smart redaction (#6195)

Co-authored-by: James Brunton <jbrunton96@gmail.com>
2026-06-03 16:16:33 +00:00
parent 7f3ca7ea70
commit 35a712a278
26 changed files with 4397 additions and 1802 deletions
@@ -0,0 +1,149 @@
+package stirling.software.SPDF.pdf.parser;
+
+import static org.assertj.core.api.Assertions.assertThat;
+
+import java.util.ArrayList;
+import java.util.List;
+
+import org.junit.jupiter.api.Test;
+
+/** Unit tests for {@link PageColumnLayout} gutter detection and column classification. */
+class PageColumnLayoutTest {
+
+    private static final float PAGE_WIDTH = 612f; // Letter portrait
+
+    // ── single-column ────────────────────────────────────────────────────────────────────────────
+
+    @Test
+    void singleColumn_oneColumnNoGutters() {
+        List<float[]> lines = List.of(lineBox(72f, 396f));
+
+        PageColumnLayout layout = PageColumnLayout.fromLineBoxes(lines, PAGE_WIDTH);
+
+        assertThat(layout.columnCount()).isEqualTo(1);
+        assertThat(layout.gutters()).isEmpty();
+    }
+
+    @Test
+    void singleColumn_classifyAnchor_returnsZero() {
+        PageColumnLayout layout =
+                PageColumnLayout.fromLineBoxes(List.of(lineBox(72f, 396f)), PAGE_WIDTH);
+
+        assertThat(layout.columnOf(100f, 200f)).isEqualTo(0);
+    }
+
+    // ── two-column ───────────────────────────────────────────────────────────────────────────────
+
+    @Test
+    void twoColumn_detectsGutter() {
+        PageColumnLayout layout =
+                PageColumnLayout.fromLineBoxes(buildTwoColumnLines(3), PAGE_WIDTH);
+
+        assertThat(layout.columnCount()).isEqualTo(2);
+        assertThat(layout.gutters()).hasSize(1);
+        float[] gutter = layout.gutters().get(0);
+        // Gutter is centered on pageWidth/2 with PageColumnLayout.MIDPOINT_SLACK_PT slack each
+        // side.
+        float pageMid = PAGE_WIDTH / 2f;
+        assertThat(gutter[0]).isBetween(pageMid - 40f, pageMid - 20f);
+        assertThat(gutter[1]).isBetween(pageMid + 20f, pageMid + 40f);
+    }
+
+    @Test
+    void twoColumn_classifyLeftAndRightAnchors() {
+        PageColumnLayout layout =
+                PageColumnLayout.fromLineBoxes(buildTwoColumnLines(3), PAGE_WIDTH);
+        assertThat(layout.columnOf(100f, 200f)).isEqualTo(0);
+        assertThat(layout.columnOf(380f, 460f)).isEqualTo(1);
+    }
+
+    @Test
+    void twoColumn_columnsCrossing_leftLineOnlyHitsLeft() {
+        PageColumnLayout layout =
+                PageColumnLayout.fromLineBoxes(buildTwoColumnLines(3), PAGE_WIDTH);
+        assertThat(layout.columnsCrossing(72f, 280f)).containsExactly(0);
+        assertThat(layout.columnsCrossing(320f, 540f)).containsExactly(1);
+    }
+
+    @Test
+    void twoColumn_spanningLine_returnsBothColumns() {
+        List<float[]> lines = new ArrayList<>(buildTwoColumnLines(3));
+        // Full-width header that crosses pageWidth/2.
+        lines.add(lineBox(72f, 396f));
+
+        PageColumnLayout layout = PageColumnLayout.fromLineBoxes(lines, PAGE_WIDTH);
+
+        assertThat(layout.columnsCrossing(72f, 540f)).containsExactly(0, 1);
+        assertThat(layout.columnsCrossing(72f, 280f)).containsExactly(0);
+    }
+
+    private static List<float[]> buildTwoColumnLines(int rowsPerColumn) {
+        List<float[]> lines = new ArrayList<>();
+        for (int i = 0; i < rowsPerColumn; i++) {
+            lines.add(lineBox(72f, 136f)); // left column body (72..208)
+            lines.add(lineBox(320f, 220f)); // right column body (320..540)
+        }
+        return lines;
+    }
+
+    // ── three-column ─────────────────────────────────────────────────────────────────────────────
+
+    @Test
+    void threeColumn_collapsesToLeftRightSplit() {
+        // The midpoint-based detector splits the page at pageWidth/2 and treats anything else as
+        // single-column or spanning. A genuine 3-column layout collapses to 2 columns; the middle
+        // column's content ends up classified by midpoint as left or right of pageMid.
