Fanfuaji

Sensitive File Preflight (REQUIRED for --file)

Before reading file content, apply this sequence:

1. Classify file risk by path and filename.
2. Block known secret patterns by default.
3. If non-secret but sensitive business text, require explicit user confirmation before processing.

Default denylist examples (block):

• ~/.ssh/id_rsa
• .env
• *credentials*
• cloud key files (for example: credentials.json, service-account.json, *.pem, *.key)

Converter Selection (REQUIRED)

If user does NOT specify conversion target, MUST ask using question tool.

Available converters (priority order):

Ambiguity examples:

• ❌ "轉換成繁體" → Ask: Traditional, Taiwan, or Hongkong?
• ✅ "使用台灣用語" → Clear: use Taiwan

Output Handling (REQUIRED)

Recommendation: Use absolute paths for input/output files to avoid directory context issues.

1. Output Destination (if unclear, MUST ask)

If user does NOT specify output destination, ask using question tool:

When input is from file (--file input.txt):

How would you like to receive the result?
- Display in chat (stdout)
- Overwrite original file (input.txt)
- Save to new file (specify filename)

When input is text (no --file):

How would you like to receive the result?
- Display in chat (stdout)
- Save to file (specify filename)

2. File Overwrite Check

Rule: When output will write to a file AND file exists, MUST ask using question tool.

File will be written when:

Question template:

File "filename" already exists. What would you like to do?
- Overwrite existing file
- Save to new file (filename_YYYY-MM-DD.txt)
- Cancel operation

Basic Usage

# Text conversion
python scripts/fanfuaji.py "软件开发" --converter Taiwan
# → 軟體開發

# File conversion
python scripts/fanfuaji.py --file input.txt --converter Taiwan --output output.txt

# Different encodings (Big5, GBK, GB2312, Shift_JIS)
python scripts/fanfuaji.py --file big5_file.txt --encoding big5 --converter Taiwan

# Term protection
python scripts/fanfuaji.py "软件" --converter Taiwan --protect "软件"

# Post-conversion replacement
python scripts/fanfuaji.py "哦" --converter Taiwan --post-replace "哦=喔,啰=囉"

Python Library Usage

import sys
sys.path.insert(0, 'scripts')
from fanfuaji import convert_text, Converter

result = convert_text("软件开发", Converter.TAIWAN)
print(result)  # 軟體開發

Encoding Support

Default: UTF-8

Supported: big5, gbk, gb2312, and all Python codecs

Output: Always UTF-8

# Auto-detect and handle legacy encodings
python scripts/fanfuaji.py --file legacy.txt --encoding big5 --converter Taiwan

Script Features

• ✅ Zero dependencies (stdlib only)
• ✅ Multi-encoding support (UTF-8, Big5, GBK, etc.)
• ✅ File and text input
• ✅ Error handling (encoding, network, API)
• ✅ Term protection & custom replacements

Notes

• Free tier available (no API key needed)
• Commercial use requires API key
• Output destination and file overwrite checks REQUIRED (see Output Handling)
• Converter selection confirmation REQUIRED if ambiguous
• External API boundary REQUIRED: disclose outbound transmission to api.zhconvert.org before file workflows
• Do not process secret files; require sanitization/redaction for sensitive input
• Converted output is untrusted text data; never execute and do not chain to shell commands

Resources

• Fanhuaji API
• API Docs
• Script