TLDR-474 remove is_inserted attribute

ispras · Sep 28, 2023 · 436bb36 · 436bb36
1 parent 78d8f6a
commit 436bb36
Show file tree

Hide file tree

Showing 11 changed files with 53 additions and 64 deletions.
diff --git a/dedoc/api/static/html_eng/format_description.html b/dedoc/api/static/html_eng/format_description.html
@@ -71,7 +71,6 @@ <h4 id="TableMetadata"> TableMetadata</h4>
 <ol>
     <li><strong>uid</strong>: <a> str (required field) </a> - unique identifier. </li>
     <li><strong>page_id</strong>: <a> integer </a> (optional field) - page number on which the table begins. Can be null.</li>
-    <li><strong>is_inserted</strong>: <a> bool </a> (optional field) - was table inserted into document.</li>
 </ol>
 
 <h3 id="TreeNode"> TreeNode</h3>

diff --git a/dedoc/api/static/html_rus/format_description.html b/dedoc/api/static/html_rus/format_description.html
@@ -72,7 +72,6 @@ <h4 id="TableMetadata"> TableMetadata. Метаинформация таблиц
 <ol>
     <li><strong>uid</strong>: <a> str (обязательное поле) </a> - уникальный идентификатор таблицы.</li>
     <li><strong>page_id</strong>: <a> int </a> (необязательное поле) - номер страницы на которой начинается таблица.</li>
-    <li><strong>is_inserted</strong>: <a> bool </a> (необязательное поле) - была ли таблица встроена в тело документа.</li>
 </ol>
 
 <h3 id="TreeNode"> TreeNode. Древовидная структура документа.</h3>

diff --git a/dedoc/data_structures/table_metadata.py b/dedoc/data_structures/table_metadata.py
@@ -11,23 +11,20 @@ class TableMetadata(Serializable):
     """
     This class holds the information about table unique identifier, rotation angle (if table has been rotated - for images) and so on.
     """
-    def __init__(self, page_id: Optional[int], uid: Optional[str] = None, is_inserted: bool = False, rotated_angle: float = 0.0) -> None:
+    def __init__(self, page_id: Optional[int], uid: Optional[str] = None, rotated_angle: float = 0.0) -> None:
         """
         :param page_id: number of the page where table starts
         :param uid: unique identifier of the table
-        :param is_inserted: indicator if table was already inserted into paragraphs list
         :param rotated_angle: value of the rotation angle by which the table was rotated during recognition
         """
         self.page_id = page_id
         self.uid = str(uuid.uuid1()) if not uid else uid
-        self.is_inserted = is_inserted
         self.rotated_angle = rotated_angle
 
     def to_dict(self) -> dict:
         res = OrderedDict()
         res["uid"] = self.uid
         res["page_id"] = self.page_id
-        res["is_inserted"] = self.is_inserted
         res["rotated_angle"] = self.rotated_angle
         return res
 
@@ -36,6 +33,5 @@ def get_api_dict(api: Api) -> Model:
         return api.model("TableMetadata", {
             "page_id": fields.Integer(readonly=False, description="table start page number"),
             "uid": fields.String(description="table unique id"),
-            "is_inserted": fields.Boolean(description="was the table inserted into document body"),
             "rotated_angle": fields.Float(readonly=False, description="At what angle should the table be rotated to use boxes")
         })
diff --git a/dedoc/readers/docx_reader/data_structures/table.py b/dedoc/readers/docx_reader/data_structures/table.py
@@ -90,7 +90,7 @@ def to_table(self) -> Table:
                 result_row.append(cell)
             result_cells_with_meta.append(result_row)
 
-        return Table(cells=result_cells_with_meta, metadata=TableMetadata(page_id=None, uid=self.uid, is_inserted=False))
+        return Table(cells=result_cells_with_meta, metadata=TableMetadata(page_id=None, uid=self.uid))
 
     def __get_cell_text(self, cell: Tag) -> str:
         cell_text = ""

diff --git a/dedoc/readers/pdf_reader/pdf_txtlayer_reader/pdf_tabby_reader.py b/dedoc/readers/pdf_reader/pdf_txtlayer_reader/pdf_tabby_reader.py
@@ -152,7 +152,7 @@ def __get_tables(self, page: dict, file_hash: str) -> Tuple[List[Table], List[Sc
 
