#!/usr/bin/env python
# -*- coding: utf-8 -*-

"""
| This file is part of the web2py Web Framework
| Copyrighted by Massimo Di Pierro <mdipierro@cs.depaul.edu>
| License: LGPLv3 (http://www.gnu.org/licenses/lgpl.html)

Template helpers
--------------------------------------------
"""

import cgi
import os
import re
import copy
import types
import urllib
import base64
import itertools
from pydal._compat import PY2, reduce, pickle, copyreg, HTMLParser, name2codepoint, iteritems, unichr, unicodeT, \
    urllib_quote, to_bytes, to_native, to_unicode, basestring, urlencode, implements_bool, text_type, long
from yatl import sanitizer
import marshal

from gluon import decoder
from gluon.storage import Storage
from gluon.utils import web2py_uuid, compare
from gluon.highlight import highlight
from gluon.validators import simple_hash


def local_html_escape(data, quote=False):
    """
    Works with bytes.
    Replace special characters "&", "<" and ">" to HTML-safe sequences.
    If the optional flag quote is true (the default), the quotation mark
    characters, both double quote (") and single quote (') characters are also
    translated.
    """
    if PY2:
        import cgi
        data = cgi.escape(data, quote)
        return data.replace("'", "&#x27;") if quote else data
    else:
        import html
        if isinstance(data, str):
            return html.escape(data, quote=quote)
        data = data.replace(b"&", b"&amp;")  # Must be done first!
        data = data.replace(b"<", b"&lt;")
        data = data.replace(b">", b"&gt;")
        if quote:
            data = data.replace(b'"', b"&quot;")
            data = data.replace(b'\'', b"&#x27;")
        return data

regex_crlf = re.compile('\r|\n')

join = ''.join

# name2codepoint is incomplete respect to xhtml (and xml): 'apos' is missing.
entitydefs = dict([(k_v[0], to_bytes(unichr(k_v[1]))) for k_v in iteritems(name2codepoint)])
entitydefs.setdefault('apos', to_bytes("'"))


__all__ = [
    'A',
    'ASSIGNJS',
    'B',
    'BEAUTIFY',
    'BODY',
    'BR',
    'BUTTON',
    'CENTER',
    'CAT',
    'CODE',
    'COL',
    'COLGROUP',
    'DIV',
    'EM',
    'EMBED',
    'FIELDSET',
    'FORM',
    'H1',
    'H2',
    'H3',
    'H4',
    'H5',
    'H6',
    'HEAD',
    'HR',
    'HTML',
    'I',
    'IFRAME',
    'IMG',
    'INPUT',
    'LABEL',
    'LEGEND',
    'LI',
    'LINK',
    'OL',
    'UL',
    'MARKMIN',
    'MENU',
    'META',
    'OBJECT',
    'ON',
    'OPTION',
    'P',
    'PRE',
    'SCRIPT',
    'OPTGROUP',
    'SELECT',
    'SPAN',
    'STRONG',
    'STYLE',
    'TABLE',
    'TAG',
    'TD',
    'TEXTAREA',
    'TH',
    'THEAD',
    'TBODY',
    'TFOOT',
    'TITLE',
    'TR',
    'TT',
    'URL',
    'XHTML',
    'XML',
    'xmlescape',
    'embed64',
]

DEFAULT_PASSWORD_DISPLAY = '*' * 8


def xmlescape(data, quote=True):
    """
    Returns an escaped string of the provided data

    Args:
        data: the data to be escaped
        quote: optional (default False)
    """

    # first try the xml function
    if hasattr(data, 'xml') and callable(data.xml):
        return to_bytes(data.xml())

    if not(isinstance(data, (text_type, bytes))):
        # i.e., integers
        data = str(data)
    data = to_bytes(data, 'utf8', 'xmlcharrefreplace')

    # ... and do the escaping
    data = local_html_escape(data, quote)
    return data


def call_as_list(f, *a, **b):
    if not isinstance(f, (list, tuple)):
        f = [f]
    for item in f:
        item(*a, **b)


def truncate_string(text, length, dots='...'):
    text = to_unicode(text)
    if len(text) > length:
        text = to_native(text[:length - len(dots)]) + dots
    return text


def URL(a=None,
        c=None,
        f=None,
        r=None,
        args=None,
        vars=None,
        anchor='',
        extension=None,
        env=None,
        hmac_key=None,
        hash_vars=True,
        salt=None,
        user_signature=None,
        scheme=None,
        host=None,
        port=None,
        encode_embedded_slash=False,
        url_encode=True,
        language=None,
        hash_extension=True
        ):
    """
    generates a url '/a/c/f' corresponding to application a, controller c
    and function f. If r=request is passed, a, c, f are set, respectively,
    to r.application, r.controller, r.function.

