DarkOrange3
+DarkOrange4
+DarkOrchid
+DarkOrchid1
+DarkOrchid2
+DarkOrchid3
+DarkOrchid4
+DarkRed
+DarkSalmon
+DarkSeaGreen
+DarkSeaGreen1
+DarkSeaGreen2
+DarkSeaGreen3
+DarkSeaGreen4
+DarkSlateBlue
+DarkSlateGray
+DarkSlateGray1
+DarkSlateGray2
+DarkSlateGray3
+DarkSlateGray4
+DarkSlateGrey
+DarkTurquoise
+DarkViolet
+DeepPink
+DeepPink1
+DeepPink2
+DeepPink3
+DeepPink4
+DeepSkyBlue
+DeepSkyBlue1
+DeepSkyBlue2
+DeepSkyBlue3
+DeepSkyBlue4
+DimGray
+DimGrey
+DodgerBlue
+DodgerBlue1
+DodgerBlue2
+DodgerBlue3
+DodgerBlue4
+firebrick
+firebrick1
+firebrick2
+firebrick3
+firebrick4
+FloralWhite
+ForestGreen
+fractal
+freeze
+fuchsia
+gainsboro
+GhostWhite
+gold
+gold1
+gold2
+gold3
+gold4
+goldenrod
+goldenrod1
+goldenrod2
+goldenrod3
+goldenrod4
+gray
+gray
+gray0
+gray1
+gray10
+gray100
+gray100
+gray11
+gray12
+gray13
+gray14
+gray15
+gray16
+gray17
+gray18
+gray19
+gray2
+gray20
+gray21
+gray22
+gray23
+gray24
+gray25
+gray26
+gray27
+gray28
+gray29
+gray3
+gray30
+gray31
+gray32
+gray33
+gray34
+gray35
+gray36
+gray37
+gray38
+gray39
+gray4
+gray40
+gray41
+gray42
+gray43
+gray44
+gray45
+gray46
+gray47
+gray48
+gray49
+gray5
+gray50
+gray51
+gray52
+gray53
+gray54
+gray55
+gray56
+gray57
+gray58
+gray59
+gray6
+gray60
+gray61
+gray62
+gray63
+gray64
+gray65
+gray66
+gray67
+gray68
+gray69
+gray7
+gray70
+gray71
+gray72
+gray73
+gray74
+gray75
+gray76
+gray77
+gray78
+gray79
+gray8
+gray80
+gray81
+gray82
+gray83
+gray84
+gray85
+gray86
+gray87
+gray88
+gray89
+gray9
+gray90
+gray91
+gray92
+gray93
+gray94
+gray95
+gray96
+gray97
+gray98
+gray99
+green
+green
+green1
+green2
+green3
+green4
+GreenYellow
+grey
+grey0
+grey1
+grey10
+grey100
+grey11
+grey12
+grey13
+grey14
+grey15
+grey16
+grey17
+grey18
+grey19
+grey2
+grey20
+grey21
+grey22
+grey23
+grey24
+grey25
+grey26
+grey27
+grey28
+grey29
+grey3
+grey30
+grey31
+grey32
+grey33
+grey34
+grey35
+grey36
+grey37
+grey38
+grey39
+grey4
+grey40
+grey41
+grey42
+grey43
+grey44
+grey45
+grey46
+grey47
+grey48
+grey49
+grey5
+grey50
+grey51
+grey52
+grey53
+grey54
+grey55
+grey56
+grey57
+grey58
+grey59
+grey6
+grey60
+grey61
+grey62
+grey63
+grey64
+grey65
+grey66
+grey67
+grey68
+grey69
+grey7
+grey70
+grey71
+grey72
+grey73
+grey74
+grey75
+grey76
+grey77
+grey78
+grey79
+grey8
+grey80
+grey81
+grey82
+grey83
+grey84
+grey85
+grey86
+grey87
+grey88
+grey89
+grey9
+grey90
+grey91
+grey92
+grey93
+grey94
+grey95
+grey96
+grey97
+grey98
+grey99
+honeydew
+honeydew1
+honeydew2
+honeydew3
+honeydew4
+HotPink
+HotPink1
+HotPink2
+HotPink3
+HotPink4
+IndianRed
+IndianRed1
+IndianRed2
+IndianRed3
+IndianRed4
+indigo
+ivory
+ivory1
+ivory2
+ivory3
+ivory4
+khaki
+khaki1
+khaki2
+khaki3
+khaki4
+lavender
+LavenderBlush
+LavenderBlush1
+LavenderBlush2
+LavenderBlush3
+LavenderBlush4
+LawnGreen
+LemonChiffon
+LemonChiffon1
+LemonChiffon2
+LemonChiffon3
+LemonChiffon4
+LightBlue
+LightBlue1
+LightBlue2
+LightBlue3
+LightBlue4
+LightCoral
+LightCyan
+LightCyan1
+LightCyan2