+        List<float[]> lines = new ArrayList<>();
+        for (int i = 0; i < 6; i++) {
+            lines.add(lineBox(72f, 150f)); // 72..222
+            lines.add(lineBox(252f, 150f)); // 252..402
+            lines.add(lineBox(432f, 150f)); // 432..582
+        }
+
+        PageColumnLayout layout = PageColumnLayout.fromLineBoxes(lines, PAGE_WIDTH);
+
+        assertThat(layout.columnCount()).isEqualTo(2);
+        assertThat(layout.gutters()).hasSize(1);
+    }
+
+    // ── empty page ───────────────────────────────────────────────────────────────────────────────
+
+    @Test
+    void emptyPage_singleColumnFallback() {
+        PageColumnLayout layout = PageColumnLayout.fromLineBoxes(List.of(), PAGE_WIDTH);
+        assertThat(layout.columnCount()).isEqualTo(1);
+        assertThat(layout.gutters()).isEmpty();
+    }
+
+    @Test
+    void onlyShortFragments_singleColumnFallback() {
+        // Page numbers / decorations — too narrow to vote either side.
+        List<float[]> lines = List.of(lineBox(300f, 6f));
+        PageColumnLayout layout = PageColumnLayout.fromLineBoxes(lines, PAGE_WIDTH);
+        assertThat(layout.columnCount()).isEqualTo(1);
+    }
+
+    // ── narrow gap should not be confused for a gutter ───────────────────────────────────────────
+
+    @Test
+    void narrowInternalGap_doesNotProduceGutter() {
+        // Both halves sit left of the page midpoint, so no line votes for a right column and
+        // detection falls back to single-column.
+        List<float[]> lines = new ArrayList<>();
+        lines.add(lineBox(72f, 100f));
+        lines.add(lineBox(180f, 100f));
+        for (int i = 0; i < 5; i++) {
+            lines.add(lineBox(72f, 208f));
+        }
+
+        PageColumnLayout layout = PageColumnLayout.fromLineBoxes(lines, PAGE_WIDTH);
+        assertThat(layout.columnCount()).isEqualTo(1);
+    }
+
+    // ── helpers ──────────────────────────────────────────────────────────────────────────────────
+
+    /** Builds a line bounding box {@code [x1, 0, x1+width, 0]}; Y is unused by detection. */
+    private static float[] lineBox(float x1, float width) {
+        return new float[] {x1, 0f, x1 + width, 0f};
+    }
+}
@@ -1,163 +0,0 @@
-package stirling.software.common.util.propertyeditor;
-
-import static org.junit.jupiter.api.Assertions.*;
-
-import java.util.List;
-
-import org.junit.jupiter.api.BeforeEach;
-import org.junit.jupiter.api.Test;
-
-import stirling.software.common.model.api.security.RedactionArea;
-
-class StringToArrayListPropertyEditorTest {
-
-    private StringToArrayListPropertyEditor<RedactionArea> editor;
-
-    @BeforeEach
-    void setUp() {
-        editor = new StringToArrayListPropertyEditor<>(RedactionArea.class);
-    }
-
-    @Test
-    void testSetAsText_ValidJson() {
-        // Arrange
-        String json =
-                "[{\"x\":10.5,\"y\":20.5,\"width\":100.0,\"height\":50.0,\"page\":1,\"color\":\"#FF0000\"}]";
-
-        // Act
-        editor.setAsText(json);
-        Object value = editor.getValue();
-
-        // Assert
-        assertNotNull(value, "Value should not be null");
-        assertInstanceOf(List.class, value, "Value should be a List");
-
-        @SuppressWarnings("unchecked")
-        List<RedactionArea> list = (List<RedactionArea>) value;
-        assertEquals(1, list.size(), "List should have 1 entry");
-
-        RedactionArea area = list.get(0);
-        assertEquals(10.5, area.getX(), "X should be 10.5");
-        assertEquals(20.5, area.getY(), "Y should be 20.5");
-        assertEquals(100.0, area.getWidth(), "Width should be 100.0");
-        assertEquals(50.0, area.getHeight(), "Height should be 50.0");
-        assertEquals(1, area.getPage(), "Page should be 1");
-        assertEquals("#FF0000", area.getColor(), "Color should be #FF0000");
-    }
-
-    @Test
-    void testSetAsText_MultipleItems() {
-        // Arrange
-        String json =
-                "["
