Add yaml header check (#184)

Add yaml header check

---------

Signed-off-by: Guilhem Allaman <40383801+gounux@users.noreply.github.com>
Co-authored-by: gounux <contact@guilhemallaman.net>
Co-authored-by: Julien <dev@ingeoveritas.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

geotribu_cli/cli.py

@@ -29,6 +29,7 @@
     parser_comments_broadcast,
     parser_comments_latest,
     parser_comments_read,
+    parser_header_check,
     parser_images_optimizer,
     parser_latest_content,
     parser_mastodon_export,
@@ -231,6 +232,16 @@ def main(args: list[str] = None):
     add_common_arguments(subcmd_upgrade)
     parser_upgrade(subcmd_upgrade)
 
+    subcmd_header_check = subparsers.add_parser(
+        "header-check",
+        aliases=["header", "check", "header-check", "metadata"],
+        help="Vérifier entête markdown",
+        formatter_class=main_parser.formatter_class,
+        prog="header-check",
+    )
+    add_common_arguments(subcmd_header_check)
+    parser_header_check(subcmd_header_check)
+
     # -- NESTED SUBPARSER : CREATE ---------------------------------------------------
     subcmd_content_manager = subparsers.add_parser(
         "creer",

geotribu_cli/content/header_check.py

@@ -0,0 +1,239 @@
+import argparse
+import logging
+import os
+from pathlib import Path
+
+import frontmatter
+
+from geotribu_cli.constants import GeotribuDefaults
+from geotribu_cli.json.json_client import JsonFeedClient
+from geotribu_cli.utils.check_image_size import get_image_dimensions_by_url
+from geotribu_cli.utils.check_path import check_path
+from geotribu_cli.utils.slugger import sluggy
+
+logger = logging.getLogger(__name__)
+defaults_settings = GeotribuDefaults()
+
+MANDATORY_KEYS = [
+    "title",
+    "authors",
+    "categories",
+    "date",
+    "description",
+    "license",
+    "tags",
+]
+
+# ############################################################################
+# ########## CLI #################
+# ################################
+
+
+def parser_header_check(
+    subparser: argparse.ArgumentParser,
+) -> argparse.ArgumentParser:
+    """Set the argument parser subcommand.
+
+    Args:
+        subparser (argparse.ArgumentParser): parser to set up
+
+    Returns:
+        argparse.ArgumentParser: parser ready to use
+    """
+    subparser.add_argument(
+        "content_path",
+        help="Chemin du fichier markdown dont l'entête est à vérifier",
+        type=Path,
+        metavar="content",
+        nargs="+",
+    )
+    subparser.add_argument(
+        "-af",
+        "--authors-folder",
+        dest="authors_folder",
+        type=Path,
+        help="Chemin qui contient les presentations markdown des auteurs/autrices",
+    )
+    subparser.add_argument(
+        "-minw",
+        "--min-width",
+        dest="min_image_width",
+        default=400,
+        type=int,
+        help="Largeur minimum de l'image à vérifier",
+    )
+    subparser.add_argument(
+        "-maxw",
+        "--max-width",
+        dest="max_image_width",
+        default=800,
+        type=int,
+        help="Largeur maximum de l'image à vérifier",
+    )
+    subparser.add_argument(
+        "-minh",
+        "--min-height",
+        dest="min_image_height",
+        default=400,
+        type=int,
+        help="Hauteur minimum de l'image à vérifier",
+    )
+    subparser.add_argument(
+        "-maxh",
+        "--max-height",
+        dest="max_image_height",
+        default=800,
+        type=int,
+        help="Hauteur maximum de l'image à vérifier",
+    )
+    subparser.add_argument(
+        "-r",
+        "--raise",
+        dest="raise_exceptions",
+        action="store_true",
+        default=False,
+        help="Lever des exceptions et donc arrêter le programme si des erreurs sont rencontrées",
+    )
+    subparser.set_defaults(func=run)
+    return subparser
+
+
+# ############################################################################
+# ########## MAIN ################
+# ################################
+
+
+def check_author_md(author: str, folder: Path) -> bool:
+    if author == "Geotribu":
+        return True
+    p = os.path.join(folder, f"{sluggy(author)}.md")
+    return os.path.exists(p)
+
+
+def check_image_size(
+    image_url: str, minw: int, maxw: int, minh: int, maxh: int
+) -> bool:
+    width, height = get_image_dimensions_by_url(image_url)
+    return minw <= width <= maxw and minh <= height <= maxh
+
+
+def get_existing_tags() -> list[str]:
+    jfc = JsonFeedClient()
+    return jfc.tags(should_sort=True)
+
+
+def check_existing_tags(tags: list[str]) -> tuple[bool, set[str], set[str]]:
