html-snapshots 5.10.0 → 6.0.0

package/lib/input-generators/_base.js

@@ -7,9 +7,9 @@
  * Licensed under the MIT license.
  */
 
-const EventEmitter = require("events").EventEmitter;
-const path = require("path");
-const urlm = require("url");
+const { EventEmitter } = require("node:events");
+const path = require("node:path");
+const urlm = require("node:url");
 const common = require("../common");
 
 // Defaults for this module
@@ -25,7 +25,7 @@ const defaults = Object.freeze({
   checkInterval: 250,
   useJQuery: false,
   verbose: false,
-  phantomjsOptions: "",
+  playwrightLaunchOptions: {},
   puppeteerLaunchOptions: {},
   debug: {
     flag: "false",
@@ -116,7 +116,8 @@ function prepOptions (options, listener) {
 
   // Prep single, global object options for normalization.
   [
-    "puppeteerLaunchOptions"
+    "puppeteerLaunchOptions",
+    "playwrightLaunchOptions"
   ]
   .forEach((optionName) => {
     prepSingleObjectOptions(optionName, options);  
@@ -126,7 +127,7 @@ function prepOptions (options, listener) {
   // The certain arguments are per-page options.
   // However, outputPath is a special case
   [
-    "selector", "timeout", "useJQuery", "verbose", "phantomjsOptions",
+    "selector", "timeout", "useJQuery", "verbose", "playwrightLaunchOptions",
     "puppeteerLaunchOptions"
   ]
   .forEach((perPageOption) => {
@@ -254,7 +255,7 @@ module.exports = {
    * @param {Number} options.checkInterval - The millisecond resolute poll interval.
    * @param {Function} options.userJQuery - Produces a jquery bool given a page url.
    * @param {Function} options.verbose - Produces a verbose bool given a page url.
-   * @param {Function} options.phantomjsOptions - Produces phantomjs options given a page url.
+   * @param {Function} options.playwrightLaunchOptions - Produces playwright launch options given a page url.
    * @param {Function} options.puppeteerLaunchOptions - Produces puppeteer launch options given a page url.
    * @param {Object} options.debug - Produces debug options given a page url.
    * @param {Function} options._inputEmitter - The EventEmitter to generate input.
@@ -288,8 +289,8 @@ module.exports = {
         useJQuery: options.useJQuery(page),
         // map the input page to a verbose flag
         verbose: options.verbose(page),
-        // map the input page to phantomJS options
-        phantomjsOptions: options.phantomjsOptions(page),
+        // map the input page to playwright launch options
+        playwrightLaunchOptions: options.playwrightLaunchOptions(page),
         // map the input page to puppeteer launch options
         puppeteerLaunchOptions: options.puppeteerLaunchOptions(page),
         // useful for puppeteer testing, debugging

package/lib/input-generators/array.js

@@ -8,7 +8,7 @@
  * Licensed under the MIT license.
  */
 
-const urlm = require("url");
+const urlm = require("node:url");
 const common = require("../common");
 const base = require("./_base");
 

package/lib/input-generators/index.js

@@ -28,7 +28,7 @@ module.exports = {
 
     if (input) {
       input = (""+input).replace(" ", "").toLowerCase();
-      hasInput = input && input.charAt(0) !== '_' && input !== "index";
+      hasInput = input && input.charAt(0) !== "_" && input !== "index";
     }
 
     try {
@@ -36,7 +36,7 @@ module.exports = {
         result = require( "./" + input);
       }
     } catch (e) {
-      console.error("Input generator load failed '" + input + "'", e)
+      console.error(`Input generator load failed '${input}'`, e);
       /* return the "null" generator on error */
     }
 

package/lib/input-generators/robots.js

@@ -8,7 +8,7 @@
  * Licensed under the MIT license.
  */
 
-const fs = require("fs").promises;
+const { promises: fs } = require("node:fs");
 const common = require("../common");
 const { sitemapCollection: smc } = require("../common/sitemap");
 const base = require("./_base");

package/lib/input-generators/textfile.js

@@ -8,7 +8,7 @@
  * Licensed under the MIT license.
  */
 
-const fs = require("fs").promises;
+const { promises: fs } = require("node:fs");
 const base = require("./_base");
 const common = require("../common");
 
