added new integrations test and fix docker build on github AB#1035

.github/workflows/build-and-push-docker.yml

@@ -1,4 +1,4 @@
-name: Build and Push Image
+name: Build and Push Images
 
 on:
   release:
@@ -33,15 +33,16 @@ jobs:
           echo "Commit SHA (short): $short_sha"
           echo "short_sha=$short_sha" >> $GITHUB_ENV
 
-      - name: Build the Docker image
-        run: |
-          docker build -t ${{ secrets.DOCKERHUB_USERNAME }}/htrflow:latest .
-
-      - name: Tag the Docker image with Git tag and commit hash
+      # Build and push the htrflow.dockerfile image with both 'latest' and Git tag + commit hash
+      - name: Build and push htrflow image
         run: |
+          docker build -f docker/htrflow.dockerfile -t ${{ secrets.DOCKERHUB_USERNAME }}/htrflow:latest .
           docker tag ${{ secrets.DOCKERHUB_USERNAME }}/htrflow:latest ${{ secrets.DOCKERHUB_USERNAME }}/htrflow:$latest_tag-$short_sha
-
-      - name: Push Docker images to Docker Hub
-        run: |
           docker push ${{ secrets.DOCKERHUB_USERNAME }}/htrflow:latest
           docker push ${{ secrets.DOCKERHUB_USERNAME }}/htrflow:$latest_tag-$short_sha
+
+      # Build and push the htrflow_openmmlab.dockerfile image with Git tag and commit hash
+      - name: Build and push htrflow_openmmlab image
+        run: |
+          docker build -f docker/htrflow_openmmlab.dockerfile -t ${{ secrets.DOCKERHUB_USERNAME }}/htrflow_openmmlab:$latest_tag-$short_sha .
+          docker push ${{ secrets.DOCKERHUB_USERNAME }}/htrflow_openmmlab:$latest_tag-$short_sha

.github/workflows/tests.yml

@@ -23,4 +23,4 @@ jobs:
           python-version: ${{ matrix.python-version }}
 
       - name: Run tests
-        run: uv run pytest -m "not gpu"
+        run: uv run pytest -m "not gpu and not openmmlab"

pyproject.toml

@@ -82,6 +82,7 @@ addopts = "-v --cov htrflow --cov-report term-missing --no-cov-on-fail"
 markers = [
     "slow: marks tests as slow (deselect with '-m \"not slow\"')",
     "gpu: marks tests as gpu (deselect with '-m \"not gpu\"')",
+    "openmmlab: marks tests as openmmlab (deselect with '-m \"not openmmlab\"')"
 ]
 pythonpath = "src"
 testpaths = ["tests/unit", "tests/integration"]

src/htrflow/models/huggingface/llava_next.py

@@ -29,7 +29,7 @@ def __init__(
         self,
         model: str = "llava-hf/llava-v1.6-mistral-7b-hf",
         processor: str = "llava-hf/llava-v1.6-mistral-7b-hf",
-        prompt: str = "[INST] <image>\Please transcribe the handwritten English text displayed in the image [/INST]",
+        prompt: str = "[INST] <image> Please transcribe the handwritten English text displayed in the image [/INST]",
         device: str | None = None,
         *model_args,
     ):
@@ -51,7 +51,9 @@ def __init__(
 
         processor = processor or model
 
-        self.processor = LlavaNextProcessor.from_pretrained(processor, cache_dir=self.cache_dir, token=True)
+        self.processor = LlavaNextProcessor.from_pretrained(
+            processor, cache_dir=self.cache_dir, token=True
+        )
 
         self.prompt = prompt
 
@@ -71,14 +73,20 @@ def __init__(
 
     def _predict(self, images: list[np.ndarray], **generation_kwargs) -> list[Result]:
         generation_kwargs = self._prepare_generation_kwargs(**generation_kwargs)
-        metadata = self.metadata | ({"generation_args": generation_kwargs} if generation_kwargs else {})
+        metadata = self.metadata | (
+            {"generation_args": generation_kwargs} if generation_kwargs else {}
+        )
 
         prompts = [self.prompt] * len(images)
 
