feat: add support for transparent groups, ensure endGroup would merge sub-canvas text/line/etc. back to primary output data. this completes the fix for #418 (#420)

Author: modesty

lib/pdf.js

@@ -359,7 +359,8 @@ export default class PDFJSClass extends EventEmitter {
           const scaledSpaceWidth = spaceWidth * textHScale;
 
           // Add spaces if gap is positive and significant (> 30% of scaled space width)
-          if (gap > scaledSpaceWidth * 0.3) {
+          // Also check that scaledSpaceWidth is valid to avoid division by zero
+          if (scaledSpaceWidth > 0 && gap > scaledSpaceWidth * 0.3) {
             const numSpaces = Math.round(gap / scaledSpaceWidth);
             prevText.str += ' '.repeat(Math.max(1, numSpaces));
           }

lib/pdfcanvas.js

@@ -436,8 +436,33 @@ export default class CanvasRenderingContext2D_ {
 	}
 
 	// Method signature matches Canvas API
+	// MQZ. 2025.01: Enhanced to support transparency groups which handled by base/display/canvas.js when invokes endGroup() 
+	// This handles transparency groups where text/lines/fills are rendered to a temporary canvas and then composited back to the main canvas via drawImage()
 	drawImage(image, var_args) {
-		//MQZ. no image drawing support for now
+		// If source has our custom data arrays (i.e., it's a PDFCanvas), merge them to this canvas
+		if (image && typeof image === "object") {
+			// Merge Texts array
+			if (image.Texts && Array.isArray(image.Texts) && image.Texts.length > 0) {
+				this.canvas.Texts.push(...image.Texts);
+				image.Texts = []; // Clear source after merging
+			}
+			// Merge HLines array
+			if (image.HLines && Array.isArray(image.HLines) && image.HLines.length > 0) {
+				this.canvas.HLines.push(...image.HLines);
+				image.HLines = [];
+			}
+			// Merge VLines array
+			if (image.VLines && Array.isArray(image.VLines) && image.VLines.length > 0) {
+				this.canvas.VLines.push(...image.VLines);
+				image.VLines = [];
+			}
+			// Merge Fills array
+			if (image.Fills && Array.isArray(image.Fills) && image.Fills.length > 0) {
+				this.canvas.Fills.push(...image.Fills);
+				image.Fills = [];
+			}
+		}
+		// Note: actual pixel drawing is not supported (no-op for regular images)
 	}
 
 	getImageData(x, y, w, h) {

package.json

@@ -1,6 +1,6 @@
 {
 	"name": "pdf2json",
-	"version": "4.0.1",
+	"version": "4.0.2",
 	"description": "PDF file parser that converts PDF binaries to JSON and text, powered by porting a fork of PDF.JS to Node.js",
 	"keywords": [
 		"pdf",
@@ -47,7 +47,7 @@
 		"parse-tb": "./bin/pdf2json.js -f ./test/pdf/misc/i242_testingWithTable.pdf -o ./test/target/misc",
 		"parse-tc": "./bin/pdf2json.js -f ./test/pdf/misc/i293_pdfpac.pdf -o ./test/target/misc",
 		"parse-rectFix": "./bin/pdf2json.js -f ./test/pdf/misc/pr298_rect_fix_from_upstream.pdf -o ./test/target/misc",
-		"parse-e": "./bin/pdf2json.js -f ./test/pdf/misc/i418_precompilato_fake.pdf -o ./test/target/misc",
+		"parse-e": "./bin/pdf2json.js -f ./test/pdf/misc/i418_precompilato_fake.pdf -o ./test/target/misc -c",
 		"build:rollup": "npx rollup -c ./rollup.config.js",
 		"build:bundle-pdfjs-base": "node rollup/bundle-pdfjs-base.js",
 		"build": "npm run build:bundle-pdfjs-base && npm run build:rollup",
@@ -121,4 +121,4 @@
 		"readme.md",
 		"license.txt"
 	]
-}
+}