@chialab/pdfjs-lib 1.0.0-alpha.35 → 1.0.0-alpha.37

package/dist/browser/index.js

@@ -30,6 +30,7 @@ import {
   VerbosityLevel,
   _isValidExplicitDest,
   assert,
+  bidi,
   convertBlackAndWhiteToRGBA,
   createValidAbsoluteUrl,
   fromBase64Util,
@@ -47,7 +48,7 @@ import {
   updateUrlHash,
   warn,
   wrapReason
-} from "./chunk-R7G44NYV.js";
+} from "./chunk-WUR3YYPJ.js";
 import {
   NodeCMapReaderFactory,
   NodeCanvasFactory,
@@ -27581,6 +27582,9 @@ function randomUUID() {
   hex[8] = hex[13] = hex[18] = hex[23] = "-";
   return hex.join("");
 }
+var OPS_DICT = Object.fromEntries(
+  Object.entries(OPS).map(([key, value]) => [value, key])
+);
 
 // src/lib/Svg.ts
 function isSvgElement(node) {
@@ -41427,14 +41431,14 @@ Font.prototype.updateFeatures = function(options) {
 };
 Font.prototype.stringToGlyphs = function(s, options) {
   var this$1 = this;
-  var bidi = new Bidi();
+  var bidi2 = new Bidi();
   var charToGlyphIndexMod = function(token) {
     return this$1.charToGlyphIndex(token.char);
   };
-  bidi.registerModifier("glyphIndex", null, charToGlyphIndexMod);
+  bidi2.registerModifier("glyphIndex", null, charToGlyphIndexMod);
   var features = options ? this.updateFeatures(options.features) : this.defaultRenderOptions.features;
-  bidi.applyFeatures(this, features);
-  var indexes = bidi.getTextGlyphs(s);
+  bidi2.applyFeatures(this, features);
+  var indexes = bidi2.getTextGlyphs(s);
   var length = indexes.length;
   var glyphs = new Array(length);
   var notdef = this.glyphs.get(0);
@@ -42318,6 +42322,10 @@ var loadTextLayerFontsMap = /* @__PURE__ */ (() => {
     return promise;
   };
 })();
+var HYPHEN_REGEX2 = /-\n+$/;
+var isNameObject = (obj) => {
+  return obj !== null && typeof obj === "object" && "name" in obj;
+};
 async function createTextLayerV2(page, {
   graphics,
   annotations: _annotations
@@ -42341,6 +42349,8 @@ async function createTextLayerV2(page, {
   const markedContentStack = [rootContainer];
   let markedContent = rootContainer;
   let currentTextItem = null;
+  let lastTextItem = null;
+  let transformMatrix = [1, 0, 0, 1, 0, 0];
   let textMatrix = [1, 0, 0, 1, 0, 0];
   let lineMatrix = [1, 0, 0, 1, 0, 0];
   let fontSize = 0;
@@ -42351,29 +42361,50 @@ async function createTextLayerV2(page, {
   let leading = 0;
   let textRise = 0;
   let textPosition = 0;
-  const createTextItem = (fontFamily2) => ({
-    glyphs: [],
-    font: page.commonObjs.get(fontFamily2)
-  });
+  const createTextItem = (fontFamily2) => {
+    const font = page.commonObjs.get(fontFamily2);
+    const textFont = fonts[font.bold || font.black ? font.italic ? 3 : 2 : font.italic ? 1 : 0];
+    const riseMatrix = [1, 0, 0, 1, 0, textRise];
+    const finalMatrix = Util.transform(
+      transformMatrix,
+      Util.transform(textMatrix, riseMatrix)
+    );
+    const [, , , , x, y] = Util.transform(transform, finalMatrix);
+    const finalFontSize = fontSize * Math.sqrt(finalMatrix[2] ** 2 + finalMatrix[3] ** 2);
+    const shift = -finalFontSize - finalFontSize * textFont.descender / textFont.unitsPerEm;
+    const angle = -Math.atan2(finalMatrix[1], finalMatrix[0]);
+    const top = y + shift * Math.cos(angle);
+    const left = x - shift * Math.sin(angle);
+    return {
+      glyphs: [],
+      font,
+      fontSize: finalFontSize,
+      top,
+      left,
+      angle
+    };
+  };
   const resetTextItem = () => {
     currentTextItem = null;
     textPosition = 0;
   };
   const closeTextItem = () => {
+    if (currentTextItem) {
