424 lines
16 KiB
Python
424 lines
16 KiB
Python
#!/usr/bin/env python3
|
||
# -*- coding: utf-8 -*-
|
||
"""
|
||
翻譯服務
|
||
|
||
Author: PANJIT IT Team
|
||
Created: 2024-01-28
|
||
Modified: 2024-01-28
|
||
"""
|
||
|
||
import hashlib
|
||
import time
|
||
from pathlib import Path
|
||
from typing import List, Dict, Any, Optional
|
||
from app.utils.logger import get_logger
|
||
from app.utils.exceptions import TranslationError, FileProcessingError
|
||
from app.services.dify_client import DifyClient
|
||
from app.models.cache import TranslationCache
|
||
from app.models.job import TranslationJob
|
||
from app.utils.helpers import generate_filename, create_job_directory
|
||
|
||
logger = get_logger(__name__)
|
||
|
||
|
||
class DocumentParser:
|
||
"""文件解析器基類"""
|
||
|
||
def __init__(self, file_path: str):
|
||
self.file_path = Path(file_path)
|
||
|
||
if not self.file_path.exists():
|
||
raise FileProcessingError(f"檔案不存在: {file_path}")
|
||
|
||
def extract_text_segments(self) -> List[str]:
|
||
"""提取文字片段"""
|
||
raise NotImplementedError
|
||
|
||
def generate_translated_document(self, translations: Dict[str, List[str]],
|
||
target_language: str, output_dir: Path) -> str:
|
||
"""生成翻譯後的文件"""
|
||
raise NotImplementedError
|
||
|
||
|
||
class DocxParser(DocumentParser):
|
||
"""DOCX 文件解析器"""
|
||
|
||
def extract_text_segments(self) -> List[str]:
|
||
"""提取 DOCX 文件的文字片段"""
|
||
try:
|
||
import docx
|
||
from docx.table import _Cell
|
||
|
||
doc = docx.Document(str(self.file_path))
|
||
text_segments = []
|
||
|
||
# 提取段落文字
|
||
for paragraph in doc.paragraphs:
|
||
text = paragraph.text.strip()
|
||
if text and len(text) > 3: # 過濾太短的文字
|
||
text_segments.append(text)
|
||
|
||
# 提取表格文字
|
||
for table in doc.tables:
|
||
for row in table.rows:
|
||
for cell in row.cells:
|
||
text = cell.text.strip()
|
||
if text and len(text) > 3:
|
||
text_segments.append(text)
|
||
|
||
logger.info(f"Extracted {len(text_segments)} text segments from DOCX")
|
||
return text_segments
|
||
|
||
except Exception as e:
|
||
logger.error(f"Failed to extract text from DOCX: {str(e)}")
|
||
raise FileProcessingError(f"DOCX 文件解析失敗: {str(e)}")
|
||
|
||
def generate_translated_document(self, translations: Dict[str, List[str]],
|
||
target_language: str, output_dir: Path) -> str:
|
||
"""生成翻譯後的 DOCX 文件"""
|
||
try:
|
||
import docx
|
||
from docx.shared import Pt
|
||
|
||
# 開啟原始文件
|
||
doc = docx.Document(str(self.file_path))
|
||
|
||
# 取得對應的翻譯
|
||
translated_texts = translations.get(target_language, [])
|
||
text_index = 0
|
||
|
||
# 處理段落
|
||
for paragraph in doc.paragraphs:
|
||
if paragraph.text.strip() and len(paragraph.text.strip()) > 3:
|
||
if text_index < len(translated_texts):
|
||
# 保留原文，添加翻譯
|
||
original_text = paragraph.text
|
||
translated_text = translated_texts[text_index]
|
||
|
||
# 清空段落
|
||
paragraph.clear()
|
||
|
||
# 添加原文
|
||
run = paragraph.add_run(original_text)
|
||
|
||
# 添加翻譯（新行，較小字體）
|
||
paragraph.add_run('\n')
|
||
trans_run = paragraph.add_run(translated_text)
|
||
trans_run.font.size = Pt(10)
|
||
trans_run.italic = True
|
||
|
||
text_index += 1
|
||
|
||
# 處理表格（簡化版本）
|
||
for table in doc.tables:
|
||
for row in table.rows:
|
||
for cell in row.cells:
|
||
if cell.text.strip() and len(cell.text.strip()) > 3:
