@gmb/bitmark-parser-generator 5.27.0 → 5.29.0

package/dist/cli/main.js

@@ -1269,6 +1269,7 @@ var propertyKeys = {
   property_bubbleTag: "@bubbleTag",
   property_extractorTag: "@extractorTag",
   property_extractorInternal: "@extractorInternal",
+  property_bitFingerprint: "@bitFingerprint",
   property_buttonCaption: "@buttonCaption",
   property_callToActionUrl: "@callToActionUrl",
   property_caption: "@caption",
@@ -1330,6 +1331,8 @@ var propertyKeys = {
   property_externalLinkText: "@externalLinkText",
   property_feedbackEngine: "@feedbackEngine",
   property_feedbackType: "@feedbackType",
+  property_userFeedbackTranslationUrl: "@userFeedbackTranslationUrl",
+  property_userFeedbackContentUrl: "@userFeedbackContentUrl",
   property_flashcardSet: "@flashcardSet",
   property_focusX: "@focusX",
   property_focusY: "@focusY",
@@ -3785,16 +3788,25 @@ var CARDSETS = {
               },
               {
                 key: ConfigKey.property_tableRowSpan,
+                // HTML-C2: contribute a `rowspan` attribute onto the enclosing
+                // cell element (`<th>`/`<td>`).
+                htmlKey: { "@el": { "@attr": { rowspan: "$" } } },
                 description: "Number of rows the cell spans.",
                 format: TagFormat.number
               },
               {
                 key: ConfigKey.property_tableColSpan,
+                // HTML-C2: contribute a `colspan` attribute onto the enclosing
+                // cell element (`<th>`/`<td>`).
+                htmlKey: { "@el": { "@attr": { colspan: "$" } } },
                 description: "Number of columns the cell spans.",
                 format: TagFormat.number
               },
               {
                 key: ConfigKey.property_tableScope,
+                // HTML-C2: contribute a `scope` attribute onto the enclosing
+                // cell element (`<th>`/`<td>`).
+                htmlKey: { "@el": { "@attr": { scope: "$" } } },
                 description: "Scope attribute for header cells.",
                 format: TagFormat.plainText
               },
@@ -3912,6 +3924,9 @@ var CARDSETS = {
                 key: ConfigKey.property_tableRowSpan,
                 jsonKey: "rowspan",
                 exportJsonKey: { rowspan: "$" },
+                // HTML-C2: contribute a `rowspan` attribute onto the enclosing
+                // cell element (`<th>`/`<td>`).
+                htmlKey: { "@el": { "@attr": { rowspan: "$" } } },
                 description: "Number of rows the cell spans.",
                 format: TagFormat.number
               },
@@ -3919,6 +3934,9 @@ var CARDSETS = {
                 key: ConfigKey.property_tableColSpan,
                 jsonKey: "colspan",
                 exportJsonKey: { colspan: "$" },
+                // HTML-C2: contribute a `colspan` attribute onto the enclosing
+                // cell element (`<th>`/`<td>`).
+                htmlKey: { "@el": { "@attr": { colspan: "$" } } },
                 description: "Number of columns the cell spans.",
                 format: TagFormat.number
               },
@@ -3926,6 +3944,9 @@ var CARDSETS = {
                 key: ConfigKey.property_tableScope,
                 jsonKey: "scope",
                 exportJsonKey: { scope: "$" },
+                // HTML-C2: contribute a `scope` attribute onto the enclosing
+                // cell element (`<th>`/`<td>`).
+                htmlKey: { "@el": { "@attr": { scope: "$" } } },
                 description: "Scope attribute for header cells.",
                 format: TagFormat.plainText
               },
@@ -4532,6 +4553,12 @@ var GROUPS = {
         format: TagFormat.coordinates,
         maxCount: Count.infinity
       },
+      {
+        key: ConfigKey.property_bitFingerprint,
+        description: "Fingerprint(s) used by the AI extractor to identify the bit (font, style, color, pixels, etc.)",
+        format: TagFormat.plainText,
+        maxCount: Count.infinity
+      },
       {
         key: ConfigKey.property_levelCEFRp,
         description: "The CEFRp level for the bit",
@@ -5736,6 +5763,16 @@ var GROUPS = {
         description: "External Id of the translated book",
         format: TagFormat.plainText
       },
+      {
+        key: ConfigKey.property_userFeedbackTranslationUrl,
+        description: "URL to an external feedback form for book translation feedback",
+        format: TagFormat.plainText
+      },
+      {
+        key: ConfigKey.property_userFeedbackContentUrl,
+        description: "URL to an external feedback form for book content feedback",
+        format: TagFormat.plainText
+      },
       {
         key: ConfigKey.property_duration,
         description: "The duration of the book",
@@ -10115,7 +10152,18 @@ var BITS = {
     since: "5.12.0",
     baseBitType: BitType._standard,
     description: "Extractor blueprint bit, used to provide blueprint information about extractors",
-    textFormatDefault: TextFormat.json
+    textFormatDefault: TextFormat.json,
+    tags: [
+      {
+        key: ConfigKey.group_resourceBitTags,
+        description: "Resource bit tags for the blueprint reference image"
+      },
+      {
+        key: ConfigKey.group_resourceImage,
+        description: "Optional uncropped reference page image, used so authoring tools can compute crops against the original page dimensions"
+      }
+    ],
+    resourceAttachmentAllowed: false
   },
   [BitType.extractorInformation]: {
     since: "3.8.0",
@@ -10838,7 +10886,9 @@ var BITS = {
       {
         key: ConfigKey.property_caption,
         description: "Caption for the table image, used to provide a description for the image",
-        format: TagFormat.bitmarkText
+        format: TagFormat.bitmarkText,
+        // HTML-10 / HTML-C2: render as the table's `<caption>` (see table bit).
