Add end-to-end English language support across skill flow, prompts, and tools - #22
Add end-to-end English language support across skill flow, prompts, and tools#22imAaryash wants to merge 6 commits into
Conversation
Introduce session-level language selection and i18n across the project. Add .gitignore and update README/SKILL to require language choice and to lock subsequent prompts to preferred_language. Update prompt files (prompts/*.md) to accept and respect preferred_language and to produce English/Chinese outputs. Extend tools with language-aware behavior: skill_writer (templates, language normalization, generation/update hooks), version_manager, wechat_decryptor, and wechat_parser (CLI --lang flags, tr helpers, localized messages and outputs). Ensure created meta stores preferred_language and CLI outputs reflect chosen language.
There was a problem hiding this comment.
Pull request overview
This PR adds bilingual (Chinese/English) support across the ex-skill workflow by introducing a preferred_language concept in the docs/prompts and extending the Python CLI tools to generate/output content in the selected language.
Changes:
- Added language selection guidance (
preferred_language) to the workflow docs and propagated language rules into prompt templates. - Extended CLI tools with
--langand bilingual runtime output (WeChat decrypt/parse, version manager, skill writer). - Added English skill template generation and persisted language metadata (
preferred_language/language) in generatedmeta.json, plus a new.gitignore.
Reviewed changes
Copilot reviewed 12 out of 13 changed files in this pull request and generated 3 comments.
Show a summary per file
| File | Description |
|---|---|
| tools/wechat_parser.py | Adds --lang, translation helper, and English-capable output formatting for extracted chats. |
| tools/wechat_decryptor.py | Adds --lang and bilingual runtime messaging for decrypt flow. |
| tools/version_manager.py | Adds --lang and bilingual output for list/rollback/cleanup. |
| tools/skill_writer.py | Adds EN/ZH templates, language normalization, and persists preferred language in metadata. |
| SKILL.md | Documents Step 0 language selection and “language-locked” execution behavior. |
| README.md | Notes new language selection behavior in the user flow. |
| prompts/intake.md | Adds preferred_language rules for bilingual intake prompts. |
| prompts/chat_analyzer.md | Adds preferred_language rules for bilingual chat analysis output. |
| prompts/persona_analyzer.md | Adds preferred_language rules for bilingual persona analysis output. |
| prompts/persona_builder.md | Adds preferred_language rules for bilingual persona generation. |
| prompts/merger.md | Adds preferred_language rules for bilingual merge reporting/output consistency. |
| prompts/correction_handler.md | Adds preferred_language rules for bilingual correction handling and writing. |
| .gitignore | Ignores Python/tool artifacts and generated skills while keeping a sample skill tracked. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| lines += [ | ||
| "---", | ||
| "", | ||
| f"## 日常闲聊（风格参考，共 {len(classified['daily_messages'])} 条，全部输出）", | ||
| f"## {(f'Daily Chat (style reference, {len(classified['daily_messages'])} messages, full output)' if is_en else f'日常闲聊（风格参考，共 {len(classified['daily_messages'])} 条，全部输出）')}", | ||
| "", |
There was a problem hiding this comment.
This f-string has a quoting issue: the inner f-strings use single quotes while also indexing classified['daily_messages'], which will raise a SyntaxError and prevent the script from running. Consider computing the daily message count into a variable (or using double quotes for dict keys / avoiding nested f-strings) and then building the header string without nested quoting conflicts.
| @@ -670,9 +687,12 @@ def main(): | |||
| parser.add_argument("--list-contacts", action="store_true", help="列出所有联系人") | |||
| parser.add_argument("--no-context", action="store_true", help="不包含完整对话片段") | |||
| parser.add_argument("--json", action="store_true", help="以 JSON 格式输出原始消息") | |||
| parser.add_argument("--lang", choices=["zh", "en"], default="zh", help="CLI/output language") | |||
There was a problem hiding this comment.