+LightCyan3
+LightCyan4
+LightGoldenrod
+LightGoldenrod1
+LightGoldenrod2
+LightGoldenrod3
+LightGoldenrod4
+LightGoldenrodYellow
+LightGray
+LightGreen
+LightGrey
+LightPink
+LightPink1
+LightPink2
+LightPink3
+LightPink4
+LightSalmon
+LightSalmon1
+LightSalmon2
+LightSalmon3
+LightSalmon4
+LightSeaGreen
+LightSkyBlue
+LightSkyBlue1
+LightSkyBlue2
+LightSkyBlue3
+LightSkyBlue4
+LightSlateBlue
+LightSlateGray
+LightSlateGrey
+LightSteelBlue
+LightSteelBlue1
+LightSteelBlue2
+LightSteelBlue3
+LightSteelBlue4
+LightYellow
+LightYellow1
+LightYellow2
+LightYellow3
+LightYellow4
+lime
+LimeGreen
+linen
+magenta
+magenta1
+magenta2
+magenta3
+magenta4
+maroon
+maroon
+maroon1
+maroon2
+maroon3
+maroon4
+matte
+MediumAquamarine
+MediumBlue
+MediumForestGreen
+MediumGoldenRod
+MediumOrchid
+MediumOrchid1
+MediumOrchid2
+MediumOrchid3
+MediumOrchid4
+MediumPurple
+MediumPurple1
+MediumPurple2
+MediumPurple3
+MediumPurple4
+MediumSeaGreen
+MediumSlateBlue
+MediumSpringGreen
+MediumTurquoise
+MediumVioletRed
+MidnightBlue
+MintCream
+MistyRose
+MistyRose1
+MistyRose2
+MistyRose3
+MistyRose4
+moccasin
+NavajoWhite
+NavajoWhite1
+NavajoWhite2
+NavajoWhite3
+NavajoWhite4
+navy
+NavyBlue
+none
+OldLace
+olive
+OliveDrab
+OliveDrab1
+OliveDrab2
+OliveDrab3
+OliveDrab4
+opaque
+orange
+orange1
+orange2
+orange3
+orange4
+OrangeRed
+OrangeRed1
+OrangeRed2
+OrangeRed3
+OrangeRed4
+orchid
+orchid1
+orchid2
+orchid3
+orchid4
+PaleGoldenrod
+PaleGreen
+PaleGreen1
+PaleGreen2
+PaleGreen3
+PaleGreen4
+PaleTurquoise
+PaleTurquoise1
+PaleTurquoise2
+PaleTurquoise3
+PaleTurquoise4
+PaleVioletRed
+PaleVioletRed1
+PaleVioletRed2
+PaleVioletRed3
+PaleVioletRed4
+PapayaWhip
+PeachPuff
+PeachPuff1
+PeachPuff2
+PeachPuff3
+PeachPuff4
+peru
+pink
+pink1
+pink2
+pink3
+pink4
+plum
+plum1
+plum2
+plum3
+plum4
+PowderBlue
+purple
+purple
+purple1
+purple2
+purple3
+purple4
+red
+red1
+red2
+red3
+red4
+RosyBrown
+RosyBrown1
+RosyBrown2
+RosyBrown3
+RosyBrown4
+RoyalBlue
+RoyalBlue1
+RoyalBlue2
+RoyalBlue3
+RoyalBlue4
+SaddleBrown
+salmon
+salmon1
+salmon2
+salmon3
+salmon4
+SandyBrown
+SeaGreen
+SeaGreen1
+SeaGreen2
+SeaGreen3
+SeaGreen4
+seashell
+seashell1
+seashell2
+seashell3
+seashell4
+sienna
+sienna1
+sienna2
+sienna3
+sienna4
+silver
+SkyBlue
+SkyBlue1
+SkyBlue2
+SkyBlue3
+SkyBlue4
+SlateBlue
+SlateBlue1
+SlateBlue2
+SlateBlue3
+SlateBlue4
+SlateGray
+SlateGray1
+SlateGray2
+SlateGray3
+SlateGray4
+SlateGrey
+snow
+snow1
+snow2
+snow3
+snow4
+SpringGreen
+SpringGreen1
+SpringGreen2
+SpringGreen3
+SpringGreen4
+SteelBlue
+SteelBlue1
+SteelBlue2
+SteelBlue3
+SteelBlue4
+tan
+tan1
+tan2
+tan3
+tan4
+teal
+thistle
+thistle1
+thistle2
+thistle3
+thistle4
+tomato
+tomato1
+tomato2
+tomato3
+tomato4
+transparent
+turquoise
+turquoise1
+turquoise2
+turquoise3
+turquoise4
+violet
+VioletRed
+VioletRed1
+VioletRed2
+VioletRed3
+VioletRed4
+wheat
+wheat1
+wheat2
+wheat3
+wheat4
+white
+WhiteSmoke
+yellow
+yellow1
+yellow2
+yellow3
+yellow4
+YellowGreen

