##############################################################################
#
# OpenERP, Open Source Business Applications
-# Copyright (C) 2012 OpenERP S.A. (<http://openerp.com>).
+# Copyright (C) 2012-2013 OpenERP S.A. (<http://openerp.com>).
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU Affero General Public License as
import cgi
import logging
import lxml.html
-import openerp.pooler as pooler
-import operator
+import lxml.html.clean as clean
import random
import re
import socket
import threading
import time
+import openerp
from openerp.loglevels import ustr
_logger = logging.getLogger(__name__)
# HTML Sanitizer
#----------------------------------------------------------
-def html_sanitize(src):
+tags_to_kill = ["script", "head", "meta", "title", "link", "style", "frame", "iframe", "base", "object", "embed"]
+tags_to_remove = ['html', 'body', 'font']
+
+
+def html_sanitize(src, silent=True):
if not src:
return src
src = ustr(src, errors='replace')
- root = lxml.html.fromstring(u"<div>%s</div>" % src)
- result = handle_element(root)
- res = []
- for element in children(result[0]):
- if isinstance(element, basestring):
- res.append(element)
- else:
- element.tail = ""
- res.append(lxml.html.tostring(element))
- return ''.join(res)
-
-# FIXME: shouldn't this be a whitelist rather than a blacklist?!
-to_remove = set(["script", "head", "meta", "title", "link", "img"])
-to_unwrap = set(["html", "body"])
-
-javascript_regex = re.compile(r"^\s*javascript\s*:.*$", re.IGNORECASE)
-
-def handle_a(el, new):
- href = el.get("href", "#")
- if javascript_regex.search(href):
- href = "#"
- new.set("href", href)
-
-special = {
- "a": handle_a,
-}
-
-def handle_element(element):
- if isinstance(element, basestring):
- return [element]
- if element.tag in to_remove:
- return []
- if element.tag in to_unwrap:
- return reduce(operator.add, [handle_element(x) for x in children(element)])
- result = lxml.html.fromstring("<%s />" % element.tag)
- for c in children(element):
- append_to(handle_element(c), result)
- if element.tag in special:
- special[element.tag](element, result)
- return [result]
-
-def children(node):
- res = []
- if node.text is not None:
- res.append(node.text)
- for child_node in node.getchildren():
- res.append(child_node)
- if child_node.tail is not None:
- res.append(child_node.tail)
- return res
-def append_to(elements, dest_node):
- for element in elements:
- if isinstance(element, basestring):
- children = dest_node.getchildren()
- if len(children) == 0:
- dest_node.text = element
- else:
- children[-1].tail = element
- else:
- dest_node.append(element)
+ logger = logging.getLogger(__name__ + '.html_sanitize')
+
+ # html encode email tags
+ part = re.compile(r"(<(([^a<>]|a[^<>\s])[^<>]*)@[^<>]+>)", re.IGNORECASE | re.DOTALL)
+ src = part.sub(lambda m: cgi.escape(m.group(1)), src)
+
+ kwargs = {
+ 'page_structure': True,
+ 'style': False, # do not remove style attributes
+ 'forms': True, # remove form tags
+ }
+ if etree.LXML_VERSION >= (2, 3, 1):
+ # kill_tags attribute has been added in version 2.3.1
+ kwargs.update({
+ 'kill_tags': tags_to_kill,
+ 'remove_tags': tags_to_remove,
+ })
+ else:
+ kwargs['remove_tags'] = tags_to_kill + tags_to_remove
+
+ if etree.LXML_VERSION >= (3, 1, 0):
+ kwargs.update({
+ 'safe_attrs_only': True,
+ 'safe_attrs': clean.defs.safe_attrs | set(['style']),
+ })
+ else:
+ # lxml < 3.1.0 does not allow to specify safe_attrs. We keep all attributes in order to keep "style"
+ kwargs['safe_attrs_only'] = False
+
+ try:
+ # some corner cases make the parser crash (such as <SCRIPT/XSS SRC=\"http://ha.ckers.org/xss.js\"></SCRIPT> in test_mail)
+ cleaner = clean.Cleaner(**kwargs)
+ cleaned = cleaner.clean_html(src)
+ except etree.ParserError:
+ if not silent:
+ raise
+ logger.warning('ParserError obtained when sanitizing %r', src, exc_info=True)
+ cleaned = '<p>ParserError when sanitizing</p>'
+ except Exception:
+ if not silent:
+ raise
+ logger.warning('unknown error obtained when sanitizing %r', src, exc_info=True)
+ cleaned = '<p>Unknown error when sanitizing</p>'
+ return cleaned
#----------------------------------------------------------
# HTML Cleaner
#----------------------------------------------------------
-def html_email_clean(html):
- """ html_email_clean: clean the html to display in the web client.
