orgmode-to-gemini-blog/enrich_html.py

#!/bin/python3
import os
import argparse
import re

parser = argparse.ArgumentParser(description="Générer un site Web à partir de fichiers HTML.")
parser.add_argument("blog_name", help="Le chemin vers le dossier contenant les fichiers HTML.")
parser.add_argument("--title", "-t", default="Mon site Web", help="Le titre du site Web.")
parser.add_argument("--style", default="templates/style_general.css", help="Le chemin vers le fichier de style CSS.")
args = parser.parse_args()

# Style CSS minimaliste
style_file = args.style


blog_name = args.blog_name
source_blog = f"sources/{blog_name}"
header_content_path = f"{source_blog}/templates/converted/header_page.html"
footer_content_path = f"{source_blog}/templates/converted/footer_page.html"
static_page_path = f"{source_blog}/templates/html/static.html"

footer_content=''
after_article=''
# TODO add footer on every article
# with open(footer_content_path, "r") as f:
#         footer_content = f.read()

# TODO make these variables overrided by configuration of a source website
# variables du template de page
BANNIERE_ENTETE=''
BLOG_TITLE='Cipher Bliss'
BLOG_SUBTITLE='Code, nouvelles technologies et entrepreneurariat par B. Lemoine'
TITLE=''
AUTHOR=''
PAGE_TITLE=''
LOCALE='fr_FR'
DESCRIPTION=''
NDD='https://www.cipherbliss.com'
EMAIL='contact@cipherbliss.com'
SITE_ICON='https://www.cipherbliss.com/wp-content/uploads/2016/12/rond.png'
SITE_ICON_TYPE='image/png'
NAVIGATION="""
<nav>
    <a href="/">Accueil</a>
    <a href="https://portfolio.cipherbliss.com">Portfolio</a>
    <a href="/feed">Flux RSS</a>
    <a href="/contact">Contact</a>
    <a href="/ressources-de-café-vie-privée">Ressources</a>
</nav>
"""
BANNIERE_ENTETE='https://www.cipherbliss.com/wp-content/uploads/2016/11/bg.jpg'
BANNIERE_ENTETE_ALT='bannière du site'
ARTICLE=''
FOOTER=''


def extract_body_content(html_content):
    pattern = r'<body[^>]*?>(.*?)</body>'
    match = re.search(pattern, html_content, re.DOTALL)
    if match:
        return match.group(1)
    else:
        return None

def remove_properties_section(text):
    pattern = r"<h1 id=\"article\">Article</h1>.+?</ul>"
    replacement = ""
    return re.sub(pattern, replacement, text, flags=re.DOTALL)

def remove_article_head_properties_orgmode(text):
    pattern = r":PROPERTIES:.+?:END:"
    replacement = ""
    return re.sub(pattern, replacement, text, flags=re.DOTALL)

def remove_hint_html(text):
    pattern = r"<p>ceci<sub>estduhtml</sub></p>"
    replacement = ""
    return re.sub(pattern, replacement, text, flags=re.DOTALL)


def enrich_one_file(file, root_path):


    print(' ----------- enrich_html: file:',os.path.join(root_path, file))
    css_content = ""

    inline_the_css=False
    # inline_the_css=True

    print(' ----------- enrich_html: CSS inline: ',inline_the_css)
    # Trouver le fichier entête 
    header_content=''
    with open(os.path.join(root_path, file), "r") as f:
        header_content = f.read()
    # Ouvrir le fichier HTML en mode lecture
    with open(os.path.join(root_path, file), "r") as f:
        html_content = f.read()

    # remove some parts
    html_content = remove_properties_section(html_content)
    html_content = remove_article_head_properties_orgmode(html_content)
    html_content = remove_hint_html(html_content)

    html_content = extract_body_content(html_content)

    if inline_the_css is True:
        print(' ----------- enrich_html: include css inline in each html page')
        with open(os.path.join(root_path, file), "r") as f:
            css_content = f.read()
            css_content = "<style type='text/css'>{css_content}</style>"

