[formrecognizer] add testcases for labeled tables - fixed/variable ro…

…ws (#18214) * add testcases for labeled tables - fixed/variable rows * fix input param * shorten test variable names * update test-resources.json env var names
Azure · Apr 22, 2021 · e135c69 · e135c69
1 parent 5ec4667
commit e135c69
Show file tree

Hide file tree

Showing 9 changed files with 1,119 additions and 4 deletions.
diff --git a/sdk/formrecognizer/azure-ai-formrecognizer/tests/preparers.py b/sdk/formrecognizer/azure-ai-formrecognizer/tests/preparers.py
@@ -25,6 +25,8 @@
     formrecognizer_multipage_storage_container_sas_url="container_sas_url",
     formrecognizer_multipage_storage_container_sas_url_2="container_sas_url",
     formrecognizer_selection_mark_storage_container_sas_url="container_sas_url",
+    formrecognizer_table_variable_rows_container_sas_url="container_sas_url",
+    formrecognizer_table_fixed_rows_container_sas_url="container_sas_url",
     formrecognizer_resource_id="/subscriptions/00000000-0000-0000-0000-000000000000/resourceGroups/rgname/providers/Microsoft.CognitiveServices/accounts/frname",
     formrecognizer_region="region"
 )

diff --git a/...mrecognizer/tests/recordings/test_custom_forms_from_url.test_label_tables_fixed_rows.yaml b/...mrecognizer/tests/recordings/test_custom_forms_from_url.test_label_tables_fixed_rows.yaml
@@ -0,0 +1,256 @@
+interactions:
+- request:
+    body: 'b''{"source": "container_sas_url", "sourceFilter": {"prefix": "", "includeSubFolders":
+      false}, "useLabelFile": true}'''
+    headers:
+      Accept:
+      - application/json
+      Accept-Encoding:
+      - gzip, deflate
+      Connection:
+      - keep-alive
+      Content-Length:
+      - '302'
+      Content-Type:
+      - application/json
+      User-Agent:
+      - azsdk-python-ai-formrecognizer/3.1.0b5 Python/3.9.0 (Windows-10-10.0.19041-SP0)
+    method: POST
+    uri: https://region.api.cognitive.microsoft.com/formrecognizer/v2.1-preview.3/custom/models
+  response:
+    body:
+      string: ''
+    headers:
+      apim-request-id:
+      - fb88ba21-8e3e-4a26-a4d2-ca2173ffc2b5
+      content-length:
+      - '0'
+      date:
+      - Thu, 22 Apr 2021 01:29:38 GMT
+      location:
+      - https://region.api.cognitive.microsoft.com/formrecognizer/v2.1-preview.3/custom/models/d55d7c8d-1f62-43ad-b083-317205e12621
+      strict-transport-security:
+      - max-age=31536000; includeSubDomains; preload
+      x-content-type-options:
+      - nosniff
+      x-envoy-upstream-service-time:
+      - '533'
+    status:
+      code: 201
+      message: Created
+- request:
+    body: null
+    headers:
+      Accept:
+      - '*/*'
+      Accept-Encoding:
+      - gzip, deflate
+      Connection:
+      - keep-alive
+      User-Agent:
+      - azsdk-python-ai-formrecognizer/3.1.0b5 Python/3.9.0 (Windows-10-10.0.19041-SP0)
+    method: GET
+    uri: https://region.api.cognitive.microsoft.com/formrecognizer/v2.1-preview.3/custom/models/d55d7c8d-1f62-43ad-b083-317205e12621?includeKeys=true
+  response:
+    body:
+      string: '{"modelInfo": {"modelId": "d55d7c8d-1f62-43ad-b083-317205e12621", "attributes":
+        {"isComposed": false}, "status": "ready", "createdDateTime": "2021-04-22T01:29:38Z",
+        "lastUpdatedDateTime": "2021-04-22T01:29:41Z"}, "trainResult": {"averageModelAccuracy":