    The more typical usage is:

        URL('index')

    that generates a url for the index function
    within the present application and controller.

    Args:
        a: application (default to current if r is given)
        c: controller (default to current if r is given)
        f: function (default to current if r is given)
        r: request (optional)
        args: any arguments (optional). Additional "path" elements
        vars: any variables (optional). Querystring elements
        anchor: anchorname, without # (optional)
        extension: force an extension
        hmac_key: key to use when generating hmac signature (optional)
        hash_vars: which of the vars to include in our hmac signature
            True (default) - hash all vars, False - hash none of the vars,
            iterable - hash only the included vars ['key1','key2']
        salt: salt hashing with this string
        user_signature: signs automatically the URL in such way that only the
            user can access the URL (use with `URL.verify` or
            `auth.requires_signature()`)
        scheme: URI scheme (True, 'http' or 'https', etc); forces absolute URL (optional)
        host: string to force absolute URL with host (True means http_host)
        port: optional port number (forces absolute URL)
        encode_embedded_slash: encode slash characters included in args
        url_encode: encode characters included in vars

    Raises:
        SyntaxError: when no application, controller or function is available
            or when a CRLF is found in the generated url

    Examples:

    >>> str(URL(a='a', c='c', f='f', args=['x', 'y', 'z'],
    ...     vars={'p':1, 'q':2}, anchor='1'))
    '/a/c/f/x/y/z?p=1&q=2#1'

    >>> str(URL(a='a', c='c', f='f', args=['x', 'y', 'z'],
    ...     vars={'p':(1,3), 'q':2}, anchor='1'))
    '/a/c/f/x/y/z?p=1&p=3&q=2#1'

    >>> str(URL(a='a', c='c', f='f', args=['x', 'y', 'z'],
    ...     vars={'p':(3,1), 'q':2}, anchor='1'))
    '/a/c/f/x/y/z?p=3&p=1&q=2#1'

    >>> str(URL(a='a', c='c', f='f', anchor='1+2'))
    '/a/c/f#1%2B2'

    >>> str(URL(a='a', c='c', f='f', args=['x', 'y', 'z'],
    ...     vars={'p':(1,3), 'q':2}, anchor='1', hmac_key='key'))
    '/a/c/f/x/y/z?p=1&p=3&q=2&_signature=a32530f0d0caa80964bb92aad2bedf8a4486a31f#1'

    >>> str(URL(a='a', c='c', f='f', args=['w/x', 'y/z']))
    '/a/c/f/w/x/y/z'

    >>> str(URL(a='a', c='c', f='f', args=['w/x', 'y/z'], encode_embedded_slash=True))
    '/a/c/f/w%2Fx/y%2Fz'

    >>> str(URL(a='a', c='c', f='f', args=['%(id)d'], url_encode=False))
    '/a/c/f/%(id)d'

    >>> str(URL(a='a', c='c', f='f', args=['%(id)d'], url_encode=True))
    '/a/c/f/%25%28id%29d'

    >>> str(URL(a='a', c='c', f='f', vars={'id' : '%(id)d' }, url_encode=False))
    '/a/c/f?id=%(id)d'

    >>> str(URL(a='a', c='c', f='f', vars={'id' : '%(id)d' }, url_encode=True))
    '/a/c/f?id=%25%28id%29d'

    >>> str(URL(a='a', c='c', f='f', anchor='%(id)d', url_encode=False))
    '/a/c/f#%(id)d'

    >>> str(URL(a='a', c='c', f='f', anchor='%(id)d', url_encode=True))
    '/a/c/f#%25%28id%29d'
    """

    from gluon.rewrite import url_out  # done here in case used not-in web2py

    if args in (None, []):
        args = []
    vars = vars or {}
    application = None
    controller = None
    function = None

    if not isinstance(args, (list, tuple)):
        args = [args]

    if not r:
        if a and not c and not f:
            (f, a, c) = (a, c, f)
        elif a and c and not f:
            (c, f, a) = (a, c, f)
        from gluon.globals import current
        if hasattr(current, 'request'):
            r = current.request

    if r:
        application = r.application
        controller = r.controller
        function = r.function
        env = r.env
        if extension is None and r.extension != 'html':
            extension = r.extension
    if a:
        application = a
    if c:
        controller = c
    if f:
        if not isinstance(f, str):
            if hasattr(f, '__name__'):
                function = f.__name__
            else:
                raise SyntaxError(
                    'when calling URL, function or function name required')
        elif '/' in f:
            if f.startswith("/"):
                f = f[1:]
            items = f.split('/')
            function = f = items[0]
            args = items[1:] + args
        else:
            function = f

