Merged in feature/postprocessing (pull request #114)

Feature/postprocessing

* tests

* passtest

* fixshorttests

* mosttests

* improvingbasedockerfile

* testspeeds

* testing

* host

* canparallel

* clean

* passfullsuite

* singlepagemax

* test

* findfeatures

* findstables

* tbls

* tablestoo

* tablestoo

* lateraltests

* tableloc

* cleanup

* inlinetable

* childids

* cleanup

* tests
This commit is contained in:
Michael McGuinness
2025-04-22 14:40:16 +00:00
parent edc50c7510
commit fee71e7740
404 changed files with 211048 additions and 2820 deletions
+238 -182
View File
@@ -1,192 +1,248 @@
{
"Blocks": [
"base": [
{
"BlockType": "PAGE",
"ColumnIndex": null,
"ColumnSpan": null,
"Confidence": null,
"EntityTypes": null,
"Geometry": {
"BoundingBox": {
"Height": 1,
"Left": 0,
"Top": 0,
"Width": 1
},
"Polygon": [
{
"X": 0.0000010227642,
"Y": 0
},
{
"X": 1,
"Y": 0.0001285361
},
{
"X": 1,
"Y": 1
},
{
"X": 0,
"Y": 1
}
]
},
"Id": "cf623e5d-d3f1-4aa8-9255-970b0a178924",
"Page": null,
"Query": null,
"Relationships": [
"AnalyzeDocumentModelVersion": "1.0",
"Blocks": [
{
"Ids": [
"44a179f8-88e8-4cdd-aebf-96be77e681e0"
"BlockType": "PAGE",
"ColumnIndex": null,
"ColumnSpan": null,
"Confidence": null,
"EntityTypes": null,
"Geometry": {
"BoundingBox": {
"Height": 1,
"Left": 0,
"Top": 0,
"Width": 1
},
"Polygon": [
{
"X": 0.0000011078263,
"Y": 0
},
{
"X": 1,
"Y": 0.00008426185
},
{
"X": 1,
"Y": 1
},
{
"X": 0,
"Y": 1
}
]
},
"Id": "c811b895-fdd3-45b2-aaea-9b03ac08e6c7",
"Page": null,
"Query": null,
"Relationships": [
{
"Ids": [
"fb57ceba-2d5a-4334-bc58-5ae5c7cd1a82",
"fdc70cec-80dc-4833-81ac-ca919e28981e"
],
"Type": "CHILD"
}
],
"Type": "CHILD"
"RowIndex": null,
"RowSpan": null,
"SelectionStatus": "",
"Text": null,
"TextType": ""
},
{
"BlockType": "LINE",
"ColumnIndex": null,
"ColumnSpan": null,
"Confidence": 99.73413,
"EntityTypes": null,
"Geometry": {
"BoundingBox": {
"Height": 0.025445992,
"Left": 0.16460413,
"Top": 0.092252895,
"Width": 0.210229
},
"Polygon": [
{
"X": 0.16462424,
"Y": 0.092252895
},
{
"X": 0.3748331,
"Y": 0.092557475
},
{
"X": 0.37481487,
"Y": 0.117698886
},
{
"X": 0.16460413,
"Y": 0.11739321
}
]
},
"Id": "fb57ceba-2d5a-4334-bc58-5ae5c7cd1a82",
"Page": null,
"Query": null,
"Relationships": [
{
"Ids": [
"979fd840-9770-44c6-92ce-0414e9de1ce3",
"007d447f-af28-4535-8416-b8268650e578"
],
"Type": "CHILD"
}
],
"RowIndex": null,
"RowSpan": null,
"SelectionStatus": "",
"Text": "Hello World!",
"TextType": ""
},
{
"BlockType": "WORD",
"ColumnIndex": null,
"ColumnSpan": null,
"Confidence": 99.91676,
"EntityTypes": null,
"Geometry": {
"BoundingBox": {
"Height": 0.025081681,
"Left": 0.16460413,
"Top": 0.0924382,
"Width": 0.087125756
},
"Polygon": [
{
"X": 0.1646241,
"Y": 0.0924382
},
{
"X": 0.25172988,
"Y": 0.09256441
},
{
"X": 0.25171068,
"Y": 0.11751988
},
{
"X": 0.16460413,
"Y": 0.11739321
}
]
},
"Id": "979fd840-9770-44c6-92ce-0414e9de1ce3",
"Page": null,
"Query": null,
"Relationships": null,
"RowIndex": null,
"RowSpan": null,
"SelectionStatus": "",
"Text": "Hello",
"TextType": "PRINTED"
},
{
"BlockType": "WORD",
"ColumnIndex": null,
"ColumnSpan": null,
"Confidence": 99.55151,
"EntityTypes": null,
"Geometry": {
"BoundingBox": {
"Height": 0.024770401,
"Left": 0.26289847,
