Closes #373: high quality PDF to image conversion before OCR

2026-08-06 14:13:11 +07:00 · 2020-02-13 17:43:07 +01:00
parent a7423caeb1
commit 90a49efa4a
5 changed files with 41 additions and 21 deletions
@@ -0,0 +1,19 @@
+package com.sismics.util.format;
+
+import com.sismics.docs.core.util.format.PdfFormatHandler;
+import org.junit.Assert;
+import org.junit.Test;
+
+import java.nio.file.Paths;
+
+public class TestPdfFormatHandler {
+    @Test
+    public void testIssue373() throws Exception {
+        PdfFormatHandler formatHandler = new PdfFormatHandler();
+        String content = formatHandler.extractContent("deu", Paths.get(ClassLoader.getSystemResource("file/issue373.pdf").toURI()));
+        Assert.assertTrue(content.contains("Aufrechterhaltung"));
+        Assert.assertTrue(content.contains("Außentemperatur"));
+        Assert.assertTrue(content.contains("Grundumsatzmessungen"));
+        Assert.assertTrue(content.contains("ermitteln"));
+    }
+}