+      lastTextItem = currentTextItem;
+    }
     if (!currentTextItem?.glyphs.length) {
       resetTextItem();
       return;
     }
-    const { font, glyphs: glyphsList } = currentTextItem;
+    const {
+      font,
+      glyphs: glyphsList,
+      fontSize: finalFontSize,
+      top,
+      left,
+      angle
+    } = currentTextItem;
     const textFont = fonts[font.bold || font.black ? font.italic ? 3 : 2 : font.italic ? 1 : 0];
-    const riseMatrix = [1, 0, 0, 1, 0, textRise];
-    const finalMatrix = Util.transform(textMatrix, riseMatrix);
-    const [, , , , x, y] = Util.transform(transform, finalMatrix);
-    const finalFontSize = fontSize * Math.sqrt(finalMatrix[2] ** 2 + finalMatrix[3] ** 2);
-    const angle = -Math.atan2(finalMatrix[1], finalMatrix[0]);
-    const shift = -finalFontSize - finalFontSize * textFont.descender / textFont.unitsPerEm;
-    const top = y + shift * Math.cos(angle);
-    const left = x - shift * Math.sin(angle);
     const glyphsBlocks = glyphsList.reduce(
       (acc, glyph) => {
         if (!glyph[0]) {
@@ -42391,12 +42422,14 @@ async function createTextLayerV2(page, {
     );
     const chunks = [];
     let currentLeft = 0;
+    let computedText = "";
     for (let i = 0; i < glyphsBlocks.length; i++) {
       const glyphs = glyphsBlocks[i];
       if (glyphs.length === 0) {
         continue;
       }
       const text = glyphs.map((g) => g[0].unicode).join("");
+      computedText += text;
       const textWidth = textFont.getAdvanceWidth(text, fontSize);
       if (!textWidth) {
         continue;
@@ -42429,30 +42462,60 @@ async function createTextLayerV2(page, {
         }
       }
       chunks.push({
-        text,
+        text: bidi(text).str,
         width: textWidth,
         margin: margin * finalFontSize,
         scale: graphicWidth / textWidth
       });
       currentLeft += margin + textWidth / fontSize;
     }
+    const dir = bidi(computedText).dir;
     markedContent.children.push({
       role: "text",
-      text: chunks,
+      text: dir === "ltr" ? chunks : chunks.reverse(),
       fontFamily: font.loadedName,
       fontFallback: "sans-serif",
-      hyphen: false,
+      hyphen: HYPHEN_REGEX2.test(computedText),
       top,
       left,
       fontSize: finalFontSize,
       fontWeight: font?.black ? 900 : font?.bold ? 700 : 400,
       fontStyle: font?.italic ? "italic" : "normal",
-      dir: "ltr",
+      dir,
       scale: 1,
       angle: angle * (180 / Math.PI)
     });
     resetTextItem();
   };
+  const createParagraphIfNeeded = (oldTextItem, newTextItem) => {
+    if (structTree) {
+      return;
+    }
+    if (oldTextItem && newTextItem && oldTextItem.fontSize === newTextItem.fontSize) {
+      return;
+    }
+    if (oldTextItem) {
+      closeMarkedContent();
+    }
+    const currentMarkedContent = markedContent || rootContainer;
+    markedContent = {
+      role: "p",
+      children: []
+    };
+    currentMarkedContent.children.push(markedContent);
+    markedContentStack.push(markedContent);
+  };
+  const closeMarkedContent = () => {
+    const oldMarkedContent = markedContentStack.pop();
+    markedContent = markedContentStack.at(-1) || rootContainer;
+    if (oldMarkedContent) {
+      if (["artifact", "figure", "formula"].includes(oldMarkedContent.role)) {
+        markedContent.children = markedContent.children.filter(
+          (c) => c !== oldMarkedContent
+        );
+      }
+    }
+  };
   for (let i = 0; i < operatorsList.fnArray.length; i++) {
     const fnId = operatorsList.fnArray[i];
