extwee 2.3.2 → 2.3.4

package/src/Twine1HTML/parse-web.js

@@ -0,0 +1,255 @@
+import Passage from '../Passage.js';
+import { Story } from '../Story.js';
+
+/**
+ * Lightweight HTML parser for web builds - specifically for Twine 1 HTML parsing
+ * This replaces node-html-parser to reduce bundle size
+ */
+class LightweightTwine1Parser {
+  constructor(html) {
+    this.html = html;
+    this.doc = null;
+    
+    // Parse HTML using browser's native DOMParser if available, otherwise fallback
+    if (typeof DOMParser !== 'undefined') {
+      const parser = new DOMParser();
+      this.doc = parser.parseFromString(html, 'text/html');
+    } else {
+      // Fallback for environments without DOMParser
+      this.doc = this.createSimpleDOM(html);
+    }
+  }
+
+  querySelector(selector) {
+    if (this.doc && this.doc.querySelector) {
+      return this.doc.querySelector(selector);
+    }
+    
+    // Simple fallback implementation
+    if (selector === '#storeArea') {
+      const match = this.html.match(/<div[^>]*id=["']storeArea["'][^>]*>/i);
+      return match ? { found: true } : null;
+    }
+    if (selector === '#store-area') {
+      const match = this.html.match(/<div[^>]*id=["']store-area["'][^>]*>/i);
+      return match ? { found: true } : null;
+    }
+    return null;
+  }
+
+  querySelectorAll(selector) {
+    if (this.doc && this.doc.querySelectorAll) {
+      return Array.from(this.doc.querySelectorAll(selector));
+    }
+    
+    // Fallback implementation for [tiddler] elements
+    if (selector === '[tiddler]') {
+      return this.extractTiddlerElements();
+    }
+    return [];
+  }
+
+  extractTiddlerElements() {
+    const tiddlerRegex = /<div[^>]*tiddler=["']([^"']+)["'][^>]*>([\s\S]*?)<\/div>/gi;
+    const elements = [];
+    let match;
+
+    while ((match = tiddlerRegex.exec(this.html)) !== null) {
+      const elementHtml = match[0];
+      const attributes = this.parseAttributes(elementHtml);
+      const textContent = this.extractTextContent(match[2]);
+
+      elements.push({
+        attributes,
+        rawText: textContent
+      });
+    }
+
+    return elements;
+  }
+
+  parseAttributes(elementHtml) {
+    const attributes = {};
+    
+    // Extract tiddler attribute
+    const tiddlerMatch = elementHtml.match(/tiddler=["']([^"']+)["']/i);
+    if (tiddlerMatch) {
+      attributes.tiddler = tiddlerMatch[1];
+    }
+
+    // Extract tags attribute
+    const tagsMatch = elementHtml.match(/tags=["']([^"']*)["']/i);
+    if (tagsMatch) {
+      attributes.tags = tagsMatch[1];
+    }
+
+    // Extract twine-position attribute
+    const positionMatch = elementHtml.match(/twine-position=["']([^"']+)["']/i);
+    if (positionMatch) {
+      attributes['twine-position'] = positionMatch[1];
+    }
+
+    // Extract modifier attribute
+    const modifierMatch = elementHtml.match(/modifier=["']([^"']+)["']/i);
+    if (modifierMatch) {
+      attributes.modifier = modifierMatch[1];
+    }
+
+    return attributes;
+  }
+
+  extractTextContent(html) {
+    // Remove HTML tags and decode basic entities
+    return html
+      .replace(/<[^>]*>/g, '') // Remove HTML tags
+      .replace(/&lt;/g, '<')
+      .replace(/&gt;/g, '>')
+      .replace(/&quot;/g, '"')
+      .replace(/&#39;/g, "'")
+      .replace(/&amp;/g, '&') // This should be last
+      .trim();
+  }
+
+  createSimpleDOM(html) {
+    // Minimal DOM-like object for fallback
+    return {
+      querySelector: (selector) => {
+        if (selector === '#storeArea' && html.includes('id="storeArea"')) {
+          return { found: true };
+        }
+        if (selector === '#store-area' && html.includes('id="store-area"')) {
+          return { found: true };
+        }
+        return null;
+      },
+      querySelectorAll: (selector) => {
+        if (selector === '[tiddler]') {
+          return this.extractTiddlerElements();
+        }
+        return [];
+      }
+    };
+  }
+}
+
+/**
+ * Web-optimized Twine 1 HTML parser with reduced dependencies
+ * Parses Twine 1 HTML into a Story object using lightweight DOM parsing
+ * @see {@link https://github.com/iftechfoundation/twine-specs/blob/master/twine-1-htmloutput-doc.md Twine 1 HTML Documentation}
+ * @function parse
+ * @param {string} content - Twine 1 HTML content to parse.