static/fonts.txt

@@ -0,0 +1,141 @@
+Courier
+Helvetica-Narrow-BoldOblique
+AvantGarde-Book
+AvantGarde-BookOblique
+AvantGarde-Demi
+AvantGarde-DemiOblique
+Bookman-Demi
+Bookman-DemiItalic
+Bookman-Light
+Bookman-LightItalic
+Courier-Bold
+Courier-BoldOblique
+Courier-Oblique
+fixed
+Helvetica
+Helvetica-Bold
+Helvetica-BoldOblique
+Helvetica-Narrow
+Helvetica-Narrow-Bold
+Helvetica-Narrow-Oblique
+Helvetica-Oblique
+NewCenturySchlbk-Bold
+NewCenturySchlbk-BoldItalic
+NewCenturySchlbk-Italic
+NewCenturySchlbk-Roman
+Palatino-Bold
+Palatino-BoldItalic
+Palatino-Italic
+Palatino-Roman
+Symbol
+Times-Bold
+Times-BoldItalic
+Times-Italic
+Times-Roman
+C059-BdIta
+C059-Bold
+C059-Bold-Italic
+C059-Italic
+C059-Roman
+Century-Schoolbook-L-Bold
+Century-Schoolbook-L-Bold-Italic
+Century-Schoolbook-L-Italic
+Century-Schoolbook-L-Roman
+D050000L
+DejaVu-Sans
+DejaVu-Sans-Bold
+DejaVu-Sans-Mono
+DejaVu-Sans-Mono-Bold
+DejaVu-Serif
+DejaVu-Serif-Bold
+Dingbats
+Droid-Sans-Fallback
+Nimbus-Mono-L
+Nimbus-Mono-L-Bold
+Nimbus-Mono-L-Bold-Oblique
+Nimbus-Mono-L-Regular-Oblique
+Nimbus-Mono-PS
+Nimbus-Mono-PS-Bold
+Nimbus-Mono-PS-Bold-Italic
+Nimbus-Mono-PS-Italic
+Nimbus-Roman
+Nimbus-Roman-Bold
+Nimbus-Roman-Bold-Italic
+Nimbus-Roman-Italic
+Nimbus-Roman-No9-L
+Nimbus-Roman-No9-L-Medium
+Nimbus-Roman-No9-L-Medium-Italic
+Nimbus-Roman-No9-L-Regular-Italic
+Nimbus-Sans
+Nimbus-Sans-Bold
+Nimbus-Sans-Bold-Italic
+Nimbus-Sans-Italic
+Nimbus-Sans-L
+Nimbus-Sans-L-Bold
+Nimbus-Sans-L-Bold-Condensed
+Nimbus-Sans-L-Bold-Condensed-Italic
+Nimbus-Sans-L-Bold-Italic
+Nimbus-Sans-L-Regular-Condensed
+Nimbus-Sans-L-Regular-Condensed-Italic
+Nimbus-Sans-L-Regular-Italic
+Nimbus-Sans-Narrow
+Nimbus-Sans-Narrow-Bold
+Nimbus-Sans-Narrow-Bold-Oblique
+Nimbus-Sans-Narrow-Oblique
+NimbusMonoPS-Bold
+NimbusMonoPS-BoldItalic
+NimbusMonoPS-Italic
+NimbusMonoPS-Regular
+NimbusRoman-Bold
+NimbusRoman-BoldItalic
+NimbusRoman-Italic
+NimbusRoman-Regular
+NimbusSans-Bold
+NimbusSans-BoldItalic
+NimbusSans-Italic
+NimbusSans-Regular
+NimbusSansNarrow-Bold
+NimbusSansNarrow-BoldOblique
+NimbusSansNarrow-Oblique
+NimbusSansNarrow-Regular
+Noto-Mono
+Noto-Sans-Mono-Bold
+Noto-Sans-Mono-Regular
+P052-Bold
+P052-Bold-Italic
+P052-BoldItalic
+P052-Italic
+P052-Roman
+Standard-Symbols-L
+Standard-Symbols-PS
+URW-Bookman-Demi
+URW-Bookman-Demi-Italic
+URW-Bookman-L-Demi-Bold
+URW-Bookman-L-Demi-Bold-Italic
+URW-Bookman-L-Light
+URW-Bookman-L-Light-Italic
+URW-Bookman-Light
+URW-Bookman-Light-Italic
+URW-Chancery-L-Medium-Italic
+URW-Gothic-Book
+URW-Gothic-Book-Oblique
+URW-Gothic-Demi
+URW-Gothic-Demi-Oblique
+URW-Gothic-L-Book
+URW-Gothic-L-Book-Oblique
+URW-Gothic-L-Demi
+URW-Gothic-L-Demi-Oblique
+URW-Palladio-L-Bold
+URW-Palladio-L-Bold-Italic
+URW-Palladio-L-Italic
+URW-Palladio-L-Roman
+URWBookman-Demi
+URWBookman-DemiItalic
+URWBookman-Light
+URWBookman-LightItalic
+URWGothic-Book
+URWGothic-BookOblique
+URWGothic-Demi
+URWGothic-DemiOblique
+Z003-Medium-Italic
+Z003-MediumItalic

