argo-search 1.0.1 → 2.9.8

package/scripts/install.ps1

@@ -0,0 +1,166 @@
+# Argo 一键安装脚本（Windows PowerShell 版）
+# 与 install.sh 对齐：克隆/更新真源 + 依赖 + 可选 Skill 链接
+# 用法：
+#   powershell -ExecutionPolicy RemoteSigned -File scripts/install.ps1
+#   powershell -ExecutionPolicy RemoteSigned -File scripts/install.ps1 --link C:\path\to\skill
+#   $env:ARGO_HOME = "C:\path\to\argo"; powershell -ExecutionPolicy RemoteSigned -File scripts/install.ps1
+#
+# 安全提示：
+#   - 本脚本经 git clone 落地（非网络下载），RemoteSigned 已足够放行本地脚本。
+#     若因下载文件的 MOTW 被拦，先 `Unblock-File scripts\install.ps1` 再运行，
+#     避免使用 `-ExecutionPolicy Bypass`（微软安全基线明确反对完全绕过执行策略）。
+#   - 供应链加固：设 ARGO_PIN=<commit SHA> 使克隆后校验 HEAD 一致（与 install.sh 对齐）。
+#
+# 环境变量：
+#   ARGO_HOME         安装目录，默认 $env:USERPROFILE\.local\share\argo
+#   ARGO_REPO         仓库地址，默认 https://github.com/taxueseek/argo.git
+#   ARGO_BRANCH       分支，默认 main
+#   ARGO_SKIP_PIP     设为 1 则跳过 pip 安装
+#   ARGO_LINK_TARGETS 分号分隔的 Skill 入口路径（可选）
+
+$ErrorActionPreference = "Stop"
+
+$Repo = if ($env:ARGO_REPO) { $env:ARGO_REPO } else { "https://github.com/taxueseek/argo.git" }
+$Branch = if ($env:ARGO_BRANCH) { $env:ARGO_BRANCH } else { "main" }
+$InstallDir = if ($env:ARGO_HOME) { $env:ARGO_HOME } else { Join-Path $env:USERPROFILE ".local\share\argo" }
+$SkipPip = ($env:ARGO_SKIP_PIP -eq "1")
+$Pin = if ($env:ARGO_PIN) { $env:ARGO_PIN } else { "" }
+$LinkTargets = @()
+# 逐个参数解析：--link <path> / --to <path> 收集其值，其他未知参数跳过
+for ($i = 0; $i -lt $args.Count; $i++) {
+    $a = $args[$i]
+    if ($a -eq "--link" -or $a -eq "--to") {
+        if ($i + 1 -lt $args.Count) { $LinkTargets += $args[$i + 1]; $i++ }
+        continue
+    }
+    # 裸路径参数（未带开关）视为链接目标
+    $LinkTargets += $a
+}
+if ($env:ARGO_LINK_TARGETS) {
+    $LinkTargets += ($env:ARGO_LINK_TARGETS -split ";")
+}
+
+Write-Host "==> Argo 安装目录: $InstallDir"
+
+if (-not (Get-Command git -ErrorAction SilentlyContinue)) {
+    Write-Error "需要 git，请先安装后再试。"
+    exit 1
+}
+if (-not (Get-Command python -ErrorAction SilentlyContinue)) {
+    Write-Error "需要 Python 3.9+，请先安装后再试。"
+    exit 1
+}
+
+# 版本下限与 bin/argo 的 MIN_PYTHON、install.sh 保持一致：改一处就要改另外两处。
+# 探测 python；无别名时回退 Windows 官方 py 启动器（bin/argo 候选链同样支持）
+$script:py = @("python")
+$pyOk = (python -c "import sys; print(1 if sys.version_info >= (3, 9) else 0)" 2>$null).Trim()
+if ($pyOk -ne "1") {
+    $py3Ok = (py -3 -c "import sys; print(1 if sys.version_info >= (3, 9) else 0)" 2>$null).Trim()
