Update file_handler.py
Browse files- file_handler.py +296 -82
file_handler.py
CHANGED
|
@@ -1,4 +1,4 @@
|
|
| 1 |
-
# file_handler.py -
|
| 2 |
|
| 3 |
import os
|
| 4 |
import tempfile
|
|
@@ -12,7 +12,7 @@ try:
|
|
| 12 |
PYDUB_AVAILABLE = True
|
| 13 |
except ImportError:
|
| 14 |
PYDUB_AVAILABLE = False
|
| 15 |
-
st.warning("⚠️ Pydub nie jest dostępny.
|
| 16 |
|
| 17 |
try:
|
| 18 |
import librosa
|
|
@@ -24,86 +24,124 @@ except ImportError:
|
|
| 24 |
from config import FILE_PROCESSING, USER_MESSAGES
|
| 25 |
|
| 26 |
class FileHandler:
|
| 27 |
-
"""Klasa do obsługi plików audio/video -
|
| 28 |
|
| 29 |
def __init__(self):
|
| 30 |
-
self.temp_files = []
|
| 31 |
self.processing_stats = {}
|
| 32 |
|
| 33 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 34 |
"""
|
| 35 |
-
Główna funkcja przetwarzania pliku
|
| 36 |
-
Returns: Lista ścieżek do plików gotowych do transkrypcji
|
| 37 |
"""
|
| 38 |
try:
|
| 39 |
file_size_mb = uploaded_file.size / (1024 * 1024)
|
| 40 |
|
| 41 |
-
# Loguj rozpoczęcie przetwarzania
|
| 42 |
st.info(f"🔄 Przetwarzam {uploaded_file.name} ({file_size_mb:.1f}MB)")
|
| 43 |
|
| 44 |
-
# Sprawdź czy plik
|
| 45 |
-
if file_size_mb
|
| 46 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 47 |
if compressed_file:
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
|
|
|
|
|
|
|
|
|
|
| 51 |
|
| 52 |
-
#
|
| 53 |
-
|
| 54 |
-
return self._split_audio_file(uploaded_file, max_chunk_size_mb)
|
| 55 |
-
else:
|
| 56 |
-
# Plik nie wymaga dzielenia - zapisz bezpośrednio
|
| 57 |
-
temp_path = self._save_temp_file(uploaded_file)
|
| 58 |
-
return [temp_path]
|
| 59 |
|
| 60 |
except Exception as e:
|
| 61 |
st.error(f"❌ Błąd przetwarzania {uploaded_file.name}: {str(e)}")
|
| 62 |
return []
|
| 63 |
|
| 64 |
-
def
|
| 65 |
-
"""
|
| 66 |
if not PYDUB_AVAILABLE:
|
| 67 |
st.warning("Pydub niedostępny - pomijam kompresję")
|
| 68 |
return None
|
| 69 |
|
| 70 |
try:
|
|
|
|
|
|
|
| 71 |
# Załaduj audio
|
| 72 |
audio_data = uploaded_file.read()
|
|
|
|
|
|
|
| 73 |
audio = AudioSegment.from_file(BytesIO(audio_data))
|
| 74 |
|
| 75 |
-
#
|
| 76 |
compressed = audio.set_channels(1) # Mono
|
| 77 |
compressed = compressed.set_frame_rate(16000) # 16kHz (wystarczy dla mowy)
|
| 78 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 79 |
# Export do BytesIO
|
| 80 |
output = BytesIO()
|
| 81 |
compressed.export(
|
| 82 |
output,
|
| 83 |
format="mp3",
|
| 84 |
-
bitrate=
|
| 85 |
-
parameters=["-ac", "1"
|
| 86 |
)
|
| 87 |
output.seek(0)
|
| 88 |
|
| 89 |
-
#
|
| 90 |
-
|
| 91 |
-
output.size = len(output.getvalue())
|
| 92 |
|
| 93 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 94 |
|
| 95 |
except Exception as e:
|
| 96 |
st.warning(f"Kompresja nieudana: {str(e)}")
|
| 97 |
return None
|
| 98 |
|
| 99 |
-
def
|
| 100 |
-
"""Dzieli plik audio na
|
| 101 |
try:
|
| 102 |
if not PYDUB_AVAILABLE:
|
| 103 |
st.error("❌ Pydub wymagany do dzielenia plików. Zainstaluj: pip install pydub")
|
| 104 |
return []
|
| 105 |
|
| 106 |
-
|
|
|
|
|
|
|
| 107 |
audio_data = uploaded_file.read()
|
| 108 |
audio = AudioSegment.from_file(BytesIO(audio_data))
|
| 109 |
|
|
@@ -111,12 +149,15 @@ class FileHandler:
|
|
| 111 |
total_duration_ms = len(audio)
|
| 112 |
file_size_mb = uploaded_file.size / (1024 * 1024)
|
| 113 |
|
| 114 |
-
#
|
| 115 |
-
|
|
|
|
|
|
|
|
|
|
| 116 |
chunk_duration_ms = total_duration_ms // estimated_parts
|
| 117 |
|
| 118 |
-
#
|
| 119 |
-
overlap_ms =
|
| 120 |
|
| 121 |
st.info(f"📂 Dzielę na {estimated_parts} części (~{chunk_duration_ms//60000:.1f} min każda)")
|
| 122 |
|
|
@@ -124,22 +165,41 @@ class FileHandler:
