Add CI workflow

.github/workflows/ci.yaml

@@ -0,0 +1,45 @@
+name: CI
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+    branches: [main]
+    paths-ignore:
+      - "README.md"
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        python-version: ["3.11"]
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install dependencies
+        run: |
+          make setup
+          make install
+
+      - name: Format
+        run: |
+          make format
+
+      - name: Lint
+        run: |
+          make lint
+
+      - name: Type check
+        run: |
+          make typecheck
+
+      # - name: Run tests
+      #   run: |
+      #     make test

Makefile

@@ -0,0 +1,31 @@
+#
+# Installation
+#
+
+.PHONY: setup
+setup:
+	pip install -U uv
+
+.PHONY: install
+install:
+	uv sync --all-extras
+
+#
+# linter/formatter/typecheck
+#
+
+.PHONY: lint
+lint: install
+	uv run ruff check --output-format=github .
+
+.PHONY: format
+format: install
+	uv run ruff format --check --diff .
+
+.PHONY: typecheck
+typecheck: install
+	uv run mypy --cache-dir=/dev/null .
+
+.PHONY: test
+test: install
+	uv run pytest -vsx --log-cli-level=INFO

src/layout_prompter/modules/rankers.py

@@ -21,8 +21,7 @@ def invoke(
         config: Optional[RunnableConfig] = None,
         **kwargs: Any,
     ) -> List[SerializedOutputData]:
-        breakpoint()
-        return super().invoke(input, config, **kwargs)
+        raise NotImplementedError
 
 
 class LayoutPrompterRanker(BaseModel, LayoutRanker):
@@ -62,6 +61,7 @@ def invoke(
 
         min_vals = np.min(metrics_arr, axis=0, keepdims=True)
         max_vals = np.max(metrics_arr, axis=0, keepdims=True)
+
         scaled_metrics = (metrics_arr - min_vals) / (max_vals - min_vals)
 
         quality = (

src/layout_prompter/modules/selectors.py

@@ -1,13 +1,10 @@
 import random
-from abc import abstractmethod
-from typing import Any, List, Optional, Tuple
+from typing import List, Optional, Tuple
 
 import cv2
 import numpy as np
 from langchain_core.example_selectors.base import BaseExampleSelector
-from langchain_core.runnables import Runnable
-from langchain_core.runnables.config import RunnableConfig
-from pydantic import BaseModel, ValidationInfo, field_validator, model_validator
+from pydantic import BaseModel, model_validator
 from typing_extensions import Self
 
 from layout_prompter.models import ProcessedLayoutData
@@ -36,7 +33,10 @@ def select_examples(  # type: ignore[override]
     ) -> List[ProcessedLayoutData]:
         raise NotImplementedError
 
-    def add_example(self, example: ProcessedLayoutData) -> Any:
+    def add_example(  # type: ignore[override]
+        self,
+        example: ProcessedLayoutData,
+    ) -> None:
         self.examples.append(example)
 
     def _is_filter(self, data: ProcessedLayoutData) -> bool:

src/layout_prompter/modules/serializers.py

@@ -1,4 +1,5 @@
 import json
+import logging
 from typing import Any, Final, List
 
 from langchain_core.prompt_values import ChatPromptValue
@@ -14,6 +15,8 @@
 
 from layout_prompter.models import Coordinates, ProcessedLayoutData, SerializedData
 
+logger = logging.getLogger(__name__)
+
 SYSTEM_PROMPT: Final[str] = """\
 Please generate a layout based on the given information. You need to ensure that the generated layout looks realistic, with elements well aligned and avoiding unnecessary overlap.
 