+        htmlKey: { "@el": "caption", "@children": "$" }
       },
       {
         key: ConfigKey.group_backgroundWallpaper,
@@ -12529,7 +12579,12 @@ var BITS = {
       {
         key: ConfigKey.property_caption,
         description: "Caption for the table, used to define the title of the table",
-        format: TagFormat.bitmarkText
+        format: TagFormat.bitmarkText,
+        // HTML-10 / HTML-C2: a bit-level caption renders as the table's
+        // `<caption>` — placed inside the idiomatic `<table>` container as
+        // its first child. Outside a table context it falls back to the
+        // residual carrier.
+        htmlKey: { "@el": "caption", "@children": "$" }
       },
       {
         key: ConfigKey.property_tableFixedHeader,
@@ -12614,7 +12669,9 @@ var BITS = {
       {
         key: ConfigKey.property_caption,
         description: "Caption for the table image, used to provide a description for the image",
-        format: TagFormat.bitmarkText
+        format: TagFormat.bitmarkText,
+        // HTML-10 / HTML-C2: render as the table's `<caption>` (see table bit).
+        htmlKey: { "@el": "caption", "@children": "$" }
       },
       {
         key: ConfigKey.group_backgroundWallpaper,
@@ -13804,7 +13861,7 @@ var instance2 = new Config();
 // src/generated/package_info.ts
 var PACKAGE_INFO = {
   "name": "@gmb/bitmark-parser-generator",
-  "version": "5.27.0",
+  "version": "5.29.0",
   "license": "ISC"};
 var Environment = {
   unknown: "",
@@ -14185,6 +14242,8 @@ var NodeType = {
   // bit
   bitType: "bitType",
   // bit type
+  bitFingerprint: "bitFingerprint",
+  bitFingerprintValue: "bitFingerprintValue",
   blockId: "blockId",
   blockIdValue: "blockIdValue",
   body: "body",
@@ -14662,6 +14721,10 @@ var NodeType = {
   relatedBookValue: "relatedBookValue",
   translationOfBook: "translationOfBook",
   translationOfBookValue: "translationOfBookValue",
+  userFeedbackTranslationUrl: "userFeedbackTranslationUrl",
+  userFeedbackTranslationUrlValue: "userFeedbackTranslationUrlValue",
+  userFeedbackContentUrl: "userFeedbackContentUrl",
+  userFeedbackContentUrlValue: "userFeedbackContentUrlValue",
   releaseDate: "releaseDate",
   releaseDateValue: "releaseDateValue",
   releaseKind: "releaseKind",
@@ -27625,6 +27688,12 @@ var Builder = class extends BaseBuilder {
         options
       ),
       sourceBB: this.normaliseSourceBB(data.sourceBB),
+      bitFingerprint: this.toAstProperty(
+        bitType,
+        ConfigKey.property_bitFingerprint,
+        data.bitFingerprint,
+        options
+      ),
       levelCEFRp: this.toAstProperty(
         bitType,
         ConfigKey.property_levelCEFRp,
@@ -27687,6 +27756,18 @@ var Builder = class extends BaseBuilder {
         data.translationOfBook,
         options
       ),
+      userFeedbackTranslationUrl: this.toAstProperty(
+        bitType,
+        ConfigKey.property_userFeedbackTranslationUrl,
+        data.userFeedbackTranslationUrl,
+        options
+      ),
+      userFeedbackContentUrl: this.toAstProperty(
+        bitType,
+        ConfigKey.property_userFeedbackContentUrl,
+        data.userFeedbackContentUrl,
+        options
+      ),
       author: this.toAstProperty(bitType, ConfigKey.property_author, data.author, options),
       subject: this.toAstProperty(bitType, ConfigKey.property_subject, data.subject, options),
       date: this.toAstProperty(bitType, ConfigKey.property_date, data.date, options),
@@ -31462,7 +31543,8 @@ var BitmarkGenerator = class extends AstWalkerGenerator {
       this.writeNL();
     }
     const content = cell.content ?? instance3.EMPTY_STRING;
-    this.writeTextOrValue(content, this.textFormat, TextLocation.body);
+    const options = this.options.noBreakscaping ? { noBreakscaping: true } : void 0;
