markdown-table-formatter 2.0.0 → 2.1.0

package/README.md

@@ -32,6 +32,7 @@ The formatting rules come from [markdown-table-prettify](https://github.com/dark
 - [Continuous Integration](#continuous-integration)
 - [pre-commit hook](#pre-commit-hook)
 - [Ignoring Code Blocks](#ignoring-code-blocks)
+- [Performance](#performance)
 - [Debug](#debug)
 - [Contribute](#contribute)
 - [Release notes](#release-notes)
@@ -132,6 +133,17 @@ To achieve this, place tags `<!-- markdown-table-prettify-ignore-start -->` and
 
 Tables inside fenced code blocks (` ``` ` or `~~~`) and indented code blocks are never formatted.
 
+## Performance
+
+Formatting the 630 markdown files (11 MB) of the [MegaLinter](https://github.com/oxsecurity/megalinter) repository, with the command of its [format-tables.sh](https://github.com/oxsecurity/megalinter/blob/main/.automation/format-tables.sh) script (`find . -name "*.md" -print0 | xargs -0 -n 200 markdown-table-formatter --verbose`), on Windows 11 with Node.js 24:
+
+| Version | Run 1  | Run 2  | Run 3  | Average   | Speed       |
+|---------|--------|--------|--------|-----------|-------------|
+| 1.7.0   | 6.38 s | 6.07 s | 5.87 s | **6.1 s** | 1x          |
+| 2.1.0   | 1.41 s | 1.44 s | 1.40 s | **1.4 s** | 4.3x faster |
+
+Tip: send all files in a single call when possible, as each call starts Node.js again.
+
 ## Debug
 
 Set environment variable `NODE_DEBUG=markdown-table-formatter` to display debug logs.

package/lib/cli.js

@@ -16,9 +16,16 @@ function findFiles(pattern) {
     EXCLUDED_DIRS.has(
       path.basename(typeof entry === "string" ? entry : entry.name),
     );
+  // Entries come with their type, so only symbolic links need a stat call
   return fs
-    .globSync(pattern, { exclude })
-    .filter((file) => fs.statSync(file).isFile())
+    .globSync(pattern, { exclude, withFileTypes: true })
+    .flatMap((entry) => {
+      const file = path.join(entry.parentPath, entry.name);
+      const isFile = entry.isSymbolicLink()
+        ? fs.statSync(file).isFile()
+        : entry.isFile();
+      return isFile ? [file] : [];
+    })
     .sort();
 }
 
@@ -62,7 +69,11 @@ class MarkdownTableFormatterCli {
       // use sent glob expression
       files = findFiles(files[0]);
     }
-    const { status, updates } = await formatter.run(files, this.options);
+    // The CLI does not use file contents: don't keep them in memory
+    const { status, updates } = await formatter.run(files, {
+      ...this.options,
+      keepContents: false,
+    });
     // Manage result
     if (status === 0 && updates.length === 0) {
       console.log("No markdown table formatting has been applied.");

package/lib/engine.js

@@ -25,6 +25,10 @@ function getWasm() {
  * @return {string} - Formatted markdown content
  */
 function formatTables(text, columnPadding = 0) {
+  // A text without pipe holds no table: the engine would return it unchanged
+  if (!text.includes("|")) {
+    return text;
+  }
   const engine = getWasm();
   const input = Buffer.from(text, "utf8");
   const inputPtr = engine.mtf_alloc(input.length);

package/lib/markdown-table-formatter.js

@@ -1,8 +1,12 @@
 #! /usr/bin/env node
 const fs = require("fs/promises");
+const path = require("path");
 const debug = require("util").debuglog("markdown-table-formatter");
 const { formatTables } = require("./engine");
 
+// Number of files read and written at the same time
+const MAX_PARALLEL_FILES = 32;
+
 class MarkdownTableFormatter {
   "use strict";
 
@@ -26,10 +30,50 @@ class MarkdownTableFormatter {
     return formatTables(markdownString, options.columnpadding || 0);
   }
 