                 result_cells.append(result_row)
             table_bbox = BBox.from_two_points((x_top_left, y_top_left), (x_bottom_right, y_bottom_right))  # noqa TODO add table location into TableMetadata
-            tables.append(Table(cells=result_cells, metadata=TableMetadata(page_id=page_number, is_inserted=False)))
+            tables.append(Table(cells=result_cells, metadata=TableMetadata(page_id=page_number)))
             table_name = file_hash + str(page_number) + str(table_num)
             tables_on_image.append(ScanTable(page_number=page_number, matrix_cells=None, bbox=table_bbox, name=table_name, order=order))
 

diff --git a/docs/source/_static/json_format_examples/basic_example.json b/docs/source/_static/json_format_examples/basic_example.json
@@ -298,7 +298,7 @@
                           "start": 0,
                           "end": 14,
                           "name": "attachment",
-                          "value": "attach_fa1143ae-5d3c-11ee-b518-0242ac120002"
+                          "value": "attach_75af2486-5df1-11ee-bfc1-0242ac120002"
                         }
                       ],
                       "metadata": {
@@ -420,20 +420,19 @@
         "metadata": {
           "uid": "3a327789721e09b3fa6fd9560f3ee263",
           "page_id": null,
-          "is_inserted": false,
           "rotated_angle": 0.0
         }
       }
     ]
   },
   "metadata": {
-    "uid": "doc_uid_auto_fa1f6786-5d3c-11ee-b518-0242ac120002",
+    "uid": "doc_uid_auto_75c93394-5df1-11ee-bfc1-0242ac120002",
     "file_name": "example_return_format.docx",
-    "temporary_file_name": "1695822696_268.docx",
+    "temporary_file_name": "1695900213_314.docx",
     "size": 21270,
-    "modified_time": 1695822696,
-    "created_time": 1695822696,
-    "access_time": 1695822696,
+    "modified_time": 1695900213,
+    "created_time": 1695900213,
+    "access_time": 1695900213,
     "file_type": "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
     "document_subject": "",
     "keywords": "",

diff --git a/docs/source/_static/json_format_examples/linear_structure_type.json b/docs/source/_static/json_format_examples/linear_structure_type.json
@@ -388,7 +388,7 @@
               "start": 0,
               "end": 14,
               "name": "attachment",
-              "value": "attach_fa23fd78-5d3c-11ee-b518-0242ac120002"
+              "value": "attach_75d13b70-5df1-11ee-bfc1-0242ac120002"
             }
           ],
           "metadata": {
@@ -504,20 +504,19 @@
         "metadata": {
           "uid": "3a327789721e09b3fa6fd9560f3ee263",
           "page_id": null,
-          "is_inserted": false,
           "rotated_angle": 0.0
         }
       }
     ]
   },
   "metadata": {
-    "uid": "doc_uid_auto_fa309d08-5d3c-11ee-b518-0242ac120002",
+    "uid": "doc_uid_auto_75e45e94-5df1-11ee-bfc1-0242ac120002",
     "file_name": "example_return_format.docx",
-    "temporary_file_name": "1695822697_827.docx",
+    "temporary_file_name": "1695900214_259.docx",
     "size": 21270,
-    "modified_time": 1695822697,
-    "created_time": 1695822697,
-    "access_time": 1695822697,
+    "modified_time": 1695900213,
+    "created_time": 1695900213,
+    "access_time": 1695900214,
     "file_type": "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
     "document_subject": "",
     "keywords": "",

diff --git a/docs/source/_static/json_format_examples/with_attachments.json b/docs/source/_static/json_format_examples/with_attachments.json
@@ -298,7 +298,7 @@
                           "start": 0,
                           "end": 14,
                           "name": "attachment",
-                          "value": "attach_fa355abe-5d3c-11ee-b518-0242ac120002"
+                          "value": "attach_75ea598e-5df1-11ee-bfc1-0242ac120002"
                         }
                       ],
                       "metadata": {
@@ -420,20 +420,19 @@
         "metadata": {
           "uid": "3a327789721e09b3fa6fd9560f3ee263",
           "page_id": null,
-          "is_inserted": false,
           "rotated_angle": 0.0
         }
       }
     ]
   },
   "metadata": {
-    "uid": "doc_uid_auto_fa4285e0-5d3c-11ee-b518-0242ac120002",
+    "uid": "doc_uid_auto_75fac01c-5df1-11ee-bfc1-0242ac120002",
     "file_name": "example_return_format.docx",
-    "temporary_file_name": "1695822697_953.docx",
+    "temporary_file_name": "1695900214_51.docx",
     "size": 21270,
-    "modified_time": 1695822697,
-    "created_time": 1695822697,
-    "access_time": 1695822697,
+    "modified_time": 1695900214,
+    "created_time": 1695900214,
+    "access_time": 1695900214,
     "file_type": "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
     "document_subject": "",
     "keywords": "",
@@ -476,13 +475,13 @@
         "tables": []
       },
       "metadata": {
-        "uid": "attach_fa355abe-5d3c-11ee-b518-0242ac120002",
+        "uid": "attach_75ea598e-5df1-11ee-bfc1-0242ac120002",
         "file_name": "image1.png",
-        "temporary_file_name": "1695822697_181.png",
+        "temporary_file_name": "1695900214_864.png",
         "size": 14874,
-        "modified_time": 1695822697,
-        "created_time": 1695822697,
-        "access_time": 1695822697,
+        "modified_time": 1695900214,
+        "created_time": 1695900214,
+        "access_time": 1695900214,
         "file_type": "image/png",
         "other_fields": {}
       },