|
|
| 124 |
base_name = os.path.splitext(uploaded_file.name)[0]
|
| 125 |
|
| 126 |
for i in range(estimated_parts):
|
| 127 |
-
start_ms = max(0, i * chunk_duration_ms - overlap_ms if i > 0 else 0)
|
| 128 |
end_ms = min(total_duration_ms, (i + 1) * chunk_duration_ms + overlap_ms)
|
| 129 |
|
| 130 |
# Wytnij część
|
| 131 |
chunk = audio[start_ms:end_ms]
|
| 132 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 133 |
# Zapisz do pliku tymczasowego
|
| 134 |
-
temp_fd, temp_path = tempfile.mkstemp(
|
|
|
|
|
|
|
|
|
|
| 135 |
os.close(temp_fd)
|
| 136 |
|
| 137 |
-
chunk.export(temp_path, format="mp3", bitrate="
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 138 |
parts.append(temp_path)
|
| 139 |
self.temp_files.append(temp_path)
|
| 140 |
|
| 141 |
-
st.success(f"✅ Część {i+1}/{estimated_parts}: {(end_ms-start_ms)//60000:.1f} min")
|
| 142 |
|
|
|
|
|
|
|
|
|
|
| 143 |
return parts
|
| 144 |
|
| 145 |
except Exception as e:
|
|
@@ -149,13 +209,16 @@ class FileHandler:
|
|
| 149 |
def _save_temp_file(self, uploaded_file) -> str:
|
| 150 |
"""Zapisuje uploaded file do pliku tymczasowego"""
|
| 151 |
try:
|
| 152 |
-
# Stwórz plik tymczasowy
|
| 153 |
suffix = f".{uploaded_file.name.split('.')[-1]}"
|
| 154 |
temp_fd, temp_path = tempfile.mkstemp(suffix=suffix)
|
| 155 |
|
| 156 |
# Zapisz dane
|
| 157 |
with os.fdopen(temp_fd, 'wb') as tmp_file:
|
| 158 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 159 |
|
| 160 |
self.temp_files.append(temp_path)
|
| 161 |
return temp_path
|
|
@@ -164,73 +227,103 @@ class FileHandler:
|
|
| 164 |
st.error(f"❌ Błąd zapisu tymczasowego: {str(e)}")
|
| 165 |
return ""
|
| 166 |
|
| 167 |
-
def
|
| 168 |
-
"""
|
| 169 |
try:
|
| 170 |
-
|
| 171 |
-
|
| 172 |
-
|
| 173 |
-
|
| 174 |
-
|
| 175 |
-
|
| 176 |
-
|
| 177 |
-
|
| 178 |
-
|
| 179 |
-
|
| 180 |
-
|
| 181 |
-
|
| 182 |
-
|
| 183 |
-
|
| 184 |
-
|
|
|
|
|
|
|
| 185 |
|
| 186 |
-
def
|
| 187 |
-
"""Walidacja pliku
|
| 188 |
try:
|
| 189 |
# Sprawdź rozmiar
|
| 190 |
file_size_mb = uploaded_file.size / (1024 * 1024)
|
| 191 |
-
|
| 192 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 193 |
|
| 194 |
# Sprawdź rozszerzenie
|
| 195 |
file_ext = uploaded_file.name.split('.')[-1].lower()
|
| 196 |
-
supported_formats =
|
| 197 |
-
FILE_PROCESSING['supported_audio_formats'] +
|
| 198 |
-
FILE_PROCESSING['supported_video_formats']
|
| 199 |
-
)
|
| 200 |
|
| 201 |
if file_ext not in supported_formats:
|
| 202 |
return False, f"Nieobsługiwany format: .{file_ext}"
|
| 203 |
|
| 204 |
-
#
|
| 205 |
-
if
|
| 206 |
-
return
|
| 207 |
|
| 208 |
return True, "OK"
|
| 209 |
|
| 210 |
except Exception as e:
|
| 211 |
return False, f"Błąd walidacji: {str(e)}"
|
| 212 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 213 |
def estimate_processing_time(self, uploaded_files: List) -> Dict:
|
| 214 |
"""Estymuj czas przetwarzania"""
|
| 215 |
total_size_mb = sum(f.size for f in uploaded_files) / (1024 * 1024)
|
| 216 |
-
total_duration_est = total_size_mb * 60
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 217 |
|
| 218 |
# Estymacja czasu transkrypcji (Whisper ~1:10 ratio)
|
| 219 |
-
transcription_time = total_duration_est *
|
| 220 |
|
| 221 |
-
# Estymacja czasu generowania raportu
|
| 222 |
-
report_time = len(uploaded_files) * 30
|
| 223 |
|
| 224 |
return {
|
| 225 |
'total_size_mb': total_size_mb,
|
| 226 |
'estimated_audio_duration': total_duration_est,
|
|
|
|
| 227 |
'estimated_transcription_time': transcription_time,
|
| 228 |
'estimated_report_time': report_time,
|
| 229 |
-
'total_estimated_time': transcription_time + report_time