+        0.995, "trainingDocuments": [{"documentName": "label_table_fixed_rows1.pdf",
+        "pages": 1, "status": "succeeded"}, {"documentName": "label_table_fixed_rows2.pdf",
+        "pages": 1, "status": "succeeded"}, {"documentName": "label_table_fixed_rows3.pdf",
+        "pages": 1, "status": "succeeded"}, {"documentName": "label_table_fixed_rows4.pdf",
+        "pages": 1, "status": "succeeded"}, {"documentName": "label_table_fixed_rows5.pdf",
+        "pages": 1, "status": "succeeded"}], "fields": [{"fieldName": "table", "accuracy":
+        0.995}], "errors": []}}'
+    headers:
+      apim-request-id:
+      - 59a83ea2-d358-4620-8d0d-5d569c20b9fd
+      content-length:
+      - '720'
+      content-type:
+      - application/json; charset=utf-8
+      date:
+      - Thu, 22 Apr 2021 01:29:43 GMT
+      strict-transport-security:
+      - max-age=31536000; includeSubDomains; preload
+      x-content-type-options:
+      - nosniff
+      x-envoy-upstream-service-time:
+      - '17'
+    status:
+      code: 200
+      message: OK
+- request:
+    body: 'b''{"source": "blob_sas_url"}'''
+    headers:
+      Accept:
+      - application/json
+      Accept-Encoding:
+      - gzip, deflate
+      Connection:
+      - keep-alive
+      Content-Length:
+      - '230'
+      Content-Type:
+      - application/json
+      User-Agent:
+      - azsdk-python-ai-formrecognizer/3.1.0b5 Python/3.9.0 (Windows-10-10.0.19041-SP0)
+    method: POST
+    uri: https://region.api.cognitive.microsoft.com/formrecognizer/v2.1-preview.3/custom/models/d55d7c8d-1f62-43ad-b083-317205e12621/analyze?includeTextDetails=false
+  response:
+    body:
+      string: ''
+    headers:
+      apim-request-id:
+      - 10f98cd9-0aa5-4987-84b1-9cb206208c54
+      content-length:
+      - '0'
+      date:
+      - Thu, 22 Apr 2021 01:29:43 GMT
+      operation-location:
+      - https://region.api.cognitive.microsoft.com/formrecognizer/v2.1-preview.3/custom/models/d55d7c8d-1f62-43ad-b083-317205e12621/analyzeresults/59b6a250-6815-45c3-9029-d67a1a4f6571
+      strict-transport-security:
+      - max-age=31536000; includeSubDomains; preload
+      x-content-type-options:
+      - nosniff
+      x-envoy-upstream-service-time:
+      - '347'
+    status:
+      code: 202
+      message: Accepted
+- request:
+    body: null
+    headers:
+      Accept:
+      - '*/*'
+      Accept-Encoding:
+      - gzip, deflate
+      Connection:
+      - keep-alive
+      User-Agent:
+      - azsdk-python-ai-formrecognizer/3.1.0b5 Python/3.9.0 (Windows-10-10.0.19041-SP0)
+    method: GET
+    uri: https://region.api.cognitive.microsoft.com/formrecognizer/v2.1-preview.3/custom/models/d55d7c8d-1f62-43ad-b083-317205e12621/analyzeresults/59b6a250-6815-45c3-9029-d67a1a4f6571
+  response:
+    body:
+      string: '{"status": "notStarted", "createdDateTime": "2021-04-22T01:29:44Z",
+        "lastUpdatedDateTime": "2021-04-22T01:29:48Z"}'
+    headers:
+      apim-request-id:
+      - 27da524d-3b74-4832-ac70-2fee5d03f1dd
+      content-length:
+      - '109'
+      content-type:
+      - application/json; charset=utf-8
+      date:
+      - Thu, 22 Apr 2021 01:29:49 GMT
+      strict-transport-security:
+      - max-age=31536000; includeSubDomains; preload
+      x-content-type-options:
+      - nosniff
+      x-envoy-upstream-service-time:
+      - '15'
+    status:
