python - Javascript variable with html code regex email matching

Question

This python script is not working to output the email address [email protected] for this case.

This was my previous post.

How can I use BeautifulSoup or Slimit on a site to output the email address from a javascript variable

#!/usr/bin/env python

from bs4 import BeautifulSoup
import re

soup = '''
<script LANGUAGE="JavaScript">
function something()
{
var ptr;
ptr = "";
ptr += "<table><td class=france></td></table>";
ptr += "<table><td class=france><a href=mail";
ptr += "to:[email protected]>email</a></td></table>";
document.all.something.innerHTML = ptr;
}
</script>
'''


soup = BeautifulSoup(soup)

for script in soup.find_all('script'):
  reg = '(<)?(w+@w+(?:.w+)+)(?(1)>)'
  reg2 = 'mailto:.*'
  secondHalf= re.search(reg, script.text)
  firstHalf= re.search(reg2, script.text)
  secondHalfEmail = secondHalf.group()
  firstHalfEmail = firstHalf.group()
  firstHalfEmail = firstHalfEmail.replace('mailto:', '')
  firstHalfEmail = firstHalfEmail.replace('";', '')
  if firstHalfEmail == secondHalfEmail:
     email = secondHalfEmail
  else:
     if ('>') not in firstHalfEmail:
        if ('>') not in secondHalfEmail:
            if firstHalfEmail != secondHalfEmail:
                email = firstHalfEmail + secondHalfEmail
        else:
            email = firstHalfEmail
    else:
        email = secondHalfEmail

    print email

It would be nice if someone can help me.

Thank you

Answer 1

Here is a rather interesting (I think) approach.

Instead of parsing this javascript code - execute it!

Get the ptr value, load it via BeautifulSoup and get the href attribute value from the a tag. Example using V8 engine:

from bs4 import BeautifulSoup
from pyv8 import PyV8

data = """
<script LANGUAGE="JavaScript">
function something()
{
var ptr;
ptr = "";
ptr += "<table><td class=france></td></table>";
ptr += "<table><td class=france><a href=mail";
ptr += "to:[email protected]>email</a></td></table>";
document.all.something.innerHTML = ptr;
}
</script>
"""

soup = BeautifulSoup(data)

# prepare the function to return a value and add a function call
js_code = soup.script.text.strip().replace('document.all.something.innerHTML = ptr;', 'return ptr;') + "; something()"

ctxt = PyV8.JSContext()
ctxt.enter()

soup = BeautifulSoup(ctxt.eval(str(js_code)))
print soup.a['href'].split('mailto:')[1]

Prints:

[email protected]