The CLI --lang flag only affects runtime prints after parse_args(), but argparse help/epilog and most help= strings remain Chinese (and the description is now always English). This means --lang en won't produce an English --help experience, and the default zh experience still shows an English description. If the goal is truly bilingual CLI UX, consider a small pre-parse of --lang (e.g., parse_known_args) to set the language before constructing the full parser, and localize description, epilog, and argument help strings accordingly (or keep them bilingual).
| elif args.action == "update": | ||
| if not args.slug: | ||
| print("错误：update 操作需要 --slug", file=sys.stderr) | ||
| err = "Error: update requires --slug" if lang == "en" else "错误：update 操作需要 --slug" | ||
| print(err, file=sys.stderr) | ||
| sys.exit(1) | ||
|
|
||
| skill_dir = base_dir / args.slug | ||
| if not skill_dir.exists(): | ||
| print(f"错误：找不到 Skill 目录 {skill_dir}", file=sys.stderr) | ||
| err = f"Error: skill directory not found: {skill_dir}" if lang == "en" else f"错误：找不到 Skill 目录 {skill_dir}" | ||
| print(err, file=sys.stderr) | ||
| sys.exit(1) | ||
|
|
||
| persona_patch = Path(args.persona_patch).read_text(encoding="utf-8") if args.persona_patch else None | ||
| new_version = update_ex_skill(skill_dir, persona_patch) | ||
| print(f"✅ Skill 已更新到 {new_version}：{skill_dir}") | ||
| new_version = update_ex_skill(skill_dir, persona_patch, language=None) | ||
| if lang == "en": |
There was a problem hiding this comment.
update can be invoked without --persona-patch (it’s optional), but update_ex_skill() will still archive the current version, bump meta.version, and rewrite SKILL.md/meta.json even when there is no patch/correction to apply. This creates “empty” versions and changes timestamps without any actual content update. Consider requiring --persona-patch for --action update (or making update_ex_skill early-return/no-op when both persona_patch and correction are missing).
Add English workflow files (SKILL_EN.md and prompts_en/*) and update README to point maintainers to the English pack. Update SKILL.md to route prompt usage based on preferred_language (zh/en). Internationalize and clean up tooling: update skill_writer.py, version_manager.py, wechat_decryptor.py, and wechat_parser.py with English help text, improved docstrings, language flags, and bilingual error/output messages so CLI and generated content can follow preferred_language. Also add safer argument/help strings and minor robustness fixes when reading files.
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 19 out of 20 changed files in this pull request and generated 6 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| lines += [ | ||
| "---", | ||
| "", | ||
| f"## 日常闲聊（风格参考，共 {len(classified['daily_messages'])} 条，全部输出）", | ||
| f"## {(f'Daily Chat (style reference, {len(classified['daily_messages'])} messages, full output)' if is_en else f'日常闲聊（风格参考，共 {len(classified['daily_messages'])} 条，全部输出）')}", | ||
| "", |
There was a problem hiding this comment.
This f-string will not parse because it nests single-quoted f-strings that also contain single-quoted dict keys (classified['daily_messages']). Please refactor to avoid nested f-strings (e.g., compute the count in a variable and use one f-string, or use different quote styles).
| except ImportError: | ||
| print("请先安装依赖：pip install psutil", file=sys.stderr) | ||
| print("Please install dependency first: pip install psutil", file=sys.stderr) | ||
| sys.exit(1) |
There was a problem hiding this comment.
Dependency error messages here ignore the selected --lang and always print English, even though this script defines tr()/CLI_LANG for bilingual output. Please route these user-facing errors through tr(...) so zh users still get Chinese output (and keep the experience consistent with the rest of the tool).
| try: | ||
| import pymem | ||
| import pymem.process | ||
| except ImportError: | ||
| print("请先安装依赖：pip install pymem", file=sys.stderr) | ||
| print("Please install dependency first: pip install pymem", file=sys.stderr) | ||
| sys.exit(1) |
There was a problem hiding this comment.
Same localization issue as above: this ImportError path always prints English and does not use tr(...), so --lang zh users get mixed-language output. Please wrap this message with tr(...) (or otherwise localize consistently).
| try: | ||
| from Crypto.Hash import HMAC, SHA1 | ||
| from Crypto.Protocol.KDF import PBKDF2 | ||
| from Crypto.Cipher import AES | ||
| except ImportError: | ||
| print("请先安装依赖：pip install pycryptodome", file=sys.stderr) | ||
| print("Please install dependency first: pip install pycryptodome", file=sys.stderr) | ||
| sys.exit(1) |
There was a problem hiding this comment.
Same localization issue as above: this ImportError path always prints English and bypasses tr(...), which breaks the tool’s bilingual output contract. Please use tr(...) for the dependency-install guidance as well.
| pref_lang = get_preferred_language(meta, lang) | ||
| skill_dir = create_ex_skill(base_dir, slug, meta, persona_content, pref_lang) | ||
| if lang == "en": |
There was a problem hiding this comment.