-        model_inputs = self.processor(prompts, images=images, return_tensors="pt", padding=True).to(self.model.device)
+        model_inputs = self.processor(
+            prompts, images=images, return_tensors="pt", padding=True
+        ).to(self.model.device)
         model_outputs = self.model.generate(**model_inputs, **generation_kwargs)
 
-        texts = self.processor.batch_decode(model_outputs.sequences, skip_special_tokens=True)
+        texts = self.processor.batch_decode(
+            model_outputs.sequences, skip_special_tokens=True
+        )
 
         scores = model_outputs.sequences_scores.tolist()
         step = generation_kwargs["num_return_sequences"]
@@ -147,10 +155,16 @@ def _create_text_results(
             texts_chunk = texts[i : i + step]
             scores_chunk = scores[i : i + step]
 
-            without_prompt_texts_chunk = [re.sub(r"\[INST\].*?\[/INST\]", "", text) for text in texts_chunk]
+            without_prompt_texts_chunk = [
+                re.sub(r"\[INST\].*?\[/INST\]", "", text) for text in texts_chunk
+            ]
 
-            recognized_text = RecognizedText(texts=without_prompt_texts_chunk, scores=scores_chunk)
-            result = Result.text_recognition_result(metadata=metadata, text=recognized_text)
+            recognized_text = RecognizedText(
+                texts=without_prompt_texts_chunk, scores=scores_chunk
+            )
+            result = Result.text_recognition_result(
+                metadata=metadata, text=recognized_text
+            )
             results.append(result)
         return results
 

tests/integration/data/test_gpu_htr_model_pipeline.yaml → tests/integration/data/pipelines/test_gpu_hf_htr_pipeline.yaml

@@ -2,20 +2,12 @@
 
 steps:
 
-- step: TextRecognition
-  settings:
-    model: Satrn 
-    model_settings:
-       model: Riksarkivet/satrn_htr
-    generation_settings:
-       batch_size: 1
-       num_beams: 1
-
 - step: TextRecognition
   settings:
     model: TrOCR 
     model_settings:
        model: Riksarkivet/trocr-base-handwritten-hist-swe-2
     generation_settings:
        batch_size: 1
-       num_beams: 1
+       num_beams: 1
+

tests/integration/data/pipelines/test_gpu_opennmlab_htr_pipeline.yaml

@@ -0,0 +1,13 @@
+# HTR pipeline
+
+steps:
+
+- step: TextRecognition
+  settings:
+    model: Satrn 
+    model_settings:
+       model: Riksarkivet/satrn_htr
+    generation_settings:
+       batch_size: 1
+       num_beams: 1
+

tests/integration/data/pipelines/test_gpu_svea_htr_pipeline.yaml

@@ -0,0 +1,52 @@
+# HTRFlow demo pipeline
+# This pipeline works well on pages with running text
+
+steps:
+
+# Region segmentation
+- step: Segmentation
+  settings:
+    model: yolo
+    model_settings:
+       model: Riksarkivet/yolov9-regions-1
+    generation_settings:
+       batch_size: 1
+# Line segementation
+- step: Segmentation
+  settings:
+    model: yolo
+    model_settings:
+       model: Riksarkivet/yolov9-lines-within-regions-1
+    generation_settings:
+        batch_size: 1
+# Text recognition
+- step: TextRecognition
+  settings:
+    model: WordLevelTrocr
+    model_settings:
+       model: Riksarkivet/trocr-base-handwritten-hist-swe-2
+    generation_settings:
+       batch_size: 1
+       num_beams: 1
+# Order regions by reading order
+- step: ReadingOrderMarginalia
+  settings:
+    two_page: always
+# Export to Alto and Page XML
+- step: Export
+  settings:
+    dest: outputs/alto
+    format: alto
+- step: Export
+  settings:
+    dest: outputs/page
+    format: page
+
+# Sets label format to regionX_lineY_wordZ
+labels:
+  level_labels:
+    - region
+    - line
+    - word
+  sep: _
+  template: "{label}{number}"