|
|
|
|
|
|
|
| 230 |
}
|
| 231 |
|
| 232 |
def get_file_info(self, uploaded_file) -> Dict:
|
| 233 |
-
"""Pobierz informacje o pliku"""
|
| 234 |
file_size_mb = uploaded_file.size / (1024 * 1024)
|
| 235 |
file_ext = uploaded_file.name.split('.')[-1].lower()
|
| 236 |
|
|
@@ -238,40 +331,161 @@ class FileHandler:
|
|
| 238 |
'name': uploaded_file.name,
|
| 239 |
'size_mb': file_size_mb,
|
| 240 |
'format': file_ext,
|
| 241 |
-
'
|
| 242 |
-
'
|
| 243 |
-
'
|
|
|
|
|
|
|
|
|
|
| 244 |
}
|
| 245 |
|
| 246 |
def cleanup_temp_files(self):
|
| 247 |
"""Wyczyść pliki tymczasowe"""
|
| 248 |
cleaned = 0
|
|
|
|
|
|
|
| 249 |
for temp_file in self.temp_files:
|
| 250 |
try:
|
| 251 |
if os.path.exists(temp_file):
|
| 252 |
os.remove(temp_file)
|
| 253 |
cleaned += 1
|
| 254 |
except Exception as e:
|
|
|
|
| 255 |
st.warning(f"Nie można usunąć {temp_file}: {e}")
|
| 256 |
|
| 257 |
self.temp_files = []
|
|
|
|
| 258 |
if cleaned > 0:
|
| 259 |
st.success(f"🧹 Wyczyszczono {cleaned} plików tymczasowych")
|
|
|
|
|
|
|
|
|
|
| 260 |
|
| 261 |
def get_processing_stats(self) -> Dict:
|
| 262 |
"""Zwróć statystyki przetwarzania"""
|
| 263 |
return {
|
| 264 |
'temp_files_count': len(self.temp_files),
|
|
|
|
|
|
|
| 265 |
'processing_stats': self.processing_stats,
|
| 266 |
'libraries_available': {
|
| 267 |
'pydub': PYDUB_AVAILABLE,
|
| 268 |
'librosa': LIBROSA_AVAILABLE
|
| 269 |
}
|
| 270 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 271 |
|
| 272 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 273 |
if __name__ == "__main__":
|
| 274 |
print("🧪 Test FileHandler")
|
| 275 |
handler = FileHandler()
|
| 276 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 277 |
print("✅ FileHandler gotowy do użycia")
|
|
|
|
| 1 |
+
# file_handler.py - Poprawiony handler plików dla Whisper API
|
| 2 |
|
| 3 |
import os
|
| 4 |
import tempfile
|
|
|
|
| 12 |
PYDUB_AVAILABLE = True
|
| 13 |
except ImportError:
|
| 14 |
PYDUB_AVAILABLE = False
|
| 15 |
+
st.warning("⚠️ Pydub nie jest dostępny. Funkcje kompresji ograniczone.")
|
| 16 |
|
| 17 |
try:
|
| 18 |
import librosa
|
|
|
|
| 24 |
from config import FILE_PROCESSING, USER_MESSAGES
|
| 25 |
|
| 26 |
class FileHandler:
|
| 27 |
+
"""Klasa do obsługi plików audio/video - zoptymalizowana dla Whisper API (max 25MB)"""
|
| 28 |
|
| 29 |
def __init__(self):
|
| 30 |
+
self.temp_files = []
|
| 31 |
self.processing_stats = {}
|
| 32 |
|
| 33 |
+
# Whisper API limits
|
| 34 |
+
self.WHISPER_MAX_SIZE_MB = 25
|
| 35 |
+
self.SAFE_CHUNK_SIZE_MB = 20 # Bezpieczny rozmiar chunka
|
| 36 |
+
|
| 37 |
+
def process_file(self, uploaded_file, max_chunk_size_mb: int = 20, auto_compress: bool = True) -> List[str]:
|
| 38 |
"""
|
| 39 |
+
Główna funkcja przetwarzania pliku dla Whisper API
|
| 40 |
+
Returns: Lista ścieżek do plików gotowych do transkrypcji (każdy <25MB)
|
| 41 |
"""
|
| 42 |
try:
|
| 43 |
file_size_mb = uploaded_file.size / (1024 * 1024)
|
| 44 |
|
|
|
|
| 45 |
st.info(f"🔄 Przetwarzam {uploaded_file.name} ({file_size_mb:.1f}MB)")
|
| 46 |
|
| 47 |
+
# Sprawdź czy plik mieści się w limicie Whisper
|
| 48 |
+
if file_size_mb <= self.WHISPER_MAX_SIZE_MB:
|
| 49 |
+
# Plik OK - zapisz bezpośrednio
|
| 50 |
+
temp_path = self._save_temp_file(uploaded_file)
|
| 51 |
+
if temp_path:
|
| 52 |
+
st.success(f"✅ Plik gotowy do transkrypcji ({file_size_mb:.1f}MB)")
|
| 53 |
+
return [temp_path]
|
| 54 |
+
else:
|
| 55 |
+
return []
|
| 56 |
+
|
| 57 |
+
# Plik za duży - wymaga przetwarzania
|
| 58 |
+
if file_size_mb > 100:
|
| 59 |
+
st.error(f"❌ Plik zbyt duży ({file_size_mb:.1f}MB). Maksymalnie 100MB.")