@@ -50,7 +50,7 @@ function generateInput (options) {
     .then(data => {
       if (data) {
         let error;
-        data.toString().split('\n').every(line => {
+        data.toString().split("\n").every(line => {
           const page = line.replace(/^\s+|\s+$/g, "");
 
           if (!base.input(options, page)) {

package/lib/playwright/index.js

@@ -0,0 +1,125 @@
+#!/usr/bin/env -S node --unhandled-rejections=strict
+/**
+ * playwright process entry point
+ *
+ * Command line arguments:
+ * 0   - /path/to/node
+ * 1   - *this script*
+ * 2   - outputFile
+ * 3   - url
+ * 4   - selector
+ * [5] - timeout
+ * [6] - filter
+ * [7] - debug
+ * [8] - slowMo
+ * [9] - arbitrary launchOptions (JSON)
+ *
+ * launchOptions may include a `browserType` key ("chromium" | "firefox" | "webkit")
+ * to select the browser engine. Defaults to "chromium".
+ *
+ * Copyright (c) 2013 - 2025 Alex Grant, LocalNerve, contributors
+ * Licensed under the MIT license.
+ */
+const path = require("node:path");
+const { promises: fs } = require("node:fs");
+const playwright = require("playwright");
+
+if (process.argv.length < 5) {
+  throw Error(`playwright script '${
+    process.argv[1]
+  }' expected these arguments:\n\
+  OUTPUTFILE URL SELECTOR [TIMEOUT FILTER DEBUG SLOWMO LAUNCHOPTIONS]`);
+}
+
+const launchOpts = process.argv[9] ? JSON.parse(process.argv[9]) : null;
+
+// browserType is a playwright concern, not a launch option — extract it before
+// passing the rest through to playwright.launch()
+const { browserType: browserTypeKey = "chromium", ...extraLaunchOpts } = launchOpts ?? {};
+
+if (!["chromium", "firefox", "webkit"].includes(browserTypeKey)) {
+  throw Error(`playwright script: unsupported browserType '${browserTypeKey}'. \
+Must be one of: chromium, firefox, webkit`);
+}
+
+const options = {
+  outputFile: process.argv[2],
+  url: process.argv[3],
+  selector: process.argv[4],
+  timeout: process.argv[5] ? parseInt(process.argv[5], 10) : 10000,
+  filter: process.argv[6] && process.argv[6] !== "false" ? process.argv[6] : null,
+  debug: process.argv[7] && process.argv[7].toLowerCase() === "true",
+  slowMo: process.argv[8] ? parseInt(process.argv[8], 10) : 500,
+};
+
+const launchOptions = {
+  ...(options.debug ? {
+    headless: false,
+    slowMo: options.slowMo,
+    devtools: browserTypeKey === "chromium"  // devtools only supported in chromium
+  } : {
+    headless: true
+  }),
+  ...extraLaunchOpts
+};
+
+/**
+ * Run the snapshot script.
+ *
+ * @returns {Promise<string>} on completion
+ */
+async function run () {
+  let timeLeft, browser;
+  const start = Date.now();
+
+  timeLeft = options.timeout - 200; // leave some room for reporting
+  if (timeLeft < 0) timeLeft = 0;
+
+  try {
+    const filter =
+      options.filter ? require(options.filter) : passthru => passthru;
+
+    const browserType = playwright[browserTypeKey];
+    browser = await browserType.launch(launchOptions);
+    const page = await browser.newPage();
+
+    await page.goto(options.url, {
+      timeout: timeLeft
+    });
+
+    timeLeft -= (Date.now() - start);
+    if (timeLeft < 0) timeLeft = 0;
+
+    await fs.mkdir(path.dirname(options.outputFile), {
+      recursive: true
+    });
+
+    await page.waitForSelector(options.selector, {
+      state: "visible",
+      timeout: timeLeft
+    });
+
+    const content = await page.content();
+    await fs.writeFile(options.outputFile, filter(content));
+    const elapsed = Date.now() - start;
+
+    await browser.close();
+
+    return `snapshot for ${options.url} finished in ${elapsed} ms\n\
+  written to ${options.outputFile}`;
+  } catch (e) {
+    const err = new Error(`failed to snapshot ${options.url}`, {
+      cause: e
+    });
+    if (browser) {
+      browser.close().finally(() => {
+        throw err;
+      });
+    }
+    throw err;
+  }
+}
+
+run().then(console.log).catch(e => {
+  throw e;
+});