static/submit_video.html

@@ -151,51 +151,104 @@
         </style>
     </head>
     <body>
-        <form action="/process_video/" enctype="multipart/form-data" method="post">
-            <div class="form-wrapper">
-                <div class="form-group">
-                    <h3>Inputs & Task Selection</h3>
-                    <label for="video_file">Video File</label>
-                    <input type="file" id="video_file" name="video_file"><br>
-                    <label for="srt_file">Subtitles File</label>
-                    <input type="file" id="srt_file" name="srt_file"><br>
-                    <label for="task">Task</label>
-                    <select id="task" name="task">
-                        <option value="transcribe">Transcribe</option>
-                        <option value="translate">Translate</option>
-                    </select>
-                    <label for="model_version">Model Version</label>
-                    <select id="model_version" name="model_version">
-                        <option value="deepdml/faster-whisper-large-v3-turbo-ct2">faster-whisper-large-v3-turbo</option>
-                        <option value="large-v3">large-v3</option>
-                    </select>
-                </div>
-                <div class="form-group">
-                    <h3>Visual Parameters</h3>
-                    <label for="max_words_per_line">Max words per line</label>
-                    <input type="number" id="max_words_per_line" name="max_words_per_line" value="6"><br>
-                    <label for="fontsize">Font size</label>
-                    <input type="number" id="fontsize" name="fontsize" value="42"><br>
-                    <label for="font">Font:</label>
-                    <input type="text" id="font" name="font" value="FuturaPTHeavy"><br>
-                    <label for="bg_color">Background color</label>
-                    <input type="text" id="bg_color" name="bg_color" value="#00FFFF00"><br>
-                    <label for="text_color">Text color</label>
-                    <input type="text" id="text_color" name="text_color" value="white"><br>
-                    <label for="caption_mode">Caption mode</label>
-                    <select id="caption_mode" name="caption_mode">
-                        <option value="desktop">Desktop</option>
-                        <option value="mobile">Mobile</option>
-                    </select>
-                </div>
+    <form id="video-form" action="/process_video/" method="post" enctype="multipart/form-data">
+        <div class="form-wrapper">
+            <div class="form-group">
+                <h3>Inputs & Task Selection</h3>
+                <label for="video_file">Video File</label>
+                <input type="file" id="video_file" name="video_file"><br>
+
+                <label for="task">Task</label>
+                <select id="task" name="task">
+                    <option value="transcribe">Transcribe</option>
+                    <option value="translate">Translate</option>
+                </select>
+
+                <label for="model_version">Model Version</label>
+                <select id="model_version" name="model_version">
+                    <option value="deepdml/faster-whisper-large-v3-turbo-ct2">faster-whisper-large-v3-turbo</option>
+                    <option value="turbo">turbo</option>
+                    <option value="large-v3">large-v3</option>
+                </select>
             </div>
-            <input type="submit" value="Submit">
-       </form>
-        <!-- Footer -->
-        <div class="footer">
-            <p>Created by:</p>
-            <a href="https://github.com/marquesafonso" class="github"><i class="fab fa-github fa-3x"></i></a>
-            <a href="https://www.linkedin.com/in/marquesafonso" class="linkedin"><i class="fab fa-linkedin fa-3x"></i></a>