- - strip email quotes (remove blockquote nodes)
- - strip signatures (remove --\n{\n)Blahblah), by replacing <br> by
- \n to avoid ignoring signatures converted into html
-
- :param string html: sanitized html; tags like html or head should not
- be present in the html string. This method therefore takes as input
- html code coming from a sanitized source, like fields.html.
+def html_email_clean(html, remove=False, shorten=False, max_length=300, expand_options=None):
+ """ html_email_clean: clean the html by doing the following steps:
+
+ - try to strip email quotes, by removing blockquotes or having some client-
+ specific heuristics
+ - try to strip signatures
+ - shorten the html to a maximum number of characters if requested
+
+ Some specific use case:
+
+ - MsOffice: ``div.style = border-top:solid;`` delimitates the beginning of
+ a quote; detecting by finding WordSection1 of MsoNormal
+ - Hotmail: ``hr.stopSpelling`` delimitates the beginning of a quote; detect
+ Hotmail by funding ``SkyDrivePlaceholder``
+
+ :param string html: sanitized html; tags like html or head should not
+ be present in the html string. This method therefore
+ takes as input html code coming from a sanitized source,
+ like fields.html.
+ :param boolean remove: remove the html code that is unwanted; otherwise it
+ is only flagged and tagged
+ :param boolean shorten: shorten the html; every excessing content will
+ be flagged as to remove
+ :param int max_length: if shortening, maximum number of characters before
+ shortening
"""
def _replace_matching_regex(regex, source, replace=''):
+ """ Replace all matching expressions in source by replace """
+ if not source:
+ return source
dest = ''
idx = 0
for item in re.finditer(regex, source):
dest += source[idx:]
return dest
+ def _create_node(tag, text, tail=None, attrs={}):
+ new_node = etree.Element(tag)
+ new_node.text = text
+ new_node.tail = tail
+ for key, val in attrs.iteritems():
+ new_node.set(key, val)
+ return new_node
+
+ def _insert_new_node(node, index, new_node_tag, new_node_text, new_node_tail=None, new_node_attrs={}):
+ new_node = _create_node(new_node_tag, new_node_text, new_node_tail, new_node_attrs)
+ node.insert(index, new_node)
+ return new_node
+
+ def _tag_matching_regex_in_text(regex, node, new_node_tag='span', new_node_attrs={}):
+ text = node.text or ''
+ if not re.search(regex, text):
+ return
+
+ cur_node = node
+ node.text = ''
+ idx, iteration = 0, 0
+ for item in re.finditer(regex, text):
+ if iteration == 0:
+ cur_node.text = text[idx:item.start()]
+ else:
+ _insert_new_node(node, (iteration - 1) * 2 + 1, new_node_tag, text[idx:item.start()])
+ new_node = _insert_new_node(node, iteration * 2, new_node_tag, text[item.start():item.end()], None, new_node_attrs)
+
+ cur_node = new_node
+ idx = item.end()
+ iteration += 1
+ new_node = _insert_new_node(node, -1, new_node_tag, text[idx:] + (cur_node.tail or ''), None, {})
+
+ if expand_options is None:
+ expand_options = {}
+
+ if not html or not isinstance(html, basestring):
+ return html
html = ustr(html)
- # 1. <br[ /]> -> \n, because otherwise the tree is obfuscated
- br_tags = re.compile(r'([<]\s*br\s*\/?[>])')
- html = _replace_matching_regex(br_tags, html, '__BR_TAG__')
- # TDE note: seems to have lots of <div><br></div> in emails... needs to be checks, could be cleaned
+ # Pre processing
+ # ------------------------------------------------------------
+ # TDE TODO: --- MAIL ORIGINAL ---: '[\-]{4,}([^\-]*)[\-]{4,}'
+
+ # html: remove encoding attribute inside tags
+ doctype = re.compile(r'(<[^>]*\s)(encoding=(["\'][^"\']*?["\']|[^\s\n\r>]+)(\s[^>]*|/)?>)', re.IGNORECASE | re.DOTALL)
+ html = doctype.sub(r"", html)
- # 2. form a tree, handle (currently ?) pure-text by enclosing them in a pre
