Update srt_utils.py

srt_utils.py

@@ -18,6 +18,28 @@ def seconds_to_srt_time(seconds):
     ms = int((seconds % 1) * 1000)
     return f"{hours:02d}:{minutes:02d}:{secs:02d},{ms:03d}"
 
+def shift_srt_timestamps(srt_content, offset_seconds):
+    """Shifts all timestamps in SRT content by offset_seconds"""
+    subs = parse_srt(srt_content)
+    if not subs:
+        return srt_content
+        
+    shifted_srt = ""
+    for i, sub in enumerate(subs, 1):
+        start = sub['start'] + offset_seconds
+        end = sub['end'] + offset_seconds
+        
+        # Ensure non-negative
+        if start < 0: start = 0
+        if end < 1e-3: end = 1e-3 # avoid 0 overlap logic issues if possible
+        
+        start_str = seconds_to_srt_time(start)
+        end_str = seconds_to_srt_time(end)
+        
+        shifted_srt += f"{i}\n{start_str} --> {end_str}\n{sub['text']}\n\n"
+        
+    return shifted_srt.strip()
+
 def parse_srt(srt_content):
     """Parses SRT content into a list of dictionaries. Returns VALIDATED list."""
     pattern = re.compile(r"(\d+)\s*\n([^-\n]+?) --> ([^-\n]+?)\s*\n((?:(?!\d+\s*\n\d{1,2}:\d{2}).+\n?)*)", re.MULTILINE)
@@ -298,7 +320,7 @@ import subprocess
 import shutil
 import os
 
-def process_audio_for_transcription(input_file: str, has_bg_music: bool = False) -> str:
+def process_audio_for_transcription(input_file: str, has_bg_music: bool = False, time_start: float = None, time_end: float = None) -> str:
     """
     Process audio to maximize speech clarity.
     
@@ -316,7 +338,13 @@ def process_audio_for_transcription(input_file: str, has_bg_music: bool = False)
     
     input_filename = os.path.basename(input_file)
     input_stem = os.path.splitext(input_filename)[0]
-    final_output = os.path.join(output_dir, f"{input_stem}.processed.mp3")
+    
+    # Adicionar sufixo se houver corte, para evitar cache/conflito incorreto
+    suffix = ""
+    if time_start is not None: suffix += f"_s{int(time_start)}"
+    if time_end is not None: suffix += f"_e{int(time_end)}"
+    
+    final_output = os.path.join(output_dir, f"{input_stem}{suffix}.processed.mp3")
 
     ffmpeg_cmd = shutil.which("ffmpeg")
     if not ffmpeg_cmd:
@@ -384,11 +412,20 @@ def process_audio_for_transcription(input_file: str, has_bg_music: bool = False)
     cmd_convert = [
         ffmpeg_cmd, "-y",
         "-i", vocals_path,
+    ]
+    
+    # Apply cutting if requested (Output seeking for accuracy)
+    if time_start is not None:
+        cmd_convert.extend(["-ss", str(time_start)])
+    if time_end is not None:
+        cmd_convert.extend(["-to", str(time_end)])
+        
+    cmd_convert.extend([
         "-ac", "1", "-ar", "16000",
         "-af", filter_chain,
         "-c:a", "libmp3lame", "-q:a", "2",
         final_output
-    ]
+    ])
     
     try:
         subprocess.run(cmd_convert, check=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)