blog.py

"""Converts markdown files and jupyter notebooks in a directory to html files

This file:
- gets config values from config.ini file
- reads all markdown files in a folder, parses front matter, saves info in a Posts dataclass
- returns a  sorted dict of Posts and a dict of tags: post_slugs 


"""

# libs built into python
import os
import shutil
import json
from pathlib import Path
from dataclasses import dataclass
from configparser import ConfigParser
import re
from collections import defaultdict
from dataclasses import dataclass
from typing import List, Optional
import argparse

# external libs, try to minimize use
import yaml

# import pandas as pd  # no real need to use this so rethink later
from mako.template import Template
from mako.lookup import TemplateLookup

# from tqdm.auto import tqdm
from datetime import datetime
import markdown
import pymdownx.emoji

# import nbconvert
# import nbformat


@dataclass
class Post:
    """class to store each individual posts data"""

    title: str
    title_id: str
    type: str  # type of content post, notebook, link, etc
    slug: str
    link: str  # relative link to post in form tag/slug.html
    path: Path
    tags: List[str]
    markdown: str
    html: str
    date: datetime
    lastmod: datetime
    toc: Optional[str]
    draft: bool = False
    github_link: str = None  # link to original markdown file on github


##################
# read config file
##################
config = ConfigParser()
config.read("config.ini")

# read in all the config stuff actually used from the config file
title: str = config["blog"]["title"]
header: str = config["blog"]["header"]
header_text: str = config["blog"]["header_text"]
baseurl: str = config["blog"]["baseurl"]
default_post_title: str = config["blog"]["default_post_title"]
default_post_type: str = config["blog"]["default_post_type"]

# paths to content
path_md: Path = Path(config["paths"]["posts"])  # md posts
path_nb: Path = Path(config["paths"]["notebooks"])  # jupyter notebooks
path_nb2md = path_nb / config["paths"]["nb2md"]  # markdown version of notebooks
path_static: Path = Path(config["paths"]["static"])  # static files
path_publish: Path = Path(config["paths"]["publish"])  # final output folder

#############################################
# configure python markdown parser
#############################################
# make enters into line breaks by adding "nl2br", add emoji extension
extensions = [
    "extra",
    "toc",
    "pymdownx.emoji",
    # "codehilite", # https://python-markdown.github.io/extensions/code_hilite/
    # "smarty"
]

# https://help.farbox.com/pygments.html - consider monokai default themes
# noclasses: True puts all the styling in the html itself. False uses css styles
extension_configs = {
    "output_format": "html5",
    "pymdownx.emoji": {"emoji_generator": pymdownx.emoji.to_alt, "alt": "html_entity"},
    "codehilite": {"noclasses": False, "linenums": False, "pygments_style": "autumn"},
}
md = markdown.Markdown(extensions=extensions, extension_configs=extension_configs)

######################
# setup Mako templates
######################
lookup = TemplateLookup(directories=["templates"])


def convert_notebooks_to_md():
    """converts jupter notebooks to markdown files and saves them in the path
    specified in config file.
    """
    # removing previously converted notebooks
    shutil.rmtree("nb2md", ignore_errors=True)

    # am I using nbconvert here?
    # exporter = nbconvert.MarkdownExporter()
    # write_file = nbconvert.writers.FilesWriter(build_directory="tmp")

    nb_paths = [
        f for f in path_nb.rglob("*.ipynb") if ".ipynb_checkpoints" not in str(f)
    ]

    print(f"Converting {len(nb_paths)} notebooks to markdown files.")

    for nb_path in nb_paths:
        try:
            os.system(
                f"jupyter nbconvert --to markdown {nb_path} --output-dir {path_nb2md}"
            )
        except:
            print(f"failed to convert {nb_path}")
        print(f"converted {nb_path}")

        # os.system(f"jupyter nbconvert --to markdown {nb_path} --output-dir tmp")
        # make folder for exported files
        # make_folder(Path("tmp") / f"{nb_path.stem}_files")
        # os.system(f"nbdev_nb2md --dest {Path('tmp')} --jekyll=True {nb_path}")

        # with open(nb_path) as f:
        #     nb_node = nbformat.read(f, as_version=4)

        # (body, resources) = exporter.from_notebook_node(nb_node)

        # write_file.write(
        #     output=body, resources=resources, notebook_name=nb_path.name[:-6]
        # )

    print(f"--------------------------\n")


def md_to_post(p: Path, post_type=default_post_type, debug=False):
    """takes in a path to md file  and returns a Post obj"""

    try:
        all_txt = p.read_text()
    except:
        print(f"Failed to read {p}")

