Merge pull request #12 from microsoft/laserprec/use_tox

Use tox in CI pipeline
microsoft · Jan 26, 2021 · d738bbb · d738bbb
2 parents 8da3807 + 1ea5dd4
commit d738bbb
Show file tree

Hide file tree

Showing 72 changed files with 145 additions and 61 deletions.
diff --git a/CODEOWNERS b/CODEOWNERS
@@ -9,6 +9,6 @@
 genalog/degradation/ @laserprec
 genalog/generation/ @laserprec
 genalog/text/ @laserprec
-genalog/ocr/ @dbanda @laserprec
+genalog/ocr/ @laserprec
 
-tests/ @laserprec @dbanda
+tests/ @laserprec
diff --git a/MANIFEST.in b/MANIFEST.in
@@ -0,0 +1,6 @@
+include *.md
+include *.txt
+include LICENSE CODEOWNERS
+include .gitignore tox.ini MANIFEST.in
+recursive-include genalog *.py *.jinja
+recursive-include tests *.py *.jinja *.jpg
diff --git a/devops/pr-gate.yml b/devops/pr-gate.yml
@@ -43,12 +43,12 @@ steps:
   displayName: 'Install dependencies'
 
 - bash: |
-    python -m flake8
+    tox -e flake8
   workingDirectory: $(Build.SourcesDirectory)
   displayName: 'Run Linter (flake8)'
 
 - bash: |
-    python -m pytest tests
+    tox -e py
   env:
     BLOB_KEY : $(BLOB_KEY)
     SEARCH_SERVICE_KEY: $(SEARCH_SERVICE_KEY)

diff --git a/genalog/ocr/README.md b/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")
     ```
 
diff --git a/requirements-dev.txt b/requirements-dev.txt
@@ -1,4 +1,5 @@
+flake8
+flake8-import-order
 pytest
 pytest-cov
-flake8
-flake8-import-order
+tox
diff --git a/setup.py b/setup.py
@@ -22,7 +22,11 @@
     long_description=long_description,
     long_description_content_type="text/markdown",
     url='https://msazure.visualstudio.com/DefaultCollection/Cognitive%20Services/_git/Tools-Synthetic-Data-Generator',
-    packages=setuptools.find_packages(),
+    packages=setuptools.find_packages(exclude=['tests', 'tests.*']),
+    package_data={'': [
+        'genalog/generation/templates/*.jinja'
+    ]},
+    include_package_data=True,
     classifiers=[
         "Programming Language :: Python :: 3",
         "Operating System :: OS Independent",

diff --git a/tests/ocr/.env → tests/.env b/tests/ocr/.env → tests/.env
diff --git a/tests/conftest.py b/tests/conftest.py
@@ -0,0 +1,19 @@
+import logging
+import os
+
+import pytest
+from dotenv import load_dotenv
+
+from tests.required_env import RequiredEnvVar
+
+ENV_FILEPATH = "tests/.env"
+
+
+@pytest.fixture(scope="session")
+def load_azure_resources():
+    # Loading the non-secrets
+    load_dotenv(ENV_FILEPATH)
+    logging.info(f"Loading .env from {ENV_FILEPATH}")
+    logging.debug("Printing environment vars: ")
+    for env in RequiredEnvVar:
+        logging.debug(f"\t{env.value}: {os.environ.get(env.value)}")
diff --git a/tests/e2e/test_anchor_e2e.py b/tests/e2e/test_anchor_e2e.py
@@ -7,11 +7,12 @@
 from genalog.text import alignment, anchor, preprocess
 
 
+@pytest.mark.slow
 @pytest.mark.parametrize(
     "gt_file, ocr_file",
     zip(
-        sorted(glob.glob("tests/text/data/gt_*.txt")),
-        sorted(glob.glob("tests/text/data/ocr_*.txt")),
+        sorted(glob.glob("tests/unit/text/data/gt_*.txt")),
+        sorted(glob.glob("tests/unit/text/data/ocr_*.txt")),
     ),
 )
 def test_align_w_anchor_and_align(gt_file, ocr_file):
@@ -34,11 +35,12 @@ def test_align_w_anchor_and_align(gt_file, ocr_file):
         )
 
