Spaces:

ulduldp
/

test-ffmpeg

Running

App Files Files Community

ulduldp commited on 13 days ago

Commit

4882d31

verified ·

1 Parent(s): 666181c

Update app.py

Browse files

Files changed (1) hide show

app.py +80 -18

app.py CHANGED Viewed

@@ -2,9 +2,9 @@ from flask import Flask, render_template_string, request, jsonify, send_from_dir
 import os
 import uuid
 import subprocess
-import textwrap
 from werkzeug.utils import secure_filename
 from faster_whisper import WhisperModel
 app = Flask(__name__)
@@ -18,6 +18,7 @@ os.makedirs(UPLOAD_FOLDER, exist_ok=True)
 os.makedirs(OUTPUT_FOLDER, exist_ok=True)
 os.makedirs(SUBTITLE_FOLDER, exist_ok=True)
 model = WhisperModel(
     "tiny",
     device="cpu",
@@ -145,7 +146,6 @@ video{
 </head>
 <body>
 <div class="container">
     <h1>Photo + Audio → Video</h1>
@@ -223,7 +223,6 @@ form.addEventListener("submit", async (e)=>{
     }
 });
 </script>
 </body>
 </html>
 """
@@ -251,26 +250,70 @@ def escape_ffmpeg_path(path: str) -> str:
         .replace("'", r"\'")
     )
-def hard_wrap_caption(text: str, max_width: int = 18, max_lines: int = 4) -> str:
     text = " ".join(text.strip().split())
     if not text:
         return ""
-    tokens = []
-    for word in text.split(" "):
-        if len(word) <= max_width:
-            tokens.append(word)
-        else:
-            # split long words so they can wrap too
-            tokens.extend([word[i:i + max_width] for i in range(0, len(word), max_width)])
     lines = []
     current = ""
     for token in tokens:
-        candidate = token if not current else f"{current} {token}"
-        if len(candidate) <= max_width:
-            current = candidate
         else:
             if current:
                 lines.append(current)
@@ -279,13 +322,20 @@ def hard_wrap_caption(text: str, max_width: int = 18, max_lines: int = 4) -> str
     if current:
         lines.append(current)
-    # keep it from becoming too tall
     if len(lines) > max_lines:
-        lines = lines[:max_lines]
     return "\n".join(lines)
 def make_ass_subtitles(segments, ass_path):
     header = """[Script Info]
 ScriptType: v4.00+
 PlayResX: 1080
@@ -296,7 +346,7 @@ WrapStyle: 2
 [V4+ Styles]
 Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
-Style: Default,Arial,40,&H00FFFFFF,&H000000FF,&H00000000,&H00000000,1,0,0,0,100,100,0,0,3,0,0,2,140,140,260,1
 [Events]
 Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
@@ -304,6 +354,9 @@ Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
     lines = [header]
     for seg in segments:
         text = seg["text"].strip()
         if not text:
@@ -312,9 +365,17 @@ Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
         start = ass_time(seg["start"])
         end = ass_time(seg["end"])
-        wrapped = hard_wrap_caption(text, max_width=18, max_lines=4)
         wrapped = ass_escape(wrapped).replace("\n", r"\N")
         dialogue = (
             f"Dialogue: 0,{start},{end},Default,,0,0,0,,"
             r"{\bord0\shad0\blur0\be0\1c&HFFFFFF&\3c&H000000&\4c&H000000&\3a&H00&\4a&H00}"
@@ -394,6 +455,7 @@ def generate():
         make_ass_subtitles(transcript, ass_path)
         safe_ass_path = escape_ffmpeg_path(os.path.abspath(ass_path))
         vf = (
             "scale=1080:1920:force_original_aspect_ratio=increase,"
             "crop=1080:1920,"