     const args = operatorsList.argsArray[i];
@@ -42472,6 +42535,7 @@ async function createTextLayerV2(page, {
       case OPS.showText:
       case OPS.showSpacedText: {
         currentTextItem ?? (currentTextItem = createTextItem(fontFamily));
+        createParagraphIfNeeded(lastTextItem, currentTextItem);
         const [chars] = args;
         for (const char of chars) {
           if (typeof char === "number") {
@@ -42526,6 +42590,7 @@ async function createTextLayerV2(page, {
       case OPS.nextLineShowText: {
         closeTextItem();
         currentTextItem = createTextItem(fontFamily);
+        createParagraphIfNeeded(lastTextItem, currentTextItem);
         textMatrix[4] = lineMatrix[4];
         textMatrix[5] = textMatrix[1] * 0 + textMatrix[3] * -leading + textMatrix[5];
         lineMatrix = [...textMatrix];
@@ -42555,6 +42620,7 @@ async function createTextLayerV2(page, {
       case OPS.nextLineSetSpacingShowText: {
         closeTextItem();
         currentTextItem = createTextItem(fontFamily);
+        createParagraphIfNeeded(lastTextItem, currentTextItem);
         const [aw, ac, text] = args;
         wordSpacing = aw;
         charSpacing = ac;
@@ -42600,29 +42666,37 @@ async function createTextLayerV2(page, {
       case OPS.beginMarkedContentProps: {
         const [role, idRef, props] = args;
         const id2 = normalizeMarkedContentId(idRef);
-        markedContent.children.push({
+        let normalizedRole = isNameObject(role) ? role.name.toLowerCase() : role?.toString().toLowerCase();
+        switch (normalizedRole) {
+          case "placedgraphic":
+            normalizedRole = "figure";
+            break;
+        }
+        const currentMarkedContent = markedContent || rootContainer;
+        markedContent = {
           id: id2,
-          role: role?.toString().toLowerCase() || "span",
+          role: normalizedRole || "span",
           attrs: props ?? void 0,
           children: []
-        });
-        markedContent = markedContent.children.at(-1) || rootContainer;
+        };
+        currentMarkedContent.children.push(markedContent);
         markedContentStack.push(markedContent);
         break;
       }
       case OPS.endMarkedContent: {
-        const oldMarkedContent = markedContentStack.pop();
-        markedContent = markedContentStack.at(-1) || rootContainer;
-        if (["artifact", "figure", "formula"].includes(oldMarkedContent.role)) {
-          markedContent.children = markedContent.children.filter(
-            (c) => c !== oldMarkedContent
-          );
-        }
+        closeMarkedContent();
+        break;
+      }
+      case OPS.transform: {
+        closeTextItem();
+        const [a, b, c, d, e, f] = args;
+        transformMatrix = Util.transform(transformMatrix, [a, b, c, d, e, f]);
         break;
       }
       case OPS.restore: {
         const state = stateStack.pop();
         if (state) {
+          transformMatrix = state.transformMatrix;
           textMatrix = state.textMatrix;
           lineMatrix = state.lineMatrix;
           fontSize = state.fontSize;
@@ -42637,6 +42711,7 @@ async function createTextLayerV2(page, {
       }
       case OPS.save: {
         stateStack.push({
+          transformMatrix: [...transformMatrix],
           textMatrix: [...textMatrix],
           lineMatrix: [...lineMatrix],
           fontSize,
@@ -42688,6 +42763,7 @@ export {
   InvalidPDFException,
   MathClamp,
   OPS,
+  OPS_DICT,
   OutputScale,
   PDFDataRangeTransport,
   PDFDateString,