 
+@pytest.mark.slow
 @pytest.mark.parametrize(
     "gt_file, ocr_file",
     zip(
-        sorted(glob.glob("tests/text/data/gt_*.txt")),
-        sorted(glob.glob("tests/text/data/ocr_*.txt")),
+        sorted(glob.glob("tests/unit/text/data/gt_*.txt")),
+        sorted(glob.glob("tests/unit/text/data/ocr_*.txt")),
     ),
 )
 @pytest.mark.parametrize("max_seg_length", [25, 50, 75, 100, 150])

diff --git a/tests/e2e/test_conll_format_e2e.py b/tests/e2e/test_conll_format_e2e.py
@@ -6,6 +6,7 @@
 from genalog.text import conll_format
 
 
+@pytest.mark.slow
 @pytest.mark.parametrize(
     "required_args", [(["tests/e2e/data/synthetic_dataset", "test_version"])]
 )
@@ -27,6 +28,7 @@ def test_conll_format(required_args, optional_args):
 basepath = "tests/e2e/data/conll_formatter/"
 
 
+@pytest.mark.slow
 @pytest.mark.parametrize(
     "clean_label_filename, ocr_text_filename",
     zip(

diff --git a/tests/e2e/test_document_generation.py b/tests/e2e/test_document_generation.py
@@ -11,7 +11,7 @@
 UNSUPPORTED_CONTENT_FORMAT = ["foo bar"]
 UNSUPPORTED_CONTENT_TYPE = CompositeContent(["foo"], [ContentType.TITLE])
 
-CUSTOM_TEMPLATE_PATH = "tests/generation/templates"
+CUSTOM_TEMPLATE_PATH = "tests/unit/generation/templates"
 CUSTOM_TEMPLATE_NAME = "mock.html.jinja"
 CUSTOM_STYLE_TEMPLATE_NAME = "font_family.html.jinja"
 MULTI_PAGE_TEMPLATE_NAME = "multipage.html.jinja"

diff --git a/tests/e2e/test_ocr_e2e.py b/tests/e2e/test_ocr_e2e.py
@@ -1,19 +1,24 @@
 import json
 
 import pytest
-from dotenv import load_dotenv
 
 from genalog.ocr.blob_client import GrokBlobClient
 from genalog.ocr.grok import Grok
 
-load_dotenv("tests/ocr/.env")
 
+@pytest.fixture(scope="module", autouse=True)
+def load_azure_config(load_azure_resources):
+    # Loading the non-secrets
+    # Assume the secrets are set in the environment variable prior
+    pass
 
+
+@pytest.mark.azure
 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
@@ -43,11 +48,12 @@ def test_upload_images(self, use_async):
         ), f"folder {dst_folder} was not deleted"
 
 
+@pytest.mark.azure
 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,

diff --git a/tests/e2e/test_pipeline.py b/tests/e2e/test_pipeline.py
@@ -4,7 +4,7 @@
 
 from genalog import pipeline
 
-EXAMPLE_TEXT_FILE = "tests/text/data/gt_1.txt"
+EXAMPLE_TEXT_FILE = "tests/unit/text/data/gt_1.txt"
 
 
 @pytest.fixture
@@ -36,7 +36,7 @@ def test_custom_generate_img(custom_analog_generator):
 
 
 def test_generate_dataset_multiprocess():
-    INPUT_TEXT_FILENAMES = glob.glob("tests/text/data/gt_*.txt")
+    INPUT_TEXT_FILENAMES = glob.glob("tests/unit/text/data/gt_*.txt")
     with pytest.deprecated_call():
         pipeline.generate_dataset_multiprocess(
             INPUT_TEXT_FILENAMES, "test_out", {}, [], "text_block.html.jinja"