package/lib/playwright/removeScripts.js

@@ -0,0 +1,11 @@
+/**
+ * Remove all script tags from content
+ * 
+ * Copyright (c) 2013 - 2025 Alex Grant, LocalNerve, contributors
+ * Licensed under the MIT license. 
+ */
+function removeScriptTags (content) {
+  return content.replace(/<script\b[^<]*(?:(?!<\/script>)<[^<]*)*<\/script>/gi, "");
+}
+
+module.exports = removeScriptTags;

package/lib/puppeteer/index.js

@@ -17,9 +17,9 @@
  * Copyright (c) 2013 - 2025 Alex Grant, LocalNerve, contributors
  * Licensed under the MIT license.
  */
-const puppeteer = require('puppeteer');
-const path = require('path');
-const { promises: fs } = require('fs');
+const path = require("node:path");
+const { promises: fs } = require("node:fs");
+const puppeteer = require("puppeteer");
 
 if (process.argv.length < 5) {
   throw Error(`puppeteer script '${
@@ -33,8 +33,8 @@ const options = {
   url: process.argv[3],
   selector: process.argv[4],
   timeout: process.argv[5] ? parseInt(process.argv[5], 10) : 10000,
-  filter: process.argv[6] && process.argv[6] !== 'false' ? process.argv[6] : null,
-  debug: process.argv[7] && process.argv[7].toLowerCase() === 'true' ? true : false,
+  filter: process.argv[6] && process.argv[6] !== "false" ? process.argv[6] : null,
+  debug: process.argv[7] && process.argv[7].toLowerCase() === "true" ? true : false,
   slowMo: process.argv[8] ? parseInt(process.argv[8], 10) : 500,
   launchOpts: process.argv[9] ? JSON.parse(process.argv[9]) : null
 };
@@ -44,7 +44,7 @@ const launchOptions = {
     slowMo: options.slowMo,
     devtools: true
   } : {
-    headless: 'new'
+    headless: true
   }),
   ...(options.launchOpts ?? {})
 };
@@ -79,7 +79,7 @@ async function run () {
     });
 
     await page.waitForSelector(options.selector, {
-      Visible: true,
+      visible: true,
       timeout: timeLeft
     });
 

package/lib/puppeteer/removeScripts.js

@@ -5,7 +5,7 @@
  * Licensed under the MIT license. 
  */
 function removeScriptTags (content) {
-  return content.replace(/<script\b[^<]*(?:(?!<\/script>)<[^<]*)*<\/script>/gi, '');
+  return content.replace(/<script\b[^<]*(?:(?!<\/script>)<[^<]*)*<\/script>/gi, "");
 }
 
 module.exports = removeScriptTags;

