refactor(completions): centralize content scanning

Add a browser-safe ContentRegistry owning path and inline content
stores, then derive completions from it. Remove the old block-processor,
directive-scanner, and link-source modules in favor of scanDoc,
deriveCompletions, and injectSparkDirectives, and register an inline
resolver for the frontend file index.

src/cli/commands/serve.ts

@@ -8,17 +8,12 @@ import { networkInterfaces } from "os";
 import { fileURLToPath } from "url";
 import { createJournalServer } from "../journal.js";
 import {
-  scanCompletions,
-  type CompletionsPayload,
-} from "../completions/index.js";
-import {
-  processBlocks,
-  type ProcessedBlocks,
-} from "../completions/block-processor.js";
-import {
-  scanDirectives,
-  type DirectiveScanResult,
-} from "../completions/directive-scanner.js";
+  scanDoc,
+  deriveCompletions,
+  injectSparkDirectives,
+  type ContentRegistry,
+} from "../content-registry.js";
+import type { CompletionsPayload } from "../completions/types.js";
 
 interface ContentIndex {
   [path: string]: string;
@@ -88,20 +83,10 @@ function getBestIP(): string {
 }
 
 /**
- * 扫描目录内的 .md 等文件，生成内容索引与块数据
+ * 扫描目录内的 .md 等文件，构建内容注册表（路径索引 + 每文档内联内容）
  */
-export function scanDirectory(dir: string): {
-  index: ContentIndex;
-  blocks: ProcessedBlocks;
-  directiveResults: DirectiveScanResult[];
-} {
-  const index: ContentIndex = {};
-  const blocks: ProcessedBlocks = {
-    declarations: [],
-    tagModifiers: [],
-  };
-  const directiveResults: DirectiveScanResult[] = [];
-  const mdFiles: { content: string; relPath: string }[] = [];
+export function buildRegistry(dir: string): ContentRegistry {
+  const registry: ContentRegistry = { pathIndex: {}, docContent: {} };
 
   function scan(currentPath: string, relativePath: string) {
     const entries = readdirSync(currentPath);
@@ -125,13 +110,11 @@ export function scanDirectory(dir: string): {
         try {
           const content = readFileSync(fullPath, "utf-8");
           if (entry.endsWith(".md")) {
-            const result = processBlocks(content, normalizedRelPath, index);
-            index[normalizedRelPath] = result.stripped;
-            blocks.declarations.push(...result.blocks.declarations);
-            blocks.tagModifiers.push(...result.blocks.tagModifiers);
-            mdFiles.push({ content: result.stripped, relPath: normalizedRelPath });
+            const result = scanDoc(content, normalizedRelPath);
+            registry.pathIndex[normalizedRelPath] = result.stripped;
+            registry.docContent[normalizedRelPath] = result.content;
           } else {
-            index[normalizedRelPath] = content;
+            registry.pathIndex[normalizedRelPath] = content;
           }
         } catch (e) {
           console.error(`读取文件失败：${fullPath}`, e);
@@ -142,31 +125,17 @@ export function scanDirectory(dir: string): {
 
   scan(dir, "");
 
-  // ---- Directive scanning pass (after all blocks processed) ----
-  const posixDir = dir.split(sep).join("/");
-  for (const { content, relPath } of mdFiles) {
-    const fileDir = posixRelDir(relPath);
-    const result = scanDirectives(content, relPath, index, fileDir);
-
-    // Apply rewritten content back to index
-    index[relPath] = result.rewritten;
-
-    // Inject new index entries (inline CSV bodies)
-    for (const [key, value] of Object.entries(result.newIndexEntries)) {
-      index[key] = value;
-    }
-
-    directiveResults.push(result);
+  // ---- Inject data-spark into real-file spark table directives ----
+  for (const [relPath, content] of Object.entries(registry.pathIndex)) {
+    if (!relPath.endsWith(".md")) continue;
+    registry.pathIndex[relPath] = injectSparkDirectives(
+      content,
+      relPath,
+      registry,
+    );
   }
 
-  return { index, blocks, directiveResults };
-}
-
-/** Get the POSIX directory of a file path */
-function posixRelDir(filePath: string): string {
-  const parts = filePath.split("/");
-  parts.pop();
-  return parts.join("/") || ".";
+  return registry;
 }
 
 /**
@@ -231,6 +200,7 @@ function createRequestHandler(
   distDir: string,
   getIndex: () => ContentIndex,
   getCompletions: () => CompletionsPayload,
+  getRegistry: () => ContentRegistry,
 ) {
   return (req: IncomingMessage, res: ServerResponse) => {
     const url = req.url || "/";
@@ -248,6 +218,12 @@ function createRequestHandler(
       return;
     }
 
