crieur 1.6.0__py3-none-any.whl → 1.8.1__py3-none-any.whl

crieur/__init__.py

@@ -1,4 +1,4 @@
 from pathlib import Path
 
-VERSION = "1.6.0"
+VERSION = "1.8.1"
 ROOT_DIR = Path(__file__).parent

crieur/cli.py

@@ -27,6 +27,7 @@ def generate(
     extra_vars: str = "",
     target_path: Path = Path() / "public",
     source_path: Path = Path() / "sources",
+    statics_path: Path = Path(__file__).parent / "statics",
     templates_path: Path = Path(__file__).parent / "templates",
     without_statics: bool = False,
     feed_limit: int = 10,
@@ -38,6 +39,7 @@ def generate(
     :extra_vars: stringified JSON extra vars passed to the templates.
     :target_path: Path where site is built (default: /public/).
     :source_path: Path where stylo source were downloaded (default: /sources/).
+    :statics_path: Path where statics are located (default: @crieur/statics/).
     :template_path: Path where templates are located (default: @crieur/templates/).
     :without_statics: Do not copy statics if True (default: False).
     :feed_limit: Number of max items in the feed (default: 10).
@@ -63,8 +65,7 @@ def generate(
     generate_feed(title, base_url, numeros, extra_vars, target_path, number=feed_limit)
 
     if not without_statics:
-        static_path_local = Path(__file__).parent / "statics"
-        shutil.copytree(static_path_local, target_path / "statics", dirs_exist_ok=True)
+        shutil.copytree(statics_path, target_path / "statics", dirs_exist_ok=True)
 
 
 @cli

crieur/generator.py

@@ -2,6 +2,7 @@ import json
 import locale
 import shutil
 from datetime import datetime, timedelta, timezone
+from pathlib import Path
 
 import mistune
 from feedgen.feed import FeedGenerator
@@ -10,6 +11,7 @@ from jinja2 import FileSystemLoader
 from slugify import slugify
 
 from . import VERSION
+from .typography import typographie
 from .utils import neighborhood
 
 locale.setlocale(locale.LC_ALL, "")
@@ -24,12 +26,22 @@ def markdown(value):
     return md(value) if value else ""
 
 
+def typography(value):
+    value = value.replace("\\ ", " ")
+    return typographie(value) if value else ""
+
+
 def generate_html(
     title, base_url, numeros, keywords, authors, extra_vars, target_path, templates_path
 ):
-    environment = Env(loader=FileSystemLoader(str(templates_path)))
+    environment = Env(
+        loader=FileSystemLoader(
+            [str(templates_path), str(Path(__file__).parent / "templates")]
+        )
+    )
     environment.filters["slugify"] = slugify_
     environment.filters["markdown"] = markdown
+    environment.filters["typography"] = typography
 
     extra_vars = json.loads(extra_vars) if extra_vars else {}
 

crieur/models.py

@@ -10,8 +10,22 @@ from PIL import Image, UnidentifiedImageError
 from slugify import slugify
 from yaml.composer import ComposerError
 
+from .typography import typographie
 
-class ImgsWithSizesRenderer(mistune.HTMLRenderer):
+
+class FrenchTypographyRenderer(mistune.HTMLRenderer):
+    """Apply French typographic rules to text."""
+
+    def text(self, text):
+        text = text.replace("\\ ", " ")
+        return typographie(super().text(text), html=True)
+
+    def block_html(self, html):
+        html = html.replace("\\ ", " ")
+        return typographie(super().block_html(html), html=True)
+
+
+class ImgsWithSizesRenderer(FrenchTypographyRenderer):
     """Renders images as <figure>s and add sizes."""
 
