Generate static website

gen.py

#!/usr/bin/env python3
import os
import re


def truncate_html_words(s, num, delimiter):
	length = int(num)
	if length <=0:
		return ''
	html4_singlets = ('br', 'col', 'link', 'base', 'img', 'param', 'area', 'hr', 'input')
	re_words = re.compile(r'&.*?;|<.*?>|([A-Za-z0-9][\w-]*)')
	re_tag = re.compile(r'<(/)?([^ ]+?)(?: (/)| .*?)?>')
	pos = 0
	ellipsis_pos = 0
	words = 0
	open_tags = []
	while words <= length:
		m = re_words.search(s, pos)
		if not m:
			break
		pos = m.end(0)
		if m.group(1):
			words +=1
			if words==length:
				ellipsis_pos = pos
			continue
		tag=re_tag.match(m.group(0))
		if not tag or ellipsis_pos:
			continue
		closing_tag, tagname, self_closing = tag.groups()
		tagname = tagname.lower() 
		if self_closing or tagname in html4_singlets:
			pass
		elif closing_tag:
			try:
				i = open_tags.index(tagname)
			except ValueError:
				pass
			else:
				open_tags = open_tags[i+1:]
		else:
			open_tags.insert(0,tagname)
	if words<=length:
		return s
	out=s[:ellipsis_pos]+delimiter
	for tag in open_tags:
		out+= '</%s>' % tag
	return out

def folder_to_name(string, language=None):
	string = string.replace('-',' ').title()
	special={
		'pt-br':r'[^^]\b(Com|Que|D[eao]|Para|[\w]|N[oa]|[UE]m)\b',
		'en': r'[^^]\b(In)\b',
	}
	pattern = re.compile(special[language if language else 'pt-br'],re.IGNORECASE)
	return pattern.sub(lambda obj: obj.group(0).lower(),string )

class Folder():
	special_files = ['index.html','conteudo', 'log']
	languages = ['pt-br','en']
	def __init__(self,path):
		self.split = path.split('/')
		self.path = j(cwd,path)
		self.url = j('/',path)
		self.all_contents = [i for i in os.listdir(self.path)]
		self.special = {i:os.path.exists(j(self.path,i)) for i in self.special_files}
		self.contents =[i for i in self.all_contents if i not in self.special_files]
		self.files=[i for i in self.all_contents if os.path.isfile(j(self.path,i))]
		self.dirs= [i for i in self.all_contents if os.path.isdir(j(self.path,i))]
		self.root = self.split[0]
		self.leave = self.split[-1]
		self.language=self.root if self.root in self.languages else 'en'
		self.name = folder_to_name(self.leave,self.language)
		self.content = os.path.exists(j(self.path,'conteudo'))
		self.updated = self.get_log().readline() if self.special['log'] else '000000000'
		self.siblings=[]
		try:
			if self.split[-2]:
				parent = self.split[0:-1]
				for i in os.listdir(j(*[cwd]+parent)) :
					brother=j(*parent+[i])
					if brother!= path and os.path.isdir(brother) and 'conteudo' in os.listdir(brother):
						self.siblings.append(j(*['/']+parent+[i]))
		except IndexError:
			pass
		if self.siblings:
			print(self.siblings)
			print(brother,path)

	def folder_list(self, contents=None):
		if not contents:
			contents=self.dirs
		text='<ul>'
		for ii in sorted(contents,key=lambda m: (open(j(self.path,m,'log')).readline() if os.path.exists(j(self.path,m,'log')) else '000000'),reverse=True) :
			text+='<li>{}</li>'.format(folder_to_name(ii))
		text+='</ul>'
		return text
		
	def get_conteudo(self):
		return open(j(self.path,'conteudo')).read() if self.special['conteudo'] else None

	def get_log(self):
		if 'log' in self.all_contents:
			return open(j(self.path,'log'),'r')

class Menu():
	def __init__(self,root,name=None):
		self.name = name
		self.items = []
		self.main_list = []
		self.generate_menu(root)
		if self.name=='main':
			self.main_list.sort()
			result = '\n'.join(self.main_list)
			main_list = open(j(cwd, 'main.list'),'w')
			main_list.write(result)
			main_list.close()

	def generate_menu(self, folder, ignore_links=True):
		contents = os.listdir(j(cwd,folder))
		for i in contents:
			uri = j(cwd,folder,i)
			if ignore_links and os.path.islink(uri):
				continue
			if os.path.isdir(uri):
				if self.name:
					self.generate_menu(j(folder,i))
				else:
					self.items.append(Link(j(folder,i), None, None,None))
			elif i == 'menu' and self.name:
				menudescription = open(j(cwd,folder, i),'r').read()
				if re.search(self.name+'\s*$',menudescription, re.MULTILINE):
					self.items.append(Link(folder,
						re.search('label:\s*(\w+)\s*$',menudescription, re.MULTILINE),
						re.search('title:\s*(\w+)\s*$',menudescription, re.MULTILINE),
						re.search('target:\s*(\w+)\s*$',menudescription, re.MULTILINE),
						))	
			elif self.name == 'main' and i == 'log':
				date = open(j(cwd,folder,i)).readline()
				self.main_list.append(str(date)[0:-1]+' '+folder)

	def __str__(self):
		result = "\n<ul>\n"
		for i in self.items:
			result+="\t<li>"+str(i)+"</li>\n"
		result+= "</ul>\n"
		return result
		