+      code: 200
+      message: OK
+- request:
+    body: null
+    headers:
+      Accept:
+      - '*/*'
+      Accept-Encoding:
+      - gzip, deflate
+      Connection:
+      - keep-alive
+      User-Agent:
+      - azsdk-python-ai-formrecognizer/3.1.0b5 Python/3.9.0 (Windows-10-10.0.19041-SP0)
+    method: GET
+    uri: https://region.api.cognitive.microsoft.com/formrecognizer/v2.1-preview.3/custom/models/d55d7c8d-1f62-43ad-b083-317205e12621/analyzeresults/59b6a250-6815-45c3-9029-d67a1a4f6571
+  response:
+    body:
+      string: '{"status": "succeeded", "createdDateTime": "2021-04-22T01:29:44Z",
+        "lastUpdatedDateTime": "2021-04-22T01:29:51Z", "analyzeResult": {"version":
+        "2.1.0", "readResults": [{"page": 1, "angle": 0, "width": 8.5, "height": 11,
+        "unit": "inch"}], "pageResults": [{"page": 1, "tables": [{"rows": 5, "columns":
+        4, "cells": [{"rowIndex": 0, "columnIndex": 0, "text": "", "boundingBox":
+        [0.9904, 1.6215, 2.297, 1.6215, 2.297, 1.8221, 0.9958, 1.8272]}, {"rowIndex":
+        0, "columnIndex": 1, "text": "Price", "boundingBox": [2.297, 1.6215, 3.5874,
+        1.6215, 3.5928, 1.8221, 2.297, 1.8221]}, {"rowIndex": 0, "columnIndex": 2,
+        "text": "Tax", "boundingBox": [3.5874, 1.6215, 4.8994, 1.6215, 4.8994, 1.8221,
+        3.5928, 1.8221]}, {"rowIndex": 0, "columnIndex": 3, "text": "Total", "boundingBox":
+        [4.8994, 1.6215, 6.1952, 1.6215, 6.1952, 1.8272, 4.8994, 1.8221]}, {"rowIndex":
+        1, "columnIndex": 0, "text": "Apple", "boundingBox": [0.9958, 1.8272, 2.297,
+        1.8221, 2.297, 2.0078, 0.9958, 2.0128]}, {"rowIndex": 1, "columnIndex": 1,
+        "text": "5.00", "boundingBox": [2.297, 1.8221, 3.5928, 1.8221, 3.5928, 2.0078,
+        2.297, 2.0078]}, {"rowIndex": 1, "columnIndex": 2, "text": "2.00", "boundingBox":
+        [3.5928, 1.8221, 4.8994, 1.8221, 4.8994, 2.0078, 3.5928, 2.0078]}, {"rowIndex":
+        1, "columnIndex": 3, "text": "7.00", "boundingBox": [4.8994, 1.8221, 6.1952,
+        1.8272, 6.1952, 2.0128, 4.8994, 2.0078]}, {"rowIndex": 2, "columnIndex": 0,
+        "text": "Orange", "boundingBox": [0.9958, 2.0128, 2.297, 2.0078, 2.297, 2.2034,
+        0.9958, 2.2084]}, {"rowIndex": 2, "columnIndex": 1, "text": "2.00", "boundingBox":
+        [2.297, 2.0078, 3.5928, 2.0078, 3.5928, 2.2034, 2.297, 2.2034]}, {"rowIndex":
+        2, "columnIndex": 2, "text": "1.00", "boundingBox": [3.5928, 2.0078, 4.8994,
+        2.0078, 4.8994, 2.2034, 3.5928, 2.2034]}, {"rowIndex": 2, "columnIndex": 3,
+        "text": "3.00", "boundingBox": [4.8994, 2.0078, 6.1952, 2.0128, 6.1952, 2.2034,
+        4.8994, 2.2034]}, {"rowIndex": 3, "columnIndex": 0, "text": "Grapefruit",
+        "boundingBox": [0.9958, 2.2084, 2.297, 2.2034, 2.297, 2.3991, 0.9958, 2.3991]},
+        {"rowIndex": 3, "columnIndex": 1, "text": "1.00", "boundingBox": [2.297, 2.2034,
+        3.5928, 2.2034, 3.5928, 2.3991, 2.297, 2.3991]}, {"rowIndex": 3, "columnIndex":
+        2, "text": "4.00", "boundingBox": [3.5928, 2.2034, 4.8994, 2.2034, 4.8994,
+        2.3991, 3.5928, 2.3991]}, {"rowIndex": 3, "columnIndex": 3, "text": "5.00",
+        "boundingBox": [4.8994, 2.2034, 6.1952, 2.2034, 6.1952, 2.3991, 4.8994, 2.3991]},
+        {"rowIndex": 4, "columnIndex": 0, "text": "Pear", "boundingBox": [0.9958,
+        2.3991, 2.297, 2.3991, 2.297, 2.5947, 0.9958, 2.5947]}, {"rowIndex": 4, "columnIndex":