    # remplir le template
    html_content = f"""
    
    <!DOCTYPE html>
<html>
    <head>
        <meta charset="UTF-8">
        <meta name="viewport" content="width=device-width, initial-scale=1">
        <meta property="og:image" content="{SITE_ICON}">
        <meta property="og:locale" content="{LOCALE}">
        <meta property="og:description" content="{BLOG_SUBTITLE}">
        <meta property="og:url" content="{NDD}">
        <meta property="og:site_name" content="Cipher Bliss">
        <link rel="alternate" type="application/rss+xml" title="Cipher Bliss » Flux" href="{NDD}/feed/">
        <link href="/style.css" rel="stylesheet">
        <meta charset="utf-8">
        <meta name="viewport" content="width=device-width, initial-scale=1">
        <title>{TITLE}</title>
        <meta name="author" content="{AUTHOR}">
        <link rel="alternate" type="application/rss+xml" title="{BLOG_TITLE} » Flux"
            href="{NDD}/feed/">
        <meta property="og:title" content="{PAGE_TITLE}">
        <meta property="og:locale" content="{LOCALE}">
        <!-- Description de la page -->
        <meta name="description" content="{PAGE_TITLE}">
        <meta name="reply-to" content="{EMAIL}">
        <link rel="icon" type="{SITE_ICON_TYPE}" href="{SITE_ICON}">
    </head>

    <body>
        <div id="page">
            <header id="masthead" class="site-header">
                <div class="header-image" style="background: url({BANNIERE_ENTETE})">
                    <a href="/">
                    <img src="{SITE_ICON}" class="site-icon img">
                    </a>
                    <h1 class="blog-title">{BLOG_TITLE}</h1>
                    <p class="blog-subtitle">{BLOG_SUBTITLE}</p>
                </div>
                    <nav class="navbar is-fixed-top is-dark" role="navigation" aria-label="main navigation">
                        <div class="navbar-brand">
                            <a class="navbar-item" href="{NDD}">
                                
                            </a>

                        </div>

                        <div id="navbarBasicExample" class="navbar-menu">
                            <div class="navbar-start">
                                <a class="navbar-item" href="{NDD}">
                                    <img src="{SITE_ICON}"
                                        class="img-fluid">
                                </a>
                                {NAVIGATION}
                            </div>
                            <div class="navbar-end">
                                <div class="navbar-item">
                                    <form role="search" method="get" class="search-form" action="/">
                                        <label>
                                            <input class="search-field" placeholder="Recherche" value="" name="s"
                                                type="search">
                                        </label>
                                        <input class="is-hidden search-submit" value="Rechercher" type="submit">
                                    </form>
                                </div>
                            </div>
                        </div>
                    </nav>
            </header>
            <main class="body-wrap boxed-container">
                <article class="content">
                    {html_content}
                    <p class="after-article">
                    {after_article}
                    </p>
                </article>
            </main>
            <footer class="site-footer has-top-divider">
                <div class="container">
                    <div class="site-footer-inner">
                       
                        
            {NAVIGATION}

                    </div>
                </div>
            </footer>
        </div>
    </body>
    <!-- généré avec orgmode-to-gemini-blog par Tykayn -->
</html>
 
"""

    html_path_enriched=os.path.join(root_path, file)
    print(' ----------- enrich_html: html_path_enriched ============> ',html_path_enriched)
    # Écrire le contenu modifié dans le fichier HTML
    with open(html_path_enriched, "w") as f:
        f.write(html_content)
        print('\n ----------- enrich_html: html écrit ', html_path_enriched)

# Parcourir tous les fichiers HTML dans le dossier
for root, _, files in os.walk(blog_name):
    # print(files)
    for file in files:
        if file.endswith(".html"):
            enrich_one_file(file, root)
change enrich template 2024-11-03 11:42:44 +01:00			`#!/bin/python3`
up readme 2024-11-03 10:29:30 +01:00			`import os`
change enrich template 2024-11-03 11:42:44 +01:00			`import argparse`
up style and template 2024-11-10 00:16:15 +01:00			`import re`
separate hard to convert files 2024-11-04 00:13:58 +01:00
change enrich template 2024-11-03 11:42:44 +01:00			`parser = argparse.ArgumentParser(description="Générer un site Web à partir de fichiers HTML.")`
replace org content, make global style for generated site 2024-11-09 00:19:43 +01:00			`parser.add_argument("blog_name", help="Le chemin vers le dossier contenant les fichiers HTML.")`
change enrich template 2024-11-03 11:42:44 +01:00			`parser.add_argument("--title", "-t", default="Mon site Web", help="Le titre du site Web.")`
up 2024-11-08 18:16:36 +01:00			`parser.add_argument("--style", default="templates/style_general.css", help="Le chemin vers le fichier de style CSS.")`
change enrich template 2024-11-03 11:42:44 +01:00			`args = parser.parse_args()`
up readme 2024-11-03 10:29:30 +01:00
separate hard to convert files 2024-11-04 00:13:58 +01:00			`# Style CSS minimaliste`
change enrich template 2024-11-03 11:42:44 +01:00			`style_file = args.style`
up 2024-11-08 18:05:08 +01:00