        # if the url gets a static resource, don't force extension
        if controller == 'static':
            extension = None
            # add static version to url
            from gluon.globals import current
            if hasattr(current, 'response'):
                response = current.response
                if response.static_version and response.static_version_urls:
                    args = [function] + args
                    function = '_' + str(response.static_version)

        if '.' in function:
            function, extension = function.rsplit('.', 1)

    # only include the extension as part of the variables for the hash if requested
    function2 = '%s.%s' % (function, extension or 'html') if hash_extension else function

    if not (application and controller and function):
        raise SyntaxError('not enough information to build the url (%s %s %s)' % (application, controller, function))

    if args:
        if url_encode:
            if encode_embedded_slash:
                other = '/' + '/'.join([urllib_quote(str(x), '') for x in args])
            else:
                other = args and urllib_quote('/' + '/'.join([str(x) for x in args]))
        else:
            other = args and ('/' + '/'.join([str(x) for x in args]))
    else:
        other = ''

    if other.endswith('/'):
        other += '/'  # add trailing slash to make last trailing empty arg explicit

    list_vars = []
    for (key, vals) in sorted(vars.items()):
        if key == '_signature':
            continue
        if not isinstance(vals, (list, tuple)):
            vals = [vals]
        for val in vals:
            list_vars.append((key, val))

    if user_signature:
        from gluon.globals import current
        if current.session.auth:
            hmac_key = current.session.auth.hmac_key

    if hmac_key:
        # generate an hmac signature of the vars & args so can later
        # verify the user hasn't messed with anything

        h_args = '/%s/%s/%s%s' % (application, controller, function2, other)

        # how many of the vars should we include in our hash?
        if hash_vars is True:  # include them all
            h_vars = list_vars
        elif hash_vars is False:  # include none of them
            h_vars = ''
        else:  # include just those specified
            if hash_vars and not isinstance(hash_vars, (list, tuple)):
                hash_vars = [hash_vars]
            h_vars = [(k, v) for (k, v) in list_vars if k in hash_vars]

        # re-assembling the same way during hash authentication
        message = h_args + '?' + urlencode(sorted(h_vars))
        sig = simple_hash(
            message, hmac_key or '', salt or '', digest_alg='sha1')
        # add the signature into vars
        list_vars.append(('_signature', sig))

    if list_vars:
        if url_encode:
            other += '?%s' % urlencode(list_vars)
        else:
            other += '?%s' % '&'.join(['%s=%s' % var[:2] for var in list_vars])
    if anchor:
        if url_encode:
            other += '#' + urllib_quote(str(anchor))
        else:
            other += '#' + (str(anchor))
    if extension:
        function += '.' + extension

    if regex_crlf.search(join([application, controller, function, other])):
        raise SyntaxError('CRLF Injection Detected')

    url = url_out(r, env, application, controller, function,
                  args, other, scheme, host, port, language=language)
    return url


def verifyURL(request, hmac_key=None, hash_vars=True, salt=None, user_signature=None, hash_extension=True):
    """
    Verifies that a request's args & vars have not been tampered with by the user

    :param request: web2py's request object
    :param hmac_key: the key to authenticate with, must be the same one previously
                    used when calling URL()
    :param hash_vars: which vars to include in our hashing. (Optional)
                    Only uses the 1st value currently
                    True (or undefined) means all, False none,
                    an iterable just the specified keys

    do not call directly. Use instead:

    URL.verify(hmac_key='...')

    the key has to match the one used to generate the URL.

        >>> r = Storage()
        >>> gv = Storage(p=(1,3),q=2,_signature='a32530f0d0caa80964bb92aad2bedf8a4486a31f')
        >>> r.update(dict(application='a', controller='c', function='f', extension='html'))
        >>> r['args'] = ['x', 'y', 'z']
        >>> r['get_vars'] = gv
        >>> verifyURL(r, 'key')
        True
        >>> verifyURL(r, 'kay')
        False
        >>> r.get_vars.p = (3, 1)
        >>> verifyURL(r, 'key')
        True
        >>> r.get_vars.p = (3, 2)
        >>> verifyURL(r, 'key')
        False

    """

    if '_signature' not in request.get_vars:
        return False  # no signature in the request URL

    # check if user_signature requires
    if user_signature:
        from gluon.globals import current
        if not current.session or not current.session.auth:
            return False
        hmac_key = current.session.auth.hmac_key
    if not hmac_key:
        return False

    # get our sig from request.get_vars for later comparison
    original_sig = request.get_vars._signature

    # now generate a new hmac for the remaining args & vars
    vars, args = request.get_vars, request.args

    # remove the signature var since it was not part of our signed message
    request.get_vars.pop('_signature')

    # join all the args & vars into one long string

    # always include all of the args
    other = args and urllib_quote('/' + '/'.join([str(x) for x in args])) or ''
    
    # decide whether the extension should be part of the hash verification
    h_extension = request.extension if hash_extension else ''