+ * @returns {Story} Story object
+ */
+function parse(content) {
+  // Create a default Story.
+  const s = new Story();
+
+  // Use lightweight parser for web builds
+  const dom = new LightweightTwine1Parser(content);
+
+  // Look for `<div id="storeArea">`.
+  let storyData = dom.querySelector('#storeArea');
+
+  // Does the `<div id="storeArea">` element exist?
+  if (storyData === null) {
+    // Look for `<div id="store-area">`.
+    storyData = dom.querySelector('#store-area');
+    // Check for null
+    if (storyData == null) {
+      // Can't find any story data.
+      throw new Error('Cannot find #storeArea or #store-area!');
+    }
+  }
+
+  // Pull out the `[tiddler]` elements.
+  const storyPassages = dom.querySelectorAll('[tiddler]');
+
+  // Move through the passages.
+  for (const passage of storyPassages) {
+    // Get the passage attributes.
+    const attr = passage.attributes;
+    // Get the passage text.
+    const text = passage.rawText;
+
+    /**
+     * twine-position: (string) Required.
+     *   Comma-separated X and Y coordinates of the passage within Twine 1.
+     */
+    // Set a default position.
+    let position = null;
+    // Does position exist?
+    if (Object.prototype.hasOwnProperty.call(attr, 'twine-position')) {
+      // Update position.
+      position = attr['twine-position'];
+    }
+
+    /**
+     * tiddler: (string) Required.
+     *   The name of the passage.
+     */
+    // Create a default value.
+    const name = attr.tiddler;
+    // Is this `StoryTitle`?
+    if (name === 'StoryTitle') {
+      // If StoryTitle exists, we accept the story name.
+      s.name = text;
+    }
+
+    /**
+     * tags: (string) Required.
+     *  Space-separated list of passages tags, if any.
+     */
+    // Create empty tag array.
+    let tags = [];
+    // Does the tags attribute exist?
+    if (Object.prototype.hasOwnProperty.call(attr, 'tags')) {
+      // Escape any tags
+      // (Attributes can, themselves, be empty strings.)
+      if (attr.tags.length > 0 && attr.tags !== '""') {
+        // Escape the tags.
+        tags = attr.tags;
+        // Split by spaces into an array.
+        tags = tags.split(' ');
+      }
+
+      // Remove any empty strings.
+      tags = tags.filter(tag => tag !== '');
+    }
+
+    // Create metadata for passage.
+    // We translate Twine 1 attribute into Twine 2 metadata.
+    const metadata = {};
+
+    // Does position exist?
+    if (position !== null) {
+      // Add the property to metadata
+      metadata.position = position;
+    }
+
+    /**
+     * modifier: (string) Optional.
+     *   Name of the tool that last edited the passage.
+     *   Generally, for versions of Twine 1, this value will be "twee".
+     *   Twee compilers may place their own name (e.g. "tweego" for Tweego).
+     */
+    if (Object.prototype.hasOwnProperty.call(attr, 'modifier')) {
+      // In Twine 2, `creator` maps to Twine 1's `modifier`.
+      s.creator = attr.modifier;
+    }
+
+    // Add the passage.
+    s.addPassage(
+      new Passage(
+        name,
+        text,
+        tags,
+        metadata
+      )
+    );
+  }
+
+  // Return story object.
