Das deutsche Python-Forum

Code: Alles auswählen

import urllib2
from xml.dom.minidom import parseString

ATTR_CLASS = 'class'
ATTR_HREF = 'href'
IGNORE_PATHS = ['.', '..']
ACLASS_NAME = 'n'


MSG_NOT_EXIST = "target directory not found"
MSG_INVALID_DIR = "target url is not representing a directory"

def listdir(url):
    try:
        u = urllib2.urlopen(url)
    except urllib2.URLError, err:
        raise OSError(MSG_NOT_EXIST)
    if not u.headers.dict['content-type'].lower() == 'text/html':
        raise OSError(MSG_INVALID_DIR)
    try:
        hp = parseString(u.read())
    except:
        raise OSError(MSG_INVALID_DIR)
    try:
        table = hp.getElementsByTagName("table")[0]
        _thead = table.getElementsByTagName("thead")[0]
        _tbody = table.getElementsByTagName("tbody")[0]
    except IndexError:
        raise OSError(MSG_INVALID_DIR)
    result = []
    for row in _tbody.getElementsByTagName("tr"):
        for col in row.getElementsByTagName("td"):
            if col.hasAttribute(ATTR_CLASS) and \
               col.getAttribute(ATTR_CLASS) == ACLASS_NAME and \
               col.firstChild.hasAttribute(ATTR_HREF):
                value = col.firstChild.getAttribute(ATTR_HREF).replace('/', '')
                if value not in IGNORE_PATHS:
                    result.append(value)
                    break
    return result

Code: Alles auswählen

>>> import htmlos
>>> htmlos.listdir("http://repo.exist.com/dist/maestro/")
[u'1.5', u'1.5.1', u'1.5.2', u'1.5.3', u'code', u'README']

>>> htmlos.listdir("http://python-forum.de")
Traceback (most recent call last):
  File "<interactive input>", line 1, in <module>
  File "D:\htmlos.py", line 23, in listdir
    raise OSError(MSG_INVALID_DIR)
OSError: target url is not representing a directory