"Top": 0.09239531,
"Width": 0.11193464
},
"Polygon": [
{
"X": 0.2629173,
"Y": 0.09239531
},
{
"X": 0.3748331,
"Y": 0.092557475
},
{
"X": 0.37481526,
"Y": 0.117165715
},
{
"X": 0.26289847,
"Y": 0.11700299
}
]
},
"Id": "007d447f-af28-4535-8416-b8268650e578",
"Page": null,
"Query": null,
"Relationships": null,
"RowIndex": null,
"RowSpan": null,
"SelectionStatus": "",
"Text": "World!",
"TextType": "PRINTED"
},
{
"BlockType": "LAYOUT_TEXT",
"ColumnIndex": null,
"ColumnSpan": null,
"Confidence": 48.242188,
"EntityTypes": null,
"Geometry": {
"BoundingBox": {
"Height": 0.025445992,
"Left": 0.16460413,
"Top": 0.092252895,
"Width": 0.210229
},
"Polygon": [
{
"X": 0.16462424,
"Y": 0.092252895
},
{
"X": 0.3748331,
"Y": 0.092557475
},
{
"X": 0.37481487,
"Y": 0.117698886
},
{
"X": 0.16460413,
"Y": 0.11739321
}
]
},
"Id": "fdc70cec-80dc-4833-81ac-ca919e28981e",
"Page": null,
"Query": null,
"Relationships": [
{
"Ids": [
"fb57ceba-2d5a-4334-bc58-5ae5c7cd1a82"
],
"Type": "CHILD"
}
],
"RowIndex": null,
"RowSpan": null,
"SelectionStatus": "",
"Text": null,
"TextType": ""
}
],
"RowIndex": null,
"RowSpan": null,
"SelectionStatus": "",
"Text": null,
"TextType": ""
},
{
"BlockType": "LINE",
"ColumnIndex": null,
"ColumnSpan": null,
"Confidence": 99.89616,
"EntityTypes": null,
"Geometry": {
"BoundingBox": {
"Height": 0.024213256,
"Left": 0.16538842,
"Top": 0.092992365,
"Width": 0.20915127
},
"Polygon": [
{
"X": 0.165407,
"Y": 0.092992365
},
{
"X": 0.3745397,
"Y": 0.09330509
},
{
"X": 0.37452286,
"Y": 0.11720562
},
{
"X": 0.16538842,
"Y": 0.11689187
}
]
"DocumentMetadata": {
"Pages": 1
},
"Id": "44a179f8-88e8-4cdd-aebf-96be77e681e0",
"Page": null,
"Query": null,
"Relationships": [
{
"Ids": [
"b2eadea5-a686-48fe-9363-a774a0a146b4",
"b5939fe7-0f0d-4acf-8582-d45edc0669f8"
],
"Type": "CHILD"
}
],
"RowIndex": null,
"RowSpan": null,
"SelectionStatus": "",
"Text": "Hello World!",
"TextType": ""
},
{
"BlockType": "WORD",
"ColumnIndex": null,
"ColumnSpan": null,
"Confidence": 99.915146,
"EntityTypes": null,
"Geometry": {
"BoundingBox": {
"Height": 0.023926819,
"Left": 0.16538842,
"Top": 0.093094304,
"Width": 0.086175606
},
"Polygon": [
{
"X": 0.16540693,
"Y": 0.093094304
},
{
"X": 0.25156403,
"Y": 0.09322314
},
{
"X": 0.25154623,
"Y": 0.11702112
},
{
"X": 0.16538842,
"Y": 0.11689187
}
]
},
"Id": "b2eadea5-a686-48fe-9363-a774a0a146b4",
"Page": null,
"Query": null,
"Relationships": null,
"RowIndex": null,
"RowSpan": null,
"SelectionStatus": "",
"Text": "Hello",
"TextType": "PRINTED"
},
{
"BlockType": "WORD",
"ColumnIndex": null,
"ColumnSpan": null,
"Confidence": 99.87717,
"EntityTypes": null,
"Geometry": {
"BoundingBox": {
"Height": 0.02385673,
"Left": 0.26209888,
"Top": 0.09313698,
"Width": 0.1124408
},
"Polygon": [
{
"X": 0.2621165,
"Y": 0.09313698
},
{
"X": 0.3745397,
"Y": 0.09330509
},
{
"X": 0.374523,
"Y": 0.1169937
},
{
"X": 0.26209888,
"Y": 0.116825044
}
]
},
"Id": "b5939fe7-0f0d-4acf-8582-d45edc0669f8",
"Page": null,
"Query": null,
"Relationships": null,
"RowIndex": null,
"RowSpan": null,
"SelectionStatus": "",
"Text": "World!",
"TextType": "PRINTED"
"HumanLoopActivationOutput": null,
"ResultMetadata": {}
}
],
"DetectDocumentTextModelVersion": "1.0",
"DocumentMetadata": {
"Pages": 1
},
"ResultMetadata": {}
"table": []
}