|
||
if text_index < len(translated_texts):
|
||
original_text = cell.text
|
||
translated_text = translated_texts[text_index]
|
||
|
||
# 清空儲存格
|
||
cell.text = f"{original_text}\n{translated_text}"
|
||
|
||
text_index += 1
|
||
|
||
# 生成輸出檔名
|
||
output_filename = generate_filename(
|
||
self.file_path.name,
|
||
'translated',
|
||
'translated',
|
||
target_language
|
||
)
|
||
output_path = output_dir / output_filename
|
||
|
||
# 儲存文件
|
||
doc.save(str(output_path))
|
||
|
||
logger.info(f"Generated translated DOCX: {output_path}")
|
||
return str(output_path)
|
||
|
||
except Exception as e:
|
||
logger.error(f"Failed to generate translated DOCX: {str(e)}")
|
||
raise FileProcessingError(f"生成翻譯 DOCX 失敗: {str(e)}")
|
||
|
||
|
||
class PdfParser(DocumentParser):
|
||
"""PDF 文件解析器（只讀）"""
|
||
|
||
def extract_text_segments(self) -> List[str]:
|
||
"""提取 PDF 文件的文字片段"""
|
||
try:
|
||
from PyPDF2 import PdfReader
|
||
|
||
reader = PdfReader(str(self.file_path))
|
||
text_segments = []
|
||
|
||
for page in reader.pages:
|
||
text = page.extract_text()
|
||
|
||
# 簡單的句子分割
|
||
sentences = text.split('.')
|
||
for sentence in sentences:
|
||
sentence = sentence.strip()
|
||
if sentence and len(sentence) > 10:
|
||
text_segments.append(sentence)
|
||
|
||
logger.info(f"Extracted {len(text_segments)} text segments from PDF")
|
||
return text_segments
|
||
|
||
except Exception as e:
|
||
logger.error(f"Failed to extract text from PDF: {str(e)}")
|
||
raise FileProcessingError(f"PDF 文件解析失敗: {str(e)}")
|
||
|
||
def generate_translated_document(self, translations: Dict[str, List[str]],
|
||
target_language: str, output_dir: Path) -> str:
|
||
"""生成翻譯文字檔（PDF 不支援直接編輯）"""
|
||
try:
|
||
translated_texts = translations.get(target_language, [])
|
||
|
||
# 生成純文字檔案
|
||
output_filename = f"{self.file_path.stem}_{target_language}_translated.txt"
|
||
output_path = output_dir / output_filename
|
||
|
||
with open(output_path, 'w', encoding='utf-8') as f:
|
||
f.write(f"翻譯結果 - {target_language}\n")
|
||
f.write("=" * 50 + "\n\n")
|
||
|
||
for i, text in enumerate(translated_texts):
|
||
f.write(f"{i+1}. {text}\n\n")
|
||
|
||
logger.info(f"Generated translated text file: {output_path}")
|
||
return str(output_path)
|
||
|
||
except Exception as e:
|
||
logger.error(f"Failed to generate translated text file: {str(e)}")
|
||
raise FileProcessingError(f"生成翻譯文字檔失敗: {str(e)}")
|
||
|
||
|
||
class TranslationService:
|
||
"""翻譯服務"""
|
||
|
||
def __init__(self):
|
||
self.dify_client = DifyClient()
|
||
|
||
# 文件解析器映射
|
||
self.parsers = {
|
||
'.docx': DocxParser,
|
||
'.doc': DocxParser, # 假設可以用 docx 處理
|
||
'.pdf': PdfParser,
|
||
# 其他格式可以稍後添加
|
||
}
|
||
|
||
def get_document_parser(self, file_path: str) -> DocumentParser:
|
||
"""取得文件解析器"""
|
||
file_ext = Path(file_path).suffix.lower()
|
||
|
||
parser_class = self.parsers.get(file_ext)
|
||
if not parser_class:
|
||
raise FileProcessingError(f"不支援的檔案格式: {file_ext}")
|
||
|
||
return parser_class(file_path)
|
||
|
||
def split_text_into_sentences(self, text: str, language: str = 'auto') -> List[str]:
|
||
"""將文字分割成句子"""
|
||
# 這裡可以使用更智能的句子分割
|
||
# 暫時使用簡單的分割方式
|
||
|
||
sentences = []
|
||
|
||
# 基本的句子分割符號
|
||
separators = ['. ', '。', '！', '？', '!', '?']