|
| 60 |
+
return []
|
| 61 |
+
|
| 62 |
+
# Strategia 1: Kompresja
|
| 63 |
+
if auto_compress and file_size_mb > self.WHISPER_MAX_SIZE_MB:
|
| 64 |
+
compressed_file = self._compress_audio_for_whisper(uploaded_file)
|
| 65 |
if compressed_file:
|
| 66 |
+
compressed_size_mb = len(compressed_file.getvalue()) / (1024 * 1024)
|
| 67 |
+
if compressed_size_mb <= self.WHISPER_MAX_SIZE_MB:
|
| 68 |
+
temp_path = self._save_bytesio_to_temp(compressed_file, uploaded_file.name)
|
| 69 |
+
if temp_path:
|
| 70 |
+
st.success(f"✅ Skompresowano: {file_size_mb:.1f}MB → {compressed_size_mb:.1f}MB")
|
| 71 |
+
return [temp_path]
|
| 72 |
|
| 73 |
+
# Strategia 2: Podział na części
|
| 74 |
+
return self._split_audio_for_whisper(uploaded_file, max_chunk_size_mb)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 75 |
|
| 76 |
except Exception as e:
|
| 77 |
st.error(f"❌ Błąd przetwarzania {uploaded_file.name}: {str(e)}")
|
| 78 |
return []
|
| 79 |
|
| 80 |
+
def _compress_audio_for_whisper(self, uploaded_file) -> Union[BytesIO, None]:
|
| 81 |
+
"""Agresywna kompresja audio dla Whisper API"""
|
| 82 |
if not PYDUB_AVAILABLE:
|
| 83 |
st.warning("Pydub niedostępny - pomijam kompresję")
|
| 84 |
return None
|
| 85 |
|
| 86 |
try:
|
| 87 |
+
st.info("🗜️ Kompresuję audio...")
|
| 88 |
+
|
| 89 |
# Załaduj audio
|
| 90 |
audio_data = uploaded_file.read()
|
| 91 |
+
uploaded_file.seek(0) # Reset dla dalszego użycia
|
| 92 |
+
|
| 93 |
audio = AudioSegment.from_file(BytesIO(audio_data))
|
| 94 |
|
| 95 |
+
# Agresywna kompresja dla Whisper (jakość mowy)
|
| 96 |
compressed = audio.set_channels(1) # Mono
|
| 97 |
compressed = compressed.set_frame_rate(16000) # 16kHz (wystarczy dla mowy)
|
| 98 |
|
| 99 |
+
# Jeszcze więcej kompresji jeśli potrzeba
|
| 100 |
+
original_size_mb = uploaded_file.size / (1024 * 1024)
|
| 101 |
+
|
| 102 |
+
if original_size_mb > 50:
|
| 103 |
+
# Bardzo duży plik - maksymalna kompresja
|
| 104 |
+
bitrate = "32k"
|
| 105 |
+
elif original_size_mb > 35:
|
| 106 |
+
# Duży plik - silna kompresja
|
| 107 |
+
bitrate = "48k"
|
| 108 |
+
else:
|
| 109 |
+
# Średni plik - umiarkowana kompresja
|
| 110 |
+
bitrate = "64k"
|
| 111 |
+
|
| 112 |
# Export do BytesIO
|
| 113 |
output = BytesIO()
|
| 114 |
compressed.export(
|
| 115 |
output,
|
| 116 |
format="mp3",
|
| 117 |
+
bitrate=bitrate,
|
| 118 |
+
parameters=["-ac", "1", "-ar", "16000"]
|
| 119 |
)
|
| 120 |
output.seek(0)
|
| 121 |
|
| 122 |
+
# Sprawdź rozmiar wyniku
|
| 123 |
+
compressed_size_mb = len(output.getvalue()) / (1024 * 1024)
|
|
|
|
| 124 |
|
| 125 |
+
if compressed_size_mb <= self.WHISPER_MAX_SIZE_MB:
|
| 126 |
+
return output
|
| 127 |
+
else:
|
| 128 |
+
st.warning(f"⚠️ Kompresja niewystarczająca ({compressed_size_mb:.1f}MB). Przechodzę do dzielenia.")
|
| 129 |
+
return None
|
| 130 |
|
| 131 |
except Exception as e:
|
| 132 |
st.warning(f"Kompresja nieudana: {str(e)}")
|
| 133 |
return None
|
| 134 |
|
| 135 |
+
def _split_audio_for_whisper(self, uploaded_file, max_chunk_size_mb: int) -> List[str]:
|
| 136 |
+
"""Dzieli plik audio na części <25MB dla Whisper"""
|
| 137 |
try:
|
| 138 |
if not PYDUB_AVAILABLE:
|
| 139 |
st.error("❌ Pydub wymagany do dzielenia plików. Zainstaluj: pip install pydub")
|
| 140 |
return []
|
| 141 |
|
| 142 |
+
st.info("✂️ Dzielę plik na części...")
