larlet-fr-david/site.py

#!/usr/bin/env python3
import json
import locale
from collections import defaultdict
from dataclasses import dataclass
from datetime import datetime, timedelta
from html import escape
from itertools import chain, groupby
from operator import attrgetter
from pathlib import Path
from string import Template
from textwrap import dedent
from time import perf_counter

import feedparser
import mistune
from jinja2 import Environment as Env
from jinja2 import FileSystemLoader
from jinja2.filters import do_striptags
from minicli import cli, run, wrap
from mistune.plugins.formatting import mark, strikethrough
from mistune.util import safe_entity
from PIL import Image
from slugify import slugify

from typography import typographie
from widont import widont

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

VERSION = "2023.09.10"
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)
PUBLICATION_BUFFER = TODAY - timedelta(days=0)
NB_ITEMS_IN_FEED = 30
SOURCES_PATH = DAVID / "2023" / "_sources"

all_tags = set()
pages_by_tags = defaultdict(list)
pages_by_url = {}


class TagsRenderer(mistune.HTMLRenderer):
    """Make the asumption each line starting with a `#` is a tag."""

    def paragraph(self, text):
        if text.startswith("#"):
            tags = " ".join(
                f'<a href="/david/2023/{slugify(tag.strip())}/">#{tag.strip()}</a>'
                for tag in text.split("#")
                if tag.strip()
            )
            return f"<nav><p>{tags}</p></nav>\n"
        return super().paragraph(text)


class FrenchTypographyRenderer(mistune.HTMLRenderer):
    """Apply French typographic rules to text."""

    def text(self, text):
        return typographie(super().text(text), html=True)

    def block_html(self, html):
        return typographie(super().block_html(html), html=True)


class InternalLinkTitleRenderer(mistune.HTMLRenderer):
    """Automatically generate the title for internal links."""

    def link(self, text, url, title=None):
        s = '<a href="' + self.safe_url(url) + '"'

        if not title and url.startswith("/david/2023/"):
            # It will not work for internal urls referencing the future.
            page = pages_by_url.get(url)
            if page:
                title = page.title

        if title:
            s += ' title="' + safe_entity(title) + '"'
        return s + ">" + text + "</a>"


class CustomAndBlockquoteLanguageRenderer(
    FrenchTypographyRenderer, InternalLinkTitleRenderer, TagsRenderer
):
    """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(CustomAndBlockquoteLanguageRenderer):
    """Renders images as <figure>s and add sizes."""

    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 super().paragraph(text)

    def _generate_size(self, src, width, height):
        src_size = src.replace(".jpg", f"_{width}x{height}.jpg")
        full_path = STATIC / Path(src[1:])
        full_path_size = STATIC / Path(src_size[1:])
        if full_path_size.exists() or "/2023/" not in src:
            return src_size

        image = Image.open(full_path)
        image.thumbnail((width, height), resample=Image.LANCZOS)
        image.save(full_path_size, icc_profile=image.info.get("icc_profile"))
        return src_size

    def _generate_webp(self, src):
        src_webp = src.replace(".jpg", ".webp")
        full_path = STATIC / Path(src[1:])
        full_path_webp = STATIC / Path(src_webp[1:])
        if full_path_webp.exists() or "/2023/" not in src:
            return src_webp

        image = Image.open(full_path)
        image.save(
            full_path_webp, format="webp", icc_profile=image.info.get("icc_profile")
        )
        # command = [
        #     "cwebp",
        #     "-q",
        #     "80",
        #     full_path,
        #     "-o",
        #     full_path_webp,
        #     "-metadata",
        #     "icc",
        # ]
        # subprocess.check_output(command, stderr=subprocess.STDOUT)

        return src_webp

    def image(self, alt, url, title=None):
        SIZES = [(660, 440), (990, 660), (1320, 880)]
        full_path = STATIC / Path(url[1:])
        image = Image.open(full_path)
        width, height = image.size
        jpg_srcs = [(url, width, height)]
        # src_webp = self._generate_webp(src)
        # webp_srcs = [(src_webp, width, height)]
        for size_width, size_height in SIZES:
            src_size = self._generate_size(url, size_width, size_height)
            jpg_srcs.append((src_size, size_width, size_height))
            # src_size_webp = self._generate_webp(src_size)
            # webp_srcs.append((src_size_webp, size_width, size_height))

