urlparse for humans

urlpath.py

#!/usr/bin/env python
"""
Public domain work by anatoly techtonik <techtonik@gmail.com>

History:

  1.0 - parse URL string and split into attributes
  2.0 - modify properties to reconstruct URL with str()

"""

class urlpath(object):

    __version__ = '2.0'

    protocol = ''   # https
    server = ''     # www.example.com:80
    host = ''       # www.example.com
    port = ''       # 80
    fullpath = ''   # some/app?and&its#params
    path = ''       # some/app
    query = ''      # and&its#params

    def __init__(self, url=''):
        if url:
            self.parse(url)

    def _expand_server(self):
        '''set host, port'''
        host = port = ''
        if ':' in self.server:
            host, port = self.server.split(':', 1)
        else:
            host = self.server
        # avoiding recursion from __setattr__()
        super(urlpath, self).__setattr__('host', host)
        super(urlpath, self).__setattr__('port', port)

    def _expand_fullpath(self):
        '''set path, query'''
        path = query = ''
        if '?' in self.fullpath:
            path, query = self.fullpath.split('?', 1)
        else:
            path = self.fullpath
        # setting attributes directly to avoid
        # recursion from __setattr__()
        super(urlpath, self).__setattr__('path', path)
        super(urlpath, self).__setattr__('query', query)
            
    def parse(self, url):
        # protocol
        if '://' in url:
            self.protocol, url2 = url.split('://', 1)
        else:
            url2 = url
        # server
        if '/' not in url2:
            self.server = url2
            #self._expand_server()
            #^ called by __setattr__
            return
        self.server, url3 = url2.split('/', 1)
        #self._expand_server()
        #^ called by __setattr__
        # fullpath
        self.fullpath = url3.lstrip('/')
        #self._expand_fullpath()
        #^ called by __setattr__

    def __str__(self):
        out = ''
        if self.protocol:
            out += self.protocol + '://'
        out += self.server + "/" + self.fullpath
        return out

    def __setattr__(self, name, value):
        super(urlpath, self).__setattr__(name, value)
        if name == 'server':
            self._expand_server()
        if name == 'fullpath':
            self._expand_fullpath()
        if name in ('host', 'port'):
            # update server value
            server = self.host
            if self.port:
                server += ':' + self.port
            super(urlpath, self).__setattr__('server', server)
        if name in ('path', 'query'):
            fullpath = self.path
            if self.query:
                fullpath += '?' + self.query
            super(urlpath, self).__setattr__('fullpath', fullpath)


if __name__ == '__main__':
    # self-test
    u = urlpath("http://localhost:8917/demo/path?with&all=this#stuff5")
    print(u)
    print("protocol: " + u.protocol)
    print("server: " + u.server)
    print("  host: " + u.host)
    print("  port: " + u.port)
    print("fullpath: " + u.fullpath)
    print("    path: " + u.path)
    print("   query: " + u.query)

    u.server = 'example.com'
    print("\nchange server to example.com")
    print("server: " + u.server)
    print("  host: " + u.host)
    print("  port: " + u.port)

    u.port = '8080'
    print("\nchange port to 8080")
    print("server: " + u.server)
    print("  host: " + u.host)
    print("  port: " + u.port)

    u.fullpath = 'production?param='
    print("\nchange fullpath to production?param=")
    print("fullpath: " + u.fullpath)
    print("    path: " + u.path)
    print("   query: " + u.query)

    u.path = ''
    print("\nchange path to ''")
    print("fullpath: " + u.fullpath)
    print("    path: " + u.path)
    print("   query: " + u.query)
   
    print(u)