    # only add a period to the extension when it exists otherwise empty
    # extensions will fail to validate
    if h_extension:
        h_extension = '.%s' % (h_extension)
        
    h_args = '/%s/%s/%s%s%s' % (request.application,
                                 request.controller,
                                 request.function,
                                 h_extension,
                                 other)

    # but only include those vars specified (allows more flexibility for use with
    # forms or ajax)

    list_vars = []
    for (key, vals) in sorted(vars.items()):
        if not isinstance(vals, (list, tuple)):
            vals = [vals]
        for val in vals:
            list_vars.append((key, val))

    # which of the vars are to be included?
    if hash_vars is True:       # include them all
        h_vars = list_vars
    elif hash_vars is False:    # include none of them
        h_vars = ''
    else:                       # include just those specified
        # wrap in a try - if the desired vars have been removed it'll fail
        try:
            if hash_vars and not isinstance(hash_vars, (list, tuple)):
                hash_vars = [hash_vars]
            h_vars = [(k, v) for (k, v) in list_vars if k in hash_vars]
        except:
            # user has removed one of our vars! Immediate fail
            return False
    # build the full message string with both args & vars
    message = h_args + '?' + urlencode(sorted(h_vars))

    # hash with the hmac_key provided
    sig = simple_hash(message, str(hmac_key), salt or '', digest_alg='sha1')

    # put _signature back in get_vars just in case a second call to URL.verify is performed
    # (otherwise it'll immediately return false)
    request.get_vars['_signature'] = original_sig

    # return whether or not the signature in the request matched the one we just generated
    # (I.E. was the message the same as the one we originally signed)

    return compare(original_sig, sig)

URL.verify = verifyURL

ON = True


class XmlComponent(object):
    """
    Abstract root for all Html components
    """

    # TODO: move some DIV methods to here

    def xml(self):
        raise NotImplementedError

    def __mul__(self, n):
        return CAT(*[self for i in range(n)])

    def __add__(self, other):
        if isinstance(self, CAT):
            components = self.components
        else:
            components = [self]
        if isinstance(other, CAT):
            components += other.components
        else:
            components += [other]
        return CAT(*components)

    def add_class(self, name):
        """
        add a class to _class attribute
        """
        c = self['_class']
        classes = (set(c.split()) if c else set()) | set(name.split())
        self['_class'] = ' '.join(classes) if classes else None
        return self

    def remove_class(self, name):
        """
        remove a class from _class attribute
        """
        c = self['_class']
        classes = (set(c.split()) if c else set()) - set(name.split())
        self['_class'] = ' '.join(classes) if classes else None
        return self


class XML(XmlComponent):
    """
    use it to wrap a string that contains XML/HTML so that it will not be
    escaped by the template

    Examples:

    >>> XML('<h1>Hello</h1>').xml()
    '<h1>Hello</h1>'
    """

    def __init__(
        self,
        text,
        sanitize=False,
        permitted_tags=[
            'a',
            'b',
            'blockquote',
            'br/',
            'i',
            'li',
            'ol',
            'ul',
            'p',
            'cite',
            'code',
            'pre',
            'img/',
            'h1', 'h2', 'h3', 'h4', 'h5', 'h6',
            'table', 'tr', 'td', 'div',
            'strong', 'span',
        ],
        allowed_attributes={
            'a': ['href', 'title', 'target'],
            'img': ['src', 'alt'],
            'blockquote': ['type'],
            'td': ['colspan'],
        },
    ):
        """
        Args:
            text: the XML text
            sanitize: sanitize text using the permitted tags and allowed
                attributes (default False)
            permitted_tags: list of permitted tags (default: simple list of
                tags)
            allowed_attributes: dictionary of allowed attributed (default
                for A, IMG and BlockQuote).
                The key is the tag; the value is a list of allowed attributes.
        """
        if isinstance(text, unicodeT):
            text = to_native(text.encode('utf8', 'xmlcharrefreplace'))
        if sanitize:
            text = sanitizer.sanitize(text, permitted_tags, allowed_attributes)
        elif isinstance(text, bytes):
            text = to_native(text)
        elif not isinstance(text, str):
            text = str(text)
        self.text = text

    def xml(self):
        return to_bytes(self.text)

    def __str__(self):
        return self.text

    __repr__ = __str__

    def __add__(self, other):
        return '%s%s' % (self, other)

    def __radd__(self, other):
        return '%s%s' % (other, self)

    def __cmp__(self, other):
        return cmp(str(self), str(other))

    def __hash__(self):
        return hash(str(self))