+  return s;
+}
+
+export { parse };

package/src/Twine2ArchiveHTML/parse-web.js

@@ -0,0 +1,134 @@
+import { parse as parseTwine2HTML } from '../Twine2HTML/parse-web.js';
+
+/**
+ * Lightweight HTML parser for web builds - specifically for Twine 2 Archive HTML parsing
+ * This replaces node-html-parser to reduce bundle size and uses browser DOM APIs
+ */
+class LightweightTwine2ArchiveParser {
+  constructor(html) {
+    this.html = html;
+    this.doc = null;
+    
+    // Parse HTML using browser's native DOMParser if available, otherwise fallback
+    if (typeof DOMParser !== 'undefined') {
+      const parser = new DOMParser();
+      this.doc = parser.parseFromString(html, 'text/html');
+    } else {
+      // Fallback for environments without DOMParser
+      this.doc = this.createSimpleDOM(html);
+    }
+  }
+
+  getElementsByTagName(tagName) {
+    if (this.doc && this.doc.getElementsByTagName) {
+      return Array.from(this.doc.getElementsByTagName(tagName));
+    }
+    
+    // Fallback implementation for tw-storydata elements
+    if (tagName === 'tw-storydata') {
+      return this.extractStoryDataElements();
+    }
+    return [];
+  }
+
+  extractStoryDataElements() {
+    // Match tw-storydata elements with their complete content
+    const storyDataRegex = /<tw-storydata[^>]*>[\s\S]*?<\/tw-storydata>/gi;
+    const elements = [];
+    let match;
+
+    while ((match = storyDataRegex.exec(this.html)) !== null) {
+      const outerHTML = match[0];
+      
+      elements.push({
+        outerHTML: outerHTML,
+        // For compatibility with the original parser interface
+        toString: () => outerHTML
+      });
+    }
+
+    return elements;
+  }
+
+  // eslint-disable-next-line no-unused-vars
+  createSimpleDOM(_htmlContent) {
+    // Minimal DOM-like object for fallback
+    return {
+      getElementsByTagName: (tagName) => {
+        if (tagName === 'tw-storydata') {
+          return this.extractStoryDataElements(this.htmlContent);
+        }
+        return [];
+      }
+    };
+  }
+}
+
+/**
+ * Web-optimized Twine 2 Archive HTML parser with reduced dependencies
+ * Parse Twine 2 Archive HTML and returns an array of story objects using browser DOM APIs.
+ * @see {@link https://github.com/iftechfoundation/twine-specs/blob/master/twine-2-archive-spec.md Twine 2 Archive Specification}
+ * @function parse
+ * @param {string} content - Content to parse for Twine 2 HTML elements.
+ * @throws {TypeError} - Content is not a string!
+ * @returns {Array} Array of stories found in content.
+ * @example
+ * const content = '<tw-storydata name="Untitled" startnode="1" creator="Twine" creator-version="2.3.9" ifid="A1B2C3D4-E5F6-G7H8-I9J0-K1L2M3N4O5P6" zoom="1" format="Harlowe" format-version="3.1.0" options="" hidden><style role="stylesheet" id="twine-user-stylesheet" type="text/twine-css"></style><script role="script" id="twine-user-script" type="text/twine-javascript"></script><tw-passagedata pid="1" name="Untitled Passage" tags="" position="0,0" size="100,100"></tw-passagedata></tw-storydata>';
+ * console.log(parse(content));
+ * // => [
+ * //     Story {
+ * //      name: 'Untitled',
+ * //      startnode: '1',
+ * //      creator: 'Twine',
+ * //      creatorVersion: '2.3.9',
+ * //      ifid: 'A1B2C3D4-E5F6-G7H8-I9J0-K1L2M3N4O5P6',
+ * //      zoom: '1',
+ * //      format: 'Harlowe',
+ * //      formatVersion: '3.1.0',
+ * //      options: '',
+ * //      hidden: '',
+ * //      passages: [
+ * //        Passage {
+ * //          pid: '1',
+ * //          name: 'Untitled Passage',
+ * //          tags: '',
+ * //          position: '0,0',
+ * //          size: '100,100',
+ * //          text: ''
+ * //        }
+ * //      ]
+ * //    }
+ * //   ]
+ */
+function parse(content) {
+  // Can only parse string values.
+  if (typeof content !== 'string') {
+    throw new TypeError('Content is not a string!');
+  }
+
+  // Use lightweight parser for web builds
+  const dom = new LightweightTwine2ArchiveParser(content);
+
+  // Array of possible story elements.
+  const outputArray = [];
+
+  // Pull out the `<tw-storydata>` element.
+  const storyDataElements = dom.getElementsByTagName('tw-storydata');
+
+  // Did we find any elements?
+  if (storyDataElements.length === 0) {
+    // Produce a warning if no Twine 2 HTML content is found.
+    console.warn('Warning: No Twine 2 HTML content found!');
+  }
+
+  // Iterate through all `<tw-storydata>` elements.
+  for (const storyElement of storyDataElements) {
+    // Convert element back into HTML text and parse using web-optimized parser.
+    outputArray.push(parseTwine2HTML(storyElement.outerHTML));
+  }
+
+  // Return array.
+  return outputArray;
+}
+
+export { parse };