feat(seo-geo): 落盘自动补齐与 143 全量回填支持

Finish/交付写盘前自动回填 mindspace-geo，扩展 storage publications 扫描与 geo meta 解析，并增加 143 生产回填脚本。

Co-authored-by: Cursor <cursoragent@cursor.com>

scripts/backfill-mindspace-geo-prod.sh

@@ -0,0 +1,77 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+# Batch backfill SEO/GEO for all MindSpace HTML on production host 143.
+# 143 运行的是 release runtime 包（无 .git），因此从本机 rsync 工具脚本后再执行回填。
+#
+# Usage (from laptop):
+#   bash scripts/backfill-mindspace-geo-prod.sh
+#   bash scripts/backfill-mindspace-geo-prod.sh --dry-run
+
+HOST="${MEMIND_PROD_HOST:-john@180.159.29.143}"
+LOCAL_REPO="$(cd "$(dirname "$0")/.." && pwd)"
+REMOTE_REPO="${MEMIND_PROD_REPO:-/Users/john/Project/Memind}"
+REMOTE_MINDSPACE_ROOT="${MEMIND_PROD_MINDSPACE_ROOT:-${REMOTE_REPO}/MindSpace}"
+REMOTE_STORAGE_ROOT="${MEMIND_PROD_STORAGE_ROOT:-/Users/john/MindSpace/data/mindspace}"
+REMOTE_PUBLIC_ROOT="${MEMIND_PROD_PUBLIC_ROOT:-${REMOTE_REPO}/public}"
+REMOTE_NODE="${MEMIND_PROD_NODE:-/opt/homebrew/Cellar/node@22/22.22.3/bin/node}"
+EXTRA_ARGS=("$@")
+
+echo "==> 143 SEO/GEO backfill"
+echo "    host=${HOST}"
+echo "    local=${LOCAL_REPO}"
+echo "    remote=${REMOTE_REPO}"
+echo "    mindspace=${REMOTE_MINDSPACE_ROOT}"
+echo "    storage=${REMOTE_STORAGE_ROOT}"
+echo "    public=${REMOTE_PUBLIC_ROOT}"
+echo "    node=${REMOTE_NODE}"
+
+SEO_FILES=(
+  chat-image-materialize.mjs
+  chat-image-turn-scope.mjs
+  html-document-injection.mjs
+  mindspace-cover-meta.mjs
+  mindspace-geo-backfill.mjs
+  mindspace-geo-meta.mjs
+  mindspace-og-tags.mjs
+  mindspace-page-tag.mjs
+  mindspace-seo-audit.mjs
+  mindspace-seo-tags.mjs
+  mindspace-thumbnail-png.mjs
+  mindspace-thumbnails.mjs
+  mindspace-workspace-thumbnails.mjs
+  scripts/backfill-mindspace-geo.mjs
+  scripts/check-mindspace-cover.mjs
+  wechat/verify/page-artifact.mjs
+  wechat/verify/share-preview.mjs
+)
+
+ssh "${HOST}" "mkdir -p '${REMOTE_REPO}/scripts' '${REMOTE_REPO}/wechat/verify'"
+for rel in "${SEO_FILES[@]}"; do
+  rsync -az "${LOCAL_REPO}/${rel}" "${HOST}:${REMOTE_REPO}/${rel}"
+done
+
+ssh "${HOST}" "set -eo pipefail
+  NODE_BIN='${REMOTE_NODE}'
+  if ! test -x \"\${NODE_BIN}\"; then
+    NODE_BIN=\$(ls /opt/homebrew/Cellar/node@22/*/bin/node 2>/dev/null | head -1)
+  fi
+  if ! test -x \"\${NODE_BIN}\"; then
+    NODE_BIN=\$(ls /Users/john/.local/share/fnm/node-versions/*/installation/bin/node 2>/dev/null | head -1)
+  fi
+  test -x \"\${NODE_BIN}\" || { echo 'node not found on 143' >&2; exit 127; }
+  cd '${REMOTE_REPO}'
+  \"\${NODE_BIN}\" scripts/backfill-mindspace-geo.mjs \\
+    --scope all \\
+    --root '${REMOTE_MINDSPACE_ROOT}' \\
+    --storage-root '${REMOTE_STORAGE_ROOT}' \\
+    --public-root '${REMOTE_PUBLIC_ROOT}' \\
+    ${EXTRA_ARGS[*]:-}
+  \"\${NODE_BIN}\" scripts/check-mindspace-cover.mjs \\
+    --scope all \\
+    --root '${REMOTE_MINDSPACE_ROOT}' \\
+    --storage-root '${REMOTE_STORAGE_ROOT}' \\
+    --public-root '${REMOTE_PUBLIC_ROOT}'
+"
+
+echo "==> done"