+    existing_tags = get_existing_tags()
+    all_exists = set(tags).issubset(existing_tags)
+    missing = set(tags).difference(existing_tags)
+    present = set(tags).intersection(existing_tags)
+    return all_exists, missing, present
+
+
+def check_tags_order(tags: list[str]) -> bool:
+    for i in range(len(tags) - 1):
+        if sluggy(tags[i].upper()) > sluggy(tags[i + 1].upper()):
+            return False
+    return True
+
+
+def check_mandatory_keys(
+    keys: list[str], mandatory: list[str] = MANDATORY_KEYS
+) -> tuple[bool, set[str]]:
+    missing = set()
+    for mk in mandatory:
+        if mk not in keys:
+            missing.add(mk)
+    return len(missing) == 0, missing
+
+
+def run(args: argparse.Namespace) -> None:
+    """Run the sub command logic.
+
+    Checks YAML header of a content
+
+    Args:
+        args (argparse.Namespace): arguments passed to the subcommand
+    """
+    logger.debug(f"Running {args.command} with {args}")
+    content_paths: list[Path] = args.content_path
+
+    for content_path in content_paths:
+        logger.info(f"Checking header of {content_path}")
+        check_path(
+            input_path=content_path,
+            must_be_a_file=True,
+            must_be_a_folder=False,
+            must_be_readable=True,
+            raise_error=True,
+        )
+
+        with content_path.open(mode="r", encoding="UTF-8") as file:
+            content = frontmatter.load(file)
+            yaml_meta = content.metadata
+            logger.debug(f"YAML metadata loaded : {yaml_meta}")
+
+            # check that image size is okay
+            if "image" in yaml_meta:
+                if not yaml_meta["image"]:
+                    logger.error("Pas d'URL pour l'image")
+                elif not check_image_size(
+                    yaml_meta["image"],
+                    args.min_image_width,
+                    args.max_image_width,
+                    args.min_image_height,
+                    args.max_image_height,
+                ):
+                    msg = (
+                        f"Les dimensions de l'image ne sont pas dans l'intervalle autorisé "
+                        f"(w:{args.min_image_width}-{args.max_image_width},"
+                        f"h:{args.min_image_height}-{args.max_image_height})"
+                    )
+                    logger.error(msg)
+                    if args.raise_exceptions:
+                        raise ValueError(msg)
+                else:
+                    logger.info("Dimensions de l'image ok")
+
+            # check that author md file is present
+            if args.authors_folder:
+                for author in yaml_meta["authors"]:
+                    author_exists = check_author_md(author, args.authors_folder)
+                    if not author_exists:
+                        msg = f"Le fichier de l'auteur/autrice '{author}' n'a pas pu être trouvé dans le répertoire"
+                        logger.error(msg)
+                        if args.raise_exceptions:
+                            raise ValueError(msg)
+                    else:
+                        logger.info(f"Markdown de l'auteur/autrice '{author}' ok")
+
+            # check that tags already exist
+            all_exists, missing, _ = check_existing_tags(yaml_meta["tags"])
+            if not all_exists:
+                msg = f"Les tags suivants n'existent pas dans les contenus Geotribu précédents : {','.join(missing)}"
+                logger.error(msg)
+                if args.raise_exceptions:
+                    raise ValueError(msg)
+            else:
+                logger.info("Existence des tags ok")
+
+            # check if tags are alphabetically sorted
+            if not check_tags_order(yaml_meta["tags"]):
+                msg = f"Les tags ne sont pas triés par ordre alphabétique : {yaml_meta['tags']}"
+                logger.error(msg)
+                if args.raise_exceptions:
+                    raise ValueError(msg)
+            else:
+                logger.info("Ordre alphabétique des tags ok")
+
+            # check that mandatory keys are present
+            all_present, missing = check_mandatory_keys(
+                yaml_meta.keys(), MANDATORY_KEYS
+            )
+            if not all_present:
+                msg = f"Les clés suivantes ne sont pas présentes dans l'entête markdown : {','.join(missing)}"
+                logger.error(msg)
+                if args.raise_exceptions:
+                    raise ValueError(msg)
+            else:
+                logger.info("Clés de l'entête ok")

