Keep the empty lines in the text fields

It fixes #18036.
mozilla · Oct 5, 2024 · 7dda9c7 · 7dda9c7
1 parent 6aef332
commit 7dda9c7
Show file tree

Hide file tree

Showing 4 changed files with 37 additions and 4 deletions.
diff --git a/src/core/annotation.js b/src/core/annotation.js
@@ -1292,6 +1292,7 @@ class Annotation {
         bbox,
         matrix
       );
+
       this.data.textContent = text;
     }
   }
@@ -2299,7 +2300,6 @@ class WidgetAnnotation extends Annotation {
     }
 
     assert(typeof value === "string", "Expected `value` to be a string.");
-    value = value.trim();
 
     if (this.data.combo) {
       // The value can be one of the exportValue or any other values.
@@ -2934,6 +2934,30 @@ class TextWidgetAnnotation extends WidgetAnnotation {
     return chunks;
   }
 
+  async extractTextContent(evaluator, task, viewBox) {
+    await super.extractTextContent(evaluator, task, viewBox);
+    const text = this.data.textContent;
+    if (!text) {
+      return;
+    }
+
+    // The text extractor doesn't handle empty lines correctly, so if the
+    // content we get is more or less (modulo whitespaces) the same as the
+    // field value we just ignore it.
+    const allText = text.join("\n");
+    if (allText === this.data.fieldValue) {
+      return;
+    }
+    const regex = allText.replaceAll(/([.*+?^${}()|[\]\\])|(\s+)/g, (_m, p1) =>
+      p1 ? `\\${p1}` : "\\s+"
+    );
+    this.data.textContent = new RegExp(`^\\s*${regex}\\s*$`).test(
+      this.data.fieldValue
+    )
+      ? this.data.fieldValue.split("\n")
+      : text;
+  }
+
   getFieldObject() {
     return {
       id: this.data.id,

diff --git a/test/pdfs/.gitignore b/test/pdfs/.gitignore
@@ -674,3 +674,4 @@
 !issue18072.pdf
 !stamps.pdf
 !issue15096.pdf
+!issue18036.pdf
diff --git a/test/pdfs/issue18036.pdf b/test/pdfs/issue18036.pdf
diff --git a/test/test_manifest.json b/test/test_manifest.json
@@ -8477,13 +8477,13 @@
         "value": "Hello World"
       },
       "33R": {
-        "value": "Hello World\nDlrow Olleh\nHello World"
+        "value": "\n\nHello World\nDlrow Olleh\nHello World"
       },
       "36R": {
-        "value": "Hello World\nDlrow Olleh\nHello World"
+        "value": "\n\nHello World\nDlrow Olleh\nHello World"
       },
       "39R": {
-        "value": "Hello World\nDlrow Olleh\nHello World"
+        "value": "\n\nHello World\nDlrow Olleh\nHello World"
       }
     }
   },
@@ -10684,5 +10684,13 @@
         "value": false
       }
     }
+  },
+  {
+    "id": "issue18036",
+    "file": "pdfs/issue18036.pdf",
+    "md5": "940ad97fc1a6f8d288a213af80313c7e",
+    "rounds": 1,
+    "type": "eq",
+    "annotations": true
   }
 ]