Python shenanigans - a collection of fun python scripts

generate_html.py

"""
General script to retrieve all HTML elements in the HTML5
spec and generate a general module
"""

from functools import reduce
import urllib.request

def create_module(module_name, functions, content):
    """
    Creates the HTML module with the specified module name,
    exports and content
    """
    header = f"module {module_name}\n\t( "
    exports = format_exports(functions)

    return f"{header}{exports}) where\n\n{content}"

def convert_to_utf8(elements):
    """
    Convert the bytes retrieved from the gists to UTF-8 text
    """
    return [element.decode("utf-8") for element in elements]

def format_content(elements):
    """
    Concatenate all the functions generated by the transform
    functions to form one content string
    """
    return reduce(lambda accumulative, element: f"{accumulative}{element}", elements)

def format_exports(elements):
    """
    Formats the exports to only export the function which
    represent the HTML generation functions
    """
    export_string = ""
    for index in range(0, len(elements)):
        element = elements[index]
        export_string += f"{element}\n\t"

        if index != len(elements) - 1:
            export_string += ", "

    return export_string

def read_elements(url):
    """
    Retrieve the text content from the gist and split it at
    the newline characters
    """
    return urllib.request.urlopen(url).read().splitlines()

def transform_node(element):
    """
    Transform the given element name into a node
    """
    return f"{element} :: forall e. Markup e -> Markup e\n{element} = node \"{element}\"\n\n"

def transform_leaf(element):
    """
    Transform the given element name into a node without any
    children
    """
    return f"{element} :: forall e. Markup e\n{element} = leaf \"{element}\"\n\n"

MODULE = "Text.Hydrogen.HTML"

ELEMENTS_URL = "https://gist.githubusercontent.com/DevNebulae/30f66ba4fe1267004fd6bafb702b58f1/raw/028d8ea65c9e304bedfe1bfbbe1c0c514089ab18/html-elements.txt"
NODES_URL = "https://gist.githubusercontent.com/DevNebulae/30f66ba4fe1267004fd6bafb702b58f1/raw/99a21ced0132da95bc8883fa7b2de7cd356a5bfe/html-elements-nodes.txt"
LEAVES_URL = "https://gist.githubusercontent.com/DevNebulae/30f66ba4fe1267004fd6bafb702b58f1/raw/99a21ced0132da95bc8883fa7b2de7cd356a5bfe/html-elements-leaves.txt"

ELEMENTS = convert_to_utf8(read_elements(ELEMENTS_URL))
NODES = convert_to_utf8(read_elements(NODES_URL))
LEAVES = convert_to_utf8(read_elements(LEAVES_URL))

def main():
    """
    Main function which executes all steps and prints the
    HTML module
    """
    nodes = list(map(transform_node, NODES))
    leaves = list(map(transform_leaf, LEAVES))
    content = format_content(sorted(nodes + leaves))

    print(create_module(MODULE, ELEMENTS, content))

main()