Spaces:
Running
Running
add fix exceed quota
Browse files- utils/utils.py +21 -29
utils/utils.py
CHANGED
|
@@ -7,7 +7,7 @@ import json
|
|
| 7 |
import time
|
| 8 |
from google.api_core.exceptions import ResourceExhausted
|
| 9 |
|
| 10 |
-
genai.configure(api_key="
|
| 11 |
|
| 12 |
|
| 13 |
def unzip_office_file(pptx_file: io.BytesIO):
|
|
@@ -26,57 +26,51 @@ def unzip_office_file(pptx_file: io.BytesIO):
|
|
| 26 |
|
| 27 |
|
| 28 |
def translate_single_text(text: str, source_lang: str = 'English', target_lang: str = "Vietnamese",
|
| 29 |
-
max_retries: int = 5,
|
| 30 |
if not text or not text.strip():
|
| 31 |
return "" # Bỏ qua nếu chuỗi rỗng hoặc chỉ chứa khoảng trắng
|
| 32 |
|
| 33 |
retries = 0
|
| 34 |
while retries <= max_retries:
|
| 35 |
try:
|
| 36 |
-
model = genai.GenerativeModel('gemini-2.0-flash')
|
| 37 |
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
|
| 42 |
|
| 43 |
user_prompt = f"Source language: {source_lang}. Target language: {target_lang}. Text to translate: {text}"
|
| 44 |
-
full_prompt =
|
| 45 |
|
| 46 |
response = model.generate_content(
|
| 47 |
contents=full_prompt,
|
| 48 |
generation_config={
|
| 49 |
-
'temperature': 0.2,
|
| 50 |
-
'top_p': 1.0,
|
| 51 |
-
'top_k': 1,
|
| 52 |
-
# 'max_output_tokens': 2048,
|
| 53 |
}
|
| 54 |
)
|
| 55 |
-
|
| 56 |
if response.candidates and response.candidates[0].content.parts:
|
| 57 |
translated_text = "".join(part.text for part in response.candidates[0].content.parts if hasattr(part, 'text')).strip()
|
| 58 |
return translated_text
|
| 59 |
else:
|
| 60 |
-
print(f"Không nhận được nội dung hợp lệ từ API cho văn bản: '{text[:50]}...'")
|
| 61 |
-
# Không thử lại với lỗi này, trả về rỗng
|
| 62 |
return ""
|
| 63 |
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
print(f"
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
retries += 1
|
| 70 |
-
else:
|
| 71 |
-
print(f"Đã vượt quá số lần thử lại tối đa ({max_retries + 1}) cho '{text[:50]}...'. Bỏ qua.")
|
| 72 |
-
return "" # Trả về rỗng sau khi đã thử lại tối đa số lần
|
| 73 |
|
| 74 |
except Exception as e:
|
| 75 |
-
print(f"Lỗi không mong muốn
|
| 76 |
return ""
|
| 77 |
|
| 78 |
-
|
| 79 |
-
|
| 80 |
|
| 81 |
def preprocess_text(text_list):
|
| 82 |
"""
|
|
@@ -141,7 +135,6 @@ def translate_text(text_dict, source_lang='English', target_lang="Vietnamese"):
|
|
| 141 |
|
| 142 |
|
| 143 |
system_prompt = f"""Translate the string values within the following JSON object .
|
| 144 |
-
|
| 145 |
Follow these instructions carefully:
|
| 146 |
1. Analyze the entire JSON object to understand the context.
|
| 147 |
2. Translate *only* the string values.
|
|
@@ -149,7 +142,6 @@ def translate_text(text_dict, source_lang='English', target_lang="Vietnamese"):
|
|
| 149 |
4. Do *not* translate non-string values (like hex color codes, numbers, or potentially proper nouns like 'CALISTOGA', 'DM SANS', 'Pexels', 'Pixabay' unless they have a common translation). Use your best judgment for proper nouns.
|
| 150 |
5. Preserve the original JSON structure perfectly.
|
| 151 |
6. Your output *must* be only the translated JSON object, without any introductory text, explanations, or markdown formatting like ```json ... ```.