diff --git a/tests/ocr/data/substitution.json b/tests/ocr/data/substitution.json
diff --git a/tests/required_env.py b/tests/required_env.py
@@ -0,0 +1,24 @@
+from enum import Enum
+from itertools import chain
+
+
+class RequiredSecrets(Enum):
+    BLOB_KEY = 'BLOB_KEY'
+    SEARCH_SERVICE_KEY = 'SEARCH_SERVICE_KEY'
+    COGNITIVE_SERVICE_KEY = 'COGNITIVE_SERVICE_KEY'
+
+
+class RequiredConstants(Enum):
+    COMPUTER_VISION_ENDPOINT = 'COMPUTER_VISION_ENDPOINT'
+    SEARCH_SERVICE_NAME = 'SEARCH_SERVICE_NAME'
+    SKILLSET_NAME = 'SKILLSET_NAME'
+    INDEX_NAME = "INDEX_NAME"
+    INDEXER_NAME = "INDEXER_NAME"
+    DATASOURCE_NAME = "DATASOURCE_NAME"
+    DATASOURCE_CONTAINER_NAME = "DATASOURCE_CONTAINER_NAME"
+    BLOB_NAME = "BLOB_NAME"
+
+
+RequiredEnvVar = Enum("RequiredEnvVar", [
+    (i.name, i.value) for i in chain(RequiredSecrets, RequiredConstants)
+])
diff --git a/tests/cases/__init__.py → tests/unit/__init__.py b/tests/cases/__init__.py → tests/unit/__init__.py
diff --git a/tests/unit/cases/__init__.py b/tests/unit/cases/__init__.py
diff --git a/tests/cases/label_propagation.py → tests/unit/cases/label_propagation.py b/tests/cases/label_propagation.py → tests/unit/cases/label_propagation.py
diff --git a/tests/cases/text_alignment.py → tests/unit/cases/text_alignment.py b/tests/cases/text_alignment.py → tests/unit/cases/text_alignment.py
diff --git a/tests/unit/degradation/__init__.py b/tests/unit/degradation/__init__.py
diff --git a/tests/degradation/test_degrader.py → tests/unit/degradation/test_degrader.py b/tests/degradation/test_degrader.py → tests/unit/degradation/test_degrader.py
diff --git a/tests/degradation/test_effect.py → tests/unit/degradation/test_effect.py b/tests/degradation/test_effect.py → tests/unit/degradation/test_effect.py
diff --git a/tests/generation/2x2.jpg → tests/unit/generation/2x2.jpg b/tests/generation/2x2.jpg → tests/unit/generation/2x2.jpg
diff --git a/tests/unit/generation/__init__.py b/tests/unit/generation/__init__.py
diff --git a/...neration/templates/font_family.html.jinja → ...neration/templates/font_family.html.jinja b/...neration/templates/font_family.html.jinja → ...neration/templates/font_family.html.jinja
diff --git a/tests/generation/templates/mock.html.jinja → ...unit/generation/templates/mock.html.jinja b/tests/generation/templates/mock.html.jinja → ...unit/generation/templates/mock.html.jinja
diff --git a/...generation/templates/multipage.html.jinja → ...generation/templates/multipage.html.jinja b/...generation/templates/multipage.html.jinja → ...generation/templates/multipage.html.jinja
diff --git a/tests/generation/test_content.py → tests/unit/generation/test_content.py b/tests/generation/test_content.py → tests/unit/generation/test_content.py
diff --git a/tests/generation/test_document.py → tests/unit/generation/test_document.py b/tests/generation/test_document.py → tests/unit/generation/test_document.py
@@ -12,12 +12,12 @@
 MOCK_TEMPLATE = MagicMock()
 MOCK_TEMPLATE.render.return_value = MOCK_COMPILED_DOCUMENT
 
-IMG_BYTES = open("tests/generation/2x2.jpg", "rb").read()
+IMG_BYTES = open("tests/unit/generation/2x2.jpg", "rb").read()
 
 FILE_DESTINATION_PDF = "sample.pdf"
 FILE_DESTINATION_PNG = "sample.png"
 