replace org content, make global style for generated site 2024-11-09 00:19:43 +01:00			`blog_name = args.blog_name`
			`source_blog = f"sources/{blog_name}"`
site builder ok 2024-11-11 00:58:44 +01:00			`header_content_path = f"{source_blog}/templates/converted/header_page.html"`
			`footer_content_path = f"{source_blog}/templates/converted/footer_page.html"`
add scss converter 2024-11-10 00:01:15 +01:00			`static_page_path = f"{source_blog}/templates/html/static.html"`

site builder ok 2024-11-11 00:58:44 +01:00			`footer_content=''`
			`after_article=''`
up displayed pages 2024-11-11 01:25:10 +01:00			`# TODO add footer on every article`
site builder ok 2024-11-11 00:58:44 +01:00			`# with open(footer_content_path, "r") as f:`
			`# footer_content = f.read()`

up displayed pages 2024-11-11 01:25:10 +01:00			`# TODO make these variables overrided by configuration of a source website`
add scss converter 2024-11-10 00:01:15 +01:00			`# variables du template de page`
			`BANNIERE_ENTETE=''`
			`BLOG_TITLE='Cipher Bliss'`
			`BLOG_SUBTITLE='Code, nouvelles technologies et entrepreneurariat par B. Lemoine'`
			`TITLE=''`
			`AUTHOR=''`
			`PAGE_TITLE=''`
more complex template html 2024-11-10 18:03:31 +01:00			`LOCALE='fr_FR'`
add scss converter 2024-11-10 00:01:15 +01:00			`DESCRIPTION=''`
more complex template html 2024-11-10 18:03:31 +01:00			`NDD='https://www.cipherbliss.com'`
add scss converter 2024-11-10 00:01:15 +01:00			`EMAIL='contact@cipherbliss.com'`
			`SITE_ICON='https://www.cipherbliss.com/wp-content/uploads/2016/12/rond.png'`
more complex template html 2024-11-10 18:03:31 +01:00			`SITE_ICON_TYPE='image/png'`
try finding article name from content 2024-11-11 00:02:50 +01:00			`NAVIGATION="""`
			`<nav>`
			`<a href="/">Accueil</a>`
			`<a href="https://portfolio.cipherbliss.com">Portfolio</a>`
			`<a href="/feed">Flux RSS</a>`
			`<a href="/contact">Contact</a>`
			`<a href="/ressources-de-café-vie-privée">Ressources</a>`
			`</nav>`
			`"""`
add scss converter 2024-11-10 00:01:15 +01:00			`BANNIERE_ENTETE='https://www.cipherbliss.com/wp-content/uploads/2016/11/bg.jpg'`
			`BANNIERE_ENTETE_ALT='bannière du site'`
			`ARTICLE=''`
			`FOOTER=''`
change enrich template 2024-11-03 11:42:44 +01:00
more complex template html 2024-11-10 18:03:31 +01:00
site builder ok 2024-11-11 00:58:44 +01:00
more complex template html 2024-11-10 18:03:31 +01:00			`def extract_body_content(html_content):`
site builder ok 2024-11-11 00:58:44 +01:00			`pattern = r'<body[^>]?>(.?)</body>'`
more complex template html 2024-11-10 18:03:31 +01:00			`match = re.search(pattern, html_content, re.DOTALL)`
			`if match:`
			`return match.group(1)`
			`else:`
			`return None`

up style and template 2024-11-10 00:16:15 +01:00			`def remove_properties_section(text):`
			`pattern = r"<h1 id=\"article\">Article</h1>.+?</ul>"`
			`replacement = ""`
			`return re.sub(pattern, replacement, text, flags=re.DOTALL)`