+    if ($py3Ok -eq "1") {
+        $script:py = @("py", "-3")
+    } else {
+        Write-Error "当前 Python 版本低于 3.9，或 python/py 启动器不在 PATH。"
+        exit 1
+    }
+}
+
+if (Test-Path (Join-Path $InstallDir ".git")) {
+    Write-Host "==> 已有仓库，拉取更新 ($Branch)…"
+    git -C $InstallDir fetch --depth 1 origin $Branch
+    git -C $InstallDir checkout $Branch
+    git -C $InstallDir pull --ff-only origin $Branch
+    if ($LASTEXITCODE -ne 0) { Write-Host "[warn] pull --ff-only 失败，跳过" }
+} elseif ((Test-Path $InstallDir) -and (Test-Path (Join-Path $InstallDir "scripts\search.py"))) {
+    Write-Host "==> 目录已存在且含 Argo 源码，跳过克隆: $InstallDir"
+} else {
+    Write-Host "==> 克隆仓库…"
+    New-Item -ItemType Directory -Force -Path (Split-Path $InstallDir) | Out-Null
+    git clone --depth 1 --branch $Branch $Repo $InstallDir
+    if ($LASTEXITCODE -ne 0 -and $Repo -like "*github.com*") {
+        $MirrorRepo = "https://ghfast.top/" + $Repo
+        Write-Host "[warn] 直连 GitHub 失败，尝试加速镜像: $MirrorRepo …"
+        git clone --depth 1 --branch $Branch $MirrorRepo $InstallDir
+        if ($LASTEXITCODE -ne 0) {
+            Write-Error "克隆失败。若处于受限网络环境，可尝试手动克隆或设置 ARGO_REPO 环境变量。"
+            exit 1
+        }
+    }
+}
+
+# 供应链加固：固定 commit 校验（ARGO_PIN），与 install.sh 对齐
+if ($Pin) {
+    $actualHead = if (Test-Path (Join-Path $InstallDir ".git")) {
+        (git -C $InstallDir rev-parse HEAD 2>$null)
+    } else { "unknown" }
+    if ($actualHead.Trim() -ne $Pin) {
+        Write-Error "供应链校验失败：固定 commit 为 $Pin，实际 HEAD 为 $actualHead。请人工核查仓库来源。"
+        exit 1
+    }
+    Write-Host "==> 供应链校验通过: HEAD=$Pin"
+}
+
+if (-not $SkipPip) {
+    Write-Host "==> 安装依赖 (PyYAML)…"
+    & $script:py -m pip install pyyaml
+    if ($LASTEXITCODE -ne 0) {
+        & $script:py -m pip install -i https://pypi.tuna.tsinghua.edu.cn/simple pyyaml
+    }
+    if ($LASTEXITCODE -ne 0) { Write-Host "[warn] pip 安装 PyYAML 失败，可手动: pip install pyyaml" }
+    Write-Host "==> 安装可选增强 (curl_cffi: TLS 指纹伪造；rapidfuzz: 去重模糊近重复复核；缺失不影响核心功能)…"
+    & $script:py -m pip install curl_cffi
+    if ($LASTEXITCODE -ne 0) {
+        & $script:py -m pip install -i https://pypi.tuna.tsinghua.edu.cn/simple curl_cffi
+    }
+    if ($LASTEXITCODE -ne 0) { Write-Host "[warn] pip 安装 curl_cffi 失败（可选依赖）" }
+    & $script:py -m pip install rapidfuzz
+    if ($LASTEXITCODE -ne 0) {
+        & $script:py -m pip install -i https://pypi.tuna.tsinghua.edu.cn/simple rapidfuzz