-CUSTOM_TEMPLATE_PATH = "tests/generation/templates"
+CUSTOM_TEMPLATE_PATH = "tests/unit/generation/templates"
 CUSTOM_TEMPLATE_NAME = "mock.html.jinja"
 DEFAULT_TEMPLATE_NAME = "text_block.html.jinja"
 DEFAULT_PACKAGE_NAME = "genalog.generation"

diff --git a/tests/unit/ocr/__init__.py b/tests/unit/ocr/__init__.py
diff --git a/tests/ocr/data/img/0.png → tests/unit/ocr/data/img/0.png b/tests/ocr/data/img/0.png → tests/unit/ocr/data/img/0.png
diff --git a/tests/ocr/data/img/1.png → tests/unit/ocr/data/img/1.png b/tests/ocr/data/img/1.png → tests/unit/ocr/data/img/1.png
diff --git a/tests/ocr/data/img/11.png → tests/unit/ocr/data/img/11.png b/tests/ocr/data/img/11.png → tests/unit/ocr/data/img/11.png
diff --git a/...n/521c38122f783673598856cd81d91c21_0.json → ...n/521c38122f783673598856cd81d91c21_0.json b/...n/521c38122f783673598856cd81d91c21_0.json → ...n/521c38122f783673598856cd81d91c21_0.json
diff --git a/...n/521c38122f783673598856cd81d91c21_1.json → ...n/521c38122f783673598856cd81d91c21_1.json b/...n/521c38122f783673598856cd81d91c21_1.json → ...n/521c38122f783673598856cd81d91c21_1.json
diff --git a/.../521c38122f783673598856cd81d91c21_11.json → .../521c38122f783673598856cd81d91c21_11.json b/.../521c38122f783673598856cd81d91c21_11.json → .../521c38122f783673598856cd81d91c21_11.json
diff --git a/tests/ocr/data/metrics.csv → tests/unit/ocr/data/metrics.csv b/tests/ocr/data/metrics.csv → 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
-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
-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
+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/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/unit/ocr/data/text/11.txt,tests/unit/ocr/data/json/521c38122f783673598856cd81d91c21_11.png.json,11.txt
diff --git a/tests/ocr/data/metrics/json/123_001.json → ...s/unit/ocr/data/metrics/json/123_001.json b/tests/ocr/data/metrics/json/123_001.json → ...s/unit/ocr/data/metrics/json/123_001.json
diff --git a/tests/ocr/data/metrics/json/123_002.json → ...s/unit/ocr/data/metrics/json/123_002.json b/tests/ocr/data/metrics/json/123_002.json → ...s/unit/ocr/data/metrics/json/123_002.json
diff --git a/tests/ocr/data/metrics/json/123_003.json → ...s/unit/ocr/data/metrics/json/123_003.json b/tests/ocr/data/metrics/json/123_003.json → ...s/unit/ocr/data/metrics/json/123_003.json
diff --git a/tests/ocr/data/metrics/metrics.csv → tests/unit/ocr/data/metrics/metrics.csv b/tests/ocr/data/metrics/metrics.csv → 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
-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
-0,1,6,0,17,0,0,5,17,2659,460,459,2653,436,437,0.9498910675381264,0.95,0.9977435125987213,tests/ocr/data/metrics/text/003.txt,tests/ocr/data/metrics/json/123_003.png.json,003.txt
+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/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/unit/ocr/data/metrics/text/003.txt,tests/unit/ocr/data/metrics/json/123_003.png.json,003.txt
diff --git a/tests/ocr/data/metrics/substitution.pkl → tests/unit/ocr/data/metrics/substitution.pkl b/tests/ocr/data/metrics/substitution.pkl → tests/unit/ocr/data/metrics/substitution.pkl
diff --git a/tests/ocr/data/metrics/text/001.txt → tests/unit/ocr/data/metrics/text/001.txt b/tests/ocr/data/metrics/text/001.txt → tests/unit/ocr/data/metrics/text/001.txt
diff --git a/tests/ocr/data/metrics/text/002.txt → tests/unit/ocr/data/metrics/text/002.txt b/tests/ocr/data/metrics/text/002.txt → tests/unit/ocr/data/metrics/text/002.txt
diff --git a/tests/ocr/data/metrics/text/003.txt → tests/unit/ocr/data/metrics/text/003.txt b/tests/ocr/data/metrics/text/003.txt → tests/unit/ocr/data/metrics/text/003.txt
diff --git a/tests/unit/ocr/data/substitution.json b/tests/unit/ocr/data/substitution.json
@@ -0,0 +1 @@
+{"tests/unit/ocr/data/text\\0.txt": {}, "tests/unit/ocr/data/text\\1.txt": {}, "tests/unit/ocr/data/text\\11.txt": {}}
diff --git a/tests/ocr/data/substitution.pkl → tests/unit/ocr/data/substitution.pkl b/tests/ocr/data/substitution.pkl → tests/unit/ocr/data/substitution.pkl
diff --git a/tests/ocr/data/text/0.txt → tests/unit/ocr/data/text/0.txt b/tests/ocr/data/text/0.txt → tests/unit/ocr/data/text/0.txt
diff --git a/tests/ocr/data/text/1.txt → tests/unit/ocr/data/text/1.txt b/tests/ocr/data/text/1.txt → tests/unit/ocr/data/text/1.txt
diff --git a/tests/ocr/data/text/11.txt → tests/unit/ocr/data/text/11.txt b/tests/ocr/data/text/11.txt → tests/unit/ocr/data/text/11.txt
diff --git a/tests/ocr/test_metrics.py → tests/unit/ocr/test_metrics.py b/tests/ocr/test_metrics.py → tests/unit/ocr/test_metrics.py
diff --git a/tests/ocr/test_ocr.py → tests/unit/ocr/test_ocr.py b/tests/ocr/test_ocr.py → tests/unit/ocr/test_ocr.py
@@ -1,13 +1,17 @@
 import json
+import os
 
