Merge branch 'main' into large-scale-controlnet

.github/workflows/build.yaml

@@ -15,15 +15,15 @@ jobs:
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@v2
 
+      - name: Set buildx alias
+        run: docker buildx install
+
       - name: Login to GitHub Container Registry
         uses: docker/login-action@v2
         with:
           registry: ghcr.io
           username: ${{ github.actor }}
           password: ${{ secrets.GITHUB_TOKEN }}
 
-      - name: Set buildx alias
-        run: docker buildx install
-
       - name: Build components
         run: ./scripts/build_components.sh --cache -t $GITHUB_SHA -t dev

.github/workflows/prep-release.yaml

@@ -0,0 +1,54 @@
+name: Build Python 🐍 distributions 📦 and publish to TestPyPI
+on:
+  push:
+    tags:
+      - '[0-9]+.[0-9]+.[0-9]+*'
+jobs:
+  build-n-publish:
+    name: Build Python 🐍 distributions 📦 and publish to TestPyPI
+    runs-on: ubuntu-latest
+    permissions:
+      id-token: write
+      packages: write
+    steps:
+      - uses: actions/checkout@master
+
+      - name: Set up Python 3.9
+        uses: actions/setup-python@v1
+        with:
+          python-version: 3.9
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v2
+
+      - name: Set buildx alias
+        run: docker buildx install
+
+      - name: Login to GitHub Container Registry
+        uses: docker/login-action@v2
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Build components
+        run: ./scripts/build_components.sh -t $GITHUB_REF_NAME
+
+      - name: Build data explorer
+        run: ./scripts/build_explorer.sh -t $GITHUB_REF_NAME
+
+      - name: Update version in pyproject.toml with tag version
+        run: sed -i "s/^version = .*/version = '${{github.ref_name}}'/" pyproject.toml
+
+      - name: Build a binary wheel and a source tarball
+        run: |
+          pip install poetry
+          ./scripts/pre-build.sh
+          poetry build
+
+      - name: Publish distribution 📦 to Test PyPI
+        uses: pypa/[email protected]
+        with:
+          name: testpypi
+          repository_url: https://test.pypi.org/legacy/
+          url: https://test.pypi.org/p/fondant

.github/workflows/release.yaml

@@ -1,14 +1,11 @@
-name: Publish Python 🐍 distributions 📦 to PyPI and TestPyPI
+name: Publish Python 🐍 distributions 📦 to PyPI
 on:
-  push:
-    tags:
-      - '[0-9]+.[0-9]+.[0-9]+*'
   release:
     types:
       - published
 jobs:
   build-n-publish:
-    name: Build and publish Python 🐍 distributions 📦 to PyPI and TestPyPI
+    name: Publish Python 🐍 distributions 📦 to PyPI
     runs-on: ubuntu-latest
     permissions:
       id-token: write
@@ -21,49 +18,33 @@ jobs:
       with:
         python-version: 3.9
 
+    - name: Set up Docker Buildx
+      uses: docker/setup-buildx-action@v2
+
+    - name: Set buildx alias
+      run: docker buildx install
+
     - name: Login to GitHub Container Registry
       uses: docker/login-action@v2
       with:
         registry: ghcr.io
         username: ${{ github.actor }}
         password: ${{ secrets.GITHUB_TOKEN }}
 
-    - name: Build components
-      if: github.event_name == 'push'
-      run: ./scripts/build_components.sh -t $GITHUB_REF_NAME
-
     - name: Tag components
-      if: github.event_name == 'release'
       run: ./scripts/tag_components.sh -o $GITHUB_REF_NAME -n latest
 
-    - name: Build data explorer
-      if: github.event_name == 'push'
-      run: ./scripts/build_explorer.sh -t $GITHUB_REF_NAME
-
     - name: Tag data explorer
-      if: github.event_name == 'release'
       run: ./scripts/tag_explorer.sh -o $GITHUB_REF_NAME -n latest
 