|
| 143 |
+
|
| 144 |
+
# Załaduj audio
|
| 145 |
audio_data = uploaded_file.read()
|
| 146 |
audio = AudioSegment.from_file(BytesIO(audio_data))
|
| 147 |
|
|
|
|
| 149 |
total_duration_ms = len(audio)
|
| 150 |
file_size_mb = uploaded_file.size / (1024 * 1024)
|
| 151 |
|
| 152 |
+
# Bezpieczny rozmiar chunka (mniejszy niż limit Whisper)
|
| 153 |
+
safe_chunk_size_mb = min(max_chunk_size_mb, self.SAFE_CHUNK_SIZE_MB)
|
| 154 |
+
|
| 155 |
+
# Estymacja liczby części
|
| 156 |
+
estimated_parts = math.ceil(file_size_mb / safe_chunk_size_mb)
|
| 157 |
chunk_duration_ms = total_duration_ms // estimated_parts
|
| 158 |
|
| 159 |
+
# Overlap między częściami (10 sekund)
|
| 160 |
+
overlap_ms = 10 * 1000
|
| 161 |
|
| 162 |
st.info(f"📂 Dzielę na {estimated_parts} części (~{chunk_duration_ms//60000:.1f} min każda)")
|
| 163 |
|
|
|
|
| 165 |
base_name = os.path.splitext(uploaded_file.name)[0]
|
| 166 |
|
| 167 |
for i in range(estimated_parts):
|
| 168 |
+
start_ms = max(0, i * chunk_duration_ms - (overlap_ms if i > 0 else 0))
|
| 169 |
end_ms = min(total_duration_ms, (i + 1) * chunk_duration_ms + overlap_ms)
|
| 170 |
|
| 171 |
# Wytnij część
|
| 172 |
chunk = audio[start_ms:end_ms]
|
| 173 |
|
| 174 |
+
# Lekka kompresja części
|
| 175 |
+
chunk = chunk.set_channels(1) # Mono
|
| 176 |
+
chunk = chunk.set_frame_rate(22050) # Dobra jakość ale kompaktowa
|
| 177 |
+
|
| 178 |
# Zapisz do pliku tymczasowego
|
| 179 |
+
temp_fd, temp_path = tempfile.mkstemp(
|
| 180 |
+
suffix=f"_part{i+1:02d}.mp3",
|
| 181 |
+
prefix=f"{base_name}_"
|
| 182 |
+
)
|
| 183 |
os.close(temp_fd)
|
| 184 |
|
| 185 |
+
chunk.export(temp_path, format="mp3", bitrate="96k")
|
| 186 |
+
|
| 187 |
+
# Sprawdź rozmiar części
|
| 188 |
+
part_size_mb = os.path.getsize(temp_path) / (1024 * 1024)
|
| 189 |
+
|
| 190 |
+
if part_size_mb > self.WHISPER_MAX_SIZE_MB:
|
| 191 |
+
st.error(f"❌ Część {i+1} nadal za duża ({part_size_mb:.1f}MB)")
|
| 192 |
+
os.remove(temp_path)
|
| 193 |
+
continue
|
| 194 |
+
|
| 195 |
parts.append(temp_path)
|
| 196 |
self.temp_files.append(temp_path)
|
| 197 |
|
| 198 |
+
st.success(f"✅ Część {i+1}/{estimated_parts}: {part_size_mb:.1f}MB, {(end_ms-start_ms)//60000:.1f} min")
|
| 199 |
|
| 200 |
+
if not parts:
|
| 201 |
+
st.error("❌ Nie udało się utworzyć żadnej prawidłowej części")
|
| 202 |
+
|
| 203 |
return parts
|
| 204 |
|
| 205 |
except Exception as e:
|
|
|
|
| 209 |
def _save_temp_file(self, uploaded_file) -> str:
|
| 210 |
"""Zapisuje uploaded file do pliku tymczasowego"""
|
| 211 |
try:
|
|
|
|
| 212 |
suffix = f".{uploaded_file.name.split('.')[-1]}"
|
| 213 |
temp_fd, temp_path = tempfile.mkstemp(suffix=suffix)
|
| 214 |
|
| 215 |
# Zapisz dane
|
| 216 |
with os.fdopen(temp_fd, 'wb') as tmp_file:
|
| 217 |
+
content = uploaded_file.read()
|
| 218 |
+
tmp_file.write(content)
|
| 219 |
+
|
| 220 |
+
# Reset pozycji dla dalszego użycia
|
| 221 |
+
uploaded_file.seek(0)
|
| 222 |
|
| 223 |
self.temp_files.append(temp_path)
|
| 224 |
return temp_path
|
|
|
|
| 227 |
st.error(f"❌ Błąd zapisu tymczasowego: {str(e)}")
|
| 228 |
return ""
|
| 229 |
|
| 230 |
+
def _save_bytesio_to_temp(self, bytes_io: BytesIO, original_name: str) -> str:
|
| 231 |
+
"""Zapisz BytesIO do pliku tymczasowego"""
|
| 232 |
try:
|
| 233 |
+
suffix = f"_compressed.mp3"
|
| 234 |
+
base_name = os.path.splitext(original_name)[0]
|
| 235 |
+
|
| 236 |
+
temp_fd, temp_path = tempfile.mkstemp(
|
| 237 |
+
suffix=suffix,
|
| 238 |
+