+ # html: ClEditor seems to love using <div><br /><div> -> replace with <br />
+ br_div_tags = re.compile(r'(<div>\s*<br\s*\/>\s*<\/div>)', re.IGNORECASE)
+ html = _replace_matching_regex(br_div_tags, html, '<br />')
+
+ # form a tree
root = lxml.html.fromstring(html)
if not len(root) and root.text is None and root.tail is None:
html = '<div>%s</div>' % html
root = lxml.html.fromstring(html)
- # 2.5 remove quoted text in nodes
- quote_tags = re.compile(r'(\n(>)+[^\n\r]*)')
+ # remove all tails and replace them by a span element, because managing text and tails can be a pain in the ass
for node in root.getiterator():
- if not node.text:
- continue
- node.text = _replace_matching_regex(quote_tags, node.text)
-
- # 3. remove blockquotes
- quotes = [el for el in root.getiterator(tag='blockquote')]
- for node in quotes:
- # copy the node tail into parent text
if node.tail:
- parent = node.getparent()
- parent.text = parent.text or '' + node.tail
- # remove the node
- node.getparent().remove(node)
-
- # 4. strip signatures
- signature = re.compile(r'([-]{2}[\s]?[\r\n]{1,2}[^\z]+)')
- for elem in root.getiterator():
- if elem.text:
- match = re.search(signature, elem.text)
- if match:
- elem.text = elem.text[:match.start()] + elem.text[match.end():]
- if elem.tail:
- match = re.search(signature, elem.tail)
- if match:
- elem.tail = elem.tail[:match.start()] + elem.tail[match.end():]
-
- # 5. \n back to <br/>
- html = etree.tostring(root, pretty_print=True)
- html = html.replace('__BR_TAG__', '<br />')
-
- # 6. Misc cleaning :
- # - ClEditor seems to love using <div><br /><div> -> replace with <br />
- br_div_tags = re.compile(r'(<div>\s*<br\s*\/>\s*<\/div>)')
- html = _replace_matching_regex(br_div_tags, html, '<br />')
+ tail_node = _create_node('span', node.tail)
+ node.tail = None
+ node.addnext(tail_node)
+
+ # form node and tag text-based quotes and signature
+ quote_tags = re.compile(r'(\n(>)+[^\n\r]*)')
+ signature = re.compile(r'([-]{2,}[\s]?[\r\n]{1,2}[^.]+)')
+ for node in root.getiterator():
+ _tag_matching_regex_in_text(quote_tags, node, 'span', {'text_quote': '1'})
+ _tag_matching_regex_in_text(signature, node, 'span', {'text_signature': '1'})
+
+ # Processing
+ # ------------------------------------------------------------
+
+ # tree: tag nodes
+ # signature_begin = False # try dynamic signature recognition
+ quote_begin = False
+ overlength = False
+ cur_char_nbr = 0
+ for node in root.getiterator():
+ # root: try to tag the client used to write the html
+ if 'WordSection1' in node.get('class', '') or 'MsoNormal' in node.get('class', ''):
+ root.set('msoffice', '1')
+ if 'SkyDrivePlaceholder' in node.get('class', '') or 'SkyDrivePlaceholder' in node.get('id', ''):
+ root.set('hotmail', '1')
+
+ # state of the parsing
+ if quote_begin:
+ node.set('in_quote', '1')
+ node.set('tail_remove', '1')
+ if overlength:
+ node.set('in_overlength', '1')
+ node.set('tail_remove', '1')
+
+ if root.get('msoffice') and node.tag == 'div' and 'border-top:solid' in node.get('style', ''):
+ quote_begin = True
+ node.set('in_quote', '1')
+ node.set('tail_remove', '1')
+ if root.get('hotmail') and node.tag == 'hr' and ('stopSpelling' in node.get('class', '') or 'stopSpelling' in node.get('id', '')):
+ quote_begin = True
+ node.set('in_quote', '1')
+ node.set('tail_remove', '1')
+
+ # shorten:
+ # 1/ truncate the text at the next available space
+ # 2/ create a 'read more' node, next to current node
+ # 3/ add the truncated text in a new node, next to 'read more' node
+ if shorten and not overlength and cur_char_nbr + len(node.text or '') > max_length:
+ overlength = True
+ # truncate text
+ innertext = node.text[0:(max_length - cur_char_nbr)]
+ outertext = node.text[(max_length - cur_char_nbr):]
+ stop_idx = outertext.find(' ')
