Is it possible to hook up a more robust HTML parser to Python mechanize?

问题

I am trying to parse and submit a form on a website using mechanize, but it appears that the built-in form parser cannot detect the form and its elements. I suspect that it is choking on poorly formed HTML, and I'd like to try pre-parsing it with a parser better designed to handle bad HTML (say lxml or BeautifulSoup) and then feeding the prettified, cleaned-up output to the form parser. I need mechanize not only for submitting the form but also for maintaining sessions (I'm working this form from within a login session.)

I'm not sure how to go about doing this, if it is indeed possible.. I'm not that familiar with the various details of the HTTP protocol, how to get various parts to work together etc. Any pointers?

回答1:

reading from the big example on the first page of the mechanize website:

# Sometimes it's useful to process bad headers or bad HTML:
response = br.response()  # this is a copy of response
headers = response.info()  # currently, this is a mimetools.Message
headers["Content-type"] = "text/html; charset=utf-8"
response.set_data(response.get_data().replace("<!---", "<!--"))
br.set_response(response)

so it seems very possible to preprocess the response with another parser which will regenerate well-formed HTML, then feed it back to mechanize for further processing.

回答2:

I had a problem where a form field was missing from a form, I couldn't find any malformed html but I figured that was the cause so I used BeautifulSoup's prettify function to parse it and it worked.

resp = br.open(url)
soup = BeautifulSoup(resp.get_data())
resp.set_data(soup.prettify())
br.set_response(resp)

I'd love to know how to this automatically.

Edit: found out how to do this automatically

class PrettifyHandler(mechanize.BaseHandler):
    def http_response(self, request, response):
        if not hasattr(response, "seek"):
            response = mechanize.response_seek_wrapper(response)
        # only use BeautifulSoup if response is html
        if response.info().dict.has_key('content-type') and ('html' in response.info().dict['content-type']):
            soup = BeautifulSoup(response.get_data())
            response.set_data(soup.prettify())
        return response

    # also parse https in the same way
    https_response = http_response

br = mechanize.Browser()
br.add_handler(PrettifyHandler())

br will now use BeautifulSoup to parse all responses where html is contained in the content type (mime type), eg text/html

回答3:

What you're looking for can be done with lxml.etree which is the xml.etree.ElementTree emulator (and replacement) provided by lxml:

First we take bad mal-formed HTML:

% cat bad.html
<html>
<HEAD>
    <TITLE>this HTML is awful</title>
</head>
<body>
    <h1>THIS IS H1</H1>
    <A HREF=MYLINK.HTML>This is a link and it is awful</a>
    <img src=yay.gif>
</body>
</html>

(Observe the mixed case between opening and closing tags, missing quotation marks).

And then parse it:

>>> from lxml import etree
>>> bad = file('bad.html').read()
>>> html = etree.HTML(bad)
>>> print etree.tostring(html)
<html><head><title>this HTML is awful</title></head><body>
    <h1>THIS IS H1</h1>
    <a href="MYLINK.HTML">This is a link and it is awful</a>
    <img src="yay.gif"/></body></html>

Observe that the tagging and quotation has been corrected for us.

If you were having problems parsing the HTML before, this might be the answer you're looking for. As for the details of HTTP, that is another matter entirely.

来源：https://stackoverflow.com/questions/1782368/is-it-possible-to-hook-up-a-more-robust-html-parser-to-python-mechanize

标签

python

mechanize