+    this.writeTextOrValue(content, this.textFormat, TextLocation.body, options);
   }
   writeTableCellProperty(name, value) {
     this.writeOPA();
@@ -32937,6 +33019,8 @@ var MARK_TO_TAG = {
   highlight: "mark",
   code: "code"
 };
+var INLINE_IMAGE_RE = /==([^=]*(?:=(?!=)[^=]*)*)==\|imageInline:([^|\s]+)((?:\|(?:@?[A-Za-z][A-Za-z0-9_-]*:[^|\s]+|#[^|\s]*))*)\|?/g;
+var MAX_IMAGE_DIMENSION = 9999;
 var HtmlTableGenerator = class {
   indentUnit = "  ";
   /**
@@ -33053,6 +33137,8 @@ var HtmlTableGenerator = class {
       case "image":
       case "imageInline":
         return this.imageToHtml(node);
+      case "latex":
+        return this.latexToHtml(node);
       default:
         return this.inlineToHtml([node]);
     }
@@ -33063,8 +33149,9 @@ var HtmlTableGenerator = class {
   inlineNodeToHtml(node) {
     if (node.type === "hardBreak") return "<br>";
     if (node.type === "image" || node.type === "imageInline") return this.imageToHtml(node);
+    if (node.type === "latex") return this.latexToHtml(node);
     if (node.type === "text") {
-      return this.wrapMarks(this.escapeText(node.text ?? ""), node.marks);
+      return this.wrapMarks(this.textToHtml(node.text ?? ""), node.marks);
     }
     if (node.content) return this.inlineToHtml(node.content);
     return this.escapeText(node.text ?? "");
@@ -33142,6 +33229,44 @@ var HtmlTableGenerator = class {
     }
     return `<img ${parts.join(" ")}>`;
   }
+  textToHtml(text) {
+    let html = "";
+    let lastIndex = 0;
+    for (const match of text.matchAll(INLINE_IMAGE_RE)) {
+      const index = match.index ?? 0;
+      html += this.escapeText(text.slice(lastIndex, index));
+      html += this.inlineImageTokenToHtml(match);
+      lastIndex = index + match[0].length;
+    }
+    html += this.escapeText(text.slice(lastIndex));
+    return html;
+  }
+  inlineImageTokenToHtml(match) {
+    const attrs = {
+      alt: match[1],
+      src: match[2]
+    };
+    const rawAttrs = match[3] ?? "";
+    for (const attr of rawAttrs.split("|")) {
+      if (!attr || attr.startsWith("#")) continue;
+      const colon = attr.indexOf(":");
+      if (colon < 1) continue;
+      const key = attr.slice(0, colon).replace(/^@/, "");
+      const value = attr.slice(colon + 1);
+      if (key === "width" || key === "height") {
+        const n = Number.parseInt(value, 10);
+        if (!Number.isNaN(n) && n > 0 && n <= MAX_IMAGE_DIMENSION) attrs[key] = n;
+      } else if (key === "class" || key === "title") {
+        attrs[key] = value;
+      }
+    }
+    return this.imageToHtml({ type: "imageInline", attrs });
+  }
+  // --- formulas -------------------------------------------------------------
+  latexToHtml(node) {
+    const formula = node.attrs?.formula ?? "";
+    return `<math alttext="${this.escapeAttr(formula)}"></math>`;
+  }
   // --- escaping -------------------------------------------------------------
   escapeText(text) {
     return text.replace(/&/g, "&amp;").replace(/</g, "&lt;").replace(/>/g, "&gt;");
@@ -33450,6 +33575,7 @@ var JsonGenerator = class extends AstWalkerGenerator {
         this.addProperty(this.bitJson, k, values, { array: true });
       }
     }
+    return false;
   }
   // bitmarkAst -> bits -> bitsValue -> cardNode -> cardBits -> cardBitsValue
   enter_cardBitsValue(node, route) {
@@ -42381,7 +42507,7 @@ var IMPLIED_CLOSE = {
   p: /* @__PURE__ */ new Set(["p"]),
   option: /* @__PURE__ */ new Set(["option"])
 };
-var TAG_RE = /<(\/)?([a-zA-Z][a-zA-Z0-9-]*)((?:\s+(?:"[^"]*"|'[^']*'|[^<>"'])*)?)(\/)?>/g;
+var TAG_RE = /<(\/)?([a-zA-Z][a-zA-Z0-9:-]*)((?:\s+(?:"[^"]*"|'[^']*'|[^<>"'])*)?)(\/)?>/g;
 var ATTR_RE = /([a-zA-Z_:][a-zA-Z0-9:._-]*)(?:\s*=\s*("[^"]*"|'[^']*'|[^\s"'>]+))?/g;
 function parseAttrs(raw) {
   const attrs = {};