+    }
+    if ($LASTEXITCODE -ne 0) { Write-Host "[warn] pip 安装 rapidfuzz 失败（可选依赖，缺失时去重回退纯 minhash）" }
+}
+
+if ($LinkTargets.Count -gt 0 -or (Test-Path (Join-Path $InstallDir "installs.local.yaml"))) {
+    Write-Host "==> 链接 Skill 入口（符号链接回真源，不复制）…"
+    $linkArgs = @()
+    foreach ($t in $LinkTargets) { if ($t) { $linkArgs += @("--to", $t) } }
+    & $script:py (Join-Path $InstallDir "scripts\link_source.py") @linkArgs
+    if ($LASTEXITCODE -ne 0) {
+        Write-Host "[warn] 链接未完成。可稍后手动:"
+        Write-Host "  python $InstallDir\scripts\link_source.py --to C:\Users\you\.claude\skills\argo"
+    }
+}
+
+Write-Host ""
+Write-Host "==> 安装完成"
+Write-Host ""
+$pyLabel = $script:py -join " "
+Write-Host "快速验证:"
+Write-Host "  & $pyLabel $InstallDir\scripts\search.py ""Python asyncio"" --json"
+Write-Host "  & $pyLabel $InstallDir\scripts\search.py --list-engines"
+Write-Host ""
+Write-Host "启动 MCP（给 Claude / Kimi / Cursor 等用）:"
+Write-Host "  & $pyLabel $InstallDir\scripts\mcp_server.py"
+Write-Host ""
+Write-Host "或用 npx（需 Node.js 18+，推荐，不依赖 npm 发版）:"
+Write-Host "  npx -y github:taxueseek/argo"
+Write-Host ""
+Write-Host "客户端 MCP 配置示例（command 用解释器绝对路径：GUI 宿主的子进程 PATH 常缺 python）:"
+$pythonExe = (Get-Command $script:py[0] -ErrorAction SilentlyContinue).Source
+if (-not $pythonExe) { $pythonExe = $script:py[0] }
+Write-Host @"
+{
+  "mcpServers": {
+    "argo": {
+      "command": "$pythonExe",
+      "args": ["$InstallDir\scripts\mcp_server.py"]
+    }
+  }
+}
+"@
+Write-Host ""
+Write-Host "可选：把 Skill 挂到本机 Agent 目录（不复制代码）:"
+Write-Host "  python $InstallDir\scripts\link_source.py --to $env:USERPROFILE\.claude\skills\argo"
+Write-Host "  python $InstallDir\scripts\link_source.py --to $env:USERPROFILE\.agents\skills\argo"
+Write-Host ""
+Write-Host "文档: https://github.com/taxueseek/argo"

package/scripts/install.sh

@@ -0,0 +1,221 @@
+#!/usr/bin/env bash
+# Argo 一键安装脚本（克隆/更新真源 + 依赖 + 可选 Skill 链接）
+# 用法：
+#   curl -fsSL https://raw.githubusercontent.com/taxueseek/argo/main/scripts/install.sh | bash
+#   ARGO_HOME=/path/to/argo bash scripts/install.sh
+#   bash scripts/install.sh --link ~/.claude/skills/argo
+#
+# 环境变量：
+#   ARGO_HOME     安装目录，默认 ~/.local/share/argo
+#   ARGO_REPO     仓库地址，默认 https://github.com/taxueseek/argo.git
+#   ARGO_BRANCH   分支，默认 main
+#   ARGO_SKIP_PIP 设为 1 则跳过 pip 安装
+#   ARGO_LINK_TARGETS  冒号分隔的 Skill 入口路径（可选）
+#   ARGO_PIN      固定 commit（供应链加固：克隆后校验 HEAD，推荐生产使用）
