refactor(inference): dependecy injection for unit testing

CHANGELOG.md

@@ -1,5 +1,31 @@
 # Changelog
 
+## v0.15.3
+
+- [001baa4](/001baa419309edc6c84a4dae1dfa2e99bb3c75b5) - doc(evaluation): add evaluation.md, 2025-10-12 by *Guillaume Chervet*
+
+
+## v0.15.2
+
+- [1f4ebf7](/1f4ebf7243d71e42666297bc032251776dd9f911) - doc(evaluation): add evaluation.md, 2025-10-12 by *Guillaume Chervet*
+
+
+## v0.15.1
+
+- [b7edb55](/b7edb551645e123b271e11f667439a7ad3edbfe3) - doc(evaluation): add evaluation.md, 2025-10-12 by *Guillaume Chervet*
+
+
+## v0.15.0
+
+- [33b1b92](/33b1b929acd5a50d6ef9f7f4c377de34cdd8c234) - Merge branch 'main' of https://www.github.com/guillaume-chervet/MLOpsPython, 2025-09-21 by *Guillaume Chervet*
+- [21f73c4](/21f73c48109465fe269583828b3f12e432ca4a6b) - feat(workshop): add library workshop, 2025-09-21 by *Guillaume Chervet*
+
+
+## v0.14.11
+
+- [ecfae26](/ecfae26c4f796ee80fbb3f6dfdf7809682667f6a) - fix(workshop): update to make it works in 2025, 2025-09-20 by *Guillaume Chervet*
+
+
 ## v0.14.10
 
 - [a0310e2](/a0310e2f8248704ee42c0b776c3dcf16a9b5db7a) - refactore(all): switch to pyttest (#46), 2025-09-08 by *Guillaume Chervet*
@@ -268,28 +294,3 @@
 - [315e72c](/315e72ca2469df17729820fcd6e1f2ef23091268) - feat(workshop): create microsoft workshop, 2023-11-24 by *Guillaume Chervet*
 
 
-## v0.3.3
-
-- [6de6e5e](/6de6e5e90c8df00578d5883cea64480d38422290) - fix: Using pip instead of poetry to init the project was causing errors (#22), 2023-10-31 by *antoinelrnld*
-
-
-## v0.3.2
-
-- [14037bc](/14037bc8cd6b65eb5dca74e00da729272ae43e2e) - fix(train): azure cluster name (#21), 2023-10-21 by *Guillaume Chervet*
-
-
-## v0.3.1
-
-- [fb2f4e2](/fb2f4e24ee6ab0c920c94b70d88d4392e7ee9063) - docs(workshop): Update workshop_cloud_nord.md, 2023-10-12 by *Guillaume Chervet*
-
-
-## v0.3.0
-
-- [e4c1a89](/e4c1a8948c9772d64526e4f954b3712d130a99fb) - feat(workshop): create cloud nord workshop (#18), 2023-10-11 by *Guillaume Chervet*
-
-
-## v0.2.1
-
-- [3997017](/3997017790f048e512b0b012f347138ed6110147) - fix: api requirements.txt doesn't work on MacOs (#19), 2023-10-11 by *Ca.phe*
-
-