scripts/backfill-mindspace-geo.mjs

@@ -20,6 +20,7 @@ function parseArgs(argv) {
   let scope = 'all';
   let root = path.join(repoRoot, DEFAULT_PUBLISH_ROOT);
   let publicRoot = path.join(repoRoot, 'public');
+  let storageRoot = null;
   let userId = null;
   let file = null;
   let dryRun = false;
@@ -40,6 +41,9 @@ function parseArgs(argv) {
     } else if (arg === '--public-root' && argv[i + 1]) {
       publicRoot = path.resolve(argv[i + 1]);
       i += 1;
+    } else if (arg === '--storage-root' && argv[i + 1]) {
+      storageRoot = path.resolve(argv[i + 1]);
+      i += 1;
     } else if (arg === '--user' && argv[i + 1]) {
       userId = argv[i + 1];
       i += 1;
@@ -52,11 +56,38 @@ function parseArgs(argv) {
     }
   }
 
-  return { scope, root, publicRoot, userId, file, dryRun, forceGeo, fixWarnings };
+  return {
+    scope,
+    root,
+    publicRoot,
+    storageRoot,
+    userId,
+    file,
+    dryRun,
+    forceGeo,
+    fixWarnings,
+  };
 }
 
-const { scope, root, publicRoot, userId, file, dryRun, forceGeo, fixWarnings } = parseArgs(process.argv);
-const targets = collectHtmlTargets({ scope, mindspaceRoot: root, publicRoot, userId, file });
+const {
+  scope,
+  root,
+  publicRoot,
+  storageRoot,
+  userId,
+  file,
+  dryRun,
+  forceGeo,
+  fixWarnings,
+} = parseArgs(process.argv);
+const targets = collectHtmlTargets({
+  scope,
+  mindspaceRoot: root,
+  publicRoot,
+  storageRoot,
+  userId,
+  file,
+});
 
 let changed = 0;
 let skipped = 0;

scripts/check-mindspace-cover.mjs

@@ -23,6 +23,7 @@ function parseArgs(argv) {
   let scope = 'all';
   let root = path.join(repoRoot, DEFAULT_PUBLISH_ROOT);
   let publicRoot = path.join(repoRoot, 'public');
+  let storageRoot = null;
   let userId = null;
   let file = null;
   let seo = true;
@@ -36,6 +37,9 @@ function parseArgs(argv) {
     } else if (argv[i] === '--public-root' && argv[i + 1]) {
       publicRoot = path.resolve(argv[i + 1]);
       i += 1;
+    } else if (argv[i] === '--storage-root' && argv[i + 1]) {
+      storageRoot = path.resolve(argv[i + 1]);
+      i += 1;
     } else if (argv[i] === '--user' && argv[i + 1]) {
       userId = argv[i + 1];
       i += 1;
@@ -51,7 +55,7 @@ function parseArgs(argv) {
       process.exit(0);
     }
   }
-  return { scope, root, publicRoot, userId, file, seo };
+  return { scope, root, publicRoot, storageRoot, userId, file, seo };
 }
 
 function isPlatformOnlyPath(htmlPath, publicRoot) {
@@ -60,8 +64,15 @@ function isPlatformOnlyPath(htmlPath, publicRoot) {
   return resolved.startsWith(`${resolvedPublicRoot}${path.sep}`);
 }
 
-const { scope, root, publicRoot, userId, file, seo } = parseArgs(process.argv);
-const targets = collectHtmlTargets({ scope, mindspaceRoot: root, publicRoot, userId, file });
+const { scope, root, publicRoot, storageRoot, userId, file, seo } = parseArgs(process.argv);
+const targets = collectHtmlTargets({
+  scope,
+  mindspaceRoot: root,
+  publicRoot,
+  storageRoot,
+  userId,
+  file,
+});
 const results = [];
 
 for (const htmlPath of targets) {