			`def remove_article_head_properties_orgmode(text):`
			`pattern = r":PROPERTIES:.+?:END:"`
			`replacement = ""`
			`return re.sub(pattern, replacement, text, flags=re.DOTALL)`
more complex template html 2024-11-10 18:03:31 +01:00
up style and template 2024-11-10 00:16:15 +01:00			`def remove_hint_html(text):`
			`pattern = r"<p>ceci<sub>estduhtml</sub></p>"`
			`replacement = ""`
			`return re.sub(pattern, replacement, text, flags=re.DOTALL)`


up 2024-11-08 18:05:08 +01:00
copy style 2024-11-08 23:19:39 +01:00			`def enrich_one_file(file, root_path):`
add scss converter 2024-11-10 00:01:15 +01:00

site builder ok 2024-11-11 00:58:44 +01:00			`print(' ----------- enrich_html: file:',os.path.join(root_path, file))`
replace org content, make global style for generated site 2024-11-09 00:19:43 +01:00			`css_content = ""`

			`inline_the_css=False`
			`# inline_the_css=True`

site builder ok 2024-11-11 00:58:44 +01:00			`print(' ----------- enrich_html: CSS inline: ',inline_the_css)`
replace org content, make global style for generated site 2024-11-09 00:19:43 +01:00			`# Trouver le fichier entête`
			`header_content=''`
			`with open(os.path.join(root_path, file), "r") as f:`
			`header_content = f.read()`
up 2024-11-08 18:05:08 +01:00			`# Ouvrir le fichier HTML en mode lecture`
copy style 2024-11-08 23:19:39 +01:00			`with open(os.path.join(root_path, file), "r") as f:`
up 2024-11-08 18:05:08 +01:00			`html_content = f.read()`

try finding article name from content 2024-11-11 00:02:50 +01:00			`# remove some parts`
site builder ok 2024-11-11 00:58:44 +01:00			`html_content = remove_properties_section(html_content)`
			`html_content = remove_article_head_properties_orgmode(html_content)`
			`html_content = remove_hint_html(html_content)`

			`html_content = extract_body_content(html_content)`
up style and template 2024-11-10 00:16:15 +01:00
site builder ok 2024-11-11 00:58:44 +01:00			`if inline_the_css is True:`
			`print(' ----------- enrich_html: include css inline in each html page')`
copy style 2024-11-08 23:19:39 +01:00			`with open(os.path.join(root_path, file), "r") as f:`
			`css_content = f.read()`
			`css_content = "<style type='text/css'>{css_content}</style>"`

try finding article name from content 2024-11-11 00:02:50 +01:00			`# remplir le template`
more complex template html 2024-11-10 18:03:31 +01:00			`html_content = f"""`
add scss converter 2024-11-10 00:01:15 +01:00
more complex template html 2024-11-10 18:03:31 +01:00			`<!DOCTYPE html>`
			`<html>`
			`<head>`
			`<meta charset="UTF-8">`
			`<meta name="viewport" content="width=device-width, initial-scale=1">`
			`<meta property="og:image" content="{SITE_ICON}">`
			`<meta property="og:locale" content="{LOCALE}">`
			`<meta property="og:description" content="{BLOG_SUBTITLE}">`
try finding article name from content 2024-11-11 00:02:50 +01:00			`<meta property="og:url" content="{NDD}">`
more complex template html 2024-11-10 18:03:31 +01:00			`<meta property="og:site_name" content="Cipher Bliss">`
site builder ok 2024-11-11 00:58:44 +01:00			`<link rel="alternate" type="application/rss+xml" title="Cipher Bliss » Flux" href="{NDD}/feed/">`
more complex template html 2024-11-10 18:03:31 +01:00			`<link href="/style.css" rel="stylesheet">`
			`<meta charset="utf-8">`
			`<meta name="viewport" content="width=device-width, initial-scale=1">`
			`<title>{TITLE}</title>`
			`<meta name="author" content="{AUTHOR}">`
			`<link rel="alternate" type="application/rss+xml" title="{BLOG_TITLE} » Flux"`
			`href="{NDD}/feed/">`
			`<meta property="og:title" content="{PAGE_TITLE}">`
			`<meta property="og:locale" content="{LOCALE}">`
			`<!-- Description de la page -->`
			`<meta name="description" content="{PAGE_TITLE}">`
			`<meta name="reply-to" content="{EMAIL}">`
			`<link rel="icon" type="{SITE_ICON_TYPE}" href="{SITE_ICON}">`
			`</head>`