package/package.json

@@ -1,13 +1,13 @@
 {
   "name": "html-snapshots",
-  "version": "5.10.0",
+  "version": "6.0.0",
   "author": {
     "name": "Alex Grant",
     "email": "alex@localnerve.com",
     "url": "https://www.localnerve.com"
   },
   "homepage": "https://github.com/localnerve/html-snapshots",
-  "description": "A selector-based html snapshot tool using Puppeteer or PhantomJS that sources sitemap.xml, robots.txt, or arbitrary input",
+  "description": "A selector-based html snapshot tool using Puppeteer or Playwright that sources sitemap.xml, robots.txt, or arbitrary input",
   "contributors": [
     {
       "name": "Alex Grant",
@@ -19,20 +19,22 @@
   },
   "scripts": {
     "lint": "eslint .",
-    "test": "mocha test/mocha/**/*.js --exit --recursive --reporter spec",
-    "test:async": "mocha test/mocha/async/test.js --exit --reporter spec",
-    "test:common": "mocha test/mocha/common/*.js --exit --reporter spec",
-    "test:common:debug": "mocha --inspect-brk test/mocha/common/*.js --exit --reporter spec",
-    "test:browsers": "mocha test/mocha/browsers/test.js --exit --reporter spec",
-    "test:browsers:cover": "c8 -- npm run test:browsers",
-    "test:input-generators": "mocha test/mocha/input-generators/*.js --exit --reporter spec",
-    "test:input-generators:debug": "mocha --inspect-brk test/mocha/input-generators/*.js --exit --reporter spec",
-    "test:html-snapshots": "mocha test/mocha/html-snapshots/*.js --exit --reporter spec",
-    "test:html-snapshots:robots": "mocha test/mocha/html-snapshots/*.js --grep \"robots\" --exit --reporter spec",
-    "test:html-snapshots:puppeteer": "mocha test/mocha/html-snapshots/*.js --grep \"puppeteer specific\" --exit --reporter spec",
-    "test:html-snapshots:debug": "mocha --inspect-brk test/mocha/html-snapshots/*.js --exit --reporter spec",
-    "test:cover": "c8 -- npm test",
-    "validate": "npm ls"
+    "_cleancov": "node -e \"const fs=require('fs');fs.rmSync('coverage',{recursive:true,force:true});fs.mkdirSync('coverage',{recursive:true})\"",
+    "_test": "node --test --experimental-test-coverage --test-reporter=spec --test-reporter-destination=stdout --test-reporter=lcov --test-reporter-destination=\"coverage/lcov.info\"",
+    "pretest": "npm run _cleancov",
+    "test": "npm run _test -- --test-coverage-exclude=\"test/**\" --test-concurrency=1 \"test/suites/**/*.spec.js\"",
+    "pretest:24": "npm run _cleancov",
+    "test:24": "npm run _test -- --test-coverage-exclude=\"test/**\" --test-isolation=none \"test/suites/**/*.spec.js\"",
+    "pretest:async": "npm run _cleancov",
+    "test:async": "npm run _test -- \"test/suites/async/*.spec.js\" --test-coverage-exclude=\"test/**\"",
+    "pretest:common": "npm run _cleancov",
+    "test:common": "npm run _test -- \"test/suites/common/*.spec.js\" --test-coverage-exclude=\"test/**\"",
+    "pretest:browsers": "npm run _cleancov",
+    "test:browsers": "npm run _test -- \"test/suites/browsers/*.spec.js\" --test-coverage-exclude=\"test/**\"",
+    "pretest:input-generators": "npm run _cleancov",
+    "test:input-generators": "npm run _test -- \"test/suites/input-generators/*.spec.js\" --test-coverage-exclude=\"test/**\"",
+    "pretest:html-snapshots": "npm run _cleancov",
+    "test:html-snapshots": "node --test --experimental-test-coverage --test-reporter=spec --test-reporter-destination=stdout --test-reporter=lcov --test-reporter-destination=\"coverage/lcov.info\" \"test/suites/html-snapshots/*.spec.js\" --test-coverage-exclude=\"test/**\""
   },
   "main": "./lib/html-snapshots",
   "repository": {
@@ -52,30 +54,20 @@
   "dependencies": {
     "async": "3.2.6",
     "async-lock": "1.4.1",
-    "combine-errors": "3.0.3",
-    "phantomjs-prebuilt": "2.1.16",
-    "puppeteer": "^24.42.0",
+    "playwright": "1.60.0",
+    "puppeteer": "25.1.0",
     "xml2js": "0.6.2"
   },
   "devDependencies": {
-    "eslint": "^10.2.1",
     "@eslint/js": "^10.0.1",
-    "globals": "^17.5.0",
+    "eslint": "^10.4.1",
     "express": "^5.2.1",
-    "mocha": "^11.7.5",
-    "c8": "^11.0.0",
+    "globals": "^17.6.0",
     "server-destroy": "^1.0.1",
     "sitemap-xml": "^0.1.0"
   },
-  "overrides": {
-    "request": "npm:@cypress/request@3.0.10",
-    "diff": "^9.0.0",
-    "glob": "^13.0.6",
-    "serialize-javascript": "^7.0.5",
-    "brace-expansion": "^5.0.5"
-  },
   "license": "MIT",
   "engines": {
-    "node": ">=20"
+    "node": ">=22"
   }
 }