+        1, "text": "", "boundingBox": [2.297, 2.3991, 3.5928, 2.3991, 3.5982, 2.5947,
+        2.297, 2.5947]}, {"rowIndex": 4, "columnIndex": 2, "text": "2.00", "boundingBox":
+        [3.5928, 2.3991, 4.8994, 2.3991, 4.8994, 2.5947, 3.5982, 2.5947]}, {"rowIndex":
+        4, "columnIndex": 3, "text": "2.00", "boundingBox": [4.8994, 2.3991, 6.1952,
+        2.3991, 6.1952, 2.5947, 4.8994, 2.5947]}], "boundingBox": [0.9986, 1.6168,
+        6.2004, 1.6179, 6.2004, 2.597, 0.9981, 2.5961]}]}], "documentResults": [{"docType":
+        "custom:d55d7c8d-1f62-43ad-b083-317205e12621", "modelId": "d55d7c8d-1f62-43ad-b083-317205e12621",
+        "pageRange": [1, 1], "fields": {"table": {"type": "object", "valueObject":
+        {"Apple": {"type": "object", "valueObject": {"Price": {"type": "string", "valueString":
+        "5.00", "text": "5.00", "page": 1, "boundingBox": [2.385, 1.87, 2.645, 1.87,
+        2.645, 1.97, 2.385, 1.97]}, "Tax": {"type": "string", "valueString": "2.00",
+        "text": "2.00", "page": 1, "boundingBox": [3.685, 1.87, 3.945, 1.87, 3.945,
+        1.97, 3.685, 1.97]}, "Total": {"type": "string", "valueString": "7.00", "text":
+        "7.00", "page": 1, "boundingBox": [4.985, 1.87, 5.245, 1.87, 5.245, 1.97,
+        4.985, 1.97]}}}, "Orange": {"type": "object", "valueObject": {"Price": {"type":
+        "string", "valueString": "2.00", "text": "2.00", "page": 1, "boundingBox":
+        [2.385, 2.065, 2.645, 2.065, 2.645, 2.165, 2.385, 2.165]}, "Tax": {"type":
+        "string", "valueString": "1.00", "text": "1.00", "page": 1, "boundingBox":
+        [3.69, 2.065, 3.945, 2.065, 3.945, 2.165, 3.69, 2.165]}, "Total": {"type":
+        "string", "valueString": "3.00", "text": "3.00", "page": 1, "boundingBox":
+        [4.985, 2.065, 5.245, 2.065, 5.245, 2.165, 4.985, 2.165]}}}, "Grapefruit":
+        {"type": "object", "valueObject": {"Price": {"type": "string", "valueString":
+        "1.00", "text": "1.00", "page": 1, "boundingBox": [2.39, 2.26, 2.645, 2.26,
+        2.645, 2.36, 2.39, 2.36]}, "Tax": {"type": "string", "valueString": "4.00",
+        "text": "4.00", "page": 1, "boundingBox": [3.68, 2.26, 3.945, 2.26, 3.945,
+        2.36, 3.68, 2.36]}, "Total": {"type": "string", "valueString": "5.00", "text":
+        "5.00", "page": 1, "boundingBox": [4.985, 2.26, 5.245, 2.26, 5.245, 2.36,
+        4.985, 2.36]}}}, "Pear": {"type": "object", "valueObject": {"Price": null,
+        "Tax": {"type": "string", "valueString": "2.00", "text": "2.00", "page": 1,
+        "boundingBox": [3.685, 2.45, 3.945, 2.45, 3.945, 2.55, 3.685, 2.55]}, "Total":
+        {"type": "string", "valueString": "2.00", "text": "2.00", "page": 1, "boundingBox":
+        [4.985, 2.45, 5.245, 2.45, 5.245, 2.55, 4.985, 2.55]}}}}}}, "docTypeConfidence":
+        0.979}], "errors": []}}'
+    headers:
+      apim-request-id:
+      - 31eb026f-ea60-4f17-8c92-c6673073fd15
+      content-length:
+      - '4533'
+      content-type:
+      - application/json; charset=utf-8
+      date:
+      - Thu, 22 Apr 2021 01:29:54 GMT
+      strict-transport-security:
+      - max-age=31536000; includeSubDomains; preload
+      x-content-type-options:
+      - nosniff
+      x-envoy-upstream-service-time:
+      - '19'
+    status:
+      code: 200
+      message: OK
+version: 1