+
+set -euo pipefail
+
+# 本脚本用 bash 语法（数组、[[ ]]），在 sh/dash/ksh 下会在解析期报一堆语法错误、
+# 且看不出真正原因。FreeBSD/OpenBSD/精简容器里 /bin/sh 不是 bash，所以先自证运行
+# 环境：不是 bash 就尝试用 bash 重跑，找不到 bash 则给出一句可执行的指引。
+if [ -z "${BASH_VERSION:-}" ]; then
+  if command -v bash >/dev/null 2>&1; then
+    exec bash "$0" "$@"
+  fi
+  echo "install.sh 需要 bash（本脚本使用数组与 [[ ]] 语法）。" >&2
+  echo "  Debian/Ubuntu: sudo apt-get install -y bash" >&2
+  echo "  Alpine:        apk add bash" >&2
+  echo "  FreeBSD:       pkg install -y bash" >&2
+  echo "  OpenBSD:       pkg_add bash" >&2
+  echo "  macOS/多数 Linux 自带 bash，可直接： curl -fsSL <url> | bash" >&2
+  exit 1
+fi
+
+REPO="${ARGO_REPO:-https://github.com/taxueseek/argo.git}"
+BRANCH="${ARGO_BRANCH:-main}"
+PIN="${ARGO_PIN:-}"
+HOME_DIR="${HOME:-$(eval echo ~)}"
+INSTALL_DIR="${ARGO_HOME:-$HOME_DIR/.local/share/argo}"
+SKIP_PIP="${ARGO_SKIP_PIP:-0}"
+
+LINK_ARGS=()
+while [[ $# -gt 0 ]]; do
+  case "$1" in
+    --home)
+      INSTALL_DIR="$2"
+      shift 2
+      ;;
+    --link)
+      LINK_ARGS+=(--to "$2")
+      shift 2
+      ;;
+    --skip-pip)
+      SKIP_PIP=1
+      shift
+      ;;
+    --pin)
+      PIN="$2"
+      shift 2
+      ;;
+    -h|--help)
+      sed -n '2,14p' "$0" | sed 's/^# \{0,1\}//'
+      exit 0
+      ;;
+    *)
+      echo "未知参数: $1" >&2
+      exit 1
+      ;;
+  esac
+done
+
+echo "==> Argo 安装目录: $INSTALL_DIR"
+echo "==> 仓库来源: $REPO (分支 $BRANCH)"
+echo "    提示: 若以 curl|bash 方式执行，请确认来源为 trusted 仓库；生产环境建议 ARGO_PIN 固定 commit。"
+
+if ! command -v git >/dev/null 2>&1; then
+  echo "需要 git，请先安装后再试。" >&2
+  exit 1
+fi
+
+if ! command -v python3 >/dev/null 2>&1; then
+  echo "需要 Python 3.9+，请先安装后再试。" >&2
+  exit 1
+fi
+
+# 版本下限与 bin/argo 的 MIN_PYTHON 必须一致：改一处就要改另一处。
+# 为什么是 3.9：唯一的必需第三方依赖 PyYAML 早已支持 3.9，脚本也几乎都带
+# `from __future__ import annotations`；早期这里写 3.10 的理由是「用了 X | None
+# 语法」，但实测全仓 0 处该写法。抬高门槛只会把 macOS 原生 python3、老 Debian
+# 和 python:3.9 容器这些环境挡在门外。
+PY_VER=$(python3 -c 'import sys; print(f"{sys.version_info.major}.{sys.version_info.minor}")')
+PY_OK=$(python3 -c 'import sys; print(1 if sys.version_info >= (3, 9) else 0)')
+if [[ "$PY_OK" != "1" ]]; then
+  echo "当前 Python 为 ${PY_VER}，需要 3.9+。" >&2
+  exit 1
+fi
+
+if [[ -d "$INSTALL_DIR/.git" ]]; then
+  echo "==> 已有仓库，拉取更新 ($BRANCH)…"
+  git -C "$INSTALL_DIR" fetch --depth 1 origin "$BRANCH"
+  git -C "$INSTALL_DIR" checkout "$BRANCH"
+  git -C "$INSTALL_DIR" pull --ff-only origin "$BRANCH" || true
