mirror of
https://github.com/morpheus65535/bazarr.git
synced 2025-06-27 17:09:34 -04:00
Improved Gemini translator with some hotfixes
This commit is contained in:
destpstrzy
GitHub
parent
3d57a1231c
commit
99f596e30b
2 changed files with 89 additions and 93 deletions
|
|
@ -2,6 +2,7 @@
|
|||
|
||||
import logging
|
||||
import datetime
|
||||
import os
|
||||
import pysubs2
|
||||
import srt
|
||||
|
||||
|
|
@ -212,13 +213,12 @@ def translate_subtitles_file_gemini(dest_srt_file, source_srt_file, to_lang, med
|
|||
filtered_params = {k: v for k, v in params.items() if v is not None}
|
||||
translator = TranslatorGemini(**filtered_params)
|
||||
translator.translate()
|
||||
|
||||
add_translator_info(dest_srt_file, f"# Subtitles translated with {settings.translator.gemini_model} # ")
|
||||
except Exception as e:
|
||||
os.remove(dest_srt_file)
|
||||
logging.error(f'translate encountered an error translating with Gemini: {str(e)}')
|
||||
show_message(f'Gemini translation error: {str(e)}')
|
||||
|
||||
add_translator_info(dest_srt_file, f"# Subtitles translated with {settings.translator.gemini_model} # ")
|
||||
|
||||
except Exception as e:
|
||||
logging.error(f'BAZARR encountered an error translating with Gemini: {str(e)}')
|
||||
return False
|
||||
|
|
|
|||
|
|
@ -117,6 +117,7 @@ class TranslatorGemini:
|
|||
|
||||
The indices must remain the same in the response as in the request.
|
||||
Dialogs must be translated as they are without any changes.
|
||||
If a line has a comma or multiple sentences, try to keep one line to about 40-50 characters.
|
||||
"""
|
||||
if description:
|
||||
instruction += "\nAdditional user instruction: '" + description + "'"
|
||||
|
|
@ -178,10 +179,6 @@ class TranslatorGemini:
|
|||
return False
|
||||
|
||||
def translate(self):
|
||||
"""
|
||||
Main translation method. Reads the input subtitle file, translates it in batches,
|
||||
and writes the translated subtitles to the output file.
|
||||
"""
|
||||
if not self.current_api_key:
|
||||
show_message("Please provide a valid Gemini API key.")
|
||||
return
|
||||
|
|
@ -196,98 +193,94 @@ class TranslatorGemini:
|
|||
|
||||
self.token_limit = self._get_token_limit()
|
||||
|
||||
# Setup signal handlers if possible
|
||||
is_main_thread = self.setup_signal_handlers()
|
||||
try:
|
||||
with open(self.input_file, "r", encoding="utf-8") as original_file:
|
||||
original_text = original_file.read()
|
||||
original_subtitle = list(srt.parse(original_text))
|
||||
|
||||
with open(self.input_file, "r", encoding="utf-8") as original_file:
|
||||
original_text = original_file.read()
|
||||
original_subtitle = list(srt.parse(original_text))
|
||||
try:
|
||||
translated_subtitle = original_subtitle.copy()
|
||||
|
||||
except FileNotFoundError:
|
||||
translated_subtitle = original_subtitle.copy()
|
||||
|
||||
if len(original_subtitle) != len(translated_subtitle):
|
||||
show_message(
|
||||
f"Number of lines of existing translated file does not match the number of lines in the original file."
|
||||
)
|
||||
return
|
||||
|
||||
translated_file = open(self.output_file, "w", encoding="utf-8")
|
||||
|
||||
if self.start_line > len(original_subtitle) or self.start_line < 1:
|
||||
show_message(f"Start line must be between 1 and {len(original_subtitle)}. Please try again.")
|
||||
return
|
||||
|
||||
if len(original_subtitle) < self.batch_size:
|
||||
self.batch_size = len(original_subtitle)
|
||||
|
||||
delay = False
|
||||
delay_time = 30
|
||||
|
||||
i = self.start_line - 1
|
||||
total = len(original_subtitle)
|
||||
batch = [SubtitleObject(index=str(i), content=original_subtitle[i].content)]
|
||||
|
||||
i += 1
|
||||
|
||||
# Save initial progress
|
||||
self._save_progress(i)
|
||||
|
||||
last_time = 0
|
||||
while (i < total or len(batch) > 0) and not self.interrupt_flag:
|
||||
if i < total and len(batch) < self.batch_size:
|
||||
batch.append(SubtitleObject(index=str(i), content=original_subtitle[i].content))
|
||||
i += 1
|
||||
continue
|
||||
try:
|
||||
if not self._validate_token_size(json.dumps(batch, ensure_ascii=False)):
|
||||
show_message(
|
||||
f"Token size ({int(self.token_count / 0.9)}) exceeds limit ({self.token_limit}) for {self.model_name}."
|
||||
)
|
||||
user_prompt = "0"
|
||||
while not user_prompt.isdigit() or int(user_prompt) <= 0:
|
||||
user_prompt = show_message(
|
||||
f"Please enter a new batch size (current: {self.batch_size}): "
|
||||
)
|
||||
if user_prompt.isdigit() and int(user_prompt) > 0:
|
||||
new_batch_size = int(user_prompt)
|
||||
decrement = self.batch_size - new_batch_size
|
||||
if decrement > 0:
|
||||
for _ in range(decrement):
|
||||
i -= 1
|
||||
batch.pop()
|
||||
self.batch_size = new_batch_size
|
||||
show_message(f"Batch size updated to {self.batch_size}.")
