Restructure tests/ocr/data

genalog/ocr/README.md

@@ -66,9 +66,9 @@ Example usage:
     ```python
     from genalog.ocr.grok import Grok
     from dotenv import load_dotenv
-    load_dotenv("tests/ocr/.env")
+    load_dotenv("tests/unit/ocr/.env")
 
     grok = Grok.create_from_env_var()
-    grok.run_grok(src_folder_path = "tests/ocr/data/img", dest_folder_path = "tests/ocr/data/json")
+    grok.run_grok(src_folder_path = "tests/unit/ocr/data/img", dest_folder_path = "tests/unit/ocr/data/json")
     ```
     

tests/e2e/test_ocr_e2e.py

@@ -18,7 +18,7 @@ class TestBlobClient:
     @pytest.mark.parametrize("use_async", [True, False])
     def test_upload_images(self, use_async):
         blob_client = GrokBlobClient.create_from_env_var()
-        subfolder = "tests/ocr/data/img"
+        subfolder = "tests/unit/ocr/data/img"
         subfolder.replace("/", "_")
         dst_folder, _ = blob_client.upload_images_to_blob(
             subfolder, use_async=use_async
@@ -53,7 +53,7 @@ class TestGROKe2e:
     @pytest.mark.parametrize("use_async", [False, True])
     def test_grok_e2e(self, tmpdir, use_async):
         grok = Grok.create_from_env_var()
-        src_folder = "tests/ocr/data/img"
+        src_folder = "tests/unit/ocr/data/img"
         grok.run_grok(
             src_folder,
             tmpdir,

tests/unit/ocr/data/metrics.csv

@@ -1,4 +1,4 @@
 edit_insert,edit_delete,edit_replace,edit_insert_spacing,edit_delete_spacing,insert,delete,replace,spacing,total_chars,total_words,total_alnum_words,matching_chars,matching_alnum_words,matching_words,alnum_word_accuracy,word_accuracy,char_accuracy,txt_path,ocr_json_path,filename
-1,0,0,1,13,1,0,0,14,1027,166,159,1025,144,150,0.9056603773584906,0.9036144578313253,0.9980525803310614,tests/ocr/data/text/0.txt,tests/ocr/data/json/521c38122f783673598856cd81d91c21_0.png.json,0.txt
+1,0,0,1,13,1,0,0,14,1027,166,159,1025,144,150,0.9056603773584906,0.9036144578313253,0.9980525803310614,tests/unit/ocr/data/text/0.txt,tests/unit/ocr/data/json/521c38122f783673598856cd81d91c21_0.png.json,0.txt
-3,0,0,0,5,3,0,0,5,958,182,176,955,165,171,0.9375,0.9395604395604396,0.9968684759916493,tests/ocr/data/text/1.txt,tests/ocr/data/json/521c38122f783673598856cd81d91c21_1.png.json,1.txt
+3,0,0,0,5,3,0,0,5,958,182,176,955,165,171,0.9375,0.9395604395604396,0.9968684759916493,tests/unit/ocr/data/text/1.txt,tests/unit/ocr/data/json/521c38122f783673598856cd81d91c21_1.png.json,1.txt
-2,0,0,0,9,2,0,0,9,1022,188,183,1020,170,175,0.9289617486338798,0.9308510638297872,0.9980430528375733,tests/ocr/data/text/11.txt,tests/ocr/data/json/521c38122f783673598856cd81d91c21_11.png.json,11.txt
+2,0,0,0,9,2,0,0,9,1022,188,183,1020,170,175,0.9289617486338798,0.9308510638297872,0.9980430528375733,tests/unit/ocr/data/text/11.txt,tests/unit/ocr/data/json/521c38122f783673598856cd81d91c21_11.png.json,11.txt

tests/unit/ocr/data/metrics/metrics.csv

@@ -1,4 +1,4 @@
 edit_insert,edit_delete,edit_replace,edit_insert_spacing,edit_delete_spacing,insert,delete,replace,spacing,total_chars,total_words,total_alnum_words,matching_chars,matching_alnum_words,matching_words,alnum_word_accuracy,word_accuracy,char_accuracy,txt_path,ocr_json_path,filename
-2,5,5,0,2,1,1,5,2,1068,176,176,1061,169,169,0.9602272727272727,0.9602272727272727,0.9934456928838952,tests/ocr/data/metrics/text/001.txt,tests/ocr/data/metrics/json/123_001.png.json,001.txt
+2,5,5,0,2,1,1,5,2,1068,176,176,1061,169,169,0.9602272727272727,0.9602272727272727,0.9934456928838952,tests/unit/ocr/data/metrics/text/001.txt,tests/unit/ocr/data/metrics/json/123_001.png.json,001.txt
-0,5,17,0,11,0,2,8,11,1789,301,301,1772,283,283,0.9401993355481728,0.9401993355481728,0.9904974846282839,tests/ocr/data/metrics/text/002.txt,tests/ocr/data/metrics/json/123_002.png.json,002.txt
+0,5,17,0,11,0,2,8,11,1789,301,301,1772,283,283,0.9401993355481728,0.9401993355481728,0.9904974846282839,tests/unit/ocr/data/metrics/text/002.txt,tests/unit/ocr/data/metrics/json/123_002.png.json,002.txt
-0,1,6,0,17,0,0,5,17,2659,460,459,2653,436,437,0.9498910675381264,0.95,0.9977435125987213,tests/ocr/data/metrics/text/003.txt,tests/ocr/data/metrics/json/123_003.png.json,003.txt
+0,1,6,0,17,0,0,5,17,2659,460,459,2653,436,437,0.9498910675381264,0.95,0.9977435125987213,tests/unit/ocr/data/metrics/text/003.txt,tests/unit/ocr/data/metrics/json/123_003.png.json,003.txt

tests/unit/ocr/data/substitution.json

@@ -1 +1 @@
-{"tests/ocr/data/text\\0.txt": {}, "tests/ocr/data/text\\1.txt": {}, "tests/ocr/data/text\\11.txt": {}}
+{"tests/unit/ocr/data/text\\0.txt": {}, "tests/unit/ocr/data/text\\1.txt": {}, "tests/unit/ocr/data/text\\11.txt": {}}

tests/unit/ocr/test_ocr.py

@@ -50,7 +50,7 @@ class MockedResponse:
                             "metadata_storage_name": "521c38122f783673598856cd81d91c21_0.png",
                             "layoutText": json.load(
                                 open(
-                                    "tests/ocr/data/json/521c38122f783673598856cd81d91c21_0.png.json",
+                                    "tests/unit/ocr/data/json/521c38122f783673598856cd81d91c21_0.png.json",
                                     "r",
                                 )
                             ),
@@ -59,7 +59,7 @@ class MockedResponse:
                             "metadata_storage_name": "521c38122f783673598856cd81d91c21_1.png",
                             "layoutText": json.load(
                                 open(
-                                    "tests/ocr/data/json/521c38122f783673598856cd81d91c21_1.png.json",
+                                    "tests/unit/ocr/data/json/521c38122f783673598856cd81d91c21_1.png.json",
                                     "r",
                                 )
                             ),
@@ -68,7 +68,7 @@ class MockedResponse:
                             "metadata_storage_name": "521c38122f783673598856cd81d91c21_11.png",
                             "layoutText": json.load(
                                 open(
-                                    "tests/ocr/data/json/521c38122f783673598856cd81d91c21_11.png.json",
+                                    "tests/unit/ocr/data/json/521c38122f783673598856cd81d91c21_11.png.json",
                                     "r",
                                 )
                             ),