prefix=f"{base_name}_"
|
| 239 |
+
)
|
| 240 |
+
|
| 241 |
+
with os.fdopen(temp_fd, 'wb') as tmp_file:
|
| 242 |
+
tmp_file.write(bytes_io.getvalue())
|
| 243 |
+
|
| 244 |
+
self.temp_files.append(temp_path)
|
| 245 |
+
return temp_path
|
| 246 |
+
|
| 247 |
+
except Exception as e:
|
| 248 |
+
st.error(f"❌ Błąd zapisu skompresowanego: {str(e)}")
|
| 249 |
+
return ""
|
| 250 |
|
| 251 |
+
def validate_file_for_whisper(self, uploaded_file) -> Tuple[bool, str]:
|
| 252 |
+
"""Walidacja pliku dla Whisper API"""
|
| 253 |
try:
|
| 254 |
# Sprawdź rozmiar
|
| 255 |
file_size_mb = uploaded_file.size / (1024 * 1024)
|
| 256 |
+
|
| 257 |
+
if file_size_mb == 0:
|
| 258 |
+
return False, "Plik jest pusty"
|
| 259 |
+
|
| 260 |
+
if file_size_mb > 100: # Rozumny limit dla przetwarzania
|
| 261 |
+
return False, f"Plik za duży: {file_size_mb:.1f}MB > 100MB"
|
| 262 |
|
| 263 |
# Sprawdź rozszerzenie
|
| 264 |
file_ext = uploaded_file.name.split('.')[-1].lower()
|
| 265 |
+
supported_formats = ['mp3', 'wav', 'mp4', 'm4a', 'aac', 'mov', 'avi']
|
|
|
|
|
|
|
|
|
|
| 266 |
|
| 267 |
if file_ext not in supported_formats:
|
| 268 |
return False, f"Nieobsługiwany format: .{file_ext}"
|
| 269 |
|
| 270 |
+
# Ostrzeżenie dla dużych plików
|
| 271 |
+
if file_size_mb > self.WHISPER_MAX_SIZE_MB:
|
| 272 |
+
return True, f"Plik wymaga przetwarzania ({file_size_mb:.1f}MB > {self.WHISPER_MAX_SIZE_MB}MB)"
|
| 273 |
|
| 274 |
return True, "OK"
|
| 275 |
|
| 276 |
except Exception as e:
|
| 277 |
return False, f"Błąd walidacji: {str(e)}"
|
| 278 |
|
| 279 |
+
def get_audio_duration(self, file_path: str) -> float:
|
| 280 |
+
"""Pobierz długość pliku audio w sekundach"""
|
| 281 |
+
try:
|
| 282 |
+
if LIBROSA_AVAILABLE:
|
| 283 |
+
duration = librosa.get_duration(filename=file_path)
|
| 284 |
+
return duration
|
| 285 |
+
elif PYDUB_AVAILABLE:
|
| 286 |
+
audio = AudioSegment.from_file(file_path)
|
| 287 |
+
return len(audio) / 1000.0
|
| 288 |
+
else:
|
| 289 |
+
# Fallback - estymacja na podstawie rozmiaru
|
| 290 |
+
file_size = os.path.getsize(file_path)
|
| 291 |
+
return file_size / (1024 * 1024) * 60
|
| 292 |
+
except:
|
| 293 |
+
file_size = os.path.getsize(file_path)
|
| 294 |
+
return file_size / (1024 * 1024) * 60
|
| 295 |
+
|
| 296 |
def estimate_processing_time(self, uploaded_files: List) -> Dict:
|
| 297 |
"""Estymuj czas przetwarzania"""
|
| 298 |
total_size_mb = sum(f.size for f in uploaded_files) / (1024 * 1024)
|
| 299 |
+
total_duration_est = total_size_mb * 60
|
| 300 |
+
|
| 301 |
+
# Czas przetwarzania (kompresja/dzielenie)
|
| 302 |
+
processing_time = 0
|
| 303 |
+
for f in uploaded_files:
|
| 304 |
+
file_size_mb = f.size / (1024 * 1024)
|
| 305 |
+
if file_size_mb > self.WHISPER_MAX_SIZE_MB:
|
| 306 |
+
processing_time += file_size_mb * 2 # ~2s per MB for processing
|
| 307 |
|
| 308 |
# Estymacja czasu transkrypcji (Whisper ~1:10 ratio)
|
| 309 |
+
transcription_time = total_duration_est * 0.1
|
| 310 |
|
| 311 |
+
# Estymacja czasu generowania raportu
|
| 312 |
+
report_time = len(uploaded_files) * 30
|
| 313 |
|
| 314 |
return {
|
| 315 |
'total_size_mb': total_size_mb,
|
| 316 |
'estimated_audio_duration': total_duration_est,
|
| 317 |
+
'estimated_processing_time': processing_time,
|
| 318 |
'estimated_transcription_time': transcription_time,
|
| 319 |
'estimated_report_time': report_time,
|
| 320 |
+
'total_estimated_time': processing_time + transcription_time + report_time,
|
| 321 |
+
