fix: implement polygon intersection properly

Signed-off-by: David Huggins-Daines <[email protected]>
Signed-off-by: David Huggins-Daines <[email protected]>

Author: David Huggins-Daines

src/v2/pdf_resources/page_cell.h

@@ -27,6 +27,10 @@ namespace pdflib
 
     bool is_adjacent_to(pdf_resource<PAGE_CELL>& other, double delta);
 
+    bool intersects(pdf_resource<PAGE_CELL>& other);
+
+    bool contains(double x, double y);
+
     bool has_same_reading_orientation(pdf_resource<PAGE_CELL>& other);
 
     bool merge_with(pdf_resource<PAGE_CELL>& other, double delta);
@@ -259,23 +263,46 @@ namespace pdflib
 
     return (num_chars>0? len/num_chars : 0.0);
   }
-  
+
+  bool pdf_resource<PAGE_CELL>::intersects(pdf_resource<PAGE_CELL>& other)
+  {
+    // Use point-in-polygon (via even-odd rule) to determine if
+    // bounding quadrilaterals intersect.
+    return contains(other.r_x0, other.r_y0)
+      or contains(other.r_x1, other.r_y1)
+      or contains(other.r_x2, other.r_y2)
+      or contains(other.r_x3, other.r_y3)
+      or other.contains(r_x0, r_y0)
+      or other.contains(r_x1, r_y1)
+      or other.contains(r_x2, r_y2)
+      or other.contains(r_x3, r_y3);
+  }
+
+  inline bool inside_plane(double x, double y, double xi, double yi, double xj, double yj)
+  {
+    return ((yi < y) != (yj > y)) and (x < (xj - xi) * (y - yi) / (yj - yi) + xi);
+  }
+
+  bool pdf_resource<PAGE_CELL>::contains(double x, double y)
+  {
+    // point-in-polygon via even-odd rule
+    bool inside = false;
+    if (inside_plane(x, y, r_x0, r_y0, r_x1, r_y1))
+      inside = not inside;
+    if (inside_plane(x, y, r_x1, r_y1, r_x2, r_y2))
+      inside = not inside;
+    if (inside_plane(x, y, r_x2, r_y2, r_x3, r_y3))
+      inside = not inside;
+    if (inside_plane(x, y, r_x3, r_y3, r_x0, r_y0))
+      inside = not inside;
+    return inside;
+  }
+
+
   bool pdf_resource<PAGE_CELL>::is_adjacent_to(pdf_resource<PAGE_CELL>& other, double eps)
   {
-    // This assumes (even for right-to-left text) that other is to the
-    // right of this.  If two cells overlap then they are obviously
-    // adjacent, otherwise the right side corners of this must be
-    // within eps of the left side corners of other.
-
-    // Intersection of bounding rectangles (FIXME: Does not actually
-    // imply overlap for rotated cells, not immediately sure the
-    // correct and efficient way to compute that)
-    double max_x0 = std::max(x0, other.x0);
-    double min_x1 = std::min(x1, other.x1);
-    double max_y0 = std::max(y0, other.y0);
-    double min_y1 = std::min(y1, other.y1);
-    if (max_x0 < min_x1 and max_y0 < min_y1)
-      return true;
+    // NOTE: This assumes (even for right-to-left text) that other is
+    // to the right of this, as the calling code seems to do that.
 
     // lower_right(this) : lower_left(other)
     double dx0 = other.r_x0 - r_x1;

src/v2/pdf_sanitators/cells.h

@@ -319,7 +319,7 @@ namespace pdflib
 		double delta_0 = cells[i].average_char_width()*space_width_factor_for_merge;
 		double delta_1 = cells[i].average_char_width()*space_width_factor_for_merge_with_space;
 
-		if(cells[i].is_adjacent_to(cells[j], delta_0))
+		if(cells[i].is_adjacent_to(cells[j], delta_0) or cells[i].intersects(cells[j]))
 		  {
 		    cells[i].merge_with(cells[j], delta_1);