-                        + "{\"x\":10.0,\"y\":20.0,\"width\":100.0,\"height\":50.0,\"page\":1,\"color\":\"#FF0000\"},"
-                        + "{\"x\":30.0,\"y\":40.0,\"width\":200.0,\"height\":150.0,\"page\":2,\"color\":\"#00FF00\"}"
-                        + "]";
-
-        // Act
-        editor.setAsText(json);
-        Object value = editor.getValue();
-
-        // Assert
-        assertNotNull(value, "Value should not be null");
-        assertInstanceOf(List.class, value, "Value should be a List");
-
-        @SuppressWarnings("unchecked")
-        List<RedactionArea> list = (List<RedactionArea>) value;
-        assertEquals(2, list.size(), "List should have 2 entries");
-
-        RedactionArea area1 = list.get(0);
-        assertEquals(10.0, area1.getX(), "X should be 10.0");
-        assertEquals(20.0, area1.getY(), "Y should be 20.0");
-        assertEquals(1, area1.getPage(), "Page should be 1");
-
-        RedactionArea area2 = list.get(1);
-        assertEquals(30.0, area2.getX(), "X should be 30.0");
-        assertEquals(40.0, area2.getY(), "Y should be 40.0");
-        assertEquals(2, area2.getPage(), "Page should be 2");
-    }
-
-    @Test
-    void testSetAsText_EmptyString() {
-        // Arrange
-        String json = "";
-
-        // Act
-        editor.setAsText(json);
-        Object value = editor.getValue();
-
-        // Assert
-        assertNotNull(value, "Value should not be null");
-        assertInstanceOf(List.class, value, "Value should be a List");
-
-        @SuppressWarnings("unchecked")
-        List<RedactionArea> list = (List<RedactionArea>) value;
-        assertTrue(list.isEmpty(), "List should be empty");
-    }
-
-    @Test
-    void testSetAsText_NullString() {
-        // Act
-        editor.setAsText(null);
-        Object value = editor.getValue();
-
-        // Assert
-        assertNotNull(value, "Value should not be null");
-        assertInstanceOf(List.class, value, "Value should be a List");
-
-        @SuppressWarnings("unchecked")
-        List<RedactionArea> list = (List<RedactionArea>) value;
-        assertTrue(list.isEmpty(), "List should be empty");
-    }
-
-    @Test
-    void testSetAsText_SingleItemAsArray() {
-        // Arrange - note this is a single object, not an array
-        String json =
-                "{\"x\":10.0,\"y\":20.0,\"width\":100.0,\"height\":50.0,\"page\":1,\"color\":\"#FF0000\"}";
-
-        // Act
-        editor.setAsText(json);
-        Object value = editor.getValue();
-
-        // Assert
-        assertNotNull(value, "Value should not be null");
-        assertInstanceOf(List.class, value, "Value should be a List");
-
-        @SuppressWarnings("unchecked")
-        List<RedactionArea> list = (List<RedactionArea>) value;
-        assertEquals(1, list.size(), "List should have 1 entry");
-
-        RedactionArea area = list.get(0);
-        assertEquals(10.0, area.getX(), "X should be 10.0");
-        assertEquals(20.0, area.getY(), "Y should be 20.0");
-    }
-
-    @Test
-    void testSetAsText_InvalidJson() {
-        // Arrange
-        String json = "invalid json";
-
-        // Act & Assert
-        assertThrows(IllegalArgumentException.class, () -> editor.setAsText(json));
-    }
-
-    @Test
-    void testSetAsText_UnknownProperties() {
-        // Arrange - this JSON contains properties not in RedactionArea
-        // With FAIL_ON_UNKNOWN_PROPERTIES disabled, this should ignore the unknown properties
-        String json = "[{\"invalid\":\"structure\"}]";
-
-        // Act
-        editor.setAsText(json);
-        Object value = editor.getValue();
-
-        // Assert
-        assertNotNull(value, "Value should not be null");
-        assertInstanceOf(List.class, value, "Value should be a List");
-
-        @SuppressWarnings("unchecked")
-        List<RedactionArea> list = (List<RedactionArea>) value;
-        assertEquals(1, list.size(), "List should have 1 entry (empty object)");
-    }
-}