geotribu_cli/subcommands/__init__.py

@@ -4,6 +4,7 @@
 from geotribu_cli.comments import parser_comments_broadcast  # noqa: F401
 from geotribu_cli.comments import parser_comments_latest  # noqa: F401
 from geotribu_cli.comments import parser_comments_read  # noqa: F401
+from geotribu_cli.content.header_check import parser_header_check  # noqa: F401
 from geotribu_cli.content.new_article import parser_new_article  # noqa: F401
 from geotribu_cli.images.images_optimizer import parser_images_optimizer  # noqa: F401
 from geotribu_cli.rss.rss_reader import parser_latest_content  # noqa: F401

geotribu_cli/utils/check_image_size.py

@@ -16,9 +16,11 @@
 from decimal import Decimal
 from pathlib import Path
 from typing import Union
+from urllib import request
 
 # 3rd party
 import imagesize
+from PIL import ImageFile
 
 # #############################################################################
 # ########## Globals ###############
@@ -61,6 +63,25 @@ def get_image_size(image_filepath: Path) -> tuple[int, int]:
     return None
 
 
+def get_image_dimensions_by_url(url: str) -> tuple[int, int]:
+    """Get image dimensions as a tuple (width,height) of an image at an URL. Return None in case of error or no data.
+
+    :param str url: url of the image
+
+    :return Tuple[int, int]: dimensions tuple (width,height)
+    """
+    with request.urlopen(url) as file:
+        parser = ImageFile.Parser()
+        while True:
+            data = file.read(1024)
+            if not data:
+                break
+            parser.feed(data)
+            if parser.image:
+                return parser.image.size
+        return None
+
+
 def get_svg_size(image_filepath: Path) -> tuple[int, int]:
     """Extract SVG width and height from a SVG file and convert them into integers. \
     Relevant and working only if the file root has width and height attributes.

tests/fixtures/content/2012-12-21_article_passe.md

@@ -0,0 +1,20 @@
+---
+title: Article supposément rédigé dans le passé
+subtitle: Article supposément rédigé dans le passé pour tests
+authors:
+    - Jane Doe
+categories:
+    - article
+comments: true
+date: 2012-12-21
+description: Article supposément rédigé dans le passé
+icon: octicons/server-16
+license: beerware
+robots: index, follow
+tags:
+    - Fromage
+    - OSM
+    - QGIS
+---
+
+# Article supposément rédigé dans le futur

tests/fixtures/content/2044-04-01_article_futur.md

@@ -0,0 +1,19 @@
+---
+title: Article supposément rédigé dans le futur
+subtitle: Article supposément rédigé dans le futur pour tests
+authors:
+    - Jane Doe
+categories:
+    - article
+comments: true
+date: 2044-04-01
+icon: octicons/server-16
+robots: index, follow
+tags:
+    - Fromage
+    - IGN
+    - QGIS
+    - OSM
+---
+
+# Article supposément rédigé dans le futur

tests/fixtures/team/jane-doe.md

@@ -0,0 +1,26 @@
+---
+title: Jane Doe
+categories:
+    - contributeur
+social:
+    - bluesky:
+    - github:
+    - gitlab:
+    - linkedin:
+    - mail:
+    - mastodon:
+        - instance:
+        - username:
+    - openstreetmap:
+    - osgeo:
+    - twitter:
+    - website:
+---
+
+# Jane Doe
+
+<!-- --8<-- [start:author-sign-block] -->
+
+Test
+
+<!-- --8<-- [end:author-sign-block] -->