#    why was this here? Break unpickling in sessions
#    def __getattr__(self, name):
#        return getattr(str(self), name)

    def __getitem__(self, i):
        return str(self)[i]

    def __getslice__(self, i, j):
        return str(self)[i:j]

    def __iter__(self):
        for c in str(self):
            yield c

    def __len__(self):
        return len(str(self))

    def flatten(self, render=None):
        """
        returns the text stored by the XML object rendered
        by the `render` function
        """
        if render:
            return render(self.text, None, {})
        return self.text

    def elements(self, *args, **kargs):
        """
        to be considered experimental since the behavior of this method
        is questionable
        another option could be `TAG(self.text).elements(*args, **kwargs)`
        """
        return []

# ## important to allow safe session.flash=T(....)


def XML_unpickle(data):
    return XML(marshal.loads(data))


def XML_pickle(data):
    return XML_unpickle, (marshal.dumps(str(data)),)
copyreg.pickle(XML, XML_pickle, XML_unpickle)


@implements_bool
class DIV(XmlComponent):
    """
    HTML helper, for easy generating and manipulating a DOM structure.
    Little or no validation is done.

    Behaves like a dictionary regarding updating of attributes.
    Behaves like a list regarding inserting/appending components.

    Examples:

    >>> DIV('hello', 'world', _style='color:red;').xml()
    '<div style=\"color:red;\">helloworld</div>'

    All other HTML helpers are derived from `DIV`.

    `_something="value"` attributes are transparently translated into
    `something="value"` HTML attributes
    """

    # name of the tag, subclasses should update this
    # tags ending with a '/' denote classes that cannot
    # contain components
    tag = 'div'

    def __init__(self, *components, **attributes):
        """
        Args:
            components: any components that should be nested in this element
            attributes: any attributes you want to give to this element

        Raises:
            SyntaxError: when a stand alone tag receives components
        """

        if self.tag[-1:] == '/' and components:
            raise SyntaxError('<%s> tags cannot have components'
                              % self.tag)
        if len(components) == 1 and isinstance(components[0], (list, tuple)):
            self.components = list(components[0])
        else:
            self.components = list(components)
        self.attributes = attributes
        self._fixup()
        # converts special attributes in components attributes
        self.parent = None
        for c in self.components:
            self._setnode(c)
        self._postprocessing()

    def update(self, **kargs):
        """
        dictionary like updating of the tag attributes
        """

        for (key, value) in iteritems(kargs):
            self[key] = value
        return self

    def append(self, value):
        """
        list style appending of components

        Examples:

        >>> a=DIV()
        >>> a.append(SPAN('x'))
        >>> print(a)
        <div><span>x</span></div>
        """
        self._setnode(value)
        ret = self.components.append(value)
        self._fixup()
        return ret

    def insert(self, i, value):
        """
        List-style inserting of components

        Examples:

        >>> a=DIV()
        >>> a.insert(0, SPAN('x'))
        >>> print(a)
        <div><span>x</span></div>
        """
        self._setnode(value)
        ret = self.components.insert(i, value)
        self._fixup()
        return ret

    def __getitem__(self, i):
        """
        Gets attribute with name 'i' or component #i.
        If attribute 'i' is not found returns None

        Args:
            i: index. If i is a string: the name of the attribute
                otherwise references to number of the component
        """

        if isinstance(i, str):
            try:
                return self.attributes[i]
            except KeyError:
                return None
        else:
            return self.components[i]

    def get(self, i):
        return self.attributes.get(i)

    def __setitem__(self, i, value):
        """
        Sets attribute with name 'i' or component #i.

        Args:
            i: index. If i is a string: the name of the attribute
                otherwise references to number of the component
            value: the new value
        """
        self._setnode(value)
        if isinstance(i, (str, unicodeT)):
            self.attributes[i] = value
        else:
            self.components[i] = value

    def __delitem__(self, i):
        """
        Deletes attribute with name 'i' or component #i.

        Args:
            i: index. If i is a string: the name of the attribute
                otherwise references to number of the component
        """

        if isinstance(i, str):
            del self.attributes[i]
        else:
            del self.components[i]

    def __len__(self):
        """
        Returns the number of included components
        """
        return len(self.components)

    def __bool__(self):
        """
        Always returns True
        """
        return True

    def _fixup(self):
        """
        Handling of provided components.

