From zenbu-powers
Multi-pattern text search, filtering, and replacement using the Aho-Corasick algorithm. Use this skill whenever the user wants to: search for multiple keywords/patterns simultaneously in text or files, filter or detect sensitive words, scan multiple files for a list of terms, or redact/replace matched content. Triggers on phrases like "掃描關鍵字", "敏感詞過濾", "多個 pattern 搜尋", "批次掃描", "redact", "replace patterns", "keyword filter", "find multiple words", "scan files for patterns", or any task involving simultaneous multi-pattern matching. Always prefer this skill over naive grep loops when more than one pattern is involved.
npx claudepluginhub zenbuapps/zenbu-powers --plugin zenbu-powersThis skill uses the workspace's default tool permissions.
高效能多 pattern 同時比對，適合關鍵字過濾、敏感詞偵測、文件掃描與 redact。
Builds, explains, and debugs regular expressions for pattern matching, validating inputs like emails/URLs/phones, and extracting data from logs/documents/text.
Searches codebase for patterns with grep and reads up to 5 matching files in parallel using Read tool, 50-70% faster for exploration, bug investigation, and implementation research.
Enhances codebase searches via semantic query parsing, regex patterns, AST analysis, symbol lookups, and optimized grep/ripgrep with relevance ranking and context.
Share bugs, ideas, or general feedback.
高效能多 pattern 同時比對，適合關鍵字過濾、敏感詞偵測、文件掃描與 redact。
pip install ahocorasick-rs --break-system-packages
環境若已安裝可跳過。
| 功能 | 說明 | 對應 script |
|---|---|---|
| search | 在文字/檔案中找出所有 pattern 的出現位置 | scripts/ac_search.py |
| filter | 判斷文字是否包含任何 pattern（回傳 True/False + 命中清單） | scripts/ac_filter.py |
| scan | 批次掃描多個檔案，回傳每個檔案的命中結果 | scripts/ac_scan.py |
| redact | 將命中的 pattern 替換成指定字串（預設 ***） | scripts/ac_redact.py |
# 從文字輸入
python scripts/ac_search.py --patterns "apple,banana,cherry" --text "I like apple pie and banana bread"
# 從檔案讀取 patterns（每行一個）
python scripts/ac_search.py --patterns-file patterns.txt --text "some text here"
# 搜尋檔案內容
python scripts/ac_search.py --patterns "error,warning,critical" --file /var/log/app.log
輸出範例：
{
"matches": [
{"pattern": "apple", "start": 7, "end": 12},
{"pattern": "banana", "start": 21, "end": 27}
],
"total": 2
}
python scripts/ac_filter.py --patterns-file blocklist.txt --text "This contains badword here"
輸出範例：
{
"flagged": true,
"matched_patterns": ["badword"],
"count": 1
}
# 掃描目錄下所有 .txt 檔
python scripts/ac_scan.py --patterns-file keywords.txt --dir ./docs --ext .txt
# 掃描指定多個檔案
python scripts/ac_scan.py --patterns "TODO,FIXME,HACK" --files file1.py file2.py file3.py
輸出範例：
{
"summary": {"files_scanned": 3, "files_with_matches": 2, "total_matches": 7},
"results": [
{"file": "file1.py", "matches": [{"pattern": "TODO", "line": 12, "context": "# TODO: fix this"}]},
{"file": "file2.py", "matches": []},
{"file": "file3.py", "matches": [{"pattern": "FIXME", "line": 5, "context": "# FIXME broken"}]}
]
}
# 預設替換成 ***
python scripts/ac_redact.py --patterns "John,secret,password" --text "John knows the secret password"
# 自訂替換字串
python scripts/ac_redact.py --patterns-file pii.txt --text "Call John at 0912-345-678" --replacement "[REDACTED]"
# 替換檔案內容並輸出
python scripts/ac_redact.py --patterns-file pii.txt --file input.txt --output redacted.txt
輸出範例：
{
"redacted_text": "*** knows the *** ***",
"replacements_made": 3
}
Claude 也可以直接在 code execution 中使用，不需要走 CLI：
import subprocess, json
result = subprocess.run(
["python", "scripts/ac_search.py", "--patterns", "error,warning", "--text", log_text],
capture_output=True, text=True
)
data = json.loads(result.stdout)
或直接 import：
import ahocorasick_rs
ac = ahocorasick_rs.AhoCorasick(["apple", "banana", "cherry"])
matches = ac.find_matches_as_indexes("I like apple and banana")
# [(0, 7, 12), (1, 17, 23)] → (pattern_index, start, end)
ahocorasick_rs 不支援空字串 pattern，會拋 ValueError，script 已做防護。--overlapping flag。