def ScrubHTMLGMail(self, body): # # Scrub preamble postamble (i.e. On xxx so-and-so wrote:) # scrub_re = '<div>\s*<span class="gmail_quote">.*?</span>\s*' + re.escape(self.cfg_snip_html) + '\s*</div>' scrub_rex = re.compile(scrub_re, re.I|re.MULTILINE|re.DOTALL) body = scrub_rex.sub('', body) # # GMail doesn't use proper blockquote # count = 0 while body.find("""style="margin-left: 40px;">""") != -1 and count < 5: count += 1 body = bm_text.snip_between( body, start_re = """<(p|div) style="margin-left: 40px;">""", end_re = "</(p|div)>", replace_inner = None, replace_start = "<blockquote><p>", replace_end = "</p></blockquote>", ) return body
def ScrubHTMLOutlookExpress(self, body): body = bm_text.snip_between_tags(body, "div", replace_inner = None, replace_start = "<p>", replace_end = "</p>") body = bm_text.snip_between(body, start_re = re.escape('<font face="Arial" size="2">'), end_re = re.escape('</font>'), replace_inner = None, replace_start = "", replace_end = "") return body