|
||||
else:
|
||||
show_message("Invalid input. Batch size must be a positive integer.")
|
||||
continue
|
||||
translated_subtitle = original_subtitle.copy()
|
||||
except FileNotFoundError:
|
||||
translated_subtitle = original_subtitle.copy()
|
||||
|
||||
start_time = time.time()
|
||||
self._process_batch(batch, translated_subtitle, total)
|
||||
end_time = time.time()
|
||||
# Use with statement for the output file too
|
||||
with open(self.output_file, "w", encoding="utf-8") as translated_file:
|
||||
if len(original_subtitle) < self.batch_size:
|
||||
self.batch_size = len(original_subtitle)
|
||||
|
||||
# Save progress after each batch
|
||||
self._save_progress(i + 1)
|
||||
delay = False
|
||||
delay_time = 30
|
||||
|
||||
if delay and (end_time - start_time < delay_time) and i < total:
|
||||
time.sleep(delay_time - (end_time - start_time))
|
||||
except Exception as e:
|
||||
i = self.start_line - 1
|
||||
total = len(original_subtitle)
|
||||
batch = [SubtitleObject(index=str(i), content=original_subtitle[i].content)]
|
||||
|
||||
i += 1
|
||||
|
||||
# Save initial progress
|
||||
self._save_progress(i)
|
||||
|
||||
while (i < total or len(batch) > 0) and not self.interrupt_flag:
|
||||
if i < total and len(batch) < self.batch_size:
|
||||
batch.append(SubtitleObject(index=str(i), content=original_subtitle[i].content))
|
||||
i += 1
|
||||
continue
|
||||
|
||||
try:
|
||||
if not self._validate_token_size(json.dumps(batch, ensure_ascii=False)):
|
||||
show_message(
|
||||
f"Token size ({int(self.token_count / 0.9)}) exceeds limit ({self.token_limit}) for {self.model_name}."
|
||||
)
|
||||
user_prompt = "0"
|
||||
while not user_prompt.isdigit() or int(user_prompt) <= 0:
|
||||
user_prompt = show_message(
|
||||
f"Please enter a new batch size (current: {self.batch_size}): "
|
||||
)
|
||||
if user_prompt.isdigit() and int(user_prompt) > 0:
|
||||
new_batch_size = int(user_prompt)
|
||||
decrement = self.batch_size - new_batch_size
|
||||
if decrement > 0:
|
||||
for _ in range(decrement):
|
||||
i -= 1
|
||||
batch.pop()
|
||||
self.batch_size = new_batch_size
|
||||
show_message(f"Batch size updated to {self.batch_size}.")
|
||||
else:
|
||||
show_message("Invalid input. Batch size must be a positive integer.")
|
||||
continue
|
||||
|
||||
start_time = time.time()
|
||||
self._process_batch(batch, translated_subtitle, total)
|
||||
end_time = time.time()
|
||||
|
||||
# Save progress after each batch
|
||||
self._save_progress(i + 1)
|
||||
|
||||
if delay and (end_time - start_time < delay_time) and i < total:
|
||||
time.sleep(delay_time - (end_time - start_time))
|
||||
|
||||
except Exception as e:
|
||||
hide_progress(id=f'translate_progress_{self.output_file}')
|
||||
self._clear_progress()
|
||||
# File will be automatically closed by the with statement
|
||||
raise e
|
||||
|
||||
# Check if we exited the loop due to an interrupt
|
||||
hide_progress(id=f'translate_progress_{self.output_file}')
|
||||
if self.interrupt_flag:
|
||||
# File will be automatically closed by the with statement
|
||||
self._clear_progress()
|
||||
|
||||
# Write the final result - this happens inside the with block
|
||||
translated_file.write(srt.compose(translated_subtitle))
|
||||
|
||||
# Clear progress file on successful completion
|
||||
self._clear_progress()
|
||||
raise e
|
||||
|
||||
# Check if we exited the loop due to an interrupt
|
||||
except Exception as e:
|
||||
hide_progress(id=f'translate_progress_{self.output_file}')
|
||||
if self.interrupt_flag:
|
||||
return
|
||||
|
||||
|
||||
translated_file.write(srt.compose(translated_subtitle))
|
||||
|
||||
# Clear progress file on successful completion
|
||||
self._clear_progress()
|
||||
raise e
|
||||
|
||||
def _get_token_limit(self) -> int:
|
||||
"""
|
||||
|
|
@ -322,6 +315,7 @@ class TranslatorGemini:
|
|||
batch: List[SubtitleObject], # Changed from list[SubtitleObject]
|
||||
translated_subtitle: List[Subtitle], # Changed from list[Subtitle]
|
||||
total: int,
|
||||
retry_num = 3
|
||||
):
|
||||
"""
|
||||
Process a batch of subtitles for translation with accurate progress tracking.
|
||||
|
|
@ -398,10 +392,12 @@ class TranslatorGemini:
|
|||
|
||||
return self.current_progress
|
||||
|
||||
except requests.RequestException as e:
|
||||
# More comprehensive error handling
|
||||
print(f"Translation request failed: {e}")
|
||||
raise e
|
||||
except Exception as e:
|
||||
if retry_num > 0:
|
||||
return self._process_batch(batch, translated_subtitle, total, retry_num - 1)
|
||||
else:
|
||||
show_message(f"Translation request failed: {e}")
|
||||
raise e
|
||||
|
||||
@staticmethod
|
||||
def _process_translated_lines(
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue