Spaces:
Build error
Build error
Update soni_translate/translate_segments.py
Browse files
soni_translate/translate_segments.py
CHANGED
@@ -7,6 +7,8 @@ from .logging_setup import logger
|
|
7 |
import re
|
8 |
import json
|
9 |
import time
|
|
|
|
|
10 |
|
11 |
TRANSLATION_PROCESS_OPTIONS = [
|
12 |
"google_translator_batch",
|
@@ -15,12 +17,15 @@ TRANSLATION_PROCESS_OPTIONS = [
|
|
15 |
"gpt-3.5-turbo-0125",
|
16 |
"gpt-4-turbo-preview_batch",
|
17 |
"gpt-4-turbo-preview",
|
|
|
|
|
18 |
"disable_translation",
|
19 |
]
|
20 |
DOCS_TRANSLATION_PROCESS_OPTIONS = [
|
21 |
"google_translator",
|
22 |
"gpt-3.5-turbo-0125",
|
23 |
"gpt-4-turbo-preview",
|
|
|
24 |
"disable_translation",
|
25 |
]
|
26 |
|
@@ -418,6 +423,74 @@ def gpt_batch(segments, model, target, token_batch_limit=900, source=None):
|
|
418 |
)
|
419 |
|
420 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
421 |
def translate_text(
|
422 |
segments,
|
423 |
target,
|
@@ -443,7 +516,7 @@ def translate_text(
|
|
443 |
)
|
444 |
case model if model in ["gpt-3.5-turbo-0125", "gpt-4-turbo-preview"]:
|
445 |
return gpt_sequential(segments, model, target, source)
|
446 |
-
case model if model in ["gpt-3.5-turbo-0125_batch", "gpt-4-turbo-preview_batch"
|
447 |
return gpt_batch(
|
448 |
segments,
|
449 |
translation_process.replace("_batch", ""),
|
@@ -451,6 +524,10 @@ def translate_text(
|
|
451 |
token_batch_limit,
|
452 |
source
|
453 |
)
|
|
|
|
|
|
|
|
|
454 |
case "disable_translation":
|
455 |
return segments
|
456 |
case _:
|
|
|
7 |
import re
|
8 |
import json
|
9 |
import time
|
10 |
+
import os
|
11 |
+
import google.generativeai as genai
|
12 |
|
13 |
TRANSLATION_PROCESS_OPTIONS = [
|
14 |
"google_translator_batch",
|
|
|
17 |
"gpt-3.5-turbo-0125",
|
18 |
"gpt-4-turbo-preview_batch",
|
19 |
"gpt-4-turbo-preview",
|
20 |
+
"gemini-pro",
|
21 |
+
"gemini-pro_batch",
|
22 |
"disable_translation",
|
23 |
]
|
24 |
DOCS_TRANSLATION_PROCESS_OPTIONS = [
|
25 |
"google_translator",
|
26 |
"gpt-3.5-turbo-0125",
|
27 |
"gpt-4-turbo-preview",
|
28 |
+
"gemini-pro",
|
29 |
"disable_translation",
|
30 |
]
|
31 |
|
|
|
423 |
)
|
424 |
|
425 |
|
426 |
+
def check_gemini_api_key():
|
427 |
+
"""Check if Gemini API key is set in environment variables."""
|
428 |
+
if not os.environ.get("GOOGLE_API_KEY"):
|
429 |
+
raise ValueError(
|
430 |
+
"Gemini API key not found. Please set the GOOGLE_API_KEY environment variable."
|
431 |
+
)
|
432 |
+
|
433 |
+
def translate_with_gemini(text, target_lang, source_lang=None):
|
434 |
+
"""Translate text using Google's Gemini API."""
|
435 |
+
check_gemini_api_key()
|
436 |
+
genai.configure(api_key=os.environ["GOOGLE_API_KEY"])
|
437 |
+
model = genai.GenerativeModel('gemini-pro')
|
438 |
+
|
439 |
+
prompt = f"""Translate the following text to {target_lang}.
|
440 |
+
Keep the same tone and style. Preserve any special characters or formatting.
|
441 |
+
|
442 |
+
Text to translate: {text}
|
443 |
+
"""
|
444 |
+
if source_lang:
|
445 |
+
prompt = f"Translate from {source_lang} to {target_lang}: {text}"
|
446 |
+
|
447 |
+
response = model.generate_content(prompt)
|
448 |
+
return response.text.strip()
|
449 |
+
|
450 |
+
def gemini_sequential(segments, target, source=None):
|
451 |
+
"""Translate segments sequentially using Gemini."""
|
452 |
+
segments_ = copy.deepcopy(segments)
|
453 |
+
|
454 |
+
for line in tqdm(range(len(segments_))):
|
455 |
+
text = segments_[line]["text"]
|
456 |
+
translated_line = translate_with_gemini(text.strip(), target, source)
|
457 |
+
segments_[line]["text"] = translated_line
|
458 |
+
|
459 |
+
return segments_
|
460 |
+
|
461 |
+
def gemini_batch(segments, target, token_batch_limit=1000, source=None):
|
462 |
+
"""Translate segments in batches using Gemini."""
|
463 |
+
segments_ = copy.deepcopy(segments)
|
464 |
+
batch_texts = []
|
465 |
+
current_batch = []
|
466 |
+
current_length = 0
|
467 |
+
|
468 |
+
# Group texts into batches
|
469 |
+
for segment in segments_:
|
470 |
+
text_length = len(segment["text"])
|
471 |
+
if current_length + text_length > token_batch_limit:
|
472 |
+
batch_texts.append(current_batch)
|
473 |
+
current_batch = []
|
474 |
+
current_length = 0
|
475 |
+
current_batch.append(segment["text"])
|
476 |
+
current_length += text_length
|
477 |
+
|
478 |
+
if current_batch:
|
479 |
+
batch_texts.append(current_batch)
|
480 |
+
|
481 |
+
# Translate each batch
|
482 |
+
for i, batch in enumerate(tqdm(batch_texts)):
|
483 |
+
batch_text = "\n---\n".join(batch)
|
484 |
+
translated_batch = translate_with_gemini(batch_text, target, source)
|
485 |
+
translated_segments = translated_batch.split("\n---\n")
|
486 |
+
|
487 |
+
# Update segments with translations
|
488 |
+
start_idx = sum(len(b) for b in batch_texts[:i])
|
489 |
+
for j, translation in enumerate(translated_segments):
|
490 |
+
segments_[start_idx + j]["text"] = translation.strip()
|
491 |
+
|
492 |
+
return segments_
|
493 |
+
|
494 |
def translate_text(
|
495 |
segments,
|
496 |
target,
|
|
|
516 |
)
|
517 |
case model if model in ["gpt-3.5-turbo-0125", "gpt-4-turbo-preview"]:
|
518 |
return gpt_sequential(segments, model, target, source)
|
519 |
+
case model if model in ["gpt-3.5-turbo-0125_batch", "gpt-4-turbo-preview_batch"]:
|
520 |
return gpt_batch(
|
521 |
segments,
|
522 |
translation_process.replace("_batch", ""),
|
|
|
524 |
token_batch_limit,
|
525 |
source
|
526 |
)
|
527 |
+
case "gemini-pro":
|
528 |
+
return gemini_sequential(segments, target, source)
|
529 |
+
case "gemini-pro_batch":
|
530 |
+
return gemini_batch(segments, target, token_batch_limit, source)
|
531 |
case "disable_translation":
|
532 |
return segments
|
533 |
case _:
|