        jpg_srcsets = ", ".join(
            f"{jpg_src} {jpg_width}w" for jpg_src, jpg_width, jpg_height in jpg_srcs
        )
        # webp_srcsets = ", ".join(
        #     f"{webp_src} {webp_width}w"
        #     for webp_src, webp_width, webp_height in webp_srcs
        # )
        return dedent(
            f"""\
                <figure>
                    <a href="{url}"
                        title="Cliquer pour une version haute résolution">
                        <img
                            src="{url}"
                            width="{width}" height="{height}"
                            srcset="{jpg_srcsets}"
                            sizes="min(100vw, calc(100vh * {width} / {height}))"
                            loading="lazy"
                            decoding="async"
                            alt="{alt}">
                    </a>
                    <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),
    plugins=[mark, strikethrough],
)
# 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),
    plugins=[mark, strikethrough],
)

# 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_file_from(source_dir, pattern="*", exclude=None):
    """Walk across the `source_dir` and return the `pattern` file paths."""
    for path in _each_path_from(source_dir, pattern=pattern, exclude=exclude):
        if path.is_file():
            yield path


def each_folder_from(source_dir, exclude=None):
    """Walk across the `source_dir` and return the folder paths."""
    for path in _each_path_from(source_dir, exclude=exclude):
        if path.is_dir():
            yield path


def _each_path_from(source_dir, pattern="*", exclude=None):
    for path in sorted(Path(source_dir).glob(pattern)):
        if exclude is not None and path.name in exclude:
            continue
        yield path


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

    def __post_init__(self):
        try:
            date_str, _ = self.file_path.split(" - ", 1)
        except ValueError:
            # Fallback for 2020 contents (search index)
            suffix = len(".md")
            prefix = len("YYYY/MM-DD") + suffix
            date_str = "2020-" + self.file_path[-prefix:-suffix]
        self.url = f"/david/{date_str.replace('-', '/')}/"
        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)
        tag_template = Template(
            f'<a href="{DOMAIN}/david/2023/$tag_slug/">#$tag_name</a>'
        )
        tag_links = " ".join(
            tag_template.substitute(tag_slug=slugify(tag), tag_name=tag)
            for tag in self.tags
        )
        self.escaped_content = escape(
            self.content.replace('href="/', f'href="{DOMAIN}/')
            .replace('src="/', f'src="{DOMAIN}/')
            .replace('href="#', f'href="{self.full_url}#')
            + f"<nav><p>{tag_links}</p></nav>"
            + '<hr/><p><a href="mailto:david@larlet.fr">Réagir ?</a></p>'
        )
        # Extract first paragraph.
        self.extract = self.content.split("</p>", 1)[0] + "</p>"
        # Create the index for the search.
        self.search_data = {
            "title": self.title,
            "url": self.url,
            "date": date_str,
            "content": do_striptags(self.content)
            .replace("\u00a0(cache)", " ")
            .replace("'", " "),
        }

    def __eq__(self, other):
        return self.url == other.url

    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 sorted(each_file_from(source, pattern="*.md")):
            result, state = md.read(file_path)
            result = widont(result, html=True)
            # 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:]
            tags = {}
            if "<nav><p>" in content:
                # Extract the tags from the generated page.
                content, tags_links = content.split("<nav><p>", 1)
                nav_closing_size = len("</p></nav>\n")
                tags_links = tags_links[:-nav_closing_size]
                try:
                    tags = sorted(
                        {
                            tag.strip().split("#", 1)[1]
                            for tag in tags_links.split("</a>")
                            if tag.strip()
                        },
                        key=lambda tag: slugify(tag),
                    )
                except IndexError:
                    # It happens for old contents, parsed for the search index.
                    pass
            page = Page(title, content, tags, file_path.name)
            pages_by_url[page.url] = page
            if not page.is_draft:
                all_tags.update(tags)
                for tag in tags:
                    if page not in pages_by_tags[tag]:
                        pages_by_tags[tag].append(page)
            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)
            > PUBLICATION_BUFFER
        )


@cli
def pages():
    """Build article pages."""
    root_path = DAVID / "2023"
    for previous, page, next_ in neighborhood(
        reversed(Page.all(source=SOURCES_PATH, only_published=False)),
        first={
            "url": "/david/2022/",
            "title": "Publications 2022",
            "is_draft": False,
        },
    ):
        template = environment.get_template("article_2020.html")
        content = template.render(page=page, prev=previous, next=next_, slugify=slugify)
        target_path = Path(page.url[1:])
        target_path.mkdir(parents=True, exist_ok=True)
        (target_path / "index.html").write_text(content)
        if page.is_draft:
            print(f"Draft: {LOCAL_DOMAIN}{page.url} ({page.title})")

    def group_by_month_year(item):
        return item.date.strftime("%B %Y").title()

    template = environment.get_template("archives_2020.html")
    page_list = reversed(Page.all(source=SOURCES_PATH))
    tags = sorted((slugify(tag), tag, len(pages_by_tags[tag])) for tag in all_tags)
    content = template.render(
        page_list=groupby(page_list, key=group_by_month_year), tags=tags
    )
    (root_path / "index.html").write_text(content)


