python · josix · Mar 1, 2023 · Jan 14, 2023 · Jan 14, 2023 · Jan 14, 2023
diff --git a/.scripts/README.md b/.scripts/README.md
@@ -0,0 +1,20 @@
+# Scripts
+
+Useful scripts for the translation.
+
+## From Google Translation
+
+Translate all untranslated entries of the given .po file with Google Translate.
+
+
+```sh
+.scripts/google_translate.sh library/csv.po
+```
+
+## From zh_CN Translation
+
+If a specific doc has been translated into Simplified Chinese (zh_CN) and you'd like to adopt it as a base, you can insert the command:
+
+```sh
+.scripts/from_cn.sh library/csv.po
+```
diff --git a/.scripts/from_cn.sh b/.scripts/from_cn.sh
@@ -0,0 +1,44 @@
+#!/bin/sh
+cd .scripts
+source utils/install_poetry.sh
+
+# check if OpenCC is installed
+if [[ ! -x "`which opencc 2>/dev/null`" ]]
+then
+    echo "You do not have OpenCC installed. Please install it first."
+    echo "Instruction: https://github.com/BYVoid/OpenCC/wiki/Download"
+    exit 1
+fi
+
+# clone pydoc zh_CN repo and pull from remote
+CN_REPO=.python-docs-zh-cn
+if [[ ! -d $CN_REPO ]]
+then
+    read -p "You do not have a clone of zh_CN repo. Clone now? (y/N)" choice
+    case "$choice" in 
+        y|Y ) git clone --depth 1 --no-single-branch https://github.com/python/python-docs-zh-cn $CN_REPO ;;
+        n|N|* ) echo "Aborted"; exit 1 ;;
+    esac
+fi
+git -C $CN_REPO checkout 3.10  # the current latest version of CN repo
+git -C $CN_REPO pull
+
+
+# convert zh_CN po content and merge into zh_TW po
+TARGET=$1
+CN_PATH=$CN_REPO/$TARGET
+TW_PATH=../$TARGET
+
+poetry lock
+poetry install
+poetry run bash -c "
+    opencc -i $CN_PATH -c s2twp.json -o /tmp/tmp.po
+    pofilter --nonotes --excludefilter unchanged --excludefilter untranslated /tmp/tmp.po | msgattrib --set-fuzzy -o /tmp/tmp.po
+    pomerge -t $CN_PATH -i /tmp/tmp.po -o /tmp/tmp.po
+
+    pofilter --nonotes --excludefilter untranslated $TW_PATH /tmp/tmp2.po
+    pomerge -t /tmp/tmp.po -i /tmp/tmp2.po -o /tmp/tmp3.po
+    msgcat --lang zh_TW /tmp/tmp3.po -o $TW_PATH
+"
+
+rm /tmp/tmp.po /tmp/tmp2.po /tmp/tmp3.po
diff --git a/.scripts/google_translate.sh b/.scripts/google_translate.sh
@@ -0,0 +1,17 @@
+#!/bin/sh
+
+WORK_DIR=.scripts
+cd $WORK_DIR
+
+source utils/install_poetry.sh
+
+TEMP=tmp.po
+TARGET=../$1
+
+poetry lock
+poetry install
+poetry run bash -c "
+    python google_translate/main.py $TARGET > $TEMP
+    pomerge -t $TARGET -i $TEMP -o $TARGET
+"
+rm $TEMP
diff --git a/.scripts/google_translate/main.py b/.scripts/google_translate/main.py
@@ -0,0 +1,51 @@
+import argparse
+import logging
+from pathlib import Path
+from typing import List
+
+import polib
+from googletrans import Translator
+
+from utils import refine_translations
+
+
+def _get_po_paths(path: Path) -> List[Path]:
+    """Find all .po files in given path"""
+    if not path.exists():
+        logging.error(f"The path '{path.absolute()}' does not exist!")
+
+    # return 1-element list if it's a file
+    if path.is_file():
+        return [path.resolve()]
+
+    # find all .po files
+    po_paths = [p.resolve() for p in path.glob("**/*.po")]
+    return po_paths
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "path",
+        help="the path of a PO file or a directory containing PO files"
+    )
+    args = parser.parse_args()
+
+    translator = Translator()
+    po_files = _get_po_paths(Path(args.path).resolve())
+    errors = []
+    for path in po_files:
+        try:
+            pofile = polib.pofile(path)
+        except OSError:
+            errors.append(f"{path} doesn't seem to be a .po file")
+            continue
+
+        for entry in pofile.untranslated_entries()[::-1]:
+            translation = translator.translate(entry.msgid, src='en', dest='zh-TW')
+
+            print(
+                '#, fuzzy\n'
+                f'msgid "{repr(entry.msgid)[1:-1]}"\n'
+                f'msgstr "{repr(refine_translations(translation.text))[1:-1]}"\n'
+            )
diff --git a/.scripts/google_translate/utils.py b/.scripts/google_translate/utils.py
@@ -0,0 +1,56 @@
+MAPPING_ZH_TW_COMMON_TRANSLATION_ERROR = {
+    '創建': '建立',  # create
+    '代碼': '程式碼',  # code
+    '信息': '資訊',  # information
+    '模塊': '模組',  # module
+    '標誌': '旗標',  # flag
+    '異常': '例外',  # exception
+    '解釋器': '直譯器',  # interpreter
+    '頭文件': '標頭檔',  # header
+    '對象': '物件',  # objetc
+    '支持': '支援',  # support
+    '默認': '預設',  # default
+    '兼容': '相容',  # compatible
+    '字符串': '字串',  # string
+    '宏': '巨集',  # macro
+    '描述符': '描述器',  # descriptor
+    '字節': '位元組',  # bytes
+    '緩存': '快取',  # cache
+    '調用': '呼叫',  # call
+    '哈希': '雜湊',  # hash
+    '類型': '型別',  # type
+    '子類': '子類別',  # subclass
+    '實現': '實作',  # implement
+    '數據': '資料',  # data
+    '返回': '回傳',  # return
+    '指針': '指標',  # pointer
+    '字段': '欄位',  # field
+    '擴展': '擴充',  # extension
+    '遞歸': '遞迴',  # recursive
+    '用戶': '使用者',  # user
+    '算法': '演算法',  # algorithm
+    '優化': '最佳化',  # optimize
+    '字符': '字元',  # character
+    '設置': '設定',  # setting/configure
+    '線程': '執行緒',  # thread
+    '進程': '行程',  # process
+    '迭代': '疊代',  # iterate
+    '內存': '記憶體',  # memory
+    '打印': '印出',  # print
+    '異步': '非同步',  # async
+    '調試': '除錯',  # debug
+    '堆棧': '堆疊',  # stack
+    '回調': '回呼',  # callback
+    '公共': '公開',  # public
+    '函數': '函式',  # function
+    '變量': '變數',  # variable
+    '常量': '常數',  # constant
+    '添加': '新增',  # add
+    '基類': '基底類別',  # base class
+}
+
+
+def refine_translations(s: str) -> str:
+    for original, target in MAPPING_ZH_TW_COMMON_TRANSLATION_ERROR.items():
+        s = s.replace(original, target)
+    return s