    # extract front matter b/w "---" lines
    n = all_txt[3:].find("---") + 3
    fm = yaml.load(all_txt[:n], Loader=yaml.FullLoader)  # front matter dict
    txt = all_txt[n + 3 :].strip()  # text excluding front matter

    _post_type = fm.get("type", post_type)  # defaults to post
    slug = fm.get("slug", p.name.split(".")[0])  # todo: check repeated slugs
    dt = fm.get("date", datetime.fromtimestamp(p.stat().st_ctime).date())
    if debug:
        print(type(dt), title, dt)
    lastmod = fm.get("lastmod", datetime.fromtimestamp(p.stat().st_mtime).date())
    draft = fm.get("draft", False)

    tags = fm.get("tags", ["untagged"])

    # remove the first h1 line from text?
    # regex to grab md "^# .+\n"gm
    # regex to grab html h1: "^<h1>.+</h1>"gm
    # warning: this is hacky fix later
    html = md.convert(txt)
    regex = re.compile("^<h1 .+</h1>")
    html = re.sub(regex, "", html, count=1)
    if debug:
        print(html[:100])

    # if first header is h1 override the title from front matter
    title = fm.get("title", default_post_title)
    title_id = None
    try:
        if md.toc_tokens[0]["level"] == 1:
            title = md.toc_tokens[0]["name"]
            title_id = md.toc_tokens[0]["id"]
    except:
        pass

    # enable toc for posts if more than 1 heading or toc: True in front matter
    # if (toc := fm.get("toc", False)) :
    #    toc = md.toc
    if len(md.toc_tokens) > 0:  # just add toc anyways if more than 1 heading.
        toc = md.toc
        if len(md.toc_tokens) == 1 and len(md.toc_tokens[0]["children"]) == 0:
            toc = False
    else:
        toc = False

    link = f"{tags[0]}/{slug}.html"

    post = Post(
        title=title,
        title_id=title_id,
        type=_post_type,
        slug=slug,
        link=link,
        path=p,
        tags=tags,
        date=dt,
        lastmod=lastmod,
        markdown=txt,
        draft=draft,
        toc=toc,
        html=html,
    )

    return post


def get_posts(debug=False, build_notebooks=True):
    """reads from disk and returns a dataframe of all posts"""

    # convert notebooks to md and save in a tmp folder
    if build_notebooks:
        convert_notebooks_to_md()
    else:
        print("Using cached markdown files for jupyter notebooks")

    # lists of md files and notebooks to convert
    # ignore files starting README
    md_paths = [f for f in path_md.rglob("*.md") if not f.name.startswith("README")]

    try:
        nb_paths = [f for f in path_nb2md.rglob("*.md")]
    except:
        print("No md files converted from jupyter notebooks found in tmp dir")

    # decide whether to use dataframe or a dict to hold all the posts
    posts = []  # list holding all the posts
    postsdict = {}  # dict to hold all posts, using slug as the key
    tagsdict = defaultdict(set)  # holds set of all posts for every tag

    for p in md_paths:
        if debug:
            print(p)
        pp = md_to_post(p, debug=debug)
        posts.append(pp)

    for p in nb_paths:
        pp = md_to_post(p, post_type="notebook", debug=debug)
        if debug:
            print(type(pp.date), pp.title, pp.date)
        posts.append(pp)

    if debug:
        for post in posts:
            print(type(post.date), post.slug)

    posts.sort(key=lambda x: x.date, reverse=True)

    # create tagsdict
    for post in posts:
        for tag in post.tags:
            tagsdict[tag].add(post.slug)

    # create a dict of posts - to make it easier to lookup a certain post
    for post in posts:
        postsdict[post.slug] = post

    return posts, postsdict, tagsdict

    # misc stuff, fix this later

    df = pd.DataFrame(data)  # .set_index("filename")
    print(f"{len(df)} posts converted and saved in dataframe")

    # make the slug the index - hopefully they are all unique
    df = df.set_index("slug")
    # most recent at top
    df = df.sort_values(by="date", ascending=False)
    return df


def make_folder(path, debug: bool = False):
    """makes parent folders if they don't exist for a path"""

    # assume the last thing is a file if there is a dot in the name
    if "." in path.name:
        path = path.parent

    try:
        path.mkdir(parents=True, exist_ok=False)
        if debug:
            print(f"{path} folder made")
    except FileExistsError:
        if debug:
            print(f"{path} folder already exists")
        pass


def search_json(posts, path=path_publish):
    """makes a json of the posts info to search, returns json string
    and writes json file to disk"""

    search_list = [
        {"title": post.title, "tags": post.tags, "link": post.link} for post in posts
    ]

    posts_json = json.dumps(search_list)
    path = path / "posts.json"
    make_folder(path)

    path.write_text(posts_json)
    return posts_json


def write_posts(posts, tmpl: str = "post.html"):
    """makes a html page for each post using list of posts passed in

