(src_filepath, dst_filepath, src_folder, dst_folder, cfg='s2twp.json')
| 35 | return text |
| 36 | |
| 37 | def convert_file(src_filepath, dst_filepath, src_folder, dst_folder, cfg='s2twp.json'): |
| 38 | print("convert %s to %s" % (src_filepath, dst_filepath)) |
| 39 | converter = opencc.OpenCC(cfg) |
| 40 | with open(src_filepath, "r", encoding='utf-8') as src, open(dst_filepath, "w+", encoding='utf-8') as dst: |
| 41 | dst.write("\n".join( |
| 42 | process_urls( |
| 43 | converter.convert(line.rstrip()) |
| 44 | .replace('一箇', '一個') |
| 45 | .replace('髮送', '傳送') |
| 46 | .replace('髮布', '釋出') |
| 47 | .replace('髮生', '發生') |
| 48 | .replace('髮出', '發出') |
| 49 | .replace('嚐試', '嘗試') |
| 50 | .replace('線上性一致', '在線性一致') # 优先按"在线"解析了? |
| 51 | .replace('復雜', '複雜') |
| 52 | .replace('討論瞭', '討論了') |
| 53 | .replace('瞭解釋', '了解釋') |
| 54 | .replace('瞭如', '了如') # 引入了如, 實現了如, 了如何, 了如果, 了如此 |
| 55 | .replace('了如指掌', '瞭如指掌') # 针对上一行的例外情况 |
| 56 | .replace('明瞭', '明了') # 闡明了, 聲明了, 指明了 |
| 57 | .replace('倒黴', '倒楣') |
| 58 | .replace('區域性性', '區域性') |
| 59 | .replace('下麵條件', '下面條件') # 优先按"面条"解析了? |
| 60 | .replace('當日志', '當日誌') # 优先按"当日"解析了? |
| 61 | .replace('真即時間', '真實時間') # 优先按"实时"解析了? |
| 62 | .replace('面向物件', '物件導向') |
| 63 | .replace('非規範化', '反正規化') |
| 64 | .replace('規範化', '正規化') |
| 65 | .replace('隻影響', '只影響'), |
| 66 | src_folder, dst_folder |
| 67 | ) |
| 68 | for line in src)) |
| 69 | |
| 70 | def convert(zh_folder, tw_folder): |
| 71 | home = os.path.abspath(os.path.join(os.path.dirname(os.path.abspath(sys.argv[0])), '..')) |
no test coverage detected