        Nothing to fixup yet. May be overridden by subclasses,
        eg for wrapping some components in another component or blocking them.
        """
        return

    def _wrap_components(self, allowed_parents,
                         wrap_parent=None,
                         wrap_lambda=None):
        """
        helper for _fixup. Checks if a component is in allowed_parents,
        otherwise wraps it in wrap_parent

        Args:
            allowed_parents: (tuple) classes that the component should be an
                instance of
            wrap_parent: the class to wrap the component in, if needed
            wrap_lambda: lambda to use for wrapping, if needed

        """
        components = []
        for c in self.components:
            if isinstance(c, (allowed_parents, CAT)):
                pass
            elif wrap_lambda:
                c = wrap_lambda(c)
            else:
                c = wrap_parent(c)
            if isinstance(c, DIV):
                c.parent = self
            components.append(c)
        self.components = components

    def _postprocessing(self):
        """
        Handling of attributes (normally the ones not prefixed with '_').

        Nothing to postprocess yet. May be overridden by subclasses
        """
        return

    def _traverse(self, status, hideerror=False):
        # TODO: docstring
        newstatus = status
        for c in self.components:
            if hasattr(c, '_traverse') and callable(c._traverse):
                c.vars = self.vars
                c.request_vars = self.request_vars
                c.errors = self.errors
                c.latest = self.latest
                c.session = self.session
                c.formname = self.formname
                if not c.attributes.get('hideerror'):
                    c['hideerror'] = hideerror or self.attributes.get('hideerror')
                newstatus = c._traverse(status, hideerror) and newstatus

        # for input, textarea, select, option
        # deal with 'value' and 'validation'

        name = self['_name']
        if newstatus:
            newstatus = self._validate()
            self._postprocessing()
        elif 'old_value' in self.attributes:
            self['value'] = self['old_value']
            self._postprocessing()
        elif name and name in self.vars:
            self['value'] = self.vars[name]
            self._postprocessing()
        if name:
            self.latest[name] = self['value']
        return newstatus

    def _validate(self):
        """
        nothing to validate yet. May be overridden by subclasses
        """
        return True

    def _setnode(self, value):
        if isinstance(value, DIV):
            value.parent = self

    def _xml(self):
        """
        Helper for xml generation. Returns separately:
        - the component attributes
        - the generated xml of the inner components

        Component attributes start with an underscore ('_') and
        do not have a False or None value. The underscore is removed.
        A value of True is replaced with the attribute name.

        Returns:
            tuple: (attributes, components)
        """

        # get the attributes for this component
        # (they start with '_', others may have special meanings)
        attr = []
        for key, value in iteritems(self.attributes):
            if key[:1] != '_':
                continue
            name = key[1:]
            if value is True:
                value = name
            elif value is False or value is None:
                continue
            attr.append((name, value))
        data = self.attributes.get('data', {})
        for key, value in iteritems(data):
            name = 'data-' + key
            value = data[key]
            attr.append((name, value))
        attr.sort()
        fa = b''
        for name, value in attr:
            fa += (b' %s="%s"') % (to_bytes(name), xmlescape(value, True))

        # get the xml for the inner components
        co = b''.join([xmlescape(component) for component in self.components])
        return (fa, co)

    def xml(self):
        """
        generates the xml for this component.
        """

        (fa, co) = self._xml()

        if not self.tag:
            return co

        tagname = to_bytes(self.tag)
        if tagname[-1:] == b'/':
            # <tag [attributes] />
            return b'<%s%s />' % (tagname[:-1], fa)

        # else: <tag [attributes]>  inner components xml </tag>
        xml_tag = b'<%s%s>%s</%s>' % (tagname, fa, co, tagname)
        return xml_tag

    def __str__(self):
        """
        str(COMPONENT) returns COMPONENT.xml()
        """
        # In PY3 __str__ cannot return bytes (TypeError: __str__ returned non-string (type bytes))
        return to_native(self.xml())

    def flatten(self, render=None):
        """
        Returns the text stored by the DIV object rendered by the render function
        the render function must take text, tagname, and attributes
        `render=None` is equivalent to `render=lambda text, tag, attr: text`

        Examples:

        >>> markdown = lambda text, tag=None, attributes={}: \
                        {None: re.sub('\s+',' ',text), \
                         'h1':'#'+text+'\\n\\n', \
                         'p':text+'\\n'}.get(tag,text)
        >>> a=TAG('<h1>Header</h1><p>this is a     test</p>')
        >>> a.flatten(markdown)
        '#Header\\n\\nthis is a test\\n'
        """

        text = ''
        for c in self.components:
            if isinstance(c, XmlComponent):
                s = c.flatten(render)
            elif render:
                s = render(to_native(c))
            else:
                s = to_native(c)
            text += s
        if render:
            text = render(text, self.tag, self.attributes)
        return text

    regex_tag = re.compile('^[\w\-\:]+')
    regex_id = re.compile('#([\w\-]+)')
    regex_class = re.compile('\.([\w\-]+)')
    regex_attr = re.compile('\[([\w\-\:]+)=(.*?)\]')

    def elements(self, *args, **kargs):
        """
        Find all components that match the supplied attribute dictionary,
        or None if nothing could be found

        All components of the components are searched.