'files_needing_processing': sum(1 for f in uploaded_files
|
| 322 |
+
if f.size / (1024 * 1024) > self.WHISPER_MAX_SIZE_MB)
|
| 323 |
}
|
| 324 |
|
| 325 |
def get_file_info(self, uploaded_file) -> Dict:
|
| 326 |
+
"""Pobierz szczegółowe informacje o pliku"""
|
| 327 |
file_size_mb = uploaded_file.size / (1024 * 1024)
|
| 328 |
file_ext = uploaded_file.name.split('.')[-1].lower()
|
| 329 |
|
|
|
|
| 331 |
'name': uploaded_file.name,
|
| 332 |
'size_mb': file_size_mb,
|
| 333 |
'format': file_ext,
|
| 334 |
+
'whisper_ready': file_size_mb <= self.WHISPER_MAX_SIZE_MB,
|
| 335 |
+
'needs_compression': file_size_mb > self.WHISPER_MAX_SIZE_MB and file_size_mb <= 50,
|
| 336 |
+
'needs_splitting': file_size_mb > 50,
|
| 337 |
+
'too_large': file_size_mb > 100,
|
| 338 |
+
'estimated_duration': file_size_mb * 60,
|
| 339 |
+
'estimated_processing_time': max(0, file_size_mb - self.WHISPER_MAX_SIZE_MB) * 2
|
| 340 |
}
|
| 341 |
|
| 342 |
def cleanup_temp_files(self):
|
| 343 |
"""Wyczyść pliki tymczasowe"""
|
| 344 |
cleaned = 0
|
| 345 |
+
errors = 0
|
| 346 |
+
|
| 347 |
for temp_file in self.temp_files:
|
| 348 |
try:
|
| 349 |
if os.path.exists(temp_file):
|
| 350 |
os.remove(temp_file)
|
| 351 |
cleaned += 1
|
| 352 |
except Exception as e:
|
| 353 |
+
errors += 1
|
| 354 |
st.warning(f"Nie można usunąć {temp_file}: {e}")
|
| 355 |
|
| 356 |
self.temp_files = []
|
| 357 |
+
|
| 358 |
if cleaned > 0:
|
| 359 |
st.success(f"🧹 Wyczyszczono {cleaned} plików tymczasowych")
|
| 360 |
+
|
| 361 |
+
if errors > 0:
|
| 362 |
+
st.warning(f"⚠️ {errors} plików nie udało się usunąć")
|
| 363 |
|
| 364 |
def get_processing_stats(self) -> Dict:
|
| 365 |
"""Zwróć statystyki przetwarzania"""
|
| 366 |
return {
|
| 367 |
'temp_files_count': len(self.temp_files),
|
| 368 |
+
'whisper_max_size_mb': self.WHISPER_MAX_SIZE_MB,
|
| 369 |
+
'safe_chunk_size_mb': self.SAFE_CHUNK_SIZE_MB,
|
| 370 |
'processing_stats': self.processing_stats,
|
| 371 |
'libraries_available': {
|
| 372 |
'pydub': PYDUB_AVAILABLE,
|
| 373 |
'librosa': LIBROSA_AVAILABLE
|
| 374 |
}
|
| 375 |
}
|
| 376 |
+
|
| 377 |
+
def analyze_upload_batch(self, uploaded_files: List) -> Dict:
|
| 378 |
+
"""Analizuj całą paczkę plików"""
|
| 379 |
+
analysis = {
|
| 380 |
+
'total_files': len(uploaded_files),
|
| 381 |
+
'total_size_mb': 0,
|
| 382 |
+
'whisper_ready': 0,
|
| 383 |
+
'need_compression': 0,
|
| 384 |
+
'need_splitting': 0,
|
| 385 |
+
'too_large': 0,
|
| 386 |
+
'estimated_parts': 0,
|
| 387 |
+
'file_details': []
|
| 388 |
+
}
|
| 389 |
+
|
| 390 |
+
for file in uploaded_files:
|
| 391 |
+
info = self.get_file_info(file)
|
| 392 |
+
analysis['file_details'].append(info)
|
| 393 |
+
analysis['total_size_mb'] += info['size_mb']
|
| 394 |
+
|
| 395 |
+
if info['whisper_ready']:
|
| 396 |
+
analysis['whisper_ready'] += 1
|
| 397 |
+
elif info['needs_compression']:
|
| 398 |
+
analysis['need_compression'] += 1
|
| 399 |
+
elif info['needs_splitting']:
|
| 400 |
+
analysis['need_splitting'] += 1
|
| 401 |
+
# Estymacja liczby części
|
| 402 |
+
parts = math.ceil(info['size_mb'] / self.SAFE_CHUNK_SIZE_MB)
|
| 403 |
+
analysis['estimated_parts'] += parts
|
| 404 |
+
elif info['too_large']:
|
| 405 |
+
analysis['too_large'] += 1
|
| 406 |
+
|
| 407 |
+
return analysis
|
| 408 |
+
|
| 409 |
+
def create_processing_plan(self, uploaded_files: List) -> str:
|
| 410 |
+
"""Stwórz plan przetwarzania dla użytkownika"""
|
| 411 |
+
analysis = self.analyze_upload_batch(uploaded_files)
|
| 412 |
+