+
+            <div class="form-group">
+                <h3>Visual Parameters</h3>
+                <label for="max_words_per_line">Max words per line</label>
+                <input type="number" id="max_words_per_line" name="max_words_per_line" value="6"><br>
+
+                <label for="fontsize">Font size</label>
+                <input type="number" id="fontsize" name="fontsize" value="42"><br>
+
+                <label for="font">Font</label>
+                <select id="font" name="font">
+                    <option>Loading fonts...</option>
+                </select><br>
+
+                <label for="bg_color">Background color</label>
+                <input type="text" id="bg_color" name="bg_color" value="#00FFFF00"><br>
+
+                <label for="text_color">Text color</label>
+                <select id="text_color" name="text_color">
+                    <option>Loading colors...</option>
+                </select><br>
+
+                <label for="caption_mode">Caption mode</label>
+                <select id="caption_mode" name="caption_mode">
+                    <option value="desktop">Desktop</option>
+                    <option value="mobile">Mobile</option>
+                </select>
+            </div>
+        </div>
+
+        <div id="loading" style="display:none; text-align: center; margin-top: 10px; font-weight: bold;">
+            <i class="fas fa-spinner fa-spin"></i> Processing, please wait...
         </div>
+
+        <input type="submit" value="Submit">
+    </form>
+    <div class="footer">
+        <p>Created by:</p>
+        <a href="https://github.com/marquesafonso" class="github"><i class="fab fa-github fa-3x"></i></a>
+        <a href="https://www.linkedin.com/in/marquesafonso" class="linkedin"><i class="fab fa-linkedin fa-3x"></i></a>
+    </div>
+    <script>
+        // Loading dropdown options
+        function populateDropdown(id, url, defaultValue = null) {
+            fetch(url)
+                .then(response => response.text())
+                .then(data => {
+                    const select = document.getElementById(id);
+                    select.innerHTML = '';
+                    const lines = data.split('\n').map(x => x.trim()).filter(Boolean);
+
+                    lines.forEach(item => {
+                        const opt = document.createElement('option');
+                        opt.value = item;
+                        opt.textContent = item;
+                        if (item === defaultValue) {
+                            opt.selected = true;
+                        }
+                        select.appendChild(opt);
+                    });
+                })
+                .catch(error => {
+                    console.error(`Error loading ${url}:`, error);
+                });
+        }
+        // default font and color:
+        const DEFAULT_FONT = "Helvetica";
+        const DEFAULT_COLOR = "white";
+        // Init dropdowns
+        populateDropdown('font', '/static/fonts.txt', DEFAULT_FONT);
+        populateDropdown('text_color', '/static/colors.txt', DEFAULT_COLOR);
+
+        // Show loading spinner on submit
+        document.getElementById('video-form').addEventListener('submit', function () {
+            document.getElementById('loading').style.display = 'block';
+        });
+    </script>
     </body>
 </html>
+
+

utils/get_moviepy_configs.py

@@ -0,0 +1,8 @@
+from moviepy.editor import TextClip
+
+with open('static/fonts.txt', 'w', encoding='utf-8') as fonts:
+    fonts.write("\n".join(TextClip.list('font')))
+
+with open('static/colors.txt', 'w', encoding='utf-8') as colors:
+    colors_list = [color.decode("utf-8") for color in TextClip.list('color')[3:]]
+    colors.write("\n".join(colors_list))

