fix: propagate cell type (#73)

* fix: propagate cell type Signed-off-by: Michele Dolfi <[email protected]> * update test results Signed-off-by: Michele Dolfi <[email protected]> --------- Signed-off-by: Michele Dolfi <[email protected]>
DS4SD · Sep 17, 2024 · 27c5852 · 27c5852
1 parent b03a7de
commit 27c5852
Show file tree

Hide file tree

Showing 6 changed files with 1,207 additions and 1,199 deletions.
diff --git a/src/andromeda/tooling/structs/elements/table_element.h b/src/andromeda/tooling/structs/elements/table_element.h
@@ -23,6 +23,7 @@ namespace andromeda
 		  std::array<uint64_t,2> col_span,
 		  std::array<float, 4> bbox,
 		  //nlohmann::json bbox,
+      std::string type,
 		  std::string orig);
 
     nlohmann::json to_json();
@@ -65,9 +66,11 @@ namespace andromeda
 			       std::array<uint64_t,2> row_span,			       
 			       std::array<uint64_t,2> col_span,
 			       std::array<float, 4> bbox,
+             std::string type,
 			       std::string orig):
     text_element(),
     i(i), j(j),
+    type(type),
 
     row_span(row_span),
     col_span(col_span),

diff --git a/src/andromeda/tooling/structs/subjects/table.h b/src/andromeda/tooling/structs/subjects/table.h
@@ -300,6 +300,12 @@ namespace andromeda
                     text = grid.at(i).at(j).at("text");
                   }
 
+                std::string type = "";
+                if (grid.at(i).at(j).count("type")==1)
+                  {
+                    type = grid.at(i).at(j).at("type").get<std::string>();
+                  }
+
 		std::array<float, 4> bbox = {0.0, 0.0, 0.0, 0.0};
 		if(grid.at(i).at(j).count("bbox")==1 and
 		   grid.at(i).at(j).at("bbox").is_array())
@@ -332,7 +338,7 @@ namespace andromeda
                       }
                   }
 
-                data.back().emplace_back(i, j, row_span, col_span, bbox, text);
+                data.back().emplace_back(i, j, row_span, col_span, bbox, type, text);
               }
           }
       }

diff --git a/tests/data/docs/1806.02284.nlp.json b/tests/data/docs/1806.02284.nlp.json
@@ -96460,7 +96460,7 @@
               ]
             ],
             "text": "",
-            "type": ""
+            "type": "col_header"
           },
           {
             "bbox": [
@@ -96613,7 +96613,7 @@
               ]
             ],
             "text": "",
-            "type": ""
+            "type": "body"
           },
           {
             "bbox": [
@@ -97069,7 +97069,7 @@
               ]
             ],
             "text": "",
-            "type": ""
+            "type": "col_header"
           },
           {
             "bbox": null,
@@ -97092,7 +97092,7 @@
               ]
             ],
             "text": "",
-            "type": ""
+            "type": "col_header"
           },
           {
             "bbox": [
@@ -97405,7 +97405,7 @@
               ]
             ],
             "text": "",
-            "type": ""
+            "type": "body"
           },
           {
             "bbox": [
@@ -97456,7 +97456,7 @@
               ]
             ],
             "text": "",
-            "type": ""
+            "type": "body"
           },
           {
             "bbox": [
@@ -98099,7 +98099,7 @@
               ]
             ],
             "text": "",
-            "type": ""
+            "type": "body"
           }
         ],
         [
@@ -98348,7 +98348,7 @@
               ]
             ],
             "text": "",
-            "type": ""
+            "type": "body"
           }
         ],
         [
@@ -98597,7 +98597,7 @@
               ]
             ],
             "text": "",
-            "type": ""
+            "type": "body"
           }
         ],
         [
@@ -98846,7 +98846,7 @@
               ]
             ],
             "text": "",
-            "type": ""
+            "type": "body"
           }
         ],
         [
@@ -99095,7 +99095,7 @@
               ]
             ],
             "text": "",
-            "type": ""
+            "type": "body"
           }
         ],
         [
@@ -99344,7 +99344,7 @@
               ]
             ],
             "text": "",
-            "type": ""
+            "type": "body"
           }
         ],
         [
@@ -99593,7 +99593,7 @@
               ]
             ],
             "text": "",
-            "type": ""
+            "type": "body"
           }
         ]
       ],