@@ -42710,6 +42836,9 @@ var HtmlTableParser = class {
     if (name === "img") {
       return [this.imageNode(el, true)];
     }
+    if (this.isMathTag(name)) {
+      return [this.latexNode(el)];
+    }
     const markType = MARK_TAGS[name];
     if (markType) {
       return this.inlineNodes(el.children, [...marks, { type: markType }]);
@@ -42826,6 +42955,43 @@ var HtmlTableParser = class {
     if (el.attrs.class) attrs.class = el.attrs.class;
     return { type: inline ? "imageInline" : "image", attrs };
   }
+  // --- formulas -------------------------------------------------------------
+  isMathTag(name) {
+    return this.localName(name) === "math";
+  }
+  latexNode(el) {
+    return {
+      type: "latex",
+      attrs: {
+        formula: this.latexFromMath(el)
+      }
+    };
+  }
+  latexFromMath(el) {
+    const alt = el.attrs.alttext ?? el.attrs.alt ?? el.attrs.title;
+    if (alt != null) return decodeEntities(alt);
+    const annotation = this.findLatexAnnotation(el);
+    if (annotation) return this.textContent(annotation).trim();
+    return this.textContent(el).trim();
+  }
+  findLatexAnnotation(el) {
+    for (const child of el.children) {
+      if (!isElement(child)) continue;
+      const lowerEncoding = (child.attrs.encoding ?? "").toLowerCase();
+      if (this.localName(child.name) === "annotation" && (/\btex\b/.test(lowerEncoding) || /\blatex\b/.test(lowerEncoding))) {
+        return child;
+      }
+      const nested = this.findLatexAnnotation(child);
+      if (nested) return nested;
+    }
+    return void 0;
+  }
+  textContent(el) {
+    return el.children.map((child) => isElement(child) ? this.textContent(child) : decodeEntities(child.value)).join("").replace(/\s+/g, " ");
+  }
+  localName(name) {
+    return name.includes(":") ? name.slice(name.indexOf(":") + 1) : name;
+  }
   // --- whitespace -----------------------------------------------------------
   /** Trim leading/trailing whitespace from inline content (edges only). */
   trimInline(nodes) {
@@ -44723,7 +44889,10 @@ var BitmarkParserGenerator = class {
         outputFile: opts.outputFile,
         fileOptions: opts.fileOptions,
         jsonOptions: opts.jsonOptions,
-        bitmarkOptions: opts.bitmarkOptions
+        bitmarkOptions: {
+          ...opts.bitmarkOptions,
+          ...opts.noBreakscaping ? { noBreakscaping: true } : {}
+        }
       });
     }
     const outputFormat = opts.outputFormat ?? "html";
@@ -45085,7 +45254,7 @@ function createConvertHtmlTableCommand() {
     )
   ).addOption(
     new Option("--tableFormat <format>", "bit type to emit for HTML input").choices(TABLE_FORMAT_CHOICES).default("table-extended")
-  ).option("--keepUnknownTags", "keep unmappable HTML tags as literal text (HTML input only)").option("-p, --pretty", "prettify the JSON output with indent").option(
+  ).option("--keepUnknownTags", "keep unmappable HTML tags as literal text (HTML input only)").option("--noBreakscaping", "do not breakscape generated bitmark table cell content").option("-p, --pretty", "prettify the JSON output with indent").option(
     "--indent <indent>",
     "prettify indent (default: 2 when --pretty is set)",
     (v) => Number.parseInt(v, 10)
@@ -45101,6 +45270,7 @@ function createConvertHtmlTableCommand() {
           options.tableFormat
         ),
         keepUnknownTags: options.keepUnknownTags,
+        noBreakscaping: options.noBreakscaping,
         jsonOptions: {
           prettify
         }
@@ -45126,7 +45296,9 @@ Examples:
 
   $ bitmark-parser convertHtmlTable input.bitmark -o output.html
 
-  $ bitmark-parser convertHtmlTable --tableFormat table input.html`
+  $ bitmark-parser convertHtmlTable --tableFormat table input.html
+
+  $ bitmark-parser convertHtmlTable --noBreakscaping input.html`
   );
   return cmd;
 }