utils/process_video.py

@@ -3,7 +3,6 @@ from utils.transcriber import transcriber
 from utils.subtitler import subtitler
 
 def process_video(invideo_file: str,
-                  srt_file: str | None,
                   task: str,
                   model_version: str,
                   max_words_per_line:int,
@@ -17,12 +16,7 @@ def process_video(invideo_file: str,
     invideo_path_parts = os.path.normpath(invideo_file).split(os.path.sep)
     VIDEO_NAME = os.path.basename(invideo_file)
     OUTVIDEO_PATH = os.path.join(os.path.normpath('/'.join(invideo_path_parts[:-1])), f"result_{VIDEO_NAME}")
-    if srt_file:
-        logging.info("Subtitling...")
-        subtitler(invideo_file, srt_file, OUTVIDEO_PATH, fontsize, font, bg_color, text_color, caption_mode)
-    else:
-        srt_file = os.path.normpath(f"{invideo_file.split('.')[0]}.srt")
-        transcriber(invideo_file, srt_file, max_words_per_line, task, model_version, config_file)
-        logging.info("Subtitling...")
-        subtitler(invideo_file, srt_file, OUTVIDEO_PATH, fontsize, font, bg_color, text_color, caption_mode)
-    return OUTVIDEO_PATH, srt_file
+    srt_string = transcriber(invideo_file, max_words_per_line, task, model_version, config_file)
+    logging.info("Subtitling...")
+    subtitler(invideo_file, srt_string, OUTVIDEO_PATH, fontsize, font, bg_color, text_color, caption_mode)
+    return OUTVIDEO_PATH, srt_string

utils/subtitler.py

@@ -1,10 +1,9 @@
 from moviepy.editor import VideoFileClip, CompositeVideoClip, TextClip
 import os
 
-def parse_srt(srt_file):
-    """Parse the SRT file and return a list of (start, end, text) for each subtitle."""
-    with open(srt_file, "r", encoding='utf-8') as file:
-        lines = file.readlines()
+def parse_srt(srt_string):
+    """Parse the SRT string and return a list of (start, end, text) for each subtitle."""
+    lines = srt_string.split("\n")
     i = 0
     subtitles = []
     while i < len(lines):
@@ -29,7 +28,7 @@ def filter_caption_width(caption_mode:str):
     return caption_width_ratio, caption_height_ratio
 
 def subtitler(video_file:str,
-            srt_path:str,
+            srt_string:str,
             output_file:str,
             fontsize:int,
             font: str,
@@ -37,12 +36,11 @@ def subtitler(video_file:str,
             text_color:str,
             caption_mode:str
             ):
-    """Add subtitles from an SRT file to a video."""
+    """Add subtitles from an SRT string to a video."""
     video_file = os.path.abspath(video_file)
-    srt_path = os.path.abspath(srt_path)
     output_file = os.path.abspath(output_file)
     clip = VideoFileClip(filename=video_file, target_resolution=None)
-    subtitles = parse_srt(srt_path)
+    subtitles = parse_srt(srt_string)
     subtitle_clips = []
     caption_width_ratio, caption_height_ratio = filter_caption_width(caption_mode)
     for start, end, text in subtitles:

utils/transcriber.py

@@ -1,19 +1,21 @@
 from gradio_client import Client, handle_file
 from utils.api_configs import api_configs
 
-def transcriber(invideo_file:str, srt_file:str,
-        max_words_per_line:int, task:str, model_version:str,
-        config_file:str):
+def transcriber(invideo_file:str, 
+                max_words_per_line:int,
+                task:str,
+                model_version:str,
+                config_file:str
+                ):
         HF_TOKEN = api_configs(config_file)["secrets"]["hf-token"]
         HF_SPACE = api_configs(config_file)["secrets"]["hf-space"]
         client = Client(HF_SPACE, hf_token=HF_TOKEN)
         result = client.predict(
-                video_input=handle_file(invideo_file),
+                file_input=handle_file(invideo_file),
+                file_type = "video",
                 max_words_per_line=max_words_per_line,
                 task=task,
                 model_version=model_version,
                 api_name="/predict"
         )
-        with open(srt_file, "w", encoding='utf-8') as file:
-                file.write(result[0])
-        return srt_file
+        return result[0]