|
||
|
||
current_text = text
|
||
for sep in separators:
|
||
parts = current_text.split(sep)
|
||
if len(parts) > 1:
|
||
sentences.extend([part.strip() + sep.rstrip() for part in parts[:-1] if part.strip()])
|
||
current_text = parts[-1]
|
||
|
||
# 添加最後一部分
|
||
if current_text.strip():
|
||
sentences.append(current_text.strip())
|
||
|
||
# 過濾太短的句子
|
||
sentences = [s for s in sentences if len(s.strip()) > 5]
|
||
|
||
return sentences
|
||
|
||
def translate_text_with_cache(self, text: str, source_language: str,
|
||
target_language: str, user_id: int = None,
|
||
job_id: int = None) -> str:
|
||
"""帶快取的文字翻譯"""
|
||
|
||
# 檢查快取
|
||
cached_translation = TranslationCache.get_translation(
|
||
text, source_language, target_language
|
||
)
|
||
|
||
if cached_translation:
|
||
logger.debug(f"Cache hit for translation: {text[:50]}...")
|
||
return cached_translation
|
||
|
||
# 呼叫 Dify API
|
||
try:
|
||
result = self.dify_client.translate_text(
|
||
text=text,
|
||
source_language=source_language,
|
||
target_language=target_language,
|
||
user_id=user_id,
|
||
job_id=job_id
|
||
)
|
||
|
||
translated_text = result['translated_text']
|
||
|
||
# 儲存到快取
|
||
TranslationCache.save_translation(
|
||
text, source_language, target_language, translated_text
|
||
)
|
||
|
||
return translated_text
|
||
|
||
except Exception as e:
|
||
logger.error(f"Translation failed for text: {text[:50]}... Error: {str(e)}")
|
||
raise TranslationError(f"翻譯失敗: {str(e)}")
|
||
|
||
def translate_document(self, job_uuid: str) -> Dict[str, Any]:
|
||
"""翻譯文件（主要入口點）"""
|
||
try:
|
||
# 取得任務資訊
|
||
job = TranslationJob.query.filter_by(job_uuid=job_uuid).first()
|
||
if not job:
|
||
raise TranslationError(f"找不到任務: {job_uuid}")
|
||
|
||
logger.info(f"Starting document translation: {job_uuid}")
|
||
|
||
# 更新任務狀態
|
||
job.update_status('PROCESSING', progress=0)
|
||
|
||
# 取得文件解析器
|
||
parser = self.get_document_parser(job.file_path)
|
||
|
||
# 提取文字片段
|
||
logger.info("Extracting text segments from document")
|
||
text_segments = parser.extract_text_segments()
|
||
|
||
if not text_segments:
|
||
raise TranslationError("文件中未找到可翻譯的文字")
|
||
|
||
# 分割成句子
|
||
logger.info("Splitting text into sentences")
|
||
all_sentences = []
|
||
for segment in text_segments:
|
||
sentences = self.split_text_into_sentences(segment, job.source_language)
|
||
all_sentences.extend(sentences)
|
||
|
||
# 去重複
|
||
unique_sentences = list(dict.fromkeys(all_sentences)) # 保持順序的去重
|
||
logger.info(f"Found {len(unique_sentences)} unique sentences to translate")
|
||
|
||
# 批次翻譯
|
||
translation_results = {}
|
||
total_sentences = len(unique_sentences)
|
||
|
||
for target_language in job.target_languages:
|
||
logger.info(f"Translating to {target_language}")
|
||
translated_sentences = []
|
||
|
||
for i, sentence in enumerate(unique_sentences):
|
||
try:
|
||
translated = self.translate_text_with_cache(