tests/integration/test_gpu_availability.py

@@ -1,9 +1,6 @@
 import pytest
 import torch
 
-SELECTOR_ARGO_SERVER_URL = "http://localhost:2746"
-SELECTOR_SERVICE_ACCOUNT = "htrflow-service-account"
-
 
 @pytest.mark.gpu
 def test_gpu_availability():

tests/integration/test_gpu_htr_model_pipeline.py

@@ -5,20 +5,14 @@
 
 runner = CliRunner()
 
-image_path = "tests/integration/data/trocr_example.png"
-pipeline_path = "tests/integration/data/test_gpu_htr_model_pipeline.yaml"
 
-
-@pytest.fixture(scope="module")
-def check_test_files():
+def run_pipeline_test(image_path, pipeline_path):
+    """Helper function to run the pipeline test."""
     assert os.path.exists(image_path), f"Test image not found: {image_path}"
     assert os.path.exists(
         pipeline_path
     ), f"Test pipeline YAML not found: {pipeline_path}"
 
-
-@pytest.mark.gpu
-def test_run_htr_pipeline(check_test_files):
     result = runner.invoke(
         app,
         [
@@ -34,4 +28,40 @@ def test_run_htr_pipeline(check_test_files):
 
     assert (
         result.exit_code == 0
-    ), f"Pipeline returns sucessfully exit code {result.exit_code}"
+    ), f"Pipeline returns successfully, exit code {result.exit_code}"
+
+
+@pytest.mark.gpu
+@pytest.mark.parametrize(
+    "image_path, pipeline_path",
+    [
+        (
+            "tests/integration/data/images/trocr_example.png",
+            "tests/integration/data/pipelines/test_gpu_hf_htr_pipeline.yaml",
+        ),
+        (
+            "tests/integration/data/images/trocr_example.png",
+            "tests/integration/data/pipelines/test_gpu_svea_htr_pipeline.yaml",
+        ),
+    ],
+)
+def test_run_htr_pipeline(image_path, pipeline_path):
+    run_pipeline_test(image_path, pipeline_path)
+
+
+@pytest.mark.openmmlab
+@pytest.mark.parametrize(
+    "image_path, pipeline_path",
+    [
+        (
+            "tests/integration/data/images/trocr_example.png",
+            "tests/integration/data/pipelines/test_gpu_hf_htr_pipeline.yaml",
+        ),
+        (
+            "tests/integration/data/images/trocr_example.png",
+            "tests/integration/data/pipelines/test_gpu_opennmlab_htr_pipeline.yaml",
+        ),
+    ],
+)
+def test_run_openmmlab_pipeline(image_path, pipeline_path):
+    run_pipeline_test(image_path, pipeline_path)

tox.ini

@@ -7,9 +7,10 @@ deps =
     pytest
     lorem
     pytest-cov
-commands = pytest -m "not gpu"
+commands = pytest -m "gpu"
 
 [testenv:openmmlab]
+usedevelop = true
 description = "Run tests with OpenMMLab packages"
 deps =
     mmcv @ https://github.com/Swedish-National-Archives-AI-lab/openmim_install/raw/main/mmcv-2.0.0-cp310-cp310-manylinux1_x86_64.whl
@@ -18,8 +19,12 @@ deps =
     mmocr==1.0.1
     yapf==0.40.1
 
+commands_pre =
+    uv pip show mmocr
+    uv pip show mmdet
+
 commands =
-    uv run pytest -m gpu {posargs}
+    uv run pytest -m "openmmlab"
 
 [testenv:py312]
 basepython = python3.12