@cli
def tags():
    """Build tags pages."""
    # Parse all pages to collect tags.
    Page.all(source=SOURCES_PATH, only_published=True)
    for tag in all_tags:
        template = environment.get_template("tag_2021.html")
        content = template.render(
            page_list=sorted(pages_by_tags[tag], reverse=True),
            tag_name=tag,
        )
        target_path = DAVID / "2023" / slugify(tag)
        target_path.mkdir(parents=True, exist_ok=True)
        (target_path / "index.html").write_text(content)


@cli
def home():
    """Build the home page with last published items."""
    template = environment.get_template("profil.html")
    page_list = Page.all(source=SOURCES_PATH, only_published=True)
    tags = sorted((slugify(tag), tag, len(pages_by_tags[tag])) for tag in all_tags)
    content = template.render(page_list=page_list, tags=tags)
    (DAVID / "index.html").write_text(content)


@dataclass
class Website:
    name: str
    url: str
    feed: str

    def __post_init__(self):
        content = feedparser.parse(self.feed)
        entries = content.get("entries")
        try:
            entries = sorted(entries, key=attrgetter("updated_parsed"), reverse=True)[
                :2
            ]
        except AttributeError:
            print("No `updated_parsed` in", entries)
        for entry in entries:
            entry.website = self
        self.entries = entries


@cli
def blogroll():
    """Build the blogroll with last published items from people."""
    template = environment.get_template("blogroll.html")
    website_list = [
        # Invalid feed, date not standard.
        # Website(
        #     name="Alex Sirac",
        #     url="https://alexsirac.com/",
        #     feed="https://alexsirac.com/feed",
        # ),
        Website(
            name="Maïtané Lenoir",
            url="https://www.maiwann.net/",
            feed="https://www.maiwann.net/feed.xml",
        ),
        Website(
            name="Fanny Cheung",
            url="https://ynote.hk/",
            feed="https://ynote.hk/feed.xml",
        ),
        Website(
            name="La Lune Mauve",
            url="https://lalunemauve.fr/",
            feed="https://lalunemauve.fr/feed/",
        ),
        Website(
            name="Eliness",
            url="https://www.hypothermia.fr/",
            feed="https://www.hypothermia.fr/rss",
        ),
        Website(
            name="Luce Carević",
            url="https://luce.carevic.eu/fr",
            feed="https://luce.carevic.eu/fr/flux",
        ),
        Website(
            name="Emma",
            url="https://emmaclit.com/",
            feed="https://emmaclit.com/feed/",
        ),
        Website(
            name="Karl Dubost",
            url="https://www.la-grange.net/",
            feed="https://www.la-grange.net/feed.atom",
        ),
        Website(
            name="Thomas Parisot",
            url="https://thom4.net/",
            feed="https://thom4.net/feed/",
        ),
        Website(
            name="Arthur Perret",
            url="https://www.arthurperret.fr/",
            feed="https://www.arthurperret.fr/feed.xml",
        ),
        Website(
            name="Antoine Fauchié",
            url="https://www.quaternum.net/",
            feed="https://www.quaternum.net/atom.xml",
        ),
        Website(
            name="Éric D.",
            url="https://n.survol.fr/",
            feed="https://n.survol.fr/feed",
        ),
        Website(
            name="Winnie Lim (en)",
            url="https://winnielim.org/",
            feed="https://winnielim.org/feed/",
        ),
    ]
    entry_list = sorted(
        chain(*[website.entries for website in website_list]),
        key=attrgetter("updated_parsed"),
        reverse=True,
    )
    content = template.render(website_list=website_list, entry_list=entry_list)
    (DAVID / "blogroll" / "index.html").write_text(content)


@cli
def toot():
    """Pre-write the Mastodon message."""
    page_list = Page.all(source=SOURCES_PATH, only_published=True)
    last_published = page_list[0]
    print(f"✍️ QUOTE? — {last_published.title}, {last_published.full_url}")
    print()
    print("#blog #larletfr #rss")
    print(" ".join([f"#{tag}" for tag in last_published.tags]))


