emailparser/render

#!/usr/bin/python

import os, argparse, hashlib, sqlite3, time, shutil, cgi, re, math
from datetime import datetime

parser = argparse.ArgumentParser(description='Renders static HTML pages and indexes from an SQLite database of emails and an attachment folder.')

parser.add_argument('-o', dest='output_dir', action='store', default='rendered_files',
                   help='path of the directory where rendered files should be stored')
                   
parser.add_argument('-t', dest='template_dir', action='store', default='templates',
                   help='path where the template files are')
                   
parser.add_argument('-i', dest='title', action='store', default='Inbox',
                   help='title for the rendered pages')
                   
parser.add_argument('-d', dest='database', action='store', default='emails.db',
                   help='path of the database that should be used to render the e-mail files')
                   
parser.add_argument('-a', dest='attachment_dir', action='store', default='attachments',
                   help='path where attachments are stored')
                   
parser.add_argument('-n', dest='per_page', action='store', default='20',
                   help='amount of entries per page')

args = parser.parse_args()
options = vars(args)

def format_size(num):
	for unit in [' bytes','KB','MB','GB']:
		if num < 1024.0:
			return "%3.1f%s" % (num, unit)
		num /= 1024.0
	return "%3.1f%s" % (num, 'TB')
	
def chunk(iterable, chunksize, fillvalue=None):
	result = []
	num_chunks = int(math.ceil(len(iterable) / (chunksize * 1.0)))
	
	for i in xrange(0, num_chunks):
		start = (chunksize * i)
		
		if (chunksize * (i + 1)) > len(iterable):
			end = len(iterable)
		else:
			end = (chunksize * (i + 1))
		
		result.append((iterable[start:end]))
	
	return result

def render_index(email_list, title, identifier):
	email_list = chunk(email_list, int(options['per_page']))
	
	current_page = 0
	
	for list_chunk in email_list:
		variables = {
			'page': title, 
			'pagenum': "Page %d of %d" % (current_page + 1, len(email_list)),
			'title': options['title'],
			'index': "../index.html",
			'items': "".join('<tr><td>%s...</td><td>%s</td><td>%s</td><td>%s</td><td>%s</td><td>%s</td></tr>' % (message[0][:8], message[3], message[4], message[5], message[6], message[7]) for message in list_chunk)
		}
		
		filename = "%s/index_%s_%d.html" % (options['output_dir'], identifier, current_page)
		
		generated = template_index % variables
		open(filename, 'w').write(generated.encode('UTF-8'))
		print "Wrote page %d for %s." % (current_page, identifier)
			
		current_page += 1
		

if os.path.isfile(options['database']) == False:
	print "Database file not found. Use the -d switch to specify a custom database path."
	exit(1)

# Connect to database
database = sqlite3.connect(options['database'])
cursor = database.cursor()

# Load templates
template_message = open('%s/message.html' % options['template_dir']).read()
template_index = open('%s/index.html' % options['template_dir']).read()

# Create output directory if necessary
try:
	os.makedirs("%s/messages" % options['output_dir'])
except OSError:
	pass

email_list = []

for message_id, sender, recipient, subject, unixtime, textbody, htmlbody, sha1_hash in cursor.execute("SELECT * FROM emails"):
	
	sender = cgi.escape(sender, True)
	recipient = cgi.escape(recipient, True)
	subject = cgi.escape(subject, True)
	message_id = cgi.escape(message_id, True)
	timestamp = datetime.fromtimestamp(unixtime).strftime("%A %B %e, %Y %H:%M:%S")
	
	attachment_list = []
	attachment_cursor = database.cursor()
	for attachment_hash, attachment_filename, attachment_type, message_hash, attachment_size in attachment_cursor.execute("SELECT * FROM attachments WHERE `message_hash` = ?", (sha1_hash,)):
		attachment_extension = os.path.splitext(attachment_filename)[1][1:]
		attachment_file = "%s.%s" % (attachment_hash, attachment_extension)
		attachment_list.append('<a href="../attachments/%s" class="attachment">%s (%s, %s)</a>' % (attachment_file, attachment_filename, attachment_type, format_size(attachment_size)))
	
	if len(attachment_list) > 0:
		attachments = "".join(attachment_list)
		print "ATTACHMENTS %s" % sha1_hash
	else:
		attachments = '<div class="attachment light">No attachments.</div>'
	
	versions = {}
	
	if textbody != "":
		available_text = True
		versions['Plaintext'] = '%s_text.html' % sha1_hash
	else:
		available_text = False
		
	if htmlbody != "":
		available_html = True
		versions['HTML'] = '%s_html.html' % sha1_hash
	else:
		available_html = False
	
	version_list = "".join('<a href="%s" class="version">%s</a>' % (value, key) for key, value in versions.viewitems())
	
	if available_text == True:
		# Text version
		variables = {
			'subject': subject, 
			'date': timestamp, 
			'from': sender, 
			'to': recipient, 
			'body': "<pre>%s</pre>" % textbody, 
			'title': options['title'], 
			'version': "Plaintext version", 
			'index': "../index.html", 
			'versions': version_list,
			'attachments': attachments
		}

		generated = template_message % variables
		open('%s/messages/%s_text.html' % (options['output_dir'], sha1_hash), 'w').write(generated.encode('UTF-8'))
		print "Successfully generated plaintext version of %s." % sha1_hash
	
	if available_html == True:
		# HTML version
		variables = {
			'subject': subject, 
			'date': timestamp, 
			'from': sender, 
			'to': recipient, 
			'body': htmlbody, 
			'title': options['title'], 
			'version': "HTML version", 
			'index': "../index.html", 
			'versions': version_list,
			'attachments': attachments
		}
		
		generated = template_message % variables
		open('%s/messages/%s_html.html' % (options['output_dir'], sha1_hash), 'w').write(generated.encode('UTF-8'))
		print "Successfully generated HTML version of %s." % sha1_hash
	
	try:
		snippet = "%s..." % re.search("^(.{0,200})\\b", textbody, re.DOTALL).group(1)
	except AttributeError:
		snippet = ""
	
	email_list.append((sha1_hash, available_text, available_html, sender, recipient, subject, timestamp, len(attachment_list), snippet, unixtime))

# Sort by timestamp, ascending
sorted_list = sorted(email_list, key=lambda email: email[9])
render_index(sorted_list, "Sorted from old to new", "date_asc")

# Sort by timestamp, descending
sorted_list.reverse()
render_index(sorted_list, "Sorted from new to old", "date_desc")

# Sort by sender, ascending
sorted_list = sorted(email_list, key=lambda email: email[3].lower())

# Sort by sender, descending
sorted_list.reverse()

# Sort by recipient, ascending
sorted_list = sorted(email_list, key=lambda email: email[4].lower())

# Sort by recipient, descending
sorted_list.reverse()

# Sort by hash, ascending
sorted_list = sorted(email_list, key=lambda email: email[0].lower())

# Sort by hash, descending
sorted_list.reverse()

# Sort by subject, ascending
sorted_list = sorted(email_list, key=lambda email: email[5].lower())

# Sort by subject, descending
sorted_list.reverse()

shutil.copy('%s/style.css' % options['template_dir'], '%s/style.css' % options['output_dir'])