+ if stop_idx == -1:
+ stop_idx = len(outertext)
+ node.text = innertext + outertext[0:stop_idx]
+ # create <span> ... <a href="#">read more</a></span> node
+ read_more_node = _create_node(
+ 'span',
+ ' ... ',
+ None,
+ {'class': expand_options.get('oe_expand_span_class', 'oe_mail_expand')}
+ )
+ read_more_link_node = _create_node(
+ 'a',
+ 'read more',
+ None,
+ {
+ 'href': expand_options.get('oe_expand_href', '#'),
+ 'class': expand_options.get('oe_expand_a_class', 'oe_mail_expand'),
+ }
+ )
+ read_more_node.append(read_more_link_node)
+ # create outertext node
+ new_node = _create_node('span', outertext[stop_idx:])
+ # add newly created nodes in dom
+ node.append(read_more_node)
+ # tag node
+ new_node.set('in_overlength', '1')
+
+ cur_char_nbr += len(node.text or '')
+
+ if node.tag == 'blockquote' or node.get('text_quote') or node.get('text_signature'):
+ node.set('in_quote', '1')
+
+ # Post processing
+ # ------------------------------------------------------------
+
+ to_remove = []
+ for node in root.getiterator():
+ if node.get('in_quote') or node.get('in_overlength'):
+ # copy the node tail into parent text
+ if node.tail and not node.get('tail_remove'):
+ parent = node.getparent()
+ parent.tail = node.tail + (parent.tail or '')
+ to_remove.append(node)
+ if node.get('tail_remove'):
+ node.tail = ''
+ for node in to_remove:
+ if remove:
+ node.getparent().remove(node)
+ else:
+ if not 'oe_mail_expand' in node.get('class', ''): # trick: read more link should be displayed even if it's in overlength
+ node_class = node.get('class', '') + ' ' + 'oe_mail_cleaned'
+ node.set('class', node_class)
+
+ # html: \n that were tail of elements have been encapsulated into <span> -> back to \n
+ html = etree.tostring(root, pretty_print=False)
+ linebreaks = re.compile(r'<span>([\s]*[\r\n]+[\s]*)<\/span>', re.IGNORECASE | re.DOTALL)
+ html = _replace_matching_regex(linebreaks, html, '\n')
return html
url_index.append(url)
html = ustr(etree.tostring(tree, encoding=encoding))
+ # \r char is converted into , must remove it
+ html = html.replace(' ', '')
html = html.replace('<strong>', '*').replace('</strong>', '*')
html = html.replace('<b>', '*').replace('</b>', '*')
# Emails
#----------------------------------------------------------
-email_re = re.compile(r"""
- ([a-zA-Z][\w\.-]*[a-zA-Z0-9] # username part
- @ # mandatory @ sign
- [a-zA-Z0-9][\w\.-]* # domain must start with a letter ... Ged> why do we include a 0-9 then?
- \.
- [a-z]{2,3} # TLD
- )
- """, re.VERBOSE)
+# matches any email in a body of text
+email_re = re.compile(r"""([a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,6})""", re.VERBOSE)
+
+# matches a string containing only one email
+single_email_re = re.compile(r"""^[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,6}$""", re.VERBOSE)
+
res_re = re.compile(r"\[([0-9]+)\]", re.UNICODE)
command_re = re.compile("^Set-([a-z]+) *: *(.+)$", re.I + re.UNICODE)
"""
# If not cr, get cr from current thread database
+ local_cr = None
if not cr:
db_name = getattr(threading.currentThread(), 'dbname', None)
if db_name:
- cr = pooler.get_db_only(db_name).cursor()
+ local_cr = cr = openerp.registry(db_name).db.cursor()
else:
raise Exception("No database cursor found, please pass one explicitly")
# Send Email
try:
- mail_server_pool = pooler.get_pool(cr.dbname).get('ir.mail_server')
+ mail_server_pool = openerp.registry(cr.dbname)['ir.mail_server']
res = False
# Pack Message into MIME Object
email_msg = mail_server_pool.build_email(email_from, email_to, subject, body, email_cc, email_bcc, reply_to,
_logger.exception("tools.email_send failed to deliver email")
return False
finally:
- cr.close()
+ if local_cr:
+ cr.close()
return res
def email_split(text):
projects / odoo / odoo.git / blobdiff