feat: added confidence score and detected languages to Page classes

google/cloud/documentai_toolbox/wrappers/page.py

@@ -171,6 +171,22 @@ def hocr_bounding_box(self) -> Optional[str]:
             self.documentai_object, self._page.documentai_object.dimension
         )
 
+    @cached_property
+    def confidence(self) -> Optional[float]:
+        """
+        Optional. The confidence score of the page element detection.
+        """
+        return getattr(self.documentai_object, "confidence", None)
+
+    @cached_property
+    def detected_languages(
+        self,
+    ) -> Optional[List[documentai.Document.Page.DetectedLanguage]]:
+        """
+        Optional. A list of detected languages for the page element.
+        """
+        return getattr(self.documentai_object, "detected_languages", None)
+
     # This field is a cached property to improve export times for hOCR
     # as outlined in https://github.com/googleapis/python-documentai-toolbox/issues/312
     @cached_property

samples/snippets/quickstart_sample.py

@@ -105,6 +105,16 @@ def quickstart_sample(
             print(line.text)
         for token in page.tokens:
             print(token.text)
+            # Print token confidence
+            print(f"\tConfidence: {token.confidence:.4f}")
+            # Print detected languages
+            if token.detected_languages:
+                print("\tDetected Languages:")
+                for lang in token.detected_languages:
+                    confidence_str = f", confidence: {lang.confidence:.4f}" if hasattr(lang, "confidence") else ""
+                    print(f"\t\t- {lang.language_code}{confidence_str}")
+            else:
+                print("\tNo language detected")
 
         # Only supported with Form Parser processor
         # https://cloud.google.com/document-ai/docs/form-parser

samples/snippets/test_quickstart_sample.py

@@ -37,6 +37,8 @@ def test_quickstart_sample_gcs_bucket_prefix(capsys: pytest.CaptureFixture) -> N
     assert "Document structure in Cloud Storage" in out
     assert "Number of Pages: 1" in out
     assert "Number of Entities: 35" in out
+    assert "Confidence:" in out
+    assert "Detected Languages:" in out or "No language detected" in out
 
 
 def test_quickstart_sample_gcs_uri(capsys: pytest.CaptureFixture) -> None:
@@ -48,6 +50,8 @@ def test_quickstart_sample_gcs_uri(capsys: pytest.CaptureFixture) -> None:
 
     assert "Number of Pages: 1" in out
     assert "Number of Entities: 35" in out
+    assert "Confidence:" in out
+    assert "Detected Languages:" in out or "No language detected" in out
 
 
 def test_quickstart_sample_document_path(capsys: pytest.CaptureFixture) -> None:
@@ -58,6 +62,8 @@ def test_quickstart_sample_document_path(capsys: pytest.CaptureFixture) -> None:
     assert "Number of Pages: 1" in out
     assert "Number of Entities: 0" in out
     assert "Form Date" in out
+    assert "Confidence:" in out
+    assert "Detected Languages:" in out or "No language detected" in out
 
 
 def test_quickstart_sample_documentai_document(capsys: pytest.CaptureFixture) -> None:
@@ -72,6 +78,8 @@ def test_quickstart_sample_documentai_document(capsys: pytest.CaptureFixture) ->
     assert "Number of Pages: 1" in out
     assert "Number of Entities: 0" in out
     assert "Form Date" in out
+    assert "Confidence:" in out
+    assert "Detected Languages:" in out or "No language detected" in out
 
 
 def test_quickstart_sample_batch_process_metadata(

tests/unit/test_page.py

@@ -252,6 +252,17 @@ def test_Block(docproto):
 
     assert block.paragraphs
 
+    # Check confidence value
+    assert isinstance(block.confidence, float)
+    assert 0.0 <= block.confidence <= 1.0
+
+    # Check detected languages
+    assert isinstance(block.detected_languages, list)
+    if block.detected_languages:
+        for language in block.detected_languages:
+            assert isinstance(language, documentai.Document.Page.DetectedLanguage)
+            assert hasattr(language, "language_code")
+
 
 def test_Paragraph(docproto):
     wrapped_page = page.Page(
@@ -268,6 +279,17 @@ def test_Paragraph(docproto):
 
     assert paragraph.lines
 
+    # Check confidence value
+    assert isinstance(paragraph.confidence, float)
+    assert 0.0 <= paragraph.confidence <= 1.0
+
+    # Check detected languages
+    assert isinstance(paragraph.detected_languages, list)
+    if paragraph.detected_languages:
+        for language in paragraph.detected_languages:
+            assert isinstance(language, documentai.Document.Page.DetectedLanguage)
+            assert hasattr(language, "language_code")
+
 
 def test_Line(docproto):
     wrapped_page = page.Page(
@@ -284,6 +306,17 @@ def test_Line(docproto):
 
     assert line.tokens
 
+    # Check confidence value
+    assert isinstance(line.confidence, float)
+    assert 0.0 <= line.confidence <= 1.0
+
+    # Check detected languages
+    assert isinstance(line.detected_languages, list)
+    if line.detected_languages:
+        for language in line.detected_languages:
+            assert isinstance(language, documentai.Document.Page.DetectedLanguage)
+            assert hasattr(language, "language_code")
+
 
 def test_Token(docproto):
     wrapped_page = page.Page(
@@ -298,6 +331,17 @@ def test_Token(docproto):
     assert token.text == "Q.\n"
     assert token.hocr_bounding_box == "bbox 585 1781 620 1818"
 
+    # Check confidence value
+    assert isinstance(token.confidence, float)
+    assert 0.0 <= token.confidence <= 1.0
+
+    # Check detected languages
+    assert isinstance(token.detected_languages, list)
+    if token.detected_languages:
+        for language in token.detected_languages:
+            assert isinstance(language, documentai.Document.Page.DetectedLanguage)
+            assert hasattr(language, "language_code")
+
     assert token.symbols == []