@helping-ai-workflow/md2doc 2.10.1 → 2.11.1

package/lib/editor/server.js

@@ -14,6 +14,15 @@ const TABLE_MD_SRC = fs.readFileSync(path.join(__dirname, 'table-md.js'), 'utf8'
 // right after it to keep the two sibling serializers grouped together.
 const LIST_MD_SRC = fs.readFileSync(path.join(__dirname, 'list-md.js'), 'utf8');
 const HISTORY_SRC = fs.readFileSync(path.join(__dirname, 'history.js'), 'utf8');
+// Task 6: spec §3.4's shift-then-clamp, a pure data transform with no
+// dependency on any other editor module — order among these is irrelevant, it
+// only has to land before client.js reads window.md2docIndentClamp.
+const INDENT_CLAMP_SRC = fs.readFileSync(path.join(__dirname, 'indent-clamp.js'), 'utf8');
+// S2 spec §3.2/§4.3: the pure marker stripper/emitter behind the 轉換成
+// submenu. Same "no dependency on any other editor module" property as
+// indent-clamp.js above — it only has to land before client.js reads
+// window.md2docConvertMd.
+const CONVERT_MD_SRC = fs.readFileSync(path.join(__dirname, 'convert-md.js'), 'utf8');
 
 function readJson(req, limitBytes = 50 * 1024 * 1024) {
   return new Promise((resolve, reject) => {
@@ -42,6 +51,25 @@ function send(res, status, obj) {
   res.end(body);
 }
 
+// The one invariant that ties the two halves of the payload together: every
+// block's line range must address a line that actually EXISTS in `lines`.
+// blockmap.js derives ranges from marked's own tokenisation while `lines`
+// comes from a regex split here, so the two can only agree while both use the
+// SAME definition of a line terminator — and when they disagree the failure is
+// silent and destructive (lineops.replaceLines() splices past the end of the
+// array, deleting every line the block map thought was there). Throwing here
+// turns that into the server route's 500 + the client's error banner, which is
+// a document that will not open rather than a document that opens and then
+// eats its own tail.
+function assertBlockRangesFit(blocks, lines) {
+  let maxEnd = 0;
+  for (const b of blocks || []) if (b.endLine > maxEnd) maxEnd = b.endLine;
+  if (maxEnd > lines.length) {
+    throw new Error('block map is out of range: endLine ' + maxEnd +
+      ' > ' + lines.length + ' lines (line-terminator handling disagrees with marked)');
+  }
+}
+
 async function createEditorServer({ files, idleTimeoutMs = 30000, clientJs = '', listenPort = 0 }) {
   const absFiles = files.map((f) => path.resolve(f));
   let idleTimer = null;
@@ -84,6 +112,14 @@ async function createEditorServer({ files, idleTimeoutMs = 30000, clientJs = '',
         // EOL 偵測與拆行：lines 內部一律不含 \r（spec §3.11）。只有
         // /api/save 會把它接回檔案原本的 EOL；/api/render 一律用 \n。
         //
+        // 三種終止符，不是兩種（T7）：marked 的 preprocess 把裸 \r 正規化成
+        // \n（實測 marked 14：'# H\rpara\r' → heading + paragraph 兩個
+        // token），所以 blockmap 會給出「第 2 行」這種行號；而 /\r\n|\n/
+        // 不拆裸 \r，`lines` 只有一個元素。行號與 lines 脫鉤之後，任何
+        // commit 的 replaceLines() 都會把 startLine 之後的內容整段吃掉——
+        // 實測 '# H\rpara\r' 編輯第一個 block 之後 'para' 直接消失。
+        // 拆行規則必須跟 marked 的換行定義一致。
+        //
         // 多數決，不是「有 CRLF 就算 CRLF」（final review I3）：save 會把
         // `lines` 全部用同一個 eol 接回去，所以一萬行的 LF 檔裡混進一行
         // CRLF，舊式偵測會在第一次存檔時把一萬行全部改寫成 CRLF ——
@@ -92,12 +128,27 @@ async function createEditorServer({ files, idleTimeoutMs = 30000, clientJs = '',
         // 用「\n 總數 − CRLF 數」算裸 LF，而不是 /(^|[^\r])\n/g：後者是
         // non-overlapping 比對，連續空行的第二個 \n 會被前一次比對吃掉的
         // 字元擋掉而漏數。減法沒有這個誤差。
+        //
+        // 裸 \r 也進多數決，理由跟上一段同一條：既然現在會拆它，一個純
+        // CR 檔（classic Mac）就會在第一次存檔時被整份改寫成 LF——正是
+        // §3.11 第 4 點禁止的事。平手一律 LF。
         const lfTotal = (mdText.match(/\n/g) || []).length;
+        const crTotal = (mdText.match(/\r/g) || []).length;
         const crlfCount = (mdText.match(/\r\n/g) || []).length;
-        const eol = crlfCount > (lfTotal - crlfCount) ? '\r\n' : '\n';
+        const bareLf = lfTotal - crlfCount;
+        const bareCr = crTotal - crlfCount;
+        // Strict > on every comparison, so ANY tie falls through to LF — which
+        // is what the paragraph above promises. `>=` against bareCr handed a
+        // CR/CRLF tie to CRLF and contradicted it. Nothing else moves: a
+        // pure-CRLF file has bareCr === 0.
+        const eol = (crlfCount > bareLf && crlfCount > bareCr) ? '\r\n'
+          : (bareCr > bareLf && bareCr > crlfCount) ? '\r'
+          : '\n';
         const { html, blocks } = await renderMarkdown(mdText, file, { editMode: true });
+        const lines = mdText.split(/\r\n|\r|\n/);
+        assertBlockRangesFit(blocks, lines);
         const payload = JSON.stringify({
-          fileId, mtimeMs, eol, lines: mdText.split(/\r\n|\n/), blocks,
+          fileId, mtimeMs, eol, lines, blocks,
         });
         const inject =
           `<script>window.__ED__ = ${payload.replace(/</g, '\\u003c')}</script>\n` +
@@ -106,6 +157,8 @@ async function createEditorServer({ files, idleTimeoutMs = 30000, clientJs = '',
           `<script>${TABLE_MD_SRC}</script>\n` +
           `<script>${LIST_MD_SRC}</script>\n` +
           `<script>${HISTORY_SRC}</script>\n` +
+          `<script>${INDENT_CLAMP_SRC}</script>\n` +
+          `<script>${CONVERT_MD_SRC}</script>\n` +
           `<script>${clientJs}</script>\n`;
         // Splice at the LAST "</body>" — the document's real closing tag.
         // The first occurrence can sit inside an inlined diagram bundle's JS
@@ -201,4 +254,4 @@ async function createEditorServer({ files, idleTimeoutMs = 30000, clientJs = '',
   };
 }
 
-module.exports = { createEditorServer };
+module.exports = { createEditorServer, assertBlockRangesFit };