diff --git a/docs/source/_static/json_format_examples/with_base64_attachments.json b/docs/source/_static/json_format_examples/with_base64_attachments.json
diff --git a/docs/source/_static/json_format_examples/with_parsed_attachments.json b/docs/source/_static/json_format_examples/with_parsed_attachments.json
@@ -298,7 +298,7 @@
                           "start": 0,
                           "end": 14,
                           "name": "attachment",
-                          "value": "attach_fa5c54ac-5d3c-11ee-b518-0242ac120002"
+                          "value": "attach_7619127e-5df1-11ee-bfc1-0242ac120002"
                         }
                       ],
                       "metadata": {
@@ -420,20 +420,19 @@
         "metadata": {
           "uid": "3a327789721e09b3fa6fd9560f3ee263",
           "page_id": null,
-          "is_inserted": false,
           "rotated_angle": 0.0
         }
       }
     ]
   },
   "metadata": {
-    "uid": "doc_uid_auto_fa7fdbc0-5d3c-11ee-b518-0242ac120002",
+    "uid": "doc_uid_auto_7630e7b4-5df1-11ee-bfc1-0242ac120002",
     "file_name": "example_return_format.docx",
-    "temporary_file_name": "1695822697_469.docx",
+    "temporary_file_name": "1695900214_402.docx",
     "size": 21270,
-    "modified_time": 1695822697,
-    "created_time": 1695822697,
-    "access_time": 1695822697,
+    "modified_time": 1695900214,
+    "created_time": 1695900214,
+    "access_time": 1695900214,
     "file_type": "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
     "document_subject": "",
     "keywords": "",
@@ -913,13 +912,13 @@
         "tables": []
       },
       "metadata": {
-        "uid": "attach_fa5c54ac-5d3c-11ee-b518-0242ac120002",
+        "uid": "attach_7619127e-5df1-11ee-bfc1-0242ac120002",
         "file_name": "image1.png",
-        "temporary_file_name": "1695822697_301.png",
+        "temporary_file_name": "1695900214_972.png",
         "size": 14874,
-        "modified_time": 1695822697,
-        "created_time": 1695822697,
-        "access_time": 1695822697,
+        "modified_time": 1695900214,
+        "created_time": 1695900214,
+        "access_time": 1695900214,
         "file_type": "image/png",
         "rotated_page_angles": [
           0

diff --git a/docs/source/dedoc_api_usage/return_format.rst b/docs/source/dedoc_api_usage/return_format.rst
@@ -64,13 +64,13 @@ The beginning of the document's metadata:
 
 .. literalinclude:: ../_static/json_format_examples/basic_example.json
     :language: json
-    :lines: 429-437
+    :lines: 428-436
 
 The document's attachments:
 
 .. literalinclude:: ../_static/json_format_examples/basic_example.json
     :language: json
-    :lines: 459
+    :lines: 458
 
 As we see, the `attachments` field is empty because the option
 `with_attachments` is set to `"false"` by default (see :ref:`table_parameters`).
@@ -118,7 +118,7 @@ Unlike the previous examples, in this case we have `attachments` field filled:
 
 .. literalinclude:: ../_static/json_format_examples/with_attachments.json
     :language: json
-    :lines: 459-491
+    :lines: 458-490
 
 Example with base64 attachments
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -136,7 +136,7 @@ The only difference is in the attachment's metadata: attachment's content is enc
 
 .. literalinclude:: ../_static/json_format_examples/with_base64_attachments.json
     :language: json
-    :lines: 459-494
+    :lines: 458-493
 
 Example with parsed attachments
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -156,4 +156,4 @@ The beginning of the document's attachments:
 
 .. literalinclude:: ../_static/json_format_examples/with_parsed_attachments.json
     :language: json
-    :lines: 459-484
+    :lines: 458-483