evaluation.md

@@ -0,0 +1,56 @@
+# Evaluation Notes
+
+- **Project:** 12 points  
+- **In-class follow-up:** 8 points
+  - UV Library Publication → link to PyPI library → 3 points 
+  - Dependency Injection for Inference → link to personal GitHub → 3 points
+  - MLOps Lab Work → link to personal GitHub → 2 points
+---
+
+## Startup Project with High Added Value (topic of your choice)
+
+The project must:
+
+- Be hosted on a **public open-source Git repository**
+- Include **Pull Requests** with **review/validation** between team members
+- Contain **Unit Tests**, ideally following a **TDD** (Test-Driven Development) approach
+- Include a **public Kanban board**
+- Contain a **preprocessing phase**, such as:
+  - In the MLOps lab, the preprocessing step is the **extraction of images from a PDF**
+  - It can also be another turnkey AI component (e.g., text detection, OCR, etc.)
+- Include a **mandatory AI training phase**:
+  - The AI can be trained **from scratch** (requires a large dataset)
+  - Or **fine-tuned** from an existing model (e.g., from **Hugging Face**)
+- Use **more than 100 data samples** (images, audio, videos, etc.) **created by you** (not downloaded from the internet) and **annotated manually**
+- Be **exposed as a real-time API**
+- Optionally include a **web front-end** (not mandatory)
+
+---
+
+## Evaluation Grid (Total: 12 points)
+
+| Item | Description | Points |
+|------|--------------|--------|
+| Teamwork, Clean Code, Pull Requests, linked commits and cards | Proper commit history | **1**  |
+| Unit-tested preprocessing phase |  | **2**  |
+| AI training phase |  | **1**  |
+| Annotation phase |  | **2**  |
+| Real-time API deployed in an environment |  | **2**  |
+| Automated integration tests |  | **2**  |
+| Final Presentation (5 minutes) |  | **2**  |
+
+---
+
+## Presentation Guidelines
+
+1. **Purpose:** What is your project’s functional value? What problem does it solve for end users? What makes it innovative?  
+2. **Demo**  
+3. **Why are you the best?**  
+   - You may include technical details about **MLOps**, **cost**, and **Time to Market**.  
+4. **What is missing** — both technically and functionally.
+
+---
+
+### Presentation Format
+
+Choose the format you believe adds the **most value** to your project presentation!

packages/mlopspython-inference/mlopspython_inference/inference_pillow.py

@@ -23,10 +23,22 @@ def load_image(filename: str|BytesIO):
 BASE_PATH = Path(__file__).resolve().parent
 
 
+class IModel():
+    def predict(self, img) -> np.ndarray:
+        pass
+
+class ModelPillow(IModel):
+    def __init__(self, model_path: str):
+        self.model = load_model(model_path)
+
+    def predict(self, img) -> np.ndarray:
+        return self.model.predict(img)
+
+
 class Inference:
-    def __init__(self, logging, model_path: str):
+    def __init__(self, logging, model: IModel):
         self.logger = logging.getLogger(__name__)
-        self.model = load_model(model_path)
+        self.model = model
 
     def execute(self, filepath:str|BytesIO):
         img = load_image(filepath)

packages/mlopspython-inference/tests/test_inference.py

@@ -1,22 +1,26 @@
 import logging
 from pathlib import Path
+from unittest.mock import MagicMock
+
 import pytest
 
+from inference_pillow import IModel
 from mlopspython_inference.inference_pillow import Inference
 
 BASE_PATH = Path(__file__).resolve().parent
 input_directory = BASE_PATH / "input"
 
 
-@pytest.mark.skip(reason="Modèle lourd / GPU non requis sur CI. Enlever ce skip si nécessaire.")
+#@pytest.mark.skip(reason="Modèle lourd / GPU non requis sur CI. Enlever ce skip si nécessaire.")
 def test_inference_runs_with_sample_model_and_image():
-    model_path = input_directory / "model" / "final_model.h5"
     image_path = input_directory / "images" / "cat.png"
 
-    assert model_path.is_file(), "Modèle de test manquant"
     assert image_path.is_file(), "Image de test manquante"
 
-    inference = Inference(logging, str(model_path))
+    model_mock = MagicMock(IModel)
+    model_mock.execute = MagicMock(return_value=[[1, 0, 0]])
+
+    inference = Inference(logging, model_mock)
     result = inference.execute(str(image_path))
 
     assert result["prediction"] in {"Cat", "Dog", "Other"}

workshop_create_library.md

@@ -3,6 +3,12 @@
 uv init --package mon_package
 cd mon_package
 
+For linux/macos :
+source .venv/bin/
+
+For windows:
+.venv\Scripts\activate
+
 # Runtime dependencies (example)
 uv add httpx
 
@@ -18,10 +24,10 @@ uv run black . --check
 uv run black .
 
 
+
 # Run tests & code coverage
 uv run pytest -q
-uv run pytest --cov=mon_package --cov-report=term-missing --cov-report=html
-
+uv run pytest -q uv run pytest --cov=mon_package --cov-report=term-missing --cov-report=html
 
 # Build wheel + sdist
 uv build