snapgrep 使用指南
利用进程内三元索引加速代码搜索,性能超越 ripgrep。
本文为派生内容(基于 wiki / faq / compatibility_json 自动组装),非 AI 即时创作。
本文由本站基于该插件已收录的字段自动派生(wiki / faq / compatibility_json / readme),非 AI 即时创作。每节末尾标源字段。
快速上手
利用进程内三元索引加速代码搜索,性能超越 ripgrep。
— 源: plugins.ai_summary
安装与验证
dsh plugin --profile web add snapgrep
复制以上命令在 DSH Web Profile 内运行。安装完成后在「插件列表」启用即可。
— 源: plugins.install
关键要点
- The index is mapped into the agent's own memory. No socket, no JSON, no serialisation. A query is a function call.
- Verification reads the mmap'd index, not the disk. Candidates are confirmed in-process against bytes that are already resident, so there is no spawn and no second read.
- The index format is tuned for this shape of data — compressed content blocks, delta-varint gram metadata — which is how the index lands at 0.38–0.91× the size of the source it indexes.
- Binary files where ripgrep's NUL-byte output semantics apply
- Regex combined with path or glob filters
— 源: plugin_wiki.readme_zh (fallback readme_raw)
踩坑提醒
安装前请审阅上游仓库;本指南基于已收录字段自动派生,可能滞后于最新版本。如发现与官方文档冲突,请以上游为准。
— 来源:通用规则