@chialab/pdfjs-lib 1.0.0-alpha.36 → 1.0.0-alpha.38

package/dist/node/{chunk-FUWEGVHM.js → chunk-XAT2FF6L.js}

@@ -1,6 +1,7 @@
 import {
   BaseException,
   FeatureTest,
+  OPS,
   Util,
   shadow,
   stringToBytes,
@@ -1688,6 +1689,9 @@ 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/NodeFilterFactory.ts
 var filtersRegistry = /* @__PURE__ */ new Map();
@@ -2075,6 +2079,7 @@ export {
   rgbToHex,
   parseRgbaColor,
   randomUUID,
+  OPS_DICT,
   filtersRegistry,
   NodeFilterFactory,
   NodeCanvasFactory,

package/dist/node/index.js

@@ -3,9 +3,10 @@ import {
   FontInfo,
   MessageHandler,
   MurmurHash3_64,
+  bidi,
   convertBlackAndWhiteToRGBA,
   wrapReason
-} from "./chunk-4Y7OZIIX.js";
+} from "./chunk-FGBCMLPZ.js";
 import {
   BaseStandardFontDataFactory,
   CSSConstants,
@@ -20,6 +21,7 @@ import {
   NodeFilterFactory,
   NodeStandardFontDataFactory,
   NodeWasmFactory,
+  OPS_DICT,
   OutputScale,
   PDFDateString,
   PageViewport,
@@ -54,7 +56,7 @@ import {
   setLayerDimensions,
   stopEvent,
   toDataUrl
-} from "./chunk-FUWEGVHM.js";
+} from "./chunk-XAT2FF6L.js";
 import {
   AbortException,
   AnnotationBorderStyleType,
@@ -11488,14 +11490,14 @@ var require_opentype = __commonJS({
       };
       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);
@@ -39834,7 +39836,7 @@ function destroySvgContext(ctx) {
 
 // src/lib/PDFPageProxy.ts
 async function loadNodeCanvasFactory() {
-  const { NodeCanvasFactory: NodeCanvasFactory2 } = await import("./NodeUtils-WMSRRHQR.js");
+  const { NodeCanvasFactory: NodeCanvasFactory2 } = await import("./NodeUtils-U752HDWU.js");
   return new NodeCanvasFactory2({});
 }
 var getAnnotations = PDFPageProxy.prototype.getAnnotations;
@@ -40960,21 +40962,7 @@ var loadTextLayerFontsMap = /* @__PURE__ */ (() => {
     return promise;
   };
 })();
-var getTextDirection = (text) => {
-  const rtlChars = /[\u0591-\u07FF\uFB1D-\uFDFD\uFE70-\uFEFC]/;
-  const ltrChars = /[A-Za-z]/;
-  let rtlCount = 0;
-  let ltrCount = 0;
-  for (const char of text) {
-    if (rtlChars.test(char)) {
-      rtlCount++;
-    }
-    if (ltrChars.test(char)) {
-      ltrCount++;
-    }
-  }
-  return rtlCount > ltrCount ? "rtl" : "ltr";
-};
+var HYPHEN_REGEX2 = /-\n+$/;
 var isNameObject = (obj) => {
   return obj !== null && typeof obj === "object" && "name" in obj;
 };
@@ -41001,6 +40989,7 @@ 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];
@@ -41012,20 +41001,8 @@ async function createTextLayerV2(page, {
   let leading = 0;
   let textRise = 0;
   let textPosition = 0;
-  const createTextItem = (fontFamily2) => ({
-    glyphs: [],
-    font: page.commonObjs.get(fontFamily2)
-  });
-  const resetTextItem = () => {
-    currentTextItem = null;
-    textPosition = 0;
-  };
-  const closeTextItem = () => {
-    if (!currentTextItem?.glyphs.length) {
-      resetTextItem();
-      return;
-    }
-    const { font, glyphs: glyphsList } = currentTextItem;
+  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(
@@ -41034,10 +41011,40 @@ async function createTextLayerV2(page, {
     );
     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 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,
+      fontSize: finalFontSize,
+      top,
+      left,
+      angle
+    } = currentTextItem;
+    const textFont = fonts[font.bold || font.black ? font.italic ? 3 : 2 : font.italic ? 1 : 0];
     const glyphsBlocks = glyphsList.reduce(
       (acc, glyph) => {
         if (!glyph[0]) {
@@ -41095,30 +41102,60 @@ 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: getTextDirection(computedText),
+      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];
@@ -41138,6 +41175,7 @@ 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") {
@@ -41192,6 +41230,7 @@ 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];
@@ -41221,6 +41260,7 @@ async function createTextLayerV2(page, {
       case OPS.nextLineSetSpacingShowText: {
         closeTextItem();
         currentTextItem = createTextItem(fontFamily);
+        createParagraphIfNeeded(lastTextItem, currentTextItem);
         const [aw, ac, text] = args;
         wordSpacing = aw;
         charSpacing = ac;
@@ -41272,24 +41312,19 @@ async function createTextLayerV2(page, {
             normalizedRole = "figure";
             break;
         }
-        markedContent.children.push({
+        const currentMarkedContent = markedContent || rootContainer;
+        markedContent = {
           id: id2,
           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: {
@@ -41368,6 +41403,7 @@ export {
   InvalidPDFException,
   MathClamp,
   OPS,
+  OPS_DICT,
   OutputScale,
   PDFDataRangeTransport,
   PDFDateString,