        Examples:

        >>> a = DIV(DIV(SPAN('x'),3,DIV(SPAN('y'))))
        >>> for c in a.elements('span', first_only=True): c[0]='z'
        >>> print(a)
        <div><div><span>z</span>3<div><span>y</span></div></div></div>
        >>> for c in a.elements('span'): c[0]='z'
        >>> print(a)
        <div><div><span>z</span>3<div><span>z</span></div></div></div>

        It also supports a syntax compatible with jQuery

        Examples:

        >>> a=TAG('<div><span><a id="1-1" u:v=$>hello</a></span><p class="this is a test">world</p></div>')
        >>> for e in a.elements('div a#1-1, p.is'): print(e.flatten())
        hello
        world
        >>> for e in a.elements('#1-1'): print(e.flatten())
        hello
        >>> a.elements('a[u:v=$]')[0].xml()
        '<a id="1-1" u:v="$">hello</a>'
        >>> a=FORM( INPUT(_type='text'), SELECT(list(range(1))), TEXTAREA() )
        >>> for c in a.elements('input, select, textarea'): c['_disabled'] = 'disabled'
        >>> a.xml()
        '<form action="#" enctype="multipart/form-data" method="post"><input disabled="disabled" type="text" /><select disabled="disabled"><option value="0">0</option></select><textarea cols="40" disabled="disabled" rows="10"></textarea></form>'

        Elements that are matched can also be replaced or removed by specifying
        a "replace" argument (note, a list of the original matching elements
        is still returned as usual).

        Examples:

        >>> a = DIV(DIV(SPAN('x', _class='abc'), DIV(SPAN('y', _class='abc'), SPAN('z', _class='abc'))))
        >>> b = a.elements('span.abc', replace=P('x', _class='xyz'))
        >>> print(a)  # We should .xml() here instead of print
        <div><div><p class="xyz">x</p><div><p class="xyz">x</p><p class="xyz">x</p></div></div></div>

        "replace" can be a callable, which will be passed the original element and
        should return a new element to replace it.

        Examples:

        >>> a = DIV(DIV(SPAN('x', _class='abc'), DIV(SPAN('y', _class='abc'), SPAN('z', _class='abc'))))
        >>> b = a.elements('span.abc', replace=lambda el: P(el[0], _class='xyz'))
        >>> print(a)
        <div><div><p class="xyz">x</p><div><p class="xyz">y</p><p class="xyz">z</p></div></div></div>

        If replace=None, matching elements will be removed completely.

        Examples:

        >>> a = DIV(DIV(SPAN('x', _class='abc'), DIV(SPAN('y', _class='abc'), SPAN('z', _class='abc'))))
        >>> b = a.elements('span', find='y', replace=None)
        >>> print(a)
        <div><div><span class="abc">x</span><div><span class="abc">z</span></div></div></div>

        If a "find_text" argument is specified, elements will be searched for text
        components that match find_text, and any matching text components will be
        replaced (find_text is ignored if "replace" is not also specified).
        Like the "find" argument, "find_text" can be a string or a compiled regex.

        Examples:

        >>> a = DIV(DIV(SPAN('x', _class='abc'), DIV(SPAN('y', _class='abc'), SPAN('z', _class='abc'))))
        >>> b = a.elements(find_text=re.compile('x|y|z'), replace='hello')
        >>> print(a)
        <div><div><span class="abc">hello</span><div><span class="abc">hello</span><span class="abc">hello</span></div></div></div>

        If other attributes are specified along with find_text, then only components
        that match the specified attributes will be searched for find_text.

        Examples:

