refactor: add pre-commit hooks to verify date consistancy (#25)

.pre-commit-config.yaml

@@ -38,3 +38,15 @@ repos:
   - id: markdownlint-fix
     name: Format Markdown files
     exclude: ^utils
+
+- repo: local
+  hooks:
+  - id: verify_data_matches_folder
+    name: verify_data_matches_folder
+    files: ^content/(blog|til)/.*/index\.md$
+    stages: [commit]
+    language: python
+    additional_dependencies:
+    - python-frontmatter==1.1.0
+    args: []
+    entry: python scripts/verify_data_matches_folder.py

content/blog/2022-11-06-autograd-engine-from-scratch/index.md

@@ -1,7 +1,7 @@
 ---
 title: "DIY auto-grad Engine: A Step-by-Step Guide to Calculating Derivatives Automatically"
 description: "Want to understand the magic of Jax, PyTorch, and TensorFlow auto-grad engines? The best to learn is to build your DIY version from scratch in Python."
-date: 2022-12-22TT16:37:27+01:00
+date: 2022-11-06T16:37:27+01:00
 publishdate: 2022-12-22T16:37:27+01:00
 tags:
 - python

content/blog/2024-03-13-how-to-avoid-orientation-bug-in-computer-vision-labeling/index.md

@@ -1,7 +1,7 @@
 ---
 title: "How to avoid orientation bugs in Computer Vision labeling?"
 description: "Image orientation metadata can cause strange bugs. This post teaches you how to identify these bugs and how to fix them."
-date: 2024-04-13T19:54:44+02:00
+date: 2024-03-13T19:54:44+02:00
 publishdate: 2024-03-13T19:54:44+02:00
 tags:
 - computer vision

content/blog/2024-04-03-how-to-upload-pre-annotations-to-roboflow/index.md

@@ -1,8 +1,8 @@
 ---
 title: "How to Upload (Pre)-Annotations to Roboflow"
 description: "Learn how to upload annotations and predictions to Roboflow using the Python SDK, allowing you to import pre-existing datasets or create active learning loops programmatically."
-date: 2024-06-13T21:44:10+02:00
-publishdate: 2024-06-13T21:44:10+02:00
+date: 2024-04-03T21:44:10+02:00
+publishdate: 2024-04-03T21:44:10+02:00
 tags:
 - computer vision
 - labeling

requirements.txt

@@ -1 +1,2 @@
-invoke
+invoke
+python-frontmatter

scripts/verify_data_matches_folder.py

@@ -0,0 +1,105 @@
+import datetime
+from argparse import Namespace, ArgumentParser
+from pathlib import Path
+import re
+import sys
+
+import frontmatter
+
+
+def main() -> None:
+    args = parse_args()
+    paths = args.paths
+
+    errors_per_path = {}
+
+    for path in paths:
+        if not path.exists() or path.is_dir() or path.suffix != ".md":
+            continue
+
+        errors = []
+        folder_date = _get_folder_data(path)
+        publish_date = _get_publish_date(path)
+        creation_data = _get_data(path)
+
+
+
+        if folder_date is None:
+            errors.append(f"Folder name does not contain a date: {path.parent.name}")
+
+        if publish_date is None:
+            errors.append("Publish date not found in metadata")
+
+        if creation_data is None:
+            errors.append("Creation date not found in metadata")
+
+        if folder_date != creation_data:
+            errors.append(f"Folder date ({folder_date}) does not match creation date ({creation_data})")
+
+
+        if publish_date is not None and creation_data is not None:
+            if creation_data > publish_date:
+                errors.append(f"Publish date ({publish_date}) is before creation date ({creation_data})")
+
+        if len(errors) > 0:
+            errors_per_path[path] = errors
+
+    if len(errors_per_path) > 0:
+        for path, errors in errors_per_path.items():
+            print(f"Errors in {path}")
+            for error in errors:
+                print(f"- {error}")
+        sys.exit(1)
+
+
+def parse_args() -> Namespace:
+    parser = ArgumentParser()
+    parser.add_argument("paths", type=Path, nargs='+', help="Paths to the files to verify")
+
+    return parser.parse_args()
+
+def _get_folder_data(path: Path) -> datetime.date:
+    folder_name = path.parent.name
+
+    match = re.match(r"(\d{4}-\d{2}-\d{2})", folder_name)
+    if match is None:
+        raise None
+
+    return datetime.datetime.strptime(match.group(1), "%Y-%m-%d").date()
+
+def _get_publish_date(path: Path) -> datetime.date:
+    assert path.suffix == ".md", "File must be a markdown file"
+
+    with path.open() as file:
+        meta_data = frontmatter.load(file)
+
+
+    if "publishdate" in meta_data:
+        try:
+            return _parse_metadata_date(meta_data["publishdate"]).date()
+        except Exception:
+            return None
+
+
+    raise None
+
+def _get_data(path: Path) -> datetime.date:
+    with path.open() as file:
+        meta_data = frontmatter.load(file)
+
+    if "date" in meta_data:
+        try:
+            return _parse_metadata_date(meta_data["date"]).date()
+        except Exception:
+            return None
+
+
+    raise None
+
+def _parse_metadata_date(date: str|datetime.datetime) -> datetime.datetime:
+    if isinstance(date, datetime.datetime):
+        return date
+    return datetime.datetime.strptime(date, "%Y-%m-%dT%H:%M:%S%z")
+
+if __name__ == "__main__":
+    main()