From c82a04c49b0efdd2028aa3fa4b22d92a35a4b00c Mon Sep 17 00:00:00 2001
From: modesty <modestyz@hotmail.com>
Date: Sat, 4 Oct 2025 13:14:51 -0700
Subject: [PATCH] fix: #408: fix text block coordinates

---
 base/display/canvas.js |  7 +++----
 package.json           |  2 +-
 test/_test_.cjs        | 43 +++++++++++++++++++++++++++++++++++++++++-
 3 files changed, 46 insertions(+), 6 deletions(-)

diff --git a/base/display/canvas.js b/base/display/canvas.js
index b48cae6..4c473db 100755
--- a/base/display/canvas.js
+++ b/base/display/canvas.js
@@ -1269,10 +1269,6 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
 //              info(nodeUtil.inspect(glyphs));
 //          }
 
-
-        ctx.restore();
-      }
-
       // Text rendering for regular fonts (Type3 fonts are handled in their own context above)
       if (str && !font.disableFontFace && !font.coded) {
           var curFontSize = fontSize * scale * textHScale + 3;
@@ -1293,6 +1289,9 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
         }
       }
 
+        ctx.restore();
+      }
+
       return canvasWidth;
     },
     showSpacedText: function CanvasGraphics_showSpacedText(arr) {
diff --git a/package.json b/package.json
index beafd7c..d005e5a 100644
--- a/package.json
+++ b/package.json
@@ -1,6 +1,6 @@
 {
 	"name": "pdf2json",
-	"version": "3.2.2",
+	"version": "3.2.3",
 	"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",
diff --git a/test/_test_.cjs b/test/_test_.cjs
index c2f507a..ad1cc51 100644
--- a/test/_test_.cjs
+++ b/test/_test_.cjs
@@ -163,6 +163,46 @@ function checkResult_pageContent(Pages, fileName) {
 	});
 }
 
+function checkResult_textCoordinates(Pages, fileName) {
+	// Verify text block coordinates are unique (issue #408 regression test)
+	Pages.forEach((page, pageIndex) => {
+		const texts = page.Texts || [];
+		if (texts.length === 0) return; // Skip pages with no text
+
+		// Collect all coordinates
+		const coords = texts.map(t => ({ x: t.x, y: t.y }));
+
+		// Create unique coordinate strings
+		const uniqueCoords = new Set(coords.map(c => `${c.x},${c.y}`));
+
+		// Check that we have more than one unique coordinate if we have multiple text elements
+		// This prevents the regression where all text elements had identical coordinates (-0.25, 48.75)
+		if (texts.length > 5) {
+			assert(
+				uniqueCoords.size > 1,
+				fileName + " page " + pageIndex +
+				" : all " + texts.length + " text elements have identical coordinates. " +
+				"This is a regression of issue #408. Found only " + uniqueCoords.size +
+				" unique coordinate(s): " + Array.from(uniqueCoords).slice(0, 3).join(", ")
+			);
+		}
+
+		// Verify coordinates are reasonable (not all the same broken value)
+		texts.forEach((text, textIndex) => {
+			assert(
+				typeof text.x === 'number' && !isNaN(text.x),
+				fileName + " page " + pageIndex + " text " + textIndex +
+				" : has invalid x coordinate: " + text.x
+			);
+			assert(
+				typeof text.y === 'number' && !isNaN(text.y),
+				fileName + " page " + pageIndex + " text " + textIndex +
+				" : has invalid y coordinate: " + text.y
+			);
+		});
+	});
+}
+
 async function parseAndVerifyOnePDF(fileName, fromBuffer, pageCount) {
 	let timeoutId;
 	let pdfParser = null;
@@ -203,12 +243,13 @@ async function parseAndVerifyOnePDF(fileName, fromBuffer, pageCount) {
 		});
 
 		const evtData = await pdfParserDataReady;
-		
+
 		expect(evtData).toBeDefined();
 		checkResult_parseStatus(null, evtData, fileName);
 		checkResult_mainFields(evtData, fileName);
 		checkResult_pageCount(evtData.Pages, pageCount, fileName);
 		checkResult_pageContent(evtData.Pages, fileName);
+		checkResult_textCoordinates(evtData.Pages, fileName);
 	} catch (error) {
 		console.error(`Error parsing PDF ${fileName}: `, error);
 		throw error; // Re-throw to ensure Jest knows the test failed