larlet-fr-david/site2020.py

#!/usr/bin/env python3
import fnmatch
import locale
import os
from dataclasses import dataclass
from datetime import datetime, timedelta
from html import escape
from pathlib import Path
from textwrap import dedent
from time import perf_counter

import mistune
from jinja2 import Environment as Env
from jinja2 import FileSystemLoader
from minicli import cli, run, wrap
from mistune.directives import DirectiveInclude
from PIL import Image
from slugify import slugify

# Useful for dates rendering within Jinja2.
locale.setlocale(locale.LC_ALL, "fr_FR.UTF-8")

HERE = Path(".")
DAVID = HERE / "david"
STATIC = HERE / ".." / "larlet-fr-static"
DOMAIN = "https://larlet.fr"
LOCAL_DOMAIN = "http://larlet.test:3579"
# Hardcoding publication at 12 in Paris timezone.
NORMALIZED_STRFTIME = "%Y-%m-%dT12:00:00+01:00"
TODAY = datetime.today() + timedelta(hours=6)


class MarkParser(mistune.InlineParser):
    """Parses `==foo==` as `<mark>foo</mark>`."""

    MARK = (
        r"(\={2})(?=[^\s*])("
        r"(?:\\[\\*]|[^*])*"
        r"(?:" + mistune.InlineParser.ESCAPE + r"|[^\s*]))\1"
    )

    RULE_NAMES = mistune.InlineParser.RULE_NAMES + ("mark",)

    def parse_mark(self, m, state):
        _ = m.group(1)
        text = m.group(2)
        return "mark", self.render(text, state)


class MarkRenderer(mistune.HTMLRenderer):
    """To use in conjunction with `MarkParser`."""

    def mark(self, text):
        return "<mark>" + text + "</mark>"


class BlockquoteLanguageRenderer(MarkRenderer):
    """Sets the English language attribute for blockquotes with `[en]` prefix."""

    def _get_language(self, text):
        if text.startswith("<p>[en] "):
            return "en", text.replace("<p>[en] ", "<p>")
        else:
            return None, text

    def block_quote(self, text):
        language, text = self._get_language(text)
        if language:
            return f'\n<blockquote lang="{language}">\n{text}</blockquote>\n'
        else:
            return f"\n<blockquote>\n{text}</blockquote>\n"


class ImgsWithSizesRenderer(BlockquoteLanguageRenderer):
    """Renders images as <figure>s and add sizes (useful for lazy loading)."""

    def paragraph(self, text):
        # In case of a figure, we do not want the (non-standard) paragraph.
        if text.strip().startswith("<figure>"):
            return text
        return f"<p>{text}</p>\n"

    def image(self, src, alt="", title=None):
        full_path = STATIC / Path(src[1:])
        image = Image.open(full_path)
        width, height = image.size
        return dedent(
            f"""\
                <figure>
                    <img src="{src}"
                         alt="{alt}"
                         loading="lazy" width="{width}" height="{height}" />
                    <figcaption>{title}</figcaption>
                </figure>
            """
        )


class H2AnchorsRenderer(ImgsWithSizesRenderer):
    """Custom renderer for H2 titles with anchors."""

    def heading(self, text, level):
        if level == 2:
            slug = slugify(text)
            return (
                f'<h2 id="{slug}">'
                f"{text} "
                f'<a href="#{slug}" title="Ancre vers cette partie">#</a>'
                f"</h2>"
            )
        else:
            return super().heading(text, level)


# We want a custom renderer to create a hash/link for each H2 headings.
markdown_with_h2_anchors = mistune.Markdown(
    renderer=H2AnchorsRenderer(escape=False),
    inline=MarkParser(H2AnchorsRenderer(escape=False)),
    plugins=[DirectiveInclude()],
)
# The second markdown is pertinent to generate articles for the feed,
# we do not need anchors in that case.
markdown_with_img_sizes = mistune.Markdown(
    renderer=ImgsWithSizesRenderer(escape=False),
    inline=MarkParser(ImgsWithSizesRenderer(escape=False)),
    plugins=[DirectiveInclude()],
)

# This is the jinja2 configuration to locate templates.
environment = Env(loader=FileSystemLoader(str(DAVID / "templates")))


def neighborhood(iterable, first=None, last=None):
    """
    Yield the (previous, current, next) items given an iterable.