 import pytest
 import requests
-from dotenv import load_dotenv
 
 from genalog.ocr.rest_client import GrokRestClient
 
 
-load_dotenv("tests/ocr/.env")
+@pytest.fixture(scope="module", autouse=True)
+def set_azure_dummy_secrets(load_azure_resources):
+    os.environ['BLOB_KEY'] = "<YOUR BLOB KEY>"
+    os.environ['SEARCH_SERVICE_KEY'] = "<YOUR SEARCH SERVICE KEY>"
+    os.environ['COGNITIVE_SERVICE_KEY'] = "<YOUR COGNITIVE SERVICE KEY>"
 
 
 @pytest.fixture(autouse=True)
@@ -46,7 +50,7 @@ def json(self):
                             "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",
                                 )
                             ),
@@ -55,7 +59,7 @@ def json(self):
                             "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",
                                 )
                             ),
@@ -64,7 +68,7 @@ def json(self):
                             "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",
                                 )
                             ),

diff --git a/tests/text/data/gt_1.txt → tests/unit/text/data/gt_1.txt b/tests/text/data/gt_1.txt → tests/unit/text/data/gt_1.txt
diff --git a/tests/text/data/gt_2.txt → tests/unit/text/data/gt_2.txt b/tests/text/data/gt_2.txt → tests/unit/text/data/gt_2.txt
diff --git a/tests/text/data/gt_3.txt → tests/unit/text/data/gt_3.txt b/tests/text/data/gt_3.txt → tests/unit/text/data/gt_3.txt
diff --git a/tests/text/data/label_generator/labels/0.tsv → ...it/text/data/label_generator/labels/0.tsv b/tests/text/data/label_generator/labels/0.tsv → ...it/text/data/label_generator/labels/0.tsv
diff --git a/tests/text/data/label_generator/labels/1.tsv → ...it/text/data/label_generator/labels/1.tsv b/tests/text/data/label_generator/labels/1.tsv → ...it/text/data/label_generator/labels/1.tsv
diff --git a/...s/text/data/label_generator/labels/11.tsv → ...t/text/data/label_generator/labels/11.tsv b/...s/text/data/label_generator/labels/11.tsv → ...t/text/data/label_generator/labels/11.tsv
diff --git a/tests/text/data/label_generator/text/0.txt → ...unit/text/data/label_generator/text/0.txt b/tests/text/data/label_generator/text/0.txt → ...unit/text/data/label_generator/text/0.txt
diff --git a/tests/text/data/label_generator/text/1.txt → ...unit/text/data/label_generator/text/1.txt b/tests/text/data/label_generator/text/1.txt → ...unit/text/data/label_generator/text/1.txt
diff --git a/tests/text/data/label_generator/text/11.txt → ...nit/text/data/label_generator/text/11.txt b/tests/text/data/label_generator/text/11.txt → ...nit/text/data/label_generator/text/11.txt
diff --git a/tests/text/data/ocr_1.txt → tests/unit/text/data/ocr_1.txt b/tests/text/data/ocr_1.txt → tests/unit/text/data/ocr_1.txt
diff --git a/tests/text/data/ocr_2.txt → tests/unit/text/data/ocr_2.txt b/tests/text/data/ocr_2.txt → tests/unit/text/data/ocr_2.txt
diff --git a/tests/text/data/ocr_3.txt → tests/unit/text/data/ocr_3.txt b/tests/text/data/ocr_3.txt → tests/unit/text/data/ocr_3.txt
diff --git a/tests/text/test_alignment.py → tests/unit/text/test_alignment.py b/tests/text/test_alignment.py → tests/unit/text/test_alignment.py
@@ -5,8 +5,8 @@
 import pytest
 