			`<body>`
			`<div id="page">`
			`<header id="masthead" class="site-header">`
style sass 2024-11-10 18:43:38 +01:00			`<div class="header-image" style="background: url({BANNIERE_ENTETE})">`
			`<a href="/">`
			`<img src="{SITE_ICON}" class="site-icon img">`
			`</a>`
			`<h1 class="blog-title">{BLOG_TITLE}</h1>`
			`<p class="blog-subtitle">{BLOG_SUBTITLE}</p>`
more complex template html 2024-11-10 18:03:31 +01:00			`</div>`
			`<nav class="navbar is-fixed-top is-dark" role="navigation" aria-label="main navigation">`
			`<div class="navbar-brand">`
			`<a class="navbar-item" href="{NDD}">`
style sass 2024-11-10 18:43:38 +01:00
more complex template html 2024-11-10 18:03:31 +01:00			`</a>`

			`</div>`

			`<div id="navbarBasicExample" class="navbar-menu">`
			`<div class="navbar-start">`
			`<a class="navbar-item" href="{NDD}">`
			`<img src="{SITE_ICON}"`
			`class="img-fluid">`
			`</a>`
			`{NAVIGATION}`
			`</div>`
			`<div class="navbar-end">`
			`<div class="navbar-item">`
			`<form role="search" method="get" class="search-form" action="/">`
			`<label>`
			`<input class="search-field" placeholder="Recherche" value="" name="s"`
			`type="search">`
			`</label>`
			`<input class="is-hidden search-submit" value="Rechercher" type="submit">`
			`</form>`
			`</div>`
			`</div>`
			`</div>`
			`</nav>`
			`</header>`
			`<main class="body-wrap boxed-container">`
			`<article class="content">`
			`{html_content}`
site builder ok 2024-11-11 00:58:44 +01:00			`<p class="after-article">`
			`{after_article}`
			`</p>`
more complex template html 2024-11-10 18:03:31 +01:00			`</article>`
			`</main>`
			`<footer class="site-footer has-top-divider">`
			`<div class="container">`
			`<div class="site-footer-inner">`

style sass 2024-11-10 18:43:38 +01:00
site builder ok 2024-11-11 00:58:44 +01:00
more complex template html 2024-11-10 18:03:31 +01:00			`{NAVIGATION}`

			`</div>`
			`</div>`
			`</footer>`
			`</div>`
replace org content, make global style for generated site 2024-11-09 00:19:43 +01:00			`</body>`
more complex template html 2024-11-10 18:03:31 +01:00			`<!-- généré avec orgmode-to-gemini-blog par Tykayn -->`
			`</html>`

add scss converter 2024-11-10 00:01:15 +01:00			`"""`
up 2024-11-08 18:05:08 +01:00
replace org content, make global style for generated site 2024-11-09 00:19:43 +01:00			`html_path_enriched=os.path.join(root_path, file)`
site builder ok 2024-11-11 00:58:44 +01:00			`print(' ----------- enrich_html: html_path_enriched ============> ',html_path_enriched)`
up 2024-11-08 18:05:08 +01:00			`# Écrire le contenu modifié dans le fichier HTML`
replace org content, make global style for generated site 2024-11-09 00:19:43 +01:00			`with open(html_path_enriched, "w") as f:`
up 2024-11-08 18:05:08 +01:00			`f.write(html_content)`
site builder ok 2024-11-11 00:58:44 +01:00			`print('\n ----------- enrich_html: html écrit ', html_path_enriched)`
replace org content, make global style for generated site 2024-11-09 00:19:43 +01:00
up readme 2024-11-03 10:29:30 +01:00			`# Parcourir tous les fichiers HTML dans le dossier`
replace org content, make global style for generated site 2024-11-09 00:19:43 +01:00			`for root, _, files in os.walk(blog_name):`
			`# print(files)`
up readme 2024-11-03 10:29:30 +01:00			`for file in files:`
			`if file.endswith(".html"):`
copy style 2024-11-08 23:19:39 +01:00			`enrich_one_file(file, root)`