In create, pref_lang = get_preferred_language(meta, lang) will always pick the CLI language because --lang defaults to a non-empty value ("zh"). That means a meta.json containing preferred_language=en/language=en will still generate a Chinese SKILL.md unless the caller explicitly passes --lang en, which undermines language persistence. Consider making --lang default to an "auto"/None mode (use meta when present), or only overriding meta when the user explicitly supplies --lang.
| Method A commands: | ||
| python tools/wechat_decryptor.py --find-key-only | ||
| python tools/wechat_parser.py --db-dir ./decrypted/ --target "<wechat_name>" --output messages.txt | ||
|
|
||
| Method B command: | ||
| python tools/wechat_parser.py --imessage --target "<phone_or_apple_id>" --output messages.txt | ||
|
|
There was a problem hiding this comment.
In this English workflow guide, the example tool commands omit --lang en even though the tools default to --lang zh. As written, following these steps will still produce Chinese CLI/output text. Consider adding --lang en to the wechat_decryptor/wechat_parser/skill_writer/version_manager command examples so the workflow stays consistently English.
Expose language options and English docs across tools: update SKILL_EN.md examples to include --lang en. Extend skill_writer.py to accept an --lang (auto/zh/en) option, handle lang override, and produce English output when requested; translate inline comments and docstrings to English and adjust creation/update flows. Update wechat_decryptor.py and wechat_parser.py with English docstrings/comments, better tr() usage for bilingual messages, improved platform/key extraction messaging, and iMessage/WeChat parsing wording. Overall this adds/enhances localization support and clarifies CLI help and runtime messages.
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 19 out of 20 changed files in this pull request and generated 6 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| CLI_LANG = "zh" | ||
|
|
||
|
|
||
| def normalize_language(language: str | None) -> str: | ||
| value = (language or "").strip().lower() | ||
| if value in {"en", "english"}: | ||
| return "en" | ||
| return "zh" |
There was a problem hiding this comment.
This file uses PEP 604 union types (e.g., str | None / int | None), which is a Python 3.10+ syntax and will raise a SyntaxError on Python 3.9. The repository README advertises Python 3.9+, so either rewrite these annotations to Optional[...]/Union[...] (or remove them) or bump the documented minimum Python version accordingly.
| CLI_LANG = "zh" | ||
|
|
||
|
|
||
| def normalize_language(language: str | None) -> str: | ||
| value = (language or "").strip().lower() | ||
| if value in {"en", "english"}: | ||
| return "en" | ||
| return "zh" |
There was a problem hiding this comment.
This file uses PEP 604 union types (e.g., str | None / int | None), which is Python 3.10+ syntax and will fail to run on Python 3.9. The repo README indicates Python 3.9+, so either change these annotations to Optional[...]/Union[...] or update the documented minimum Python version.
| MAX_VERSIONS = 10 | ||
|
|
||
|
|
||
| def normalize_language(language: str | None) -> str: | ||
| value = (language or "").strip().lower() | ||
| if value in {"en", "english"}: | ||
| return "en" | ||
| return "zh" |
There was a problem hiding this comment.
normalize_language uses str | None (PEP 604), which requires Python 3.10+. If this repo targets Python 3.9+ (as indicated in README), this script will fail with a SyntaxError on 3.9. Consider switching to Optional[str]/Union[str, None] (or raising the minimum Python version).
| elif args.action == "update": | ||
| if not args.slug: | ||
| print("错误：update 操作需要 --slug", file=sys.stderr) | ||
| err = "Error: update requires --slug" if lang == "en" else "错误：update 操作需要 --slug" | ||
| print(err, file=sys.stderr) | ||
| sys.exit(1) | ||
|
|
||
| skill_dir = base_dir / args.slug | ||
| if not skill_dir.exists(): | ||
| print(f"错误：找不到 Skill 目录 {skill_dir}", file=sys.stderr) | ||
| err = f"Error: skill directory not found: {skill_dir}" if lang == "en" else f"错误：找不到 Skill 目录 {skill_dir}" | ||
| print(err, file=sys.stderr) | ||
| sys.exit(1) | ||
|
|
||
| persona_patch = Path(args.persona_patch).read_text(encoding="utf-8") if args.persona_patch else None | ||
| new_version = update_ex_skill(skill_dir, persona_patch) | ||
| print(f"✅ Skill 已更新到 {new_version}：{skill_dir}") | ||
| new_version = update_ex_skill(skill_dir, persona_patch, language=None) | ||
| if lang == "en": | ||
| print(f"✅ Skill updated to {new_version}: {skill_dir}") | ||
| else: | ||
| print(f"✅ Skill 已更新到 {new_version}：{skill_dir}") |
There was a problem hiding this comment.