    Args:
        tmpl: name of templete
        posts: list of posts
    """

    template = lookup.get_template(tmpl)

    print(f"\nWriting html pages for {len(posts)} posts.")

    for post in posts:
        html = template.render(
            post=post, header=header, header_text=header_text
        ).strip()

        # publish each post inside its first tag folder
        if post.tags:
            path = path_publish / post.tags[0]
        else:
            path = path_publish

        make_folder(path)  # make folder in case

        # copy over post_files to the post folder.
        _path = path_nb2md / f"{post.slug}_files"
        if _path.is_dir():
            shutil.copytree(_path, path / f"{post.slug}_files", dirs_exist_ok=True)

        path = path / f"{post.slug}.html"

        path.write_text(html)
        print(f"wrote {post.slug} to {path}")


def write_index_page(posts, tags=None, tmpl: str = "index.html", foldername=None):
    """makes a html index page in the foldername using list of posts passed in

    Args:
        tmpl: name of templete
        foldername: name of folder to create
        posts: list of posts
    """

    template = lookup.get_template(tmpl)
    path = path_publish

    if foldername:
        path = path / foldername

    make_folder(path)  # make folder in case it doesn't exist

    # get json for fusejs search object
    # todo: modify js to read json from disk instead of passing obj
    postsjson = search_json(posts, path)

    html = template.render(
        posts=posts,
        tags=tags,
        title=title,
        postsjson=postsjson,
        header=header,
        header_text=header_text,
    ).strip()

    # write to disk
    path = path / tmpl
    path.write_text(html)
    print(f"wrote {tmpl} to {path}")


def write_tags(posts, tags):
    """writes an index.html file for each tag"""
    print(f"\nWriting an index page for {len(tags)} tags.")
    for tag in tags.keys():
        # filter posts
        tag_posts = [post for post in posts if tag in post.tags]
        # using the same index page for now
        write_index_page(foldername=tag, posts=tag_posts)


def write_all(posts: list, tags: dict):
    """this writes all the html pages to disk"""
    # posts, postsdict, tags = get_posts()
    write_index_page(posts=posts, tags=tags, foldername="")
    write_tags(posts=posts, tags=tags)
    write_posts(posts=posts)


def copy_static():
    """copies over the contents of the static folder to the public folder"""

    try:
        shutil.copytree(path_static, path_publish, dirs_exist_ok=True)
        print(f"copied contents of {path_static} over to {path_publish}")
    except:
        print("Failed to copy static files")

    # copy notebook files
    print(f"TODO: copied notebook files over to {path_publish}")


def start_server(directory="public", PORT=8000):
    """runs basic python server, serving up the public dir"""
    import http.server
    import socketserver
    import functools

    # using functools.partial to pass in the dir to the http server
    # otherwise it doesn't take in the dir argument
    Handler = functools.partial(
        http.server.SimpleHTTPRequestHandler, directory=directory
    )

    with socketserver.TCPServer(("", PORT), Handler) as httpd:
        try:
            print("serving at port", PORT)
            httpd.serve_forever()
        except KeyboardInterrupt:
            print(f"stopping localhost websert at port {PORT}")
            httpd.shutdown()


def get_parser():
    """sets up and returns cli parser"""

    # setup cli options
    parser = argparse.ArgumentParser(
        description="build html site from posts and notebooks, and optionally, serve website."
    )
    parser.add_argument(
        "-s",
        "--serve",
        action="store_true",
        help="start local website server, defaults to False",
    )
    parser.add_argument(
        "-bn",
        "--buildnotebooks",
        action="store_true",
        help="use cached markdown files for jupyter notebooks by default",
    )
    parser.add_argument(
        "-p",
        "--posts",
        action="store_false",
        help="don't build posts, defaults to building",
    )
    return parser


def app():
    """pulls in all the things to build the blog"""

    args = get_parser().parse_args()

    if args.posts:
        if args.buildnotebooks:
            print("using previously converted markdown files for notebooks.")
        # get posts
        posts, postsdict, tags = get_posts(build_notebooks=args.buildnotebooks)

        # copy static files over to the publish dir
        copy_static()

        # write all the things to the publish dir
        write_all(posts, tags)

        print("Website generated.")

    if args.serve:
        print(f"Starting server in dir {path_publish}")
        start_server()


if __name__ == "__main__":
    app()