     def __init__(

crieur/typography.py

@@ -0,0 +1,93 @@
+import html.entities
+import unicodedata
+from dataclasses import dataclass
+
+import regex  # pour le support de "\p{}"
+
+
+@dataclass
+class Caractere:
+    unicode: str
+    html: str
+
+    def __init__(self, name: str):
+        self.unicode = unicodedata.lookup(name)
+        codepoint = ord(self.unicode)
+        html_name = html.entities.codepoint2name.get(codepoint, f"#{codepoint}")
+        self.html = f"&{html_name};"
+
+
+ESPACE_INSECABLE = Caractere(name="NO-BREAK SPACE")
+ESPACE_FINE_INSECABLE = Caractere(name="NARROW NO-BREAK SPACE")
+
+
+def assemble_regexes(*regexes):
+    return "|".join(regexes)
+
+
+def build_regex(avant, apres):
+    # \p{} permet de reconnaître un caractère par sa catégorie Unicode
+    # "Zs" est la catégorie "Separator, space".
+    return (
+        rf"((?P<avant>{avant})"
+        + rf"(\p{{Zs}}|{ESPACE_INSECABLE.html})"
+        + rf"(?P<apres>{apres}))"
+        + r"(?!(.(?!<svg))*<\/svg>)"
+    )
+
+
+RE_ESPACE_FINE_INSECABLE = regex.compile(
+    assemble_regexes(
+        build_regex(r"\w?", r"[;\?!]"),  # Ponctuations doubles.
+        build_regex(
+            r"\d", r"([ghj]|min|sec|images|mm|hab|kg|mg|µg|L|km|°C|GHz)(\b|$)"
+        ),  # Unités.
+        build_regex(r"\d", r"(Mo|Ko|Go|Mb|Kb|Gb)(\b|$)"),  # Tailles de fichiers.
+        build_regex(r"\d", r"%"),  # Pourcentages.
+        build_regex(r"\d", r"€"),  # Symboles monétaires.
+        build_regex(r"\d", r"\d"),  # Séparateurs de milliers.
+    )
+)
+
+
+def insere_espaces_fines_insecables(texte):
+    return RE_ESPACE_FINE_INSECABLE.sub(
+        r"\g<avant>" + ESPACE_FINE_INSECABLE.unicode + r"\g<apres>", texte
+    )
+
+
+RE_ESPACE_INSECABLE = regex.compile(
+    assemble_regexes(
+        build_regex(r"\w?", r":"),  # Deux points.
+        build_regex(r"«", r""),  # Guillemets en chevrons.
+        build_regex(r"", r"»"),  # Guillemets en chevrons.
+        build_regex(
+            rf"\b(\d|{ESPACE_FINE_INSECABLE.html})+", r"(?!\d)\w"
+        ),  # Nombre suivi de lettres.
+        build_regex(r"(M\.|Mme)", r"\w"),  # Titres (Monsieur, Madame).
+    )
+)
+
+
+def insere_espaces_insecables(texte):
+    return RE_ESPACE_INSECABLE.sub(
+        r"\g<avant>" + ESPACE_INSECABLE.unicode + r"\g<apres>", texte
+    )
+
+
+def encode_espaces_insecables_en_html(texte):
+    for caractere in (ESPACE_INSECABLE, ESPACE_FINE_INSECABLE):
+        texte = texte.replace(caractere.unicode, caractere.html)
+    return texte
+
+
+def typographie(texte, html=False):
+    """
+    Utilise les espaces insécables fines ou normales lorsque c’est approprié
+
+    https://fr.wikipedia.org/wiki/Espace_ins%C3%A9cable#En_France
+    """
+    res = insere_espaces_fines_insecables(insere_espaces_insecables(texte))
+    if html:
+        res = encode_espaces_insecables_en_html(res)
+    return res