-    - name: Update version in pyproject.toml with tag version
-      run: sed -i "s/^version = .*/version = '${{github.ref_name}}'/" pyproject.toml
-
-    - name: Build a binary wheel and a source tarball
+    - name: Download distributions from test.PyPI
       run: |
-        pip install poetry
-        ./scripts/pre-build.sh
-        poetry build
-
-    - name: Publish distribution 📦 to Test PyPI
-      if: github.event_name == 'push'
-      uses: pypa/[email protected]
-      with:
-        name: testpypi
-        repository_url: https://test.pypi.org/legacy/
-        url: https://test.pypi.org/p/fondant
+        pip install --upgrade pip
+        pip download fondant==$GITHUB_REF_NAME -d dist --index-url https://test.pypi.org/simple/ --no-deps --only-binary fondant
+        pip download fondant==$GITHUB_REF_NAME -d dist --index-url https://test.pypi.org/simple/ --no-deps --no-binary fondant
 
     - name: Publish distribution 📦 to PyPI if triggered by release
-      if: github.event_name == 'release'
       uses: pypa/[email protected]
       with:
         name: pypi
-        url: https://pypi.org/p/fondant
+        url: https://pypi.org/p/fondant

README.md

@@ -48,6 +48,10 @@ And create high quality datasets to fine-tune your own foundation models.
 
 <p align="right">(<a href="#chocolate_bar-fondant">back to top</a>)</p>
 
+## 💨 Getting Started
+
+Anxious to get started? Here's is a [step by step guide](https://fondant.readthedocs.io/en/latest/getting_started) to get your first pipeline up and running.
+
 ## 🪄 Example pipelines
 
 Curious to see what Fondant can do? Have a look at our example pipelines:
@@ -90,6 +94,7 @@ point to create datasets for training code assistants.
 
 <p align="right">(<a href="#chocolate_bar-fondant">back to top</a>)</p>
 
+
 ## 🧩 Reusable components
 
 Fondant comes with a library of reusable components, which can jumpstart your pipeline.

components/image_resolution_extraction/fondant_component.yaml

@@ -11,9 +11,9 @@ consumes:
 produces:
   images:
     fields:
+      data:
+        type: binary
       width:
         type: int16
       height:
-        type: int16
-      data:
-        type: binary
+        type: int16

components/image_resolution_extraction/src/main.py

@@ -39,8 +39,9 @@ def transform(self, dataframe: pd.DataFrame) -> pd.DataFrame:
         """
         logger.info("Filtering dataset...")
 
-        dataframe[[("images", "width"), ("images", "height")]] = \
-            dataframe[[("images", "data")]].map(extract_dimensions)
+        dataframe[[("images", "width"), ("images", "height")]] = dataframe[
+            [("images", "data")]
+        ].apply(lambda x: extract_dimensions(x.images.data), axis=1)
 
         return dataframe
 

components/language_filter/Dockerfile

@@ -0,0 +1,18 @@
+FROM --platform=linux/amd64 python:3.8-slim
+
+## System dependencies
+RUN apt-get update && \
+    apt-get upgrade -y && \
+    apt-get install git -y
+
+# install requirements
+COPY requirements.txt /
+RUN pip3 install --no-cache-dir -r requirements.txt
+
+# Set the working directory to the component folder
+WORKDIR /component/src
+
+# Copy over src-files
+COPY src/ .
+
+ENTRYPOINT ["python", "main.py"]

components/language_filter/README.md

@@ -0,0 +1,7 @@
+# Language filter
+
+## Description
+This component is based on the `TransformComponent` and is used to filter a dataframe based on language. 
+It allows you to remove rows that do not match the provided language, thus providing a way to focus 
+on specific languages within your data.
+

components/language_filter/fondant_component.yaml

@@ -0,0 +1,14 @@
+name: Filter languages
+description: A component that filters text based on the language.
+image: ghcr.io/ml6team/filter_language:latest
+
+consumes:
+  text:
+    fields:
+      data:
+        type: string
+
+args:
+  language:
+    description: A valid language code or identifier (e.g., "en", "fr", "de").
+    type: str

components/language_filter/requirements.txt

@@ -0,0 +1,4 @@
+git+https://github.com/ml6team/fondant@main
+pyarrow>=7.0
+gcsfs==2023.4.00
+fasttext-wheel==0.9.2