|
| 413 |
+
plan = f"""
|
| 414 |
+
📋 **PLAN PRZETWARZANIA**
|
| 415 |
+
|
| 416 |
+
📊 **Podsumowanie:**
|
| 417 |
+
- Plików: {analysis['total_files']} ({analysis['total_size_mb']:.1f}MB)
|
| 418 |
+
- Gotowych do transkrypcji: {analysis['whisper_ready']}
|
| 419 |
+
- Wymagających kompresji: {analysis['need_compression']}
|
| 420 |
+
- Wymagających dzielenia: {analysis['need_splitting']}
|
| 421 |
+
- Za dużych: {analysis['too_large']}
|
| 422 |
|
| 423 |
+
"""
|
| 424 |
+
|
| 425 |
+
if analysis['estimated_parts'] > 0:
|
| 426 |
+
plan += f"- Szacowana liczba części: {analysis['estimated_parts']}\n"
|
| 427 |
+
|
| 428 |
+
if analysis['too_large'] > 0:
|
| 429 |
+
plan += f"\n❌ **PLIKI ZA DUŻE (>100MB):**\n"
|
| 430 |
+
for info in analysis['file_details']:
|
| 431 |
+
if info['too_large']:
|
| 432 |
+
plan += f"- {info['name']}: {info['size_mb']:.1f}MB\n"
|
| 433 |
+
|
| 434 |
+
if analysis['need_splitting'] > 0:
|
| 435 |
+
plan += f"\n✂️ **PLIKI DO PODZIELENIA:**\n"
|
| 436 |
+
for info in analysis['file_details']:
|
| 437 |
+
if info['needs_splitting']:
|
| 438 |
+
parts = math.ceil(info['size_mb'] / self.SAFE_CHUNK_SIZE_MB)
|
| 439 |
+
plan += f"- {info['name']}: {info['size_mb']:.1f}MB → ~{parts} części\n"
|
| 440 |
+
|
| 441 |
+
if analysis['need_compression'] > 0:
|
| 442 |
+
plan += f"\n🗜️ **PLIKI DO KOMPRESJI:**\n"
|
| 443 |
+
for info in analysis['file_details']:
|
| 444 |
+
if info['needs_compression']:
|
| 445 |
+
plan += f"- {info['name']}: {info['size_mb']:.1f}MB\n"
|
| 446 |
+
|
| 447 |
+
# Estymacja czasów
|
| 448 |
+
times = self.estimate_processing_time(uploaded_files)
|
| 449 |
+
plan += f"""
|
| 450 |
+
⏱️ **ESTYMACJA CZASÓW:**
|
| 451 |
+
- Przetwarzanie plików: ~{times['estimated_processing_time']:.1f}s
|
| 452 |
+
- Transkrypcja: ~{times['estimated_transcription_time']:.1f}s
|
| 453 |
+
- Generowanie raportu: ~{times['estimated_report_time']:.1f}s
|
| 454 |
+
- **ŁĄCZNIE: ~{times['total_estimated_time']:.1f}s ({times['total_estimated_time']/60:.1f} min)**
|
| 455 |
+
"""
|
| 456 |
+
|
| 457 |
+
return plan
|
| 458 |
+
|
| 459 |
+
# Funkcje pomocnicze
|
| 460 |
+
def check_file_size_for_whisper(file_path: str) -> Tuple[bool, float]:
|
| 461 |
+
"""Sprawdź czy plik mieści się w limicie Whisper"""
|
| 462 |
+
try:
|
| 463 |
+
size_mb = os.path.getsize(file_path) / (1024 * 1024)
|
| 464 |
+
return size_mb <= 25, size_mb
|
| 465 |
+
except:
|
| 466 |
+
return False, 0
|
| 467 |
+
|
| 468 |
+
def estimate_compression_ratio(file_ext: str) -> float:
|
| 469 |
+
"""Estymuj współczynnik kompresji dla różnych formatów"""
|
| 470 |
+
ratios = {
|
| 471 |
+
'wav': 0.1, # WAV kompresuje się bardzo dobrze
|
| 472 |
+
'aac': 0.7, # AAC już skompresowany
|
| 473 |
+
'mp3': 0.8, # MP3 już skompresowany
|
| 474 |
+
'm4a': 0.7, # M4A już skompresowany
|
| 475 |
+
'mp4': 0.5, # Video można mocno skompresować audio
|
| 476 |
+
'mov': 0.5,
|
| 477 |
+
'avi': 0.4
|
| 478 |
+
}
|
| 479 |
+
return ratios.get(file_ext.lower(), 0.6)
|
| 480 |
+
|
| 481 |
+
# Test modułu
|
| 482 |
if __name__ == "__main__":
|
| 483 |
print("🧪 Test FileHandler")
|
| 484 |
handler = FileHandler()
|
| 485 |
+
|
| 486 |
+
stats = handler.get_processing_stats()
|
| 487 |
+
print(f"📊 Biblioteki: {stats['libraries_available']}")
|
| 488 |
+
print(f"🎯 Limit Whisper: {stats['whisper_max_size_mb']}MB")
|
| 489 |
+
print(f"🔒 Bezpieczny chunk: {stats['safe_chunk_size_mb']}MB")
|
| 490 |
+
|
| 491 |
print("✅ FileHandler gotowy do użycia")
|