+  /**
+   * Formats one file
+   * @param {string} file - File path
+   * @param {object} options - Run options (see run)
+   * @return {Promise<object>} - File result (file, updated, before, after)
+   */
+  async formatFile(file, options) {
+    if (debug.enabled) {
+      debug(`[file] ${file}`);
+    }
+    const markdownString = await fs.readFile(file, "utf8");
+    if (debug.enabled) {
+      debug(`[file before]\n${markdownString}`);
+    }
+    const markdownStringPrettified = MarkdownTableFormatter.formatText(
+      markdownString,
+      options,
+    );
+    const updated = markdownStringPrettified !== markdownString;
+    // Update file if necessary
+    if (updated && !this.checkOnly) {
+      await fs.writeFile(file, markdownStringPrettified);
+      debug(`[updated] ${file} with:\n${markdownStringPrettified}`);
+      if (options.verbose) {
+        console.info(`- Updated file ${file}`);
+      }
+    }
+    const keepContents = updated && options.keepContents !== false;
+    return {
+      file,
+      updated,
+      before: keepContents ? markdownString : null,
+      after: keepContents ? markdownStringPrettified : null,
+    };
+  }
+
   /**
    * Runs markdown table formatter
    * @param {string[]} [files] - List of files to format
-   * @return {Promise<any>} - Command result (status, stdout, stderr, childJavaProcess)
+   * @param {object} [options] - Run options
+   * @param {number} [options.columnpadding] - Number of spaces around cell values
+   * @param {boolean} [options.verbose] - Log analyzed and updated files
+   * @param {boolean} [options.keepContents] - Set to false to return null before / after contents (saves memory)
+   * @return {Promise<any>} - Command result (status, fileResults, updates)
    */
   async run(files, options = {}) {
     if (options.verbose) {
@@ -38,38 +82,31 @@ class MarkdownTableFormatter {
         console.info(`- ${file}`);
       }
     }
-    const fileResults = [];
-    // Read files and call prettify on each of them
-    for (const file of files) {
-      if (debug.enabled) {
-        debug(`[file] ${file}`);
-      }
-      const markdown = await fs.readFile(file);
-      const markdownString = markdown.toString();
-      if (debug.enabled) {
-        debug(`[file before]\n${markdownString}`);
-      }
-      const markdownStringPrettified = MarkdownTableFormatter.formatText(
-        markdownString,
-        options,
-      );
-      const updated = markdownStringPrettified !== markdownString;
-      // Update file if necessary
-      if (updated && !this.checkOnly) {
-        await fs.writeFile(file, markdownStringPrettified);
-        debug(`[updated] ${file} with:\n${markdownStringPrettified}`);
-        if (options.verbose) {
-          console.info(`- Updated file ${file}`);
+    // Several files are read and written at the same time; results keep the order of files
+    const fileResults = new Array(files.length);
+    // A file listed twice is processed after its previous occurrence, never at the same time
+    const pending = new Map();
+    let next = 0;
+    let failed = false;
+    const worker = async () => {
+      while (next < files.length && !failed) {
+        const index = next++;
+        const file = files[index];
+        const key = path.resolve(file);
+        const result = (pending.get(key) || Promise.resolve())
+          .catch(() => {})
+          .then(() => this.formatFile(file, options));
+        pending.set(key, result);
+        try {
+          fileResults[index] = await result;
+        } catch (error) {
+          failed = true;
+          throw error;
         }
       }
-      const fileResult = {
-        file,
-        updated,
-        before: updated ? markdownString : null,
-        after: updated ? markdownStringPrettified : null,
-      };
-      fileResults.push(fileResult);
-    }
+    };
+    const workerCount = Math.min(MAX_PARALLEL_FILES, files.length);
+    await Promise.all(Array.from({ length: workerCount }, worker));
     const updates = fileResults
       .filter((fileResult) => fileResult.updated)
       .map((fileResult) => fileResult.file);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "markdown-table-formatter",
-  "version": "2.0.0",
+  "version": "2.1.0",
   "description": "Format markdown tables in files (Rust engine compiled to WebAssembly, no dependencies)",
   "main": "./lib/index.js",
   "files": [