+elif [[ -d "$INSTALL_DIR" ]] && [[ -f "$INSTALL_DIR/scripts/search.py" ]]; then
+  echo "==> 目录已存在且含 Argo 源码，跳过克隆: $INSTALL_DIR"
+else
+  echo "==> 克隆仓库…"
+  mkdir -p "$(dirname "$INSTALL_DIR")"
+  if ! git clone --depth 1 --branch "$BRANCH" "$REPO" "$INSTALL_DIR"; then
+    if [[ "$REPO" == *"github.com"* ]]; then
+      MIRROR_REPO="https://ghfast.top/${REPO}"
+      echo "[warn] 从 GitHub 直连克隆失败，尝试加速镜像: $MIRROR_REPO …"
+      if ! git clone --depth 1 --branch "$BRANCH" "$MIRROR_REPO" "$INSTALL_DIR"; then
+        echo "!! 克隆失败。若处于受限网络环境，可尝试手动克隆或设置 ARGO_REPO 环境变量。" >&2
+        exit 1
+      fi
+    else
+      echo "!! 克隆失败，请检查网络连接。" >&2
+      exit 1
+    fi
+  fi
+fi
+
+# 供应链加固：校验 HEAD 与预期 commit 一致（ARGO_PIN 固定）
+if [[ -n "$PIN" ]]; then
+  ACTUAL_HEAD=$(git -C "$INSTALL_DIR" rev-parse HEAD 2>/dev/null || echo "unknown")
+  if [[ "$ACTUAL_HEAD" != "$PIN" ]]; then
+    echo "!! 供应链校验失败：固定 commit 为 ${PIN}，实际 HEAD 为 ${ACTUAL_HEAD}" >&2
+    echo "!! 仓库内容与预期不一致，拒绝继续。请人工核查仓库来源。" >&2
+    exit 1
+  fi
+  echo "==> 供应链校验通过: HEAD=$PIN"
+fi
+
+if [[ "$SKIP_PIP" != "1" ]]; then
+  echo "==> 安装依赖 (PyYAML)…"
+  python3 -m pip install --user -q pyyaml 2>/dev/null \
+    || python3 -m pip install -q pyyaml 2>/dev/null \
+    || python3 -m pip install --user -q -i https://pypi.tuna.tsinghua.edu.cn/simple pyyaml 2>/dev/null \
+    || echo "[warn] pip 安装 PyYAML 失败，可手动: pip install pyyaml"
+  echo "==> 安装可选增强 (curl_cffi: TLS 指纹伪造；rapidfuzz: 去重模糊近重复复核；缺失不影响核心功能)…"
+  python3 -m pip install --user -q curl_cffi 2>/dev/null \
+    || python3 -m pip install -q curl_cffi 2>/dev/null \
+    || python3 -m pip install --user -q -i https://pypi.tuna.tsinghua.edu.cn/simple curl_cffi 2>/dev/null \
+    || echo "[warn] pip 安装 curl_cffi 失败（可选依赖）。需要反爬 TLS 指纹伪造时手动: pip install curl_cffi"
+  python3 -m pip install --user -q rapidfuzz 2>/dev/null \
+    || python3 -m pip install -q rapidfuzz 2>/dev/null \
+    || python3 -m pip install --user -q -i https://pypi.tuna.tsinghua.edu.cn/simple rapidfuzz 2>/dev/null \
+    || echo "[warn] pip 安装 rapidfuzz 失败（可选依赖）。缺失时去重回退纯 minhash，需要模糊近重复复核时手动: pip install rapidfuzz"
+fi
+
+# ── pre-commit 快门禁 hook（2026-10-16 流程洞实证后新增） ──────────────────
+# 背景：并发代理接线没跑测试，6 道门禁红 + 阻塞全量收集，故障在树里躺了几小时
+# 才被人工发现。拦截点放在提交层：只跑四道快门禁（~8s），全量回归留给手动。
+# 链式而非覆盖：hooks/pre-commit 可能已是 privacy-guard 等用户自有防线
+# （2026-10-17 实测本机就是），argo 门禁装成独立文件 pre-commit-argo，再向
+# 既有 hook 追一行调用（幂等：grep 到标记行才追加）。既有 hook 先exit 1 时