    You can specify a `first` and/or `last` item for bounds.
    """
    iterator = iter(iterable)
    previous = first
    current = next(iterator)  # Throws StopIteration if empty.
    for next_ in iterator:
        yield (previous, current, next_)
        previous = current
        current = next_
    yield (previous, current, last)


def each_markdown_from(source_dir, file_name="*.md"):
    """Walk across the `source_dir` and return the md file paths."""
    for filename in fnmatch.filter(os.listdir(source_dir), file_name):
        yield os.path.join(source_dir, filename)


@dataclass
class Page:
    title: str
    content: str
    file_path: str
    lang: str = "fr"

    def __post_init__(self):
        suffix = len(".md")
        prefix = len("YYYY/MM-DD") + suffix
        date_str = self.file_path[-prefix:-suffix].replace("-", "/")
        self.url = f"/david/{date_str}/"
        self.date = datetime.strptime(date_str, "%Y/%m/%d").date()
        self.full_url = f"{DOMAIN}{self.url}"
        self.normalized_date = self.date.strftime(NORMALIZED_STRFTIME)
        self.escaped_title = escape(self.title)
        self.escaped_content = escape(
            self.content.replace('href="/', f'href="{DOMAIN}/')
            .replace('src="/', f'src="{DOMAIN}/')
            .replace('href="#', f'href="{self.full_url}#')
            + '<hr/><p><a href="mailto:david@larlet.fr">Réagir ?</a></p>'
        )
        # Extract first paragraph.
        self.extract = self.content.split("</p>", 1)[0] + "</p>"

    def __lt__(self, other: "Page"):
        if not isinstance(other, Page):
            return NotImplemented
        return self.date < other.date

    @staticmethod
    def all(source: Path, only_published=True, with_h2_anchors=True):
        """Retrieve all pages sorted by desc."""
        page_list = []
        md = markdown_with_h2_anchors if with_h2_anchors else markdown_with_img_sizes
        for file_path in each_markdown_from(source):
            result = md.read(file_path)
            # Extract (and remove) the title from the generated page.
            title, content = result.split("</h1>", 1)
            h1_opening_size = len("<h1>")
            title = title[h1_opening_size:]
            page = Page(title, content, file_path)
            if only_published and page.is_draft:
                continue
            page_list.append(page)
        return sorted(page_list, reverse=True)

    @property
    def is_draft(self):
        return (
            datetime(year=self.date.year, month=self.date.month, day=self.date.day)
            > TODAY
        )


@cli
def orphans():
    """Print out fragments not linked to any page."""
    linked_fragments = []
    for file_path in each_markdown_from(DAVID / "2020"):
        for line in open(file_path).readlines():
            if line.startswith(".. include:: fragments/"):
                linked_fragments.append(line[len(".. include:: fragments/") : -1])
    all_fragments = []
    for file_path in each_markdown_from(DAVID / "2020" / "fragments"):
        all_fragments.append(file_path[len("david/2020/fragments/") :])
    for fragment_filename in set(all_fragments) - set(linked_fragments):
        # Prepending path for easy command+click from fish.
        print(f"Orphan: {DAVID / '2020' / 'fragments' / fragment_filename}")


@cli
def pages():
    """Build the agregations from fragments."""
    root_path = DAVID / "2020"
    for previous, page, next_ in neighborhood(
        reversed(Page.all(source=root_path, only_published=False)),
        first={
            "url": "/david/stream/",
            "title": "Streams 2009-2019",
            "is_draft": False,
        },
    ):
        template = environment.get_template("article_2020.html")
        content = template.render(
            page=page,
            prev=previous,
            next=next_,
        )
        target_path = Path(page.url[1:])
        target_path.mkdir(parents=True, exist_ok=True)
        open(target_path / "index.html", "w").write(content)
        if page.is_draft:
            print(f"Draft: {LOCAL_DOMAIN}{page.url} ({page.title})")
    template = environment.get_template("archives_2020.html")
    content = template.render(page_list=reversed(Page.all(source=root_path)))
    open(root_path / "index.html", "w").write(content)


@cli
def home():
    """Build the home page with last published items."""
    template = environment.get_template("profil.html")
    content = template.render(
        page_list=reversed(Page.all(source=DAVID / "2020")),
    )
    open(DAVID / "index.html", "w").write(content)


@cli
def feed():
    """Generate a feed from last published items."""
    template = environment.get_template("feed.xml")
    content = template.render(
        page_list=Page.all(source=DAVID / "2020", with_h2_anchors=False),
        current_dt=TODAY.strftime(NORMALIZED_STRFTIME),
        BASE_URL=f"{DOMAIN}/david/",
    )
    open(DAVID / "log" / "index.xml", "w").write(content)


@wrap
def perf_wrapper():
    start = perf_counter()
    yield
    elapsed = perf_counter() - start
    print(f"Done in {elapsed:.5f} seconds.")


if __name__ == "__main__":
    run()