@cli
def search():
    """Build the static search page with custom index."""
    template = environment.get_template("recherche.html")
    page_list_2023 = Page.all(
        source=SOURCES_PATH, only_published=True, with_h2_anchors=False
    )
    page_list_2022 = Page.all(
        source=DAVID / "2022" / "_sources", only_published=True, with_h2_anchors=False
    )
    page_list_2021 = Page.all(
        source=DAVID / "2021" / "sources", only_published=True, with_h2_anchors=False
    )
    page_list_2020 = Page.all(
        source=DAVID / "2020", only_published=True, with_h2_anchors=False
    )
    blog_page_list_2019 = BlogPage.all(source=DAVID / "blog" / "2019")
    blog_page_list_2018 = BlogPage.all(source=DAVID / "blog" / "2018")
    blog_page_list_2017 = BlogPage.all(source=DAVID / "blog" / "2017")
    stream_page_list_2019 = StreamPage.all(source=DAVID / "stream" / "2019")
    stream_page_list_2018 = StreamPage.all(source=DAVID / "stream" / "2018")
    page_list = (
        page_list_2023
        + page_list_2022
        + page_list_2021
        + page_list_2020
        + blog_page_list_2019
        + blog_page_list_2018
        + blog_page_list_2017
        + stream_page_list_2019
        + stream_page_list_2018
    )
    search_index = json.dumps([page.search_data for page in page_list], indent=2)
    content = template.render(search_index=search_index)
    (DAVID / "recherche" / "index.html").write_text(content)


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


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


# Below are legacy blog contents, still useful for search indexation.
@dataclass
class BlogPage:
    title: str
    content: str
    file_path: str
    date_str: str

    def __post_init__(self):
        self.date = datetime.strptime(self.date_str, "%Y-%m-%d").date()
        self.url = f"/{self.file_path}/"
        # Create the index for the search.
        self.search_data = {
            "title": self.title,
            "url": self.url,
            "date": self.date_str,
            "content": do_striptags(self.content)
            .replace("\u00a0(cache)", " ")
            .replace("'", " ")
            .replace("<", "&lt;")
            .replace(">", "&gt;"),
        }

    def __eq__(self, other):
        return self.url == other.url

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

    @staticmethod
    def all(source: Path):
        """Retrieve all pages sorted by desc."""
        page_list = []
        for folder in each_folder_from(source):
            for path in each_file_from(folder, pattern="*.md"):
                metadata, content = path.read_text().split("\n\n", 1)
                if "lang:" in metadata:
                    title, slug, date_, chapo, lang = metadata.split("\n")
                else:
                    title, slug, date_, chapo = metadata.split("\n")
                title = title[len("title: ") :].strip()
                date_str = date_[len("date: ") :].strip()
                content = markdown_with_img_sizes(content)
                page = BlogPage(title, content, path.parent, date_str)
                page_list.append(page)
        return sorted(page_list, reverse=True)


@dataclass
class StreamPage:
    title: str
    content: str
    file_path: str
    date_str: str

    def __post_init__(self):
        self.date = datetime.strptime(self.date_str, "%Y/%m/%d").date()
        self.url = f"/{self.file_path}/"
        # Create the index for the search.
        self.search_data = {
            "title": self.title,
            "url": self.url,
            "date": self.date.isoformat(),
            "content": do_striptags(self.content)
            .replace("\u00a0(cache)", " ")
            .replace("'", " ")
            .replace("<", "&lt;")
            .replace(">", "&gt;"),
        }

    def __eq__(self, other):
        return self.url == other.url

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

    @staticmethod
    def all(source: Path):
        """Retrieve all pages sorted by desc."""
        page_list = []
        for folder in each_folder_from(source):
            for subfolder in each_folder_from(folder):
                for path in each_file_from(subfolder, pattern="*.md"):
                    metadata, content = path.read_text().split("\n\n", 1)
                    if "lang:" in metadata:
                        title, lang = metadata.split("\n")
                    else:
                        title = metadata.strip()
                    title = title[len("title: ") :].strip()
                    date_str = str(path.parent)[-len("YYYY/MM/DD") :]
                    content = markdown_with_img_sizes(content)
                    page = StreamPage(title, content, path.parent, date_str)
                    page_list.append(page)
        return sorted(page_list, reverse=True)


if __name__ == "__main__":
    run()