armyneo commited on
Commit
8b26d94
·
verified ·
1 Parent(s): cfbff1e

fix app.py access token error

Browse files
Files changed (1) hide show
  1. app.py +22 -8
app.py CHANGED
@@ -2,21 +2,34 @@ import re
2
  import io
3
  import zipfile
4
  from pathlib import Path
 
5
 
6
  import gradio as gr
7
  from docx import Document
8
  from docx.oxml import OxmlElement
9
  from docx.oxml.ns import qn
10
- from transformers import pipeline
 
 
 
 
 
11
 
12
 
13
  # ----------------------------------------------------
14
- # 1) ÇEVİRİ MODELİ (Helsinki-NLP / opus-mt-en-tr)
15
  # ----------------------------------------------------
16
 
 
 
 
 
 
 
17
  translator = pipeline(
18
  "translation",
19
- model="Helsinki-NLP/opus-mt-en-tr"
 
20
  )
21
 
22
 
@@ -182,7 +195,7 @@ def style_header_cell(cell, text: str):
182
  shd.set(qn("w:fill"), "D9D9D9") # light grey
183
 
184
 
185
- def srt_to_docx_bytes(srt_path: Path, translate_to_tr: bool) -> tuple[bytes, str]:
186
  """
187
  Tek SRT -> styled DOCX (bytes, filename)
188
  """
@@ -244,7 +257,8 @@ def process_srt_files(files, translate_to_tr: bool):
244
  if not files:
245
  return None
246
 
247
- paths = [Path(f) for f in files]
 
248
 
249
  zip_buffer = io.BytesIO()
250
  with zipfile.ZipFile(zip_buffer, "w", zipfile.ZIP_DEFLATED) as zf:
@@ -271,10 +285,10 @@ with gr.Blocks() as demo:
271
 
272
  - Bir veya birden fazla **.srt** yükle.
273
  - Her satır için:
274
- - **Character**: `WOMAN:`, `LEWIS:`, `NURSE:` gibi isimler çıkarılır.
275
  - **TC**: sadece **MM.SS** (start time'dan).
276
  - **TEXT**: `NAME:` prefix'leri atılmış metin.
277
- - İstersen TEXT'i **Helsinki-NLP/opus-mt-en-tr** ile Türkçe'ye çevir.
278
  - Çıktı: Tüm DOCX'leri içeren tek bir **ZIP**.
279
  """
280
  )
@@ -284,7 +298,7 @@ with gr.Blocks() as demo:
284
  label="Upload .srt files",
285
  file_types=[".srt"],
286
  file_count="multiple",
287
- type="filepath", # Gradio -> str path list
288
  )
289
 
290
  translate_chk = gr.Checkbox(
 
2
  import io
3
  import zipfile
4
  from pathlib import Path
5
+ from typing import Tuple
6
 
7
  import gradio as gr
8
  from docx import Document
9
  from docx.oxml import OxmlElement
10
  from docx.oxml.ns import qn
11
+
12
+ from transformers import (
13
+ AutoTokenizer,
14
+ AutoModelForSeq2SeqLM,
15
+ pipeline,
16
+ )
17
 
18
 
19
  # ----------------------------------------------------
20
+ # 1) ÇEVİRİ MODELİ (Helsinki-NLP / opus-mt-en-tr) - token=None
21
  # ----------------------------------------------------
22
 
23
+ MODEL_NAME = "Helsinki-NLP/opus-mt-en-tr"
24
+
25
+ # token=None => hiçbir HF token kullanılmasın, anonim indir.
26
+ tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, token=None)
27
+ model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_NAME, token=None)
28
+
29
  translator = pipeline(
30
  "translation",
31
+ model=model,
32
+ tokenizer=tokenizer,
33
  )
34
 
35
 
 
195
  shd.set(qn("w:fill"), "D9D9D9") # light grey
196
 
197
 
198
+ def srt_to_docx_bytes(srt_path: Path, translate_to_tr: bool) -> Tuple[bytes, str]:
199
  """
200
  Tek SRT -> styled DOCX (bytes, filename)
201
  """
 
257
  if not files:
258
  return None
259
 
260
+ # Gradio type="filepath" -> direkt string path listesi
261
+ paths = [Path(p) for p in files]
262
 
263
  zip_buffer = io.BytesIO()
264
  with zipfile.ZipFile(zip_buffer, "w", zipfile.ZIP_DEFLATED) as zf:
 
285
 
286
  - Bir veya birden fazla **.srt** yükle.
287
  - Her satır için:
288
+ - **Character**: `WOMAN:`, `LEWIS:`, `NURSE:` gibi isimler çıkarılır (çeviri yok).
289
  - **TC**: sadece **MM.SS** (start time'dan).
290
  - **TEXT**: `NAME:` prefix'leri atılmış metin.
291
+ - İstersen TEXT'i **Helsinki-NLP/opus-mt-en-tr** ile Türkçe'ye çevir (Character asla çevrilmez).
292
  - Çıktı: Tüm DOCX'leri içeren tek bir **ZIP**.
293
  """
294
  )
 
298
  label="Upload .srt files",
299
  file_types=[".srt"],
300
  file_count="multiple",
301
+ type="filepath", # Gradio -> string path list
302
  )
303
 
304
  translate_chk = gr.Checkbox(