|
| 152 |
-
|
| 153 |
"""
|
| 154 |
# 3. Construct User Prompt
|
| 155 |
user_prompt = f"Source language: {source_lang}. Target language: {target_lang}. JSON String: {json_input_string} \n\n Translated JSON Output:"
|
|
|
|
| 7 |
import time
|
| 8 |
from google.api_core.exceptions import ResourceExhausted
|
| 9 |
|
| 10 |
+
genai.configure(api_key="AIzaSyDInJcxzqBvsh1avs4Zkxb4ZGBooNzOyEM")
|
| 11 |
|
| 12 |
|
| 13 |
def unzip_office_file(pptx_file: io.BytesIO):
|
|
|
|
| 26 |
|
| 27 |
|
| 28 |
def translate_single_text(text: str, source_lang: str = 'English', target_lang: str = "Vietnamese",
|
| 29 |
+
max_retries: int = 5, base_delay: float = 5.0) -> str:
|
| 30 |
if not text or not text.strip():
|
| 31 |
return "" # Bỏ qua nếu chuỗi rỗng hoặc chỉ chứa khoảng trắng
|
| 32 |
|
| 33 |
retries = 0
|
| 34 |
while retries <= max_retries:
|
| 35 |
try:
|
| 36 |
+
model = genai.GenerativeModel('gemini-2.0-flash') # hoặc 'gemini-1.5-flash'
|
| 37 |
|
| 38 |
+
system_prompt = f"""You are a translation engine.
|
| 39 |
+
Translate the following text accurately from {source_lang} to {target_lang}.
|
| 40 |
+
Provide *only* the translated text as a single string.
|
| 41 |
+
Do NOT add any extra formatting, delimiters like '#', introductory phrases, or explanations."""
|
| 42 |
|
| 43 |
user_prompt = f"Source language: {source_lang}. Target language: {target_lang}. Text to translate: {text}"
|
| 44 |
+
full_prompt = system_prompt.strip() + "\n\n" + user_prompt.strip()
|
| 45 |
|
| 46 |
response = model.generate_content(
|
| 47 |
contents=full_prompt,
|
| 48 |
generation_config={
|
| 49 |
+
'temperature': 0.2,
|
| 50 |
+
'top_p': 1.0,
|
| 51 |
+
'top_k': 1,
|
|
|
|
| 52 |
}
|
| 53 |
)
|
| 54 |
+
|
| 55 |
if response.candidates and response.candidates[0].content.parts:
|
| 56 |
translated_text = "".join(part.text for part in response.candidates[0].content.parts if hasattr(part, 'text')).strip()
|
| 57 |
return translated_text
|
| 58 |
else:
|
| 59 |
+
print(f"[!] Không nhận được nội dung hợp lệ từ API cho văn bản: '{text[:50]}...'")
|
|
|
|
| 60 |
return ""
|
| 61 |
|
| 62 |
+
except ResourceExhausted as e:
|
| 63 |
+
wait_time = base_delay * (2 ** retries)
|
| 64 |
+
print(f"[429] Quota exceeded khi dịch '{text[:50]}...'. Thử lại sau {wait_time:.1f}s (lần {retries + 1}/{max_retries + 1}).")
|
| 65 |
+
time.sleep(wait_time)
|
| 66 |
+
retries += 1
|
|
|
|
|
|
|
|
|
|
|
|
|
| 67 |
|
| 68 |
except Exception as e:
|
| 69 |
+
print(f"[!] Lỗi không mong muốn khi dịch '{text[:50]}...': {e}")
|
| 70 |
return ""
|
| 71 |
|
| 72 |
+
print(f"[x] Bỏ qua sau {max_retries + 1} lần thử không thành công cho '{text[:50]}...'.")
|
| 73 |
+
return ""
|
| 74 |
|
| 75 |
def preprocess_text(text_list):
|
| 76 |
"""
|
|
|
|
| 135 |
|
| 136 |
|
| 137 |
system_prompt = f"""Translate the string values within the following JSON object .
|
|
|
|
| 138 |
Follow these instructions carefully:
|
| 139 |
1. Analyze the entire JSON object to understand the context.
|
| 140 |
2. Translate *only* the string values.
|
|
|
|
| 142 |
4. Do *not* translate non-string values (like hex color codes, numbers, or potentially proper nouns like 'CALISTOGA', 'DM SANS', 'Pexels', 'Pixabay' unless they have a common translation). Use your best judgment for proper nouns.
|
| 143 |
5. Preserve the original JSON structure perfectly.
|
| 144 |
6. Your output *must* be only the translated JSON object, without any introductory text, explanations, or markdown formatting like ```json ... ```.
|
|
|
|
| 145 |
"""
|
| 146 |
# 3. Construct User Prompt
|
| 147 |
user_prompt = f"Source language: {source_lang}. Target language: {target_lang}. JSON String: {json_input_string} \n\n Translated JSON Output:"
|