The --lang flag is documented as affecting “CLI and generation language”, but in the update path the override is never passed through (language=None). This makes --lang en ineffective for update, and also prevents intentionally switching a skill’s language during an update. Consider passing language=lang_override (or updating the flag/help text if override is intentionally unsupported).
| args = parser.parse_args() | ||
| base_dir = Path(args.base_dir).expanduser() | ||
| lang_override = None if args.lang == "auto" else normalize_language(args.lang) | ||
| lang = lang_override or "zh" | ||
|
|
There was a problem hiding this comment.
With --lang auto, lang is forced to zh, so CLI output language does not auto-detect from existing skill metadata (e.g., updating an English skill will still print Chinese status lines). Consider inferring lang from meta.json when action is update (and possibly list) and args.lang=auto.
| def main(): | ||
| if not IS_WINDOWS and not IS_MACOS: | ||
| print("错误：此工具仅支持 Windows 和 macOS", file=sys.stderr) | ||
| print("Error: this tool supports only Windows and macOS", file=sys.stderr) | ||
| sys.exit(1) |
There was a problem hiding this comment.
This error path always prints English, but the script otherwise supports --lang-controlled bilingual output. Consider printing a bilingual message here (or parsing --lang before the platform check) so users on unsupported platforms still get consistent language behavior.
|
@copilot apply changes based on the comments in this thread |
Make language normalization and various function signatures use typing.Optional for clearer type hints. Enhance skill_writer to determine update language from meta.json (when lang_override is unset), require --persona-patch for updates, pass the lang_override to update_ex_skill, and show localized success/error messages. Add a pre-parser in wechat_decryptor to set CLI_LANG early and use localized error text for unsupported platforms. Small cleanup: import Optional where needed and adjust other signatures to return Optional types.
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 19 out of 20 changed files in this pull request and generated 4 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| 收到 `/create-ex` 后，按以下流程运行： | ||
|
|
||
| ``` | ||
| Step 1 → 基础信息录入 （参考 prompts/intake.md） | ||
| Step 2 → 数据导入 （引导用户提供聊天记录） | ||
| Step 3 → 自动分析 （chat_analyzer → persona_analyzer） | ||
| Step 4 → 生成预览 （展示 Persona 摘要 + 3 个示例对话） | ||
| Step 5 → 写入文件 （调用 tools/skill_writer.py） | ||
| ``` |
There was a problem hiding this comment.
The top-level flow block still starts at Step 1, but Step 0 (language selection) is now mandatory. Update the flow snippet to include Step 0 (and keep the step numbering/order consistent) so readers don’t miss the required first step.
| 语言选择后，保存状态变量 `preferred_language`（`zh` 或 `en`），后续所有用户可见内容都必须严格使用该语言。 | ||
|
|
There was a problem hiding this comment.
The doc says all user-visible output must strictly follow preferred_language, but the tool commands later in the workflow default to Chinese unless --lang is passed. Consider explicitly stating that when invoking tools/wechat_decryptor.py, tools/wechat_parser.py, tools/skill_writer.py, and tools/version_manager.py, the orchestrator should forward --lang {preferred_language} (or equivalent) to keep CLI/tool output consistent.
| f"# {source} {('Chat Extraction Result' if is_en else '聊天记录提取结果')}", | ||
| f"{('Target' if is_en else '目标人物')}：{target_name}", | ||
| f"{('Messages sent by TA' if is_en else 'TA 发送的消息数')}：{classified['total_their_count']}", | ||
| f"{('Total messages' if is_en else '对话总消息数')}：{classified['total_count']}", |
There was a problem hiding this comment.
In English mode, the header lines still use the full-width Chinese colon ： (e.g., Target：...). For cleaner English output (and easier downstream parsing), use the ASCII : when language is en.
| A) WeChat automatic extraction | ||
| - Keep WeChat desktop logged in | ||
| - Run tools/wechat_decryptor.py --find-key-only | ||
| - Run tools/wechat_parser.py --db-dir ./decrypted/ --target "<wechat_name>" --output messages.txt | ||
|
|
||
| B) iMessage automatic extraction (macOS) | ||
| - Run tools/wechat_parser.py --imessage --db ~/Library/Messages/chat.db --target "<phone_or_apple_id>" --output messages.txt |
There was a problem hiding this comment.
