formatferry 1.0.38 → 1.0.40

package/dist/index.js

@@ -58942,6 +58942,12 @@ h${level}. ${trimmed}
 }
 
 // ../../src/shared/conversion-engine.ts
+var TableMarkdownLimitError = class extends Error {
+  constructor() {
+    super("Table Markdown exceeds the conversion limit");
+    this.name = "TableMarkdownLimitError";
+  }
+};
 var ConversionEngine = class {
   githubService;
   commonmarkService;
@@ -58953,6 +58959,10 @@ var ConversionEngine = class {
   confluenceService;
   dom;
   domSanitizer;
+  /** Scoped to one synchronous Turndown pass; cloned cells do not change while table rules read them. */
+  tableCellCache;
+  /** Table Markdown characters this conversion may still emit (maxTableMarkdownChars). */
+  tableMarkdownBudget = Number.POSITIVE_INFINITY;
   constructor(options = {}) {
     this.dom = options.dom;
     this.initializeServices();
@@ -58967,7 +58977,7 @@ var ConversionEngine = class {
         this.slackService,
         this.confluenceService
       ]) {
-        this.parseWithInjectedDom(service, this.dom);
+        this.parseWithInjectedDom(service, this.dom, service === this.githubService);
       }
     }
   }
@@ -58978,7 +58988,7 @@ var ConversionEngine = class {
    * passes the element, which Turndown accepts in place of a string. Covers
    * the nested `service.turndown(el.outerHTML)` calls in the list rules too.
    */
-  parseWithInjectedDom(service, dom) {
+  parseWithInjectedDom(service, dom, batchParagraphs) {
     const turndown = service.turndown.bind(service);
     service.turndown = (input) => {
       if (typeof input !== "string") return turndown(input);
@@ -58989,9 +58999,64 @@ var ConversionEngine = class {
       const root = doc.getElementById("turndown-root");
       root?.normalize();
       if (!root) throw new Error("Injected DOM did not produce the turndown root element");
-      return turndown(root);
+      if (batchParagraphs) {
+        this.batchParagraphSiblings(root);
+        this.batchTableRows(root);
+      }
+      return turndown(this.ownedDomInput(root));
     };
   }
+  /** Turndown consumes a fresh private tree synchronously; caller-owned DOM never uses this adapter. */
+  ownedDomInput(root) {
+    return { nodeType: 1, cloneNode: () => root };
+  }
+  /**
+   * GFM paragraphs and generic divs both contribute two separating newlines. Group long runs in the
+   * private parsed DOM so Turndown joins a few hundred outputs at a time, instead of flattening a growing
+   * string tens of thousands of times (#180). Keep DOM-sensitive table/list/quote/pre rules untouched.
+   */
+  batchParagraphSiblings(root) {
+    const containers = [root, ...Array.from(root.querySelectorAll("body,main,article,section,div"))];
+    for (const container of containers) {
+      if (container.childNodes.length < 512 || container.closest("table,ul,ol,pre,blockquote")) continue;
+      const children = Array.from(container.childNodes);
+      if (!children.every((child) => child.nodeType === 1 ? child.nodeName === "P" : child.nodeType === 8 || child.nodeType === 3 && !child.textContent?.trim())) continue;
+      let group;
+      let count = 0;
+      for (const child of children) {
+        if (!group) {
+          group = container.ownerDocument.createElement("div");
+          container.insertBefore(group, child);
+        }
+        group.appendChild(child);
+        if (child.nodeType === 1 && ++count === 256) {
+          group = void 0;
+          count = 0;
+        }
+      }
+    }
+  }
+  /** The table rule discards preliminary row Markdown and queries all descendant rows itself. */
+  batchTableRows(root) {
+    for (const container of Array.from(root.querySelectorAll("table,thead,tbody,tfoot"))) {
+      if (container.childNodes.length < 512) continue;
+      const children = Array.from(container.childNodes);
+      if (!children.every((child) => child.nodeType === 1 ? child.nodeName === "TR" : child.nodeType === 8 || child.nodeType === 3 && !child.textContent?.trim())) continue;
+      let group;
+      let count = 0;
+      for (const child of children) {
+        if (!group) {
+          group = container.ownerDocument.createElement("tbody");
+          container.insertBefore(group, child);
+        }
+        group.appendChild(child);
+        if (child.nodeType === 1 && ++count === 256) {