package/lib/phantom/customFilter.js

@@ -1,24 +0,0 @@
-/*
- * customFilter.js
- *
- * Produce a single snapshot for a web page.
- * Snapshot taken when selector is(:visible).
- * Applies a custom filter to the snapshot output.
- *
- * This is a phantomJS script that runs in phantomjs.
- *
- * Copyright (c) 2013 - 2025 Alex Grant, LocalNerve, contributors
- * Licensed under the MIT license.
- */
-var selectorDetect = require("./modules/selectorDetect");
-var cli = require("./modules/cli");
-var detectors = require("./modules/detectors");
-
-var options = cli.selector();
-var customFilter = require(options.module);
-
-selectorDetect(
-  options,
-  options.useJQuery ? detectors.jquerySelector : detectors.querySelector,
-  customFilter
-);

package/lib/phantom/default.js

@@ -1,21 +0,0 @@
-/*
- * default.js
- *
- * Produce a single snapshot for a web page.
- * Snapshot taken when a selector is detected visible.
- *
- * This is a phantomJS script that runs in phantomjs.
- *
- * Copyright (c) 2013 - 2025 Alex Grant, LocalNerve, contributors
- * Licensed under the MIT license.
- */
-var selectorDetect = require("./modules/selectorDetect");
-var cli = require("./modules/cli");
-var detectors = require("./modules/detectors");
-
-var options = cli.selector();
-
-selectorDetect(
-  options,
-  options.useJQuery ? detectors.jquerySelector : detectors.querySelector
-);

package/lib/phantom/modules/cli.js

@@ -1,100 +0,0 @@
-/*
- * cli.js
- *
- * Handles phantomjs command line interface.
- * Reads the command line arguments and returns options.
- *
- * This is a module for a phantomJS script that runs in phantomjs.
- *
- * Copyright (c) 2013 - 2025 Alex Grant, LocalNerve, contributors
- * Licensed under the MIT license.
- */
-
-var system = require("system");
-var globals = require("./globals");
-
-module.exports = {
-
-  /**
-   * command line argument handling for selector based snapshot scripts
-   *  args: PHANTOM_SCRIPT OUTPUTFILE URL SELECTOR [TIMEOUT INTERVAL USEJQUERY FILTER]
-   */
-  selector: function () {
-
-    var options = {};
-    var defaults = {
-
-      // outputFile (required)
-      // The file to write the snapshot to
-      // encoding is the same as served from url
-      outputFile: "snapshot.html",
-
-      // url (required)
-      // The url to produce the snapshot from
-      url: "http://localhost/",
-
-      // selector (required)
-      // The selector to wait for before taking the snapshot
-      // When the selector is visible, this defines document readiness
-      selector: "body",
-
-      // timeout (optional)
-      // The maximum amount of time (ms) to wait before giving up on the snapshot
-      timeout: 10000,
-
-      // checkInterval (optional)
-      // The frequency (ms) to check for document readiness
-      checkInterval: 250,
-
-      // useJQuery (optional)
-      // flag to indicate use jQuery selector or not. jQuery must already exist in page.
-      useJQuery: false,
-
-      // verbose (optional)
-      // flag to indicate verbose output is desired.
-      verbose: false
-
-      // module (optional)
-      // An external module to load
-      // (default undefined)
-    };
-
-    if (system.args.length < 4) {
-      globals.exit(3, "phantomjs script '"+system.args[0]+"' expected these arguments: OUTPUTFILE URL SELECTOR [TIMEOUT INTERVAL JQUERY MODULE]");
-    } else {
-
-      options.outputFile = system.args[1];
-      options.url = system.args[2];
-      options.selector = system.args[3];
-      if (system.args[4]) {
-        options.timeout = parseInt(system.args[4], 10);
-      }
-      if (system.args[5]) {
-        options.checkInterval = parseInt(system.args[5], 10);
-      }
-      if (system.args[6]) {
-        var useJQuery = system.args[6].toLowerCase();
-        options.useJQuery =
-          (useJQuery === "true" || useJQuery === "yes" || useJQuery === "1");
-      }
-      if (system.args[7]) {
-        var verbose = system.args[7].toLowerCase();
-        options.verbose =
-          (verbose === "true" || verbose === "yes" || verbose === "1");
-      }
-      if (system.args[8] && system.args[8] !== "undefined") {
-        options.module = system.args[8];
-      }
-
-      // ensure defaults, replaces false with false in two cases (useJQuery, verbose)
-      Object.keys(defaults).forEach(function(prop) {
-        if (!options[prop]) {
-          options[prop] = defaults[prop];
-        }
-      });
-    }
-
-    return options;
-
-  }
-};