This English intake script suggests running the tooling commands without --lang en. Since the tools default to Chinese output, the command examples here should include --lang en (and ideally mention forwarding the session’s preferred_language) to keep the workflow language-consistent.
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 19 out of 20 changed files in this pull request and generated 6 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| parser = argparse.ArgumentParser( | ||
| description="微信 PC/Mac 端数据库解密工具", | ||
| description="WeChat desktop database decryptor", | ||
| formatter_class=argparse.RawDescriptionHelpFormatter, | ||
| epilog=""" | ||
| 示例： | ||
| # 自动从内存提取密钥并解密所有数据库 | ||
| python wechat_decryptor.py --db-dir <MSG目录> --output ./decrypted/ | ||
| Examples: | ||
| # Extract key from memory and decrypt all databases | ||
| python wechat_decryptor.py --db-dir <MSG_DIR> --output ./decrypted/ | ||
|
|
||
| # 只打印密钥 | ||
| # Print key only | ||
| python wechat_decryptor.py --find-key-only | ||
|
|
||
| # 用已知密钥解密单个文件 | ||
| # Decrypt one DB with a known key | ||
| python wechat_decryptor.py --key "abcdef1234..." --db "./MSG0.db" --output "./out/" | ||
|
|
||
| # 验证密钥是否正确 | ||
| # Validate key against a DB | ||
| python wechat_decryptor.py --key "abcdef1234..." --test-db "./MSG0.db" | ||
| """ |
There was a problem hiding this comment.
--lang is parsed and CLI_LANG/tr() exist, but the argparse description/epilog are always English. This means --lang zh (or default zh) still shows English help/usage text, which undermines the bilingual CLI goal. Consider selecting description/epilog via tr() (similar to tools/wechat_parser.py) after a pre-parse of --lang.
| parser.add_argument("--db-dir", help="Directory containing WeChat message databases") | ||
| parser.add_argument("--db", help="Path to a single database file") | ||
| parser.add_argument("--output", default="./decrypted", help="Output directory for decrypted files (default: ./decrypted)") | ||
| parser.add_argument("--key", help="Known key in hex format (skip memory extraction)") | ||
| parser.add_argument("--find-key-only", action="store_true", help="Print extracted key only; do not decrypt") | ||
| parser.add_argument("--test-db", help="Validate key against one DB file (use with --key)") | ||
| parser.add_argument("--lang", choices=["zh", "en"], default="zh", help="CLI language") |
There was a problem hiding this comment.
Argparse option help strings here are hard-coded in English, so --lang zh doesn’t localize the CLI help output. If bilingual help is intended, wrap these help= values with tr(...) (or build the parser after setting CLI_LANG via a pre-parser).
| """ | ||
| 微信 PC 端数据库解密工具 | ||
| WeChat desktop database decryptor (Windows + macOS). | ||
|
|
||
| 支持： | ||
| - Windows：微信 3.x（SQLCipher 加密，从 WeChatWin.dll 内存提取密钥） | ||
| - macOS：微信 Mac 版（SQLCipher 加密，从 WeChat 进程内存提取密钥） | ||
|
|
||
| 解密原理： | ||
| 微信 PC/Mac 端将聊天数据库用 SQLCipher 加密存储。 | ||
| 加密密钥在微信运行时驻留在进程内存中，可通过特征码扫描提取。 | ||
| 提取后用 SQLCipher 的 PRAGMA key 解密数据库。 | ||
|
|
||
| 用法： | ||
| python wechat_decryptor.py --find-key-only | ||
| python wechat_decryptor.py --db-dir <MSG目录> --output ./decrypted/ | ||
| python wechat_decryptor.py --key "abcd1234" --db "./MSG0.db" --output "./decrypted/" | ||
|
|
||
| 依赖： | ||
| pip install pycryptodome psutil | ||
| Windows 额外：pip install pymem | ||
| macOS 额外：无（使用 lldb） | ||
|
|
||
| 注意： | ||
| - 运行时微信客户端必须处于登录状态（需从内存读取密钥） | ||
| - macOS 可能需要关闭 SIP 或授予终端 Full Disk Access 权限 | ||
| - 解密后的数据库仅用于个人读取，不要分发 | ||
| Extracts SQLCipher keys from a running WeChat process and decrypts local | ||
| message databases for personal export workflows. | ||
| """ |
There was a problem hiding this comment.