 import os
 import uuid
 import subprocess
 from werkzeug.utils import secure_filename
 from faster_whisper import WhisperModel
+from PIL import ImageFont
 app = Flask(__name__)
 os.makedirs(OUTPUT_FOLDER, exist_ok=True)
 os.makedirs(SUBTITLE_FOLDER, exist_ok=True)
+# Fast CPU model
 model = WhisperModel(
     "tiny",
     device="cpu",
 </head>
 <body>
 <div class="container">
     <h1>Photo + Audio → Video</h1>
     }
 });
 </script>
 </body>
 </html>
 """
         .replace("'", r"\'")
     )
+def find_font_path():
+    candidates = [
+        "/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf",
+        "/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf",
+        "/usr/share/fonts/truetype/liberation2/LiberationSans-Bold.ttf",
+        "/usr/share/fonts/truetype/liberation2/LiberationSans-Regular.ttf",
+        "/usr/share/fonts/truetype/freefont/FreeSansBold.ttf",
+        "/usr/share/fonts/truetype/freefont/FreeSans.ttf",
+    ]
+    for path in candidates:
+        if os.path.exists(path):
+            return path
+    return None
+FONT_PATH = find_font_path()
+def measure_text_width(font, text: str) -> int:
+    bbox = font.getbbox(text)
+    return bbox[2] - bbox[0]
+def pixel_wrap_text(text: str, font_path: str, font_size: int, max_width_px: int, max_lines: int = 5) -> str:
+    """
+    Wrap text based on actual pixel width, not character count.
+    Also splits long words if they exceed max_width_px.
+    """
     text = " ".join(text.strip().split())
     if not text:
         return ""
+    if font_path:
+        font = ImageFont.truetype(font_path, font_size)
+    else:
+        font = ImageFont.load_default()
+    words = text.split(" ")
     lines = []
     current = ""
+    def split_long_word(word: str):
+        if measure_text_width(font, word) <= max_width_px:
+            return [word]
+        parts = []
+        chunk = ""
+        for ch in word:
+            trial = chunk + ch
+            if measure_text_width(font, trial) <= max_width_px:
+                chunk = trial
+            else:
+                if chunk:
+                    parts.append(chunk)
+                chunk = ch
+        if chunk:
+            parts.append(chunk)
+        return parts
+    tokens = []
+    for word in words:
+        tokens.extend(split_long_word(word))
     for token in tokens:
+        trial = token if not current else f"{current} {token}"
+        if measure_text_width(font, trial) <= max_width_px:
+            current = trial
         else:
             if current:
                 lines.append(current)
     if current:
         lines.append(current)
     if len(lines) > max_lines:
+        # last line gets the rest so text doesn't disappear
+        kept = lines[:max_lines - 1]
+        rest = " ".join(lines[max_lines - 1:])
+        kept.append(rest)
+        lines = kept
     return "\n".join(lines)
 def make_ass_subtitles(segments, ass_path):
+    """
+    Solid black box behind white text.
+    Font size reduced and wrap based on pixel width.
+    """
     header = """[Script Info]
 ScriptType: v4.00+
 PlayResX: 1080
 [V4+ Styles]
 Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
+Style: Default,Arial,38,&H00FFFFFF,&H000000FF,&H00000000,&H00000000,1,0,0,0,100,100,0,0,3,0,0,2,120,120,220,1
 [Events]
 Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
     lines = [header]
+    # Available width inside 1080 frame with margins
+    max_width_px = 820
     for seg in segments:
         text = seg["text"].strip()
         if not text:
         start = ass_time(seg["start"])
         end = ass_time(seg["end"])
+        wrapped = pixel_wrap_text(
+            text=text,
+            font_path=FONT_PATH,
+            font_size=38,
+            max_width_px=max_width_px,
+            max_lines=5
+        )
         wrapped = ass_escape(wrapped).replace("\n", r"\N")
+        # BorderStyle=3 gives the opaque black box background
         dialogue = (
             f"Dialogue: 0,{start},{end},Default,,0,0,0,,"
             r"{\bord0\shad0\blur0\be0\1c&HFFFFFF&\3c&H000000&\4c&H000000&\3a&H00&\4a&H00}"
         make_ass_subtitles(transcript, ass_path)
         safe_ass_path = escape_ffmpeg_path(os.path.abspath(ass_path))
+        # IMPORTANT: crop first, then burn subtitles
         vf = (
             "scale=1080:1920:force_original_aspect_ratio=increase,"
             "crop=1080:1920,"