components/language_filter/src/main.py

@@ -0,0 +1,72 @@
+"""A component that filters text based on the language."""
+import logging
+
+import fasttext
+import pandas as pd
+
+from fondant.component import PandasTransformComponent
+
+logger = logging.getLogger(__name__)
+
+
+class LanguageIdentification:
+    """A class for language detection using FastText."""
+
+    def __init__(self, language, model_path: str = "lid.176.ftz"):
+        """
+        Initializes the LanguageDetect class.
+
+        Args:
+           language (str): language to filter on
+           model_path (str): The path to the FastText language identification model.
+        """
+        pretrained_lang_model_weight_path = model_path
+        self.language = language
+        self.model = fasttext.load_model(pretrained_lang_model_weight_path)
+
+    def predict_lang(self, text: str):
+        """
+        Detects the language of a text sequence.
+
+        Args:
+            text (str): The text for language detection.
+
+        Returns:
+            str: The predicted language label.
+        """
+        predictions = self.model.predict(text, k=1)
+        return predictions[0][0]
+
+    def is_language(self, row):
+        """Predict if text of a row is written in the defined language."""
+        return self.language in self.predict_lang(row["text"])
+
+
+class LanguageFilterComponent(PandasTransformComponent):
+    """Component that filter columns based on provided language."""
+
+    def setup(self, *, language):
+        """Setup language filter component.
+
+        Args:
+            language: Only keep text passages which are in the provided language.
+        """
+        self.lang_detector = LanguageIdentification(language)
+
+
+    def transform(self, dataframe: pd.DataFrame) -> pd.DataFrame:
+        """
+        Args:
+            dataframe: Pandas dataframe.
+
+        Returns:
+            Pandas dataframe
+        """
+        mask = dataframe.apply(self.lang_detector.is_language, axis=1)
+
+        return dataframe[mask]
+
+
+if __name__ == "__main__":
+    component = LanguageFilterComponent.from_args()
+    component.run()

components/language_filter/tests/language_filter_component_test.py

@@ -0,0 +1,54 @@
+"""Unit test for language filter component."""
+import pandas as pd
+
+from components.language_filter.src.main import LanguageFilterComponent
+from fondant.component_spec import ComponentSpec
+
+
+def test_run_component_test():
+    """Test language filter component."""
+    # Given: Dataframe with text in different languages
+    data = [{"text": "Das hier ist ein Satz in deutscher Sprache"},
+            {"text": "This is a sentence in English"},
+            {"text": "Dit is een zin in het Nederlands"}]
+    dataframe = pd.DataFrame(data)
+
+    # When: The language filter component proceed the dataframe
+    # and filter out all entries which are not written in german
+    spec = ComponentSpec.from_file("../fondant_component.yaml")
+
+    component = LanguageFilterComponent(spec, input_manifest_path="./dummy_input_manifest.json",
+                                        output_manifest_path="./dummy_input_manifest.json",
+                                        metadata={},
+                                        user_arguments={"language": "de"},
+                                        )
+    component.setup(language="de")
+    dataframe = component.transform(dataframe=dataframe)
+
+    # Then: dataframe only contains one german row
+    assert len(dataframe) == 1
+    assert dataframe.loc[0]["text"] == "Das hier ist ein Satz in deutscher Sprache"
+
+
+def test_run_component_test_filter_out_all():
+    """Test language filter component."""
+    # Given: Dataframe with text in different languages
+    data = [{"text": "Das hier ist ein Satz in deutscher Sprache"},
+            {"text": "This is a sentence in English"},
+            {"text": "Dit is een zin in het Nederlands"}]
+    dataframe = pd.DataFrame(data)
+
+    # When: The language filter component proceed the dataframe
+    # and filter out all entries which are not written in french
+    spec = ComponentSpec.from_file("../fondant_component.yaml")
+
+    component = LanguageFilterComponent(spec, input_manifest_path="./dummy_input_manifest.json",
+                                        output_manifest_path="./dummy_input_manifest.json",
+                                        metadata={},
+                                        user_arguments={"language": "fr"},
+                                        )
+    component.setup()
+    dataframe = component.transform(dataframe=dataframe)
+
+    # Then: dataframe should contain no rows anymore
+    assert len(dataframe) == 0

components/text_normalization/Dockerfile

@@ -0,0 +1,18 @@
+FROM --platform=linux/amd64 python:3.8-slim
+
+## System dependencies
+RUN apt-get update && \
+    apt-get upgrade -y && \
+    apt-get install git -y
+
+# install requirements
+COPY requirements.txt /
+RUN pip3 install --no-cache-dir -r requirements.txt
+
+# Set the working directory to the component folder
+WORKDIR /component/src
+
+# Copy over src-files
+COPY src/ .
+
+ENTRYPOINT ["python", "main.py"]