|
||
text=sentence,
|
||
source_language=job.source_language,
|
||
target_language=target_language,
|
||
user_id=job.user_id,
|
||
job_id=job.id
|
||
)
|
||
translated_sentences.append(translated)
|
||
|
||
# 更新進度
|
||
progress = (i + 1) / total_sentences * 100 / len(job.target_languages)
|
||
current_lang_index = job.target_languages.index(target_language)
|
||
total_progress = (current_lang_index * 100 + progress) / len(job.target_languages)
|
||
job.update_status('PROCESSING', progress=total_progress)
|
||
|
||
# 短暫延遲避免過快請求
|
||
time.sleep(0.1)
|
||
|
||
except Exception as e:
|
||
logger.error(f"Failed to translate sentence: {sentence[:50]}... Error: {str(e)}")
|
||
# 翻譯失敗時保留原文
|
||
translated_sentences.append(f"[翻譯失敗] {sentence}")
|
||
|
||
translation_results[target_language] = translated_sentences
|
||
|
||
# 生成翻譯文件
|
||
logger.info("Generating translated documents")
|
||
output_dir = Path(job.file_path).parent
|
||
output_files = {}
|
||
|
||
for target_language, translations in translation_results.items():
|
||
try:
|
||
# 重建翻譯映射
|
||
translation_mapping = {target_language: translations}
|
||
|
||
output_file = parser.generate_translated_document(
|
||
translations=translation_mapping,
|
||
target_language=target_language,
|
||
output_dir=output_dir
|
||
)
|
||
|
||
output_files[target_language] = output_file
|
||
|
||
# 記錄翻譯檔案到資料庫
|
||
file_size = Path(output_file).stat().st_size
|
||
job.add_translated_file(
|
||
language_code=target_language,
|
||
filename=Path(output_file).name,
|
||
file_path=output_file,
|
||
file_size=file_size
|
||
)
|
||
|
||
except Exception as e:
|
||
logger.error(f"Failed to generate translated document for {target_language}: {str(e)}")
|
||
raise TranslationError(f"生成 {target_language} 翻譯文件失敗: {str(e)}")
|
||
|
||
# 計算總成本（從 API 使用統計中取得）
|
||
total_cost = self._calculate_job_cost(job.id)
|
||
|
||
# 更新任務狀態為完成
|
||
job.update_status('COMPLETED', progress=100)
|
||
job.total_cost = total_cost
|
||
job.total_tokens = len(unique_sentences) # 簡化的 token 計算
|
||
|
||
from app import db
|
||
db.session.commit()
|
||
|
||
logger.info(f"Document translation completed: {job_uuid}")
|
||
|
||
return {
|
||
'success': True,
|
||
'job_uuid': job_uuid,
|
||
'output_files': output_files,
|
||
'total_sentences': len(unique_sentences),
|
||
'total_cost': float(total_cost),
|
||
'target_languages': job.target_languages
|
||
}
|
||
|
||
except TranslationError:
|
||
raise
|
||
except Exception as e:
|
||
logger.error(f"Document translation failed: {job_uuid}. Error: {str(e)}")
|
||
raise TranslationError(f"文件翻譯失敗: {str(e)}")
|
||
|
||
def _calculate_job_cost(self, job_id: int) -> float:
|
||
"""計算任務總成本"""
|
||
from app import db
|
||
from sqlalchemy import func
|
||
|
||
total_cost = db.session.query(
|
||
func.sum(APIUsageStats.cost)
|
||
).filter_by(job_id=job_id).scalar()
|
||
|
||
return float(total_cost) if total_cost else 0.0 |