Merge pull request #38 from kensho-technologies/libin/add_support_to_images

[kensho-kenverters] Add support of image type.

Author: libinliang866

kensho_kenverters/CHANGELOG.md

@@ -1,5 +1,11 @@
 # Changelog
 
+## v2.1.0
+
+### Added
+
+* Add support of image type.  
+
 ## v2.0.0
 
 * Add parsing of table structure.  

kensho_kenverters/constants.py

@@ -61,6 +61,7 @@ class ContentCategory(Enum):
     LINE_PLOT = "LINE_PLOT"
     PIE_CHART = "PIE_CHART"
     SCATTER_PLOT = "SCATTER_PLOT"
+    IMAGE = "IMAGE"
 
 
 ELEMENT_TITLE_CONTENT_CATEGORIES = {

kensho_kenverters/tests/data/extract_output.json

@@ -123,6 +123,7 @@ def test_convert_output_to_items(self) -> None:
                 "to as predictive analytics.",
             },
             {"category": "figure", "text": ""},
+            {"category": "image", "text": ""},
             {"category": "title", "text": "Recommendation: BUY"},
             {"category": "text", "text": "42"},
             {"category": "text", "text": "test noise string at bottom"},
@@ -272,6 +273,19 @@ def test_convert_output_to_items(self) -> None:
                     )
                 ],
             },
+            {
+                "category": "image",
+                "text": "",
+                "locations": [
+                    LocationModel(
+                        height=0.02,
+                        width=0.15,
+                        x=0.5,
+                        y=0.87,
+                        page_number=0,
+                    )
+                ],
+            },
             {
                 "category": "title",
                 "text": "Recommendation: BUY",

kensho_kenverters/tests/test_convert_output.py

@@ -100,6 +100,7 @@ def test_extract_organized_sections(self) -> None:
                     ),
                 },
                 {"category": "figure", "text": ""},
+                {"category": "image", "text": ""},
             ],
             [
                 {"category": "title", "text": "Recommendation: BUY"},
@@ -975,6 +976,21 @@ def test_extract_organized_sections(self) -> None:
                         "type": "FIGURE",
                         "uid": "33",
                     },
+                    {
+                        "children": [],
+                        "content": "",
+                        "locations": [
+                            {
+                                "height": 0.02,
+                                "page_number": 0,
+                                "width": 0.15,
+                                "x": 0.5,
+                                "y": 0.87,
+                            }
+                        ],
+                        "type": "IMAGE",
+                        "uid": "38",
+                    },
                     {
                         "children": [],
                         "content": "Recommendation: BUY",
@@ -1909,6 +1925,21 @@ def test_extract_organized_sections(self) -> None:
                         "type": "FIGURE",
                         "uid": "33",
                     },
+                    {
+                        "children": [],
+                        "content": "",
+                        "locations": [
+                            {
+                                "height": 0.02,
+                                "page_number": 0,
+                                "width": 0.15,
+                                "x": 0.5,
+                                "y": 0.87,
+                            }
+                        ],
+                        "type": "IMAGE",
+                        "uid": "38",
+                    },
                     {
                         "children": [],
                         "content": "Recommendation: BUY",
@@ -2007,6 +2038,7 @@ def test_extract_organized_sections(self) -> None:
                     ),
                 },
                 {"category": "figure", "text": ""},
+                {"category": "image", "text": ""},
             ],
             [{"category": "title", "text": "Recommendation: BUY"}],
         ]

kensho_kenverters/tests/test_output_to_sections.py

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "kensho_kenverters"
-version = "2.0.0"
+version = "2.1.0"
 description = "Extract Output Translator Tools"
 readme = "README.md"
 authors = ["Valerie Faucon-Morin <valerie.fauconmorin@kensho.com>"]