+          group = void 0;
+          count = 0;
+        }
+      }
+    }
+  }
   initializeServices() {
     this.githubService = new import_turndown2.default({
       headingStyle: "atx",
@@ -59627,10 +59692,22 @@ ${plainText}
         if (rows.length === 0) return content;
         let markdown = "\n";
         let headerProcessed = false;
+        let tableLength = 0;
         rows.forEach((row, index) => {
           const cells = Array.from(row.querySelectorAll("th, td"));
-          const cellContents = cells.map((cell) => this.escapeTableCell(this.renderTableCell(cell)));
+          const cellContents = cells.map((cell) => {
+            let rendered = this.tableCellCache?.get(cell);
+            if (rendered === void 0) {
+              rendered = this.escapeTableCell(this.renderTableCell(cell));
+              this.chargeTableMarkdown(rendered.length);
+              this.tableCellCache?.set(cell, rendered);
+            }
+            return rendered;
+          });
           if (cellContents.length > 0) {
+            tableLength += 1 + 3 * cellContents.length;
+            for (const value of cellContents) tableLength += value.length;
+            if (tableLength > this.tableMarkdownBudget) throw new TableMarkdownLimitError();
             markdown += "| " + cellContents.join(" | ") + " |\n";
             if (!headerProcessed && (row.querySelector("th") || index === 0)) {
               const separator = cellContents.map(() => "---").join(" | ");
@@ -59643,6 +59720,11 @@ ${plainText}
       }
     });
   }
+  /** Deduct a rendered cell, which the per-conversion cache keeps alive until the conversion ends. */
+  chargeTableMarkdown(chars) {
+    this.tableMarkdownBudget -= chars;
+    if (this.tableMarkdownBudget < 0) throw new TableMarkdownLimitError();
+  }
   renderTableCell(cell) {
     const ELEMENT_NODE2 = 1;
     const blocks = [];
@@ -59700,8 +59782,17 @@ ${plainText}
   }
   renderCodeSpan(content) {
     if (!content.includes("`")) return `\`${content}\``;
-    const runs = content.match(/`+/g) || ["`"];
-    const delimiter = "`".repeat(Math.max(...runs.map((run) => run.length)) + 1);
+    let longest = 0;
+    let current = 0;
+    for (let i = 0; i < content.length; i++) {
+      if (content.charCodeAt(i) === 96) {
+        current++;
+        if (current > longest) longest = current;
+      } else {
+        current = 0;
+      }
+    }
+    const delimiter = "`".repeat(longest + 1);
     return `${delimiter} ${content} ${delimiter}`;
   }
   escapeMarkdownLinkText(text) {
@@ -59711,7 +59802,8 @@ ${plainText}
     return href.replace(/</g, "%3C").replace(/>/g, "%3E");
   }
   escapeTableCell(content) {
-    return content.replace(/\|/g, "\\|").replace(/\s*<br>\s*/g, "<br>").trim();
+    const escaped = content.replace(/\|/g, "\\|");
+    return escaped.includes("<br>") ? escaped.split("<br>").map((part) => part.trim()).join("<br>") : escaped.trim();
   }
   /** A space after a checkbox marker unless the label text already starts with one. */
   checkboxSpacer(node) {
@@ -59766,6 +59858,12 @@ ${plainText}
     });
   }
   addListRules(service) {
+    if (this.dom && service === this.githubService) {
+      service.addRule("discardPreliminaryListItems", {
+        filter: (node) => node.nodeName === "LI" && (node.parentNode?.nodeName === "UL" || node.parentNode?.nodeName === "OL"),
+        replacement: () => ""
+      });
+    }
     service.addRule("enhancedList", {
       filter: ["ul", "ol"],
       replacement: (content, node) => {
@@ -59792,10 +59890,12 @@ ${plainText}
     let markdown = "";
     items.forEach((item, index) => {
       const marker = isOrdered ? `${index + 1}.` : bullet;
-      const indent = " ".repeat(marker.length + 1);
       const body = this.processListItem(item, service);
-      const lines = body.split("\n");
-      const indented = lines.map((line, i) => i === 0 || line === "" ? line : indent + line).join("\n");
+      let indented = body;
+      if (body.includes("\n")) {
+        const indent = " ".repeat(marker.length + 1);
+        indented = body.split("\n").map((line, i) => i === 0 || line === "" ? line : indent + line).join("\n");
+      }
       markdown += `${marker} ${indented}
 `;
     });
@@ -59808,13 +59908,21 @@ ${plainText}
    * multiple paragraphs stay separated.