		
class Link():
	def __init__(self, folder, label, title=None, target=None):
		if not label:
			self.label = folder.split('/')[-1].replace('-',' ')
		else:
			self.label= label
		self.href= '/'+folder
		self.title = title
		self.target=target

	def __str__(self):
		resultado = '<a href="'+self.href+'" '
		if self.title:
			resultado += 'title="'+self.title+'" '
		if self.target:
			resultado += 'target="'+self.target+'" '
		resultado += '>'+self.label+'</a>'
		return resultado

	
def generate_index(folder,toc):
	folder_contents = os.listdir(j(cwd,folder))
	link_list = ''
	index = open(j(cwd,folder,'index.html'),'w') 
	pagetitle = folder_to_name(folder.split('/')[-1])
	if 'conteudo' in folder_contents:
		conteudo = open(j(cwd,folder,'conteudo'),'r').read()
		content=html_filter(conteudo) if conteudo else ""
		template = main_template
		index.write(template.format(
			main_menu = str(main_menu),
			pagetitle= pagetitle,
			title=pagetitle,
			toc = toc_template.format(str(toc)) if toc else "",
			content=content,
			link_list=link_list
			))
	else:
		template=front_template
		contents = []
		for subfolder in sorted([Folder(j(folder,a)) for a in folder_contents if os.path.isdir(j(cwd,folder,a))], key=lambda m: m.updated, reverse=True):
			if subfolder.special['conteudo']:
				text = truncate_html_words(subfolder.get_conteudo(), 150, ' <a class="read-more" href="'+subfolder.url+'">Leia mais</a>')
			else:
				text = subfolder.folder_list()
			contents.append(subfolder.updated+"$$$$"+post_item_template.format(
				title='<a href="'+subfolder.url+'">'+subfolder.name+'</a>',
				contents=text
			))
			contents = sorted(contents, reverse=True)
		content = '\n'.join([i.split('$$$$')[1] for i in contents])
		index.write(template.format(
			main_menu = str(main_menu),
			pagetitle= pagetitle,
			title=pagetitle,
			toc = toc_template.format(str(toc)) if toc else "",
			content=content,
			link_list=link_list
			))
	index.close()

def html_filter(content):
	replacements = {
		r'\n\s*\n\s*([\w\W]+?)(?=\n\s*\n)':r'\n\n<P>\1</P>',
		r'(</h[1-9]>)\n([\w\W]+)(?=\n\s*\n)':r'\1\n\n<P>\2</P>',
		r'^([\w\W]+[?!"\w\.])(?=\n\s*\n)':r'<P>\1</P>',
		r'<blockcode\s+language\s*=\s*"(\w+)">([\w\W]+?)</blockcode>':r'<pre class="\1"><code>\2</code></pre>',
		r'<code>([\w\W\s]+?)(?=</code>)':replace_code,
	}
	for k,v in replacements.items():
		content = re.sub(k, v, content, re.MULTILINE)
	return content

def replace_code(matchobj):
	return '<code>'+matchobj.group(1).replace('<','&lt;').replace('>','&gt;')

def varrer_diretorio(folder, ignore_links=True):
	contents = os.listdir(j(cwd,folder))
	toc = None
	if folder not in ("sites",):
		for i in contents:
			uri = j(cwd,folder,i)
			if ignore_links and os.path.islink(uri):
				continue
			if os.path.isdir(uri):
				if not toc:
					toc = Menu(folder,None)
				varrer_diretorio(j(folder,i))
		generate_index(folder, toc)

def create_front_page():
	posts_per_page = 15
	contents = os.listdir(cwd)
	list_title = [
		'Posts Recentes', 'Posts anteriores'
	]
	link_list = '<a href="/">⌂</a> '
	for i in range(1,(len(main_menu.main_list)//posts_per_page) ):
		link_list+='<a href="/older'+str(i)+'.html">'+str(i)+'</a>'
		
	if 'main.list' in contents:
		counting = 0
		while len(main_menu.main_list) > 0:
			posts = []
			for i in range(counting, counting+posts_per_page):
				try:
					posts.append(main_menu.main_list.pop())
				except IndexError:
					break
			to_open = 'older'+str(counting) if counting else 'index'
			index = open(j(cwd,to_open+'.html'),'w')
			if not counting:
				index = open(j(cwd,'index.html'),'w')
			else:
				index = open(j(cwd, 'older'+str(counting)+'.html'),'w')
			contents = []
			for i in posts:
				link = i.split(' ')[1]
				name = i.split('/')[-1] if '/' in i else i.split(' ')[-1]
				name = name.replace('-',' ')
				contents.append(post_item_template.format(
					title='<a href="'+link+'">'+name+'</a>',
					contents=truncate_html_words(open(j(cwd,i.split(' ')[1],'conteudo'), 'r').read(), 150, ' <a class="read-more" href="'+link+'">Leia mais</a>')
				))
			index.write(front_template.format(
				main_menu = str(main_menu),
				title = list_title[0] if not counting else list_title[1],
				content = html_filter('\n'.join(contents)),
				link_list = link_list
			))
			index.close()
			counting +=1

j = os.path.join
cwd = os.getcwd()
main_template = open(j(cwd, 'template.html'),'r').read()
toc_template = open(j(cwd, 'toc.html'),'r').read()
post_item_template = open(j(cwd,'post.item.html'),'r').read()
front_template = open(j(cwd,'front.html'),'r').read()
main_menu = Menu('','main')
varrer_diretorio('')
create_front_page()