The top-level docstring was reduced to a brief English summary and no longer documents key prerequisites/safety notes (e.g., required dependencies like psutil/pycryptodome/pymem, WeChat login requirement, macOS Full Disk Access/SIP caveats, and personal-use warning). Restoring those details (in zh/en) would make the tool safer and easier to run without reading the source.
| def main(): | ||
| parser = argparse.ArgumentParser(description="前任 Skill 版本管理器") | ||
| parser = argparse.ArgumentParser(description="Ex Skill version manager") | ||
| parser.add_argument("--action", required=True, choices=["list", "rollback", "cleanup"]) | ||
| parser.add_argument("--slug", required=True, help="前任 slug") | ||
| parser.add_argument("--version", help="目标版本号（rollback 时使用）") | ||
| parser.add_argument("--base-dir", default="./exes", help="前任 Skill 根目录") | ||
| parser.add_argument("--slug", required=True, help="Ex skill slug") | ||
| parser.add_argument("--version", help="Target version for rollback (e.g. v2)") | ||
| parser.add_argument("--base-dir", default="./exes", help="Root ex skill directory") | ||
| parser.add_argument("--lang", choices=["zh", "en"], default="zh", help="CLI language") |
There was a problem hiding this comment.
--lang is parsed only after the argparse parser is constructed, so --help (and argparse-generated errors) will always be English. If the goal is bilingual CLI output, consider a small pre-parse for --lang (parse_known_args) and then build the main parser with localized description/help strings based on that value (as done in tools/wechat_parser.py).
| def main() -> None: | ||
| parser = argparse.ArgumentParser(description="前任 Skill 文件写入器") | ||
| parser = argparse.ArgumentParser(description="Ex Skill file writer") | ||
| parser.add_argument("--action", required=True, choices=["create", "update", "list"]) | ||
| parser.add_argument("--slug", help="前任 slug（用于目录名）") | ||
| parser.add_argument("--name", help="前任称呼") | ||
| parser.add_argument("--meta", help="meta.json 文件路径") | ||
| parser.add_argument("--persona", help="persona.md 内容文件路径") | ||
| parser.add_argument("--persona-patch", help="persona.md 增量更新内容文件路径") | ||
| parser.add_argument("--slug", help="Ex skill slug (folder name)") | ||
| parser.add_argument("--name", help="Display name for the ex skill") | ||
| parser.add_argument("--meta", help="Path to meta.json") | ||
| parser.add_argument("--persona", help="Path to persona.md content file") | ||
| parser.add_argument("--persona-patch", help="Path to incremental persona patch file") | ||
| parser.add_argument( | ||
| "--base-dir", | ||
| default="./exes", | ||
| help="前任 Skill 根目录（默认：./exes）", | ||
| help="Ex Skill root directory (default: ./exes)", | ||
| ) | ||
| parser.add_argument( | ||
| "--lang", | ||
| choices=["auto", "zh", "en"], | ||
| default="auto", | ||
| help="CLI and generation language (auto, zh, or en)", | ||
| ) |
There was a problem hiding this comment.
--lang affects runtime output, but argparse help/usage text is always English because the parser is built before language selection. If bilingual CLI output is a requirement, add a pre-parse for --lang and localize description and help= strings accordingly (similar to tools/wechat_parser.py).
| ## Step 2：数据导入 | ||
|
|
||
| 引导用户选择导入方式： | ||
| 引导用户选择导入方式（按 `preferred_language` 输出）： | ||
|
|
||
| ``` |
There was a problem hiding this comment.
This step now requires all guidance to follow preferred_language, but the example commands below don’t pass --lang {preferred_language} (English sessions will otherwise default to zh output). Also, the WeChat flow shows wechat_decryptor.py --find-key-only and then immediately parses ./decrypted/, but --find-key-only exits without decrypting; the instructions should include an actual decrypt invocation (e.g., --db-dir ... --output ... or using --key to decrypt) before running wechat_parser.py --db-dir ....
Introduce multilingual CLI support and update docs to use {preferred_language}. SKILL.md and SKILL_EN.md: add --lang/{preferred_language} to wechat_decryptor/parser and skill/version commands, and document a manual-key fallback flow for decryption. tools/wechat_decryptor.py: add bilingual header (EN/ZH) with prerequisites/safety notes, localized epilog examples, and localized argument help strings; add --lang argument handling and switch epilog/help text based on language.
|
Hii @titanwings can you review this? |
Summary
This PR adds full bilingual support (Chinese + English) to the ex-skill workflow, including language selection at the start, language-consistent prompt execution, and English-capable CLI tooling.
What Changed
Validation
Impact