@@ -25,22 +28,19 @@
 {layout_domain} layout
 
 ## Canvas Size
-{canvas_width}px x {canvas_height}px
-"""
+{canvas_width}px x {canvas_height}px"""
 
 CONTENT_AWARE_CONSTRAINT: Final[str] = """\
 # Constraints
 ## Content Constraint
 {content_constraint}
 
 ## Element Type Constraint
-{type_constraint}
-"""
+{type_constraint}"""
 
 SERIALIZED_LAYOUT: Final[str] = """\
 # Serialized Layout
-{serialized_layout}
-"""
+{serialized_layout}"""
 
 
 class LayoutSerializerInput(BaseModel):
@@ -97,16 +97,18 @@ def _get_content_constraint(self, data: ProcessedLayoutData) -> str:
         return self._convert_to_double_bracket(content_constraint)
 
     def _get_type_constraint(self, data: ProcessedLayoutData) -> str:
+        assert data.labels is not None
         type_constraint = json.dumps(
             {idx: label for idx, label in enumerate(data.labels)}
         )
         return self._convert_to_double_bracket(type_constraint)
 
     def _get_serialized_layout(self, data: ProcessedLayoutData) -> str:
-        assert len(data.labels) == len(data.discrete_gold_bboxes)
+        assert data.labels is not None and data.discrete_bboxes is not None
+        labels, discrete_gold_bboxes = data.labels, data.discrete_bboxes
 
         serialized_data_list = []
-        for class_name, bbox in zip(data.labels, data.discrete_gold_bboxes):
+        for class_name, bbox in zip(labels, discrete_gold_bboxes):
             left, top, width, height = bbox
 
             serialized_data = SerializedData(
@@ -173,4 +175,6 @@ def invoke(
             }
         )
         assert isinstance(final_prompt, ChatPromptValue)
+        logger.debug(final_prompt.to_messages())
+
         return final_prompt

src/layout_prompter/transforms/discretize_bboxes.py

@@ -43,6 +43,8 @@ def invoke(
         config: RunnableConfig | None = None,
         **kwargs: Any,
     ) -> ProcessedLayoutData:
+        assert input.bboxes is not None and input.labels is not None
+
         canvas_size = input.canvas_size.model_dump()
 
         bboxes, labels = copy.deepcopy(input.bboxes), copy.deepcopy(input.labels)

src/layout_prompter/transforms/label_dict_sort.py

@@ -16,6 +16,8 @@ def invoke(
         config: RunnableConfig | None = None,
         **kwargs: Any,
     ) -> ProcessedLayoutData:
+        assert input.bboxes is not None and input.labels is not None
+
         canvas_size = input.canvas_size
         bboxes, labels = copy.deepcopy(input.bboxes), copy.deepcopy(input.labels)
         content_bboxes = (

src/layout_prompter/transforms/lexicographic_sort.py

@@ -16,6 +16,8 @@ def invoke(
         config: RunnableConfig | None = None,
         **kwargs: Any,
     ) -> ProcessedLayoutData:
+        assert input.bboxes is not None and input.labels is not None
+
         canvas_size = input.canvas_size
         bboxes, labels = copy.deepcopy(input.bboxes), copy.deepcopy(input.labels)
         content_bboxes = (
@@ -40,6 +42,7 @@ def invoke(
         )
 
         # Extract left and top coordinates from bboxes
+        assert input.bboxes is not None
         left, top, _, _ = input.bboxes.T
 
         # Get the indices of the sorted bboxes based on left and top coordinates

src/layout_prompter/transforms/saliency_map_to_bboxes.py

@@ -3,7 +3,6 @@
 import cv2
 import numpy as np
 from langchain_core.runnables import Runnable
-from PIL import Image
 from pydantic import BaseModel
 
 from layout_prompter.typehints import PilImage
@@ -51,6 +50,6 @@ def invoke(  # type: ignore[override]
         contours, _ = cv2.findContours(
             thresholded_map, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE
         )
-        bboxes = self.get_filtered_bboxes(contours)
+        bboxes = self.get_filtered_bboxes(contours)  # type: ignore[arg-type]
 
         return bboxes if len(bboxes) != 0 else None