    */
   processListItem(item, service) {
+    if (this.dom && service === this.githubService && service.options.linkStyle !== "referenced") {
+      const child = item.firstChild;
+      if (child?.nodeType === 3 && child.nextSibling === null) {
+        const text = child.textContent ?? "";
+        if (!/\s/.test(text)) return service.escape(text).trim();
+      }
+    }
     const parts = [];
     const ELEMENT_NODE2 = 1;
     const doc = item.ownerDocument;
     let buffer = doc.createElement("div");
     const flush = () => {
       if (buffer.childNodes.length > 0) {
-        const md = service.turndown(buffer).trim();
+        const input = this.dom && service === this.githubService ? this.ownedDomInput(buffer) : buffer;
+        const md = service.turndown(input).trim();
         if (md) parts.push(md);
       }
       buffer = doc.createElement("div");
@@ -59936,7 +60044,17 @@ ${plainText}
         break;
     }
     const htmlForConversion = options.preserveTables ? sanitizedHtml : this.flattenTables(sanitizedHtml);
-    let markdown = service.turndown(htmlForConversion);
+    const previousCache = this.tableCellCache;
+    const previousBudget = this.tableMarkdownBudget;
+    this.tableCellCache = /* @__PURE__ */ new WeakMap();
+    this.tableMarkdownBudget = options.maxTableMarkdownChars ?? Number.POSITIVE_INFINITY;
+    let markdown;
+    try {
+      markdown = service.turndown(htmlForConversion);
+    } finally {
+      this.tableCellCache = previousCache;
+      this.tableMarkdownBudget = previousBudget;
+    }
     markdown = this.applyFormatPostProcessing(markdown, options.format);
     const endTime = performance.now();
     const metadata = {
@@ -64297,7 +64415,7 @@ async function binaryToMarkdown(buffer, type, format) {
         preserveLists: true,
         customRules: []
       });
-      return { markdown: converted.markdown, warnings: result.warnings, warningCodes: result.warningCodes };
+      return { markdown: converted.markdown, warnings: result.warnings };
     }
     case "pdf": {
       const result = await pdfAdapter.parseToText(toArrayBuffer());
@@ -64320,7 +64438,7 @@ async function binaryToMarkdown(buffer, type, format) {
         preserveLists: true,
         customRules: []
       });
-      return { markdown: converted.markdown, warnings: result.warnings, warningCodes: result.warningCodes };
+      return { markdown: converted.markdown, warnings: result.warnings };
     }
     case "json": {
       const result = await dataAdapter.parseJsonToHtml(toArrayBuffer());
@@ -64331,7 +64449,7 @@ async function binaryToMarkdown(buffer, type, format) {
         preserveLists: true,
         customRules: []
       });
-      return { markdown: converted.markdown, warnings: result.warnings, warningCodes: result.warningCodes };
+      return { markdown: converted.markdown, warnings: result.warnings };
     }
     case "xml": {
       const result = await dataAdapter.parseXmlToHtml(toArrayBuffer());
@@ -64342,7 +64460,7 @@ async function binaryToMarkdown(buffer, type, format) {
         preserveLists: true,
         customRules: []
       });
-      return { markdown: converted.markdown, warnings: result.warnings, warningCodes: result.warningCodes };
+      return { markdown: converted.markdown, warnings: result.warnings };
     }
     case "pptx": {
       const result = await pptxAdapter.parseToHtml(toArrayBuffer());
@@ -64353,7 +64471,7 @@ async function binaryToMarkdown(buffer, type, format) {
         preserveLists: true,
         customRules: []
       });
-      return { markdown: converted.markdown, warnings: result.warnings, warningCodes: result.warningCodes };
+      return { markdown: converted.markdown, warnings: result.warnings };
     }
   }
 }
@@ -64909,8 +65027,8 @@ import { readFileSync as readFileSync5 } from "fs";
 import { resolve, dirname as dirname3 } from "path";
 import { fileURLToPath } from "url";
 var getVersion = () => {
-  if ("1.0.38") {
-    return "1.0.38";
+  if ("1.0.40") {
+    return "1.0.40";
   }
   try {
     const __dirname2 = dirname3(fileURLToPath(import.meta.url));

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "formatferry",
-  "version": "1.0.38",
+  "version": "1.0.40",
   "description": "Format Ferry CLI — Convert HTML, DOCX, PDF, XLSX, CSV, JSON, XML, PPTX to Markdown",
   "keywords": [
     "markdown",