package/lib/phantom/modules/detectors.js

@@ -1,48 +0,0 @@
-/*
- * detectors.js
- *
- * A collection of detector functions that detect when a page is ready for a snapshot.
- *
- * The detector functions are run in PhantomJS page.evaluate in the context of a webpage,
- * in a sandboxed execution context.
- * Other special rules apply: http://phantomjs.org/api/webpage/method/evaluate.html
- *
- * The detector functions return true when the page is ready, false if not.
- *
- * This is a module for a phantomJS script that runs in phantomjs.
- *
- * Copyright (c) 2013 - 2025 Alex Grant, LocalNerve, contributors
- * Licensed under the MIT license.
- */
-/* global document, $ */
-module.exports = {
-
-  // Use plain Web API detection, with standard selectors
-  // https://developer.mozilla.org/en-US/docs/Web/Guide/CSS/Getting_Started/Selectors
-  // Returns true on element visibility
-  querySelector: function(options) {
-    var result = false;
-    var el = document.querySelector(options.selector);
-
-    if (el) {
-      // el must be an HTMLElement that is visible
-      result = el.offsetWidth && el.offsetHeight;
-    }
-
-    return result;
-  },
-
-  // Use JQuery selectors
-  // Returns true on element visibility
-  jquerySelector: function(options) {
-    var result = false;
-
-    if (typeof $ !== "undefined") {
-      // the definition of document readiness is visibility
-      result = $(options.selector).is(":visible");
-    }
-
-    return result;
-  }
-
-};

package/lib/phantom/modules/globals.js

@@ -1,43 +0,0 @@
-/*
- * globals.js
- *
- * Globals and module utilities.
- *
- * This is a module for a phantomJS script that runs in phantomjs.
- *
- * Copyright (c) 2013 - 2025 Alex Grant, LocalNerve, contributors
- * Licensed under the MIT license.
- */
-/* global phantom */
-/**
- * Exit phantomJS
- */
-function _exit(code, msg) {
-  if (code !== 0) {
-    console.error(msg);
-  } else {
-    console.log(msg);
-  }
-  phantom.exit(code);
-}
-
-/**
- * Global error handler
- */
-phantom.onError = function(msg, trace) {
-  var msgStack = ["PhantomJS error: " + msg];
-  if (trace && trace.length) {
-    msgStack.push("Trace:");
-    trace.forEach(function(t) {
-      msgStack.push(" -> " + (t.file || t.sourceURL) + ": " + t.line + (t.function ? " (in function " + t.function +")" : ""));
-    });
-  }
-  // Don't call _exit from here, causes strange crash w/linux rebuilt binary.
-  // https://github.com/ariya/phantomjs/issues/14132
-  console.log(msgStack.join("\n"));
-  phantom.exit(-1);
-};
-
-module.exports = {
-  exit: _exit
-};