+    // 1c. 处理 /__CONTENT_REGISTRY.json（含每文档内联内容，供运行时解析）
+    if (filePath === "/__CONTENT_REGISTRY.json") {
+      sendJson(res, getRegistry());
+      return;
+    }
+
     // 2. 处理 /static/ 目录（从 dist/web）
     if (filePath.startsWith("/static/")) {
       if (tryServeStatic(res, filePath, distDir)) {
@@ -308,12 +284,7 @@ export function createContentServer(
   distPath: string = distDir,
   host: string = "0.0.0.0",
 ): ContentServer {
-  let contentIndex: ContentIndex = {};
-  let collectedBlocks: ProcessedBlocks = {
-    declarations: [],
-    tagModifiers: [],
-  };
-  let directiveResults: DirectiveScanResult[] = [];
+  let registry: ContentRegistry = { pathIndex: {}, docContent: {} };
   let completionsIndex: CompletionsPayload = {
     dice: [],
     links: [],
@@ -322,21 +293,18 @@ export function createContentServer(
     tagModifiers: [],
   };
 
-  /** 从当前内容索引和已收集的块重新扫描补全数据 */
+  /** 从当前注册表重新派生补全数据 */
   function recomputeCompletions(): void {
-    completionsIndex = scanCompletions(contentIndex, collectedBlocks, directiveResults);
+    completionsIndex = deriveCompletions(registry);
     console.log(
       `[completions] dice=${completionsIndex.dice.length} links=${completionsIndex.links.length} sparkTables=${completionsIndex.sparkTables.length} declarations=${completionsIndex.declarations.length} tagModifiers=${completionsIndex.tagModifiers.length}`,
     );
   }
 
-  // 扫描内容目录生成索引
+  // 扫描内容目录生成注册表
   console.log("正在扫描内容目录...");
-  const scanResult = scanDirectory(contentDir);
-  contentIndex = scanResult.index;
-  collectedBlocks = scanResult.blocks;
-  directiveResults = scanResult.directiveResults;
-  console.log(`已索引 ${Object.keys(contentIndex).length} 个文件`);
+  registry = buildRegistry(contentDir);
+  console.log(`已索引 ${Object.keys(registry.pathIndex).length} 个文件`);
   recomputeCompletions();
 
   // 监听文件变化
@@ -356,20 +324,10 @@ export function createContentServer(
         path.endsWith(".svg")
       ) {
         try {
-          const content = readFileSync(path, "utf-8");
-          const relPath = "/" + relative(contentDir, path).split(sep).join("/");
-          if (relPath.endsWith(".md")) {
-            const result = processBlocks(content, relPath, contentIndex);
-            contentIndex[relPath] = result.stripped;
-            // Re-scan to get fresh blocks (simpler than per-file merge)
-            const rescan = scanDirectory(contentDir);
-            collectedBlocks = rescan.blocks;
-            directiveResults = rescan.directiveResults;
-            recomputeCompletions();
-          } else {
-            contentIndex[relPath] = content;
-          }
-          console.log(`[新增] ${relPath}`);
+          // 全量重建注册表以刷新跨文件派生（spark 注入）
+          registry = buildRegistry(contentDir);
+          recomputeCompletions();
+          console.log(`[新增] ${path}`);
         } catch (e) {
           console.error(`读取新增文件失败：${path}`, e);
         }
@@ -383,19 +341,9 @@ export function createContentServer(
         path.endsWith(".svg")
       ) {
         try {
-          const content = readFileSync(path, "utf-8");
-          const relPath = "/" + relative(contentDir, path).split(sep).join("/");
-          if (relPath.endsWith(".md")) {
-            const result = processBlocks(content, relPath, contentIndex);
-            contentIndex[relPath] = result.stripped;
-            const rescan = scanDirectory(contentDir);
-            collectedBlocks = rescan.blocks;
-            directiveResults = rescan.directiveResults;
-            recomputeCompletions();
-          } else {
-            contentIndex[relPath] = content;
-          }
-          console.log(`[更新] ${relPath}`);
+          registry = buildRegistry(contentDir);
+          recomputeCompletions();
+          console.log(`[更新] ${path}`);
         } catch (e) {
           console.error(`读取更新文件失败：${path}`, e);
         }
@@ -408,15 +356,9 @@ export function createContentServer(
         path.endsWith(".yarn") ||
         path.endsWith(".svg")
       ) {
-        const relPath = "/" + relative(contentDir, path).split(sep).join("/");
-        delete contentIndex[relPath];
-        console.log(`[删除] ${relPath}`);
-        if (relPath.endsWith(".md")) {
-          const rescan = scanDirectory(contentDir);
-          collectedBlocks = rescan.blocks;
-          directiveResults = rescan.directiveResults;
-          recomputeCompletions();
-        }
+        registry = buildRegistry(contentDir);
+        recomputeCompletions();
+        console.log(`[删除] ${path}`);
       }
     });
 
@@ -427,8 +369,9 @@ export function createContentServer(
   const handleRequest = createRequestHandler(
     contentDir,
     distPath,
-    () => contentIndex,
+    () => registry.pathIndex,
     () => completionsIndex,
+    () => registry,
   );
   const server = createServer(handleRequest);
 
@@ -452,7 +395,7 @@ export function createContentServer(
   return {
     server,
     watcher,
-    index: contentIndex,
+    index: registry.pathIndex,
     completions: completionsIndex,
     close() {
       console.log("正在关闭内容服务器...");