{crieur-1.6.0.dist-info → crieur-1.8.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: crieur
-Version: 1.6.0
+Version: 1.8.1
 Summary: A Static Revue Generator.
 Project-URL: Homepage, https://gitlab.huma-num.fr/ecrinum/crieur
 Project-URL: Issues, https://gitlab.huma-num.fr/ecrinum/crieur/-/issues
@@ -693,6 +693,7 @@ Requires-Dist: pillow
 Requires-Dist: python-slugify
 Requires-Dist: pytz>=2025.2
 Requires-Dist: pyyaml>=6.0.2
+Requires-Dist: regex>=2024.11.6
 Requires-Dist: tzdata>=2025.2
 Requires-Dist: unidecode
 Provides-Extra: dev
@@ -754,6 +755,7 @@ cog.out(f"```\n{help}\n```")
 ```
                        [--extra-vars EXTRA_VARS] [--target-path TARGET_PATH]
                        [--source-path SOURCE_PATH]
+                       [--statics-path STATICS_PATH]
                        [--templates-path TEMPLATES_PATH] [--without-statics]
                        [--feed-limit FEED_LIMIT]
 
@@ -768,6 +770,9 @@ options:
   --source-path SOURCE_PATH
                         Path where stylo source were downloaded (default:
                         /sources/).
+  --statics-path STATICS_PATH
+                        Path where statics are located (default:
+                        @crieur/statics/).
   --templates-path TEMPLATES_PATH
   --without-statics     Do not copy statics if True (default: False).
   --feed-limit FEED_LIMIT

{crieur-1.6.0.dist-info → crieur-1.8.1.dist-info}/RECORD

@@ -1,8 +1,9 @@
-crieur/__init__.py,sha256=OTKpyo3mBk05AW9jx7vHsWhKpHveyDwTU6fnJyxLxWo,77
+crieur/__init__.py,sha256=O8R0eSSNMbXSylnPd6kYLTSZhVo0917hSfC31AX71f0,77
 crieur/__main__.py,sha256=bYt9eEaoRQWdejEHFD8REx9jxVEdZptECFsV7F49Ink,30
-crieur/cli.py,sha256=MyqGylQE691ILFnR_58925erUbluC30FrOLuok0PAB8,5737
-crieur/generator.py,sha256=wbMeE0jCEy7kw8jhjscoeTjj8nyRPv6jqA76ZQeCdg8,4386
-crieur/models.py,sha256=KrW6gy7z0RzDBdUAd2VgwzcNkMOpdRPnu9JxvkNGZ8c,8727
+crieur/cli.py,sha256=trYYSwcYsyPoWluZSVLoSA-8w86uQy3Wnw_w6724g8k,5809
+crieur/generator.py,sha256=5vryuODzvfvVQv7eUrMzIsjPWBAY3ZOFplFn6VW6SE8,4688
+crieur/models.py,sha256=I9lOLvNxCyQq-Yot69TkWsZGRAUYdv68i-qzjQ8nevk,9136
+crieur/typography.py,sha256=zZHJLZ8aUslMuiNX6z4buKNDZuj7gcrUxcIGACWmIHw,2759
 crieur/utils.py,sha256=kIdxpd5LgVv13Lx2aEXzjQttBDtcppRlwNsH0vwX8f0,1566
 crieur/statics/pico.css,sha256=VdrimW9PLcEIzqJ__s062OrwBj_Jb6jZIwbtdqOtM-w,93407
 crieur/templates/article.html,sha256=AgqMIM0HG9IDFZ8tArsCLgbLmx86oY-HC6Jj2lgzUMg,1281
@@ -11,8 +12,8 @@ crieur/templates/base.html,sha256=4ZOLAnmle0_m8Y3lWT6wcH8f-_7SymxEDeIUzDQNnks,16
 crieur/templates/homepage.html,sha256=7YG7kA4AFuyrSuqWeFAVj09ogwsybE7w0-NKMLWms5s,2994
 crieur/templates/keyword.html,sha256=Hv3Ep3R6oN5pBw14gfQT-aeqEiuFiatmVZLWn5hR1e4,428
 crieur/templates/numero.html,sha256=F7hCaAHJ1WRWxD_zfhJzyLaiwXblJWJF7DUTy41Mnu8,1849
-crieur-1.6.0.dist-info/METADATA,sha256=OCG3y13RZyHppOrUhjvC5v_WO7e1bJj6Vd1TedJYRNk,45046
-crieur-1.6.0.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-crieur-1.6.0.dist-info/entry_points.txt,sha256=edmbmPxs9QXyvSMpPJBDPGw3vZOJEKqXJhysYNx3QSM,43
-crieur-1.6.0.dist-info/licenses/LICENSE,sha256=F5acw9_laHeyi4wPmQyf_ttyz81VqCIwScwO8C1FhXU,34519
-crieur-1.6.0.dist-info/RECORD,,
+crieur-1.8.1.dist-info/METADATA,sha256=HORg94MhKI8ug2soyYvj9RtLYaJD9ZUGuk3ITxx4A5c,45269
+crieur-1.8.1.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+crieur-1.8.1.dist-info/entry_points.txt,sha256=edmbmPxs9QXyvSMpPJBDPGw3vZOJEKqXJhysYNx3QSM,43
+crieur-1.8.1.dist-info/licenses/LICENSE,sha256=F5acw9_laHeyi4wPmQyf_ttyz81VqCIwScwO8C1FhXU,34519
+crieur-1.8.1.dist-info/RECORD,,