 from genalog.text import alignment
-from tests.cases.text_alignment import ALIGNMENT_REGRESSION_TEST_CASES
-from tests.cases.text_alignment import PARSE_ALIGNMENT_REGRESSION_TEST_CASES
+from tests.unit.cases.text_alignment import ALIGNMENT_REGRESSION_TEST_CASES
+from tests.unit.cases.text_alignment import PARSE_ALIGNMENT_REGRESSION_TEST_CASES
 
 RANDOM_INT = randint(1, 100)
 MOCK_ALIGNMENT_RESULT = [("X", "X", 0, 0, 1)]

diff --git a/tests/text/test_anchor.py → tests/unit/text/test_anchor.py b/tests/text/test_anchor.py → tests/unit/text/test_anchor.py
@@ -4,7 +4,7 @@
 import pytest
 
 from genalog.text import alignment, anchor, preprocess
-from tests.cases.text_alignment import ALIGNMENT_REGRESSION_TEST_CASES
+from tests.unit.cases.text_alignment import ALIGNMENT_REGRESSION_TEST_CASES
 
 
 @pytest.mark.parametrize(
@@ -200,8 +200,8 @@ def test_find_anchor_recur_fixed_seg_len(
 @pytest.mark.parametrize(
     "gt_file, ocr_file",
     zip(
-        sorted(glob.glob("tests/text/data/gt_1.txt")),
-        sorted(glob.glob("tests/text/data/ocr_1.txt")),
+        sorted(glob.glob("tests/unit/text/data/gt_1.txt")),
+        sorted(glob.glob("tests/unit/text/data/ocr_1.txt")),
     ),
 )
 @pytest.mark.parametrize("max_seg_length", [75])

diff --git a/tests/text/test_conll_format.py → tests/unit/text/test_conll_format.py b/tests/text/test_conll_format.py → tests/unit/text/test_conll_format.py
diff --git a/tests/text/test_lcs.py → tests/unit/text/test_lcs.py b/tests/text/test_lcs.py → tests/unit/text/test_lcs.py
diff --git a/tests/text/test_ner_label.py → tests/unit/text/test_ner_label.py b/tests/text/test_ner_label.py → tests/unit/text/test_ner_label.py
@@ -1,7 +1,7 @@
 import pytest
 
 from genalog.text import ner_label
-from tests.cases.label_propagation import LABEL_PROPAGATION_REGRESSION_TEST_CASES
+from tests.unit.cases.label_propagation import LABEL_PROPAGATION_REGRESSION_TEST_CASES
 
 
 @pytest.mark.parametrize(

diff --git a/tests/text/test_preprocess.py → tests/unit/text/test_preprocess.py b/tests/text/test_preprocess.py → tests/unit/text/test_preprocess.py
diff --git a/tests/text/test_utf8.py → tests/unit/text/test_utf8.py b/tests/text/test_utf8.py → tests/unit/text/test_utf8.py
@@ -5,7 +5,7 @@
 
 from genalog.text import alignment
 from genalog.text.alignment import GAP_CHAR
-from tests.cases.text_alignment import ALIGNMENT_REGRESSION_TEST_CASES
+from tests.unit.cases.text_alignment import ALIGNMENT_REGRESSION_TEST_CASES
 
 
 def random_utf8_char(byte_len=1):