+# 我们的不会跑到——隐私/密钥防线优先，正确。
+if command -v git >/dev/null 2>&1 && { [ -d "$INSTALL_DIR/.git" ] || [ -f "$INSTALL_DIR/.git" ]; }; then
+  _common=$(cd "$INSTALL_DIR" && git rev-parse --git-common-dir 2>/dev/null || true)
+  case "$_common" in
+    /*) ;;
+    "") _common="" ;;
+    *)  _common="$INSTALL_DIR/$_common" ;;
+  esac
+  if [ -n "$_common" ] && [ -d "$_common" ]; then
+    _hooks_dir="$_common/hooks"
+    mkdir -p "$_hooks_dir"
+    cp "$INSTALL_DIR/scripts/pre-commit" "$_hooks_dir/pre-commit-argo" \
+      && chmod +x "$_hooks_dir/pre-commit-argo" || {
+        echo "[warn] pre-commit-argo 安装失败（不影响 argo 本体使用）"
+      }
+    if [ ! -e "$_hooks_dir/pre-commit" ]; then
+      cp "$_hooks_dir/pre-commit-argo" "$_hooks_dir/pre-commit"
+      echo "==> 已装 pre-commit 快门禁 hook（~8s；git commit --no-verify 可跳过）"
+    elif grep -q "pre-commit-argo" "$_hooks_dir/pre-commit" 2>/dev/null; then
+      echo "==> pre-commit 已接入 argo 快门禁（无需重复追加）"
+    else
+      printf '\n# argo 快门禁（2026-10-17 追加；独立文件，删掉下面一行即可解链）\n[ -x "$(dirname "$0")/pre-commit-argo" ] && "$(dirname "$0")/pre-commit-argo"\n' >> "$_hooks_dir/pre-commit" \
+        && echo "==> argo 快门禁已链入既有 pre-commit（未覆盖你的 hook）"
+    fi
+  fi
+fi
+
+
+if [[ ${#LINK_ARGS[@]} -gt 0 ]] || [[ -n "${ARGO_LINK_TARGETS:-}" ]] || [[ -f "$INSTALL_DIR/installs.local.yaml" ]]; then
+  echo "==> 链接 Skill 入口（符号链接回真源，不复制）…"
+  python3 "$INSTALL_DIR/scripts/link_source.py" "${LINK_ARGS[@]+"${LINK_ARGS[@]}"}" || {
+    echo "[warn] 链接未完成。可稍后手动:"
+    echo "  python3 $INSTALL_DIR/scripts/link_source.py --to ~/.claude/skills/argo"
+  }
+fi
+
+echo ""
+echo "==> 安装完成"
+echo ""
+echo "快速验证:"
+echo "  python3 $INSTALL_DIR/scripts/search.py \"Python asyncio\" --json"
+echo "  python3 $INSTALL_DIR/scripts/search.py --list-engines"
+echo ""
+echo "启动 MCP（给 Claude / Kimi / Cursor 等用）:"
+echo "  python3 $INSTALL_DIR/scripts/mcp_server.py"
+echo ""
+echo "或用 npx（需 Node.js 18+）:"
+echo "  npx -y github:taxueseek/argo"
+echo ""
+echo "客户端 MCP 配置示例（路径请按本机替换）:"
+cat <<EOF
+{
+  "mcpServers": {
+    "argo": {
+      "command": "python3",
+      "args": ["$INSTALL_DIR/scripts/mcp_server.py"]
+    }
+  }
+}
+EOF
+echo ""
+echo "可选：把 Skill 挂到本机 Agent 目录（不复制代码）:"
+echo "  python3 $INSTALL_DIR/scripts/link_source.py --to ~/.claude/skills/argo"
+echo "  python3 $INSTALL_DIR/scripts/link_source.py --to ~/.agents/skills/argo"
+echo ""
+echo "文档: https://github.com/taxueseek/argo"