        >>> a = DIV(DIV(SPAN('x', _class='abc'), DIV(SPAN('y', _class='efg'), SPAN('z', _class='abc'))))
        >>> b = a.elements('span.efg', find_text=re.compile('x|y|z'), replace='hello')
        >>> print(a)
        <div><div><span class="abc">x</span><div><span class="efg">hello</span><span class="abc">z</span></div></div></div>
        """
        if len(args) == 1:
            args = [a.strip() for a in args[0].split(',')]
        if len(args) > 1:
            subset = [self.elements(a, **kargs) for a in args]
            return reduce(lambda a, b: a + b, subset, [])
        elif len(args) == 1:
            items = args[0].split()
            if len(items) > 1:
                subset = [a.elements(' '.join(
                    items[1:]), **kargs) for a in self.elements(items[0])]
                return reduce(lambda a, b: a + b, subset, [])
            else:
                item = items[0]
                if '#' in item or '.' in item or '[' in item:
                    match_tag = self.regex_tag.search(item)
                    match_id = self.regex_id.search(item)
                    match_class = self.regex_class.search(item)
                    match_attr = self.regex_attr.finditer(item)
                    args = []
                    if match_tag:
                        args = [match_tag.group()]
                    if match_id:
                        kargs['_id'] = match_id.group(1)
                    if match_class:
                        kargs['_class'] = re.compile('(?<!\w)%s(?!\w)' %
                                                     match_class.group(1).replace('-', '\\-').replace(':', '\\:'))
                    for item in match_attr:
                        kargs['_' + item.group(1)] = item.group(2)
                    return self.elements(*args, **kargs)
        # make a copy of the components
        matches = []
        # check if the component has an attribute with the same
        # value as provided
        tag = to_native(getattr(self, 'tag')).replace('/', '')
        check = not (args and tag not in args)
        for (key, value) in iteritems(kargs):
            if key not in ['first_only', 'replace', 'find_text']:
                if isinstance(value, (str, int)):
                    if str(self[key]) != str(value):
                        check = False
                elif key in self.attributes:
                    if not value.search(str(self[key])):
                        check = False
                else:
                    check = False
        if 'find' in kargs:
            find = kargs['find']
            is_regex = not isinstance(find, (str, int))
            for c in self.components:
                if (isinstance(c, str) and ((is_regex and find.search(c)) or
                   (str(find) in c))):
                    check = True
        # if found, return the component
        if check:
            matches.append(self)

        first_only = kargs.get('first_only', False)
        replace = kargs.get('replace', False)
        find_text = replace is not False and kargs.get('find_text', False)
        is_regex = not isinstance(find_text, (str, int, bool))
        find_components = not (check and first_only)

        def replace_component(i):
            if replace is None:
                del self[i]
                return i
            else:
                self[i] = replace(self[i]) if callable(replace) else replace
                return i + 1
        # loop the components
        if find_text or find_components:
            i = 0
            while i < len(self.components):
                c = self[i]
                j = i + 1
                if check and find_text and isinstance(c, str) and \
                        ((is_regex and find_text.search(c)) or (str(find_text) in c)):
                    j = replace_component(i)
                elif find_components and isinstance(c, XmlComponent):
                    child_matches = c.elements(*args, **kargs)
                    if len(child_matches):
                        if not find_text and replace is not False and child_matches[0] is c:
                            j = replace_component(i)
                        if first_only:
                            return child_matches
                        matches.extend(child_matches)
                i = j
        return matches

    def element(self, *args, **kargs):
        """
        Finds the first component that matches the supplied attribute dictionary,
        or None if nothing could be found

        Also the components of the components are searched.
        """
        kargs['first_only'] = True
        elements = self.elements(*args, **kargs)
        if not elements:
            # we found nothing
            return None
        return elements[0]

    def siblings(self, *args, **kargs):
        """
        Finds all sibling components that match the supplied argument list
        and attribute dictionary, or None if nothing could be found
        """
        sibs = [s for s in self.parent.components if not s == self]
        matches = []
        first_only = False
        if 'first_only' in kargs:
            first_only = kargs.pop('first_only')
        for c in sibs:
            try:
                check = True
                tag = getattr(c, 'tag').replace("/", "")
                if args and tag not in args:
                        check = False
                for (key, value) in iteritems(kargs):
                    if c[key] != value:
                            check = False
                if check:
                    matches.append(c)
                    if first_only:
                        break
            except:
                pass
        return matches

    def sibling(self, *args, **kargs):
        """
        Finds the first sibling component that match the supplied argument list
        and attribute dictionary, or None if nothing could be found
        """
        kargs['first_only'] = True
        sibs = self.siblings(*args, **kargs)
        if not sibs:
            return None
        return sibs[0]


class CAT(DIV):

    tag = ''


def TAG_unpickler(data):
    return pickle.loads(data)


def TAG_pickler(data):
    d = DIV()
    d.__dict__ = data.__dict__
    marshal_dump = pickle.dumps(d, pickle.HIGHEST_PROTOCOL)
    return (TAG_unpickler, (marshal_dump,))


class __tag_div__(DIV):
    def __init__(self, name, *a, **b):
        DIV.__init__(self, *a, **b)
        self.tag = name

copyreg.pickle(__tag_div__, TAG_pickler, TAG_unpickler)


class __TAG__(XmlComponent):

    """
    TAG factory

    Examples:

    >>> print(TAG.first(TAG.second('test'), _key = 3))
    <first key=\"3\"><second>test</second></first>

    """

    def __getitem__(self, name):
        return self.__getattr__(name)

    def __getattr__(self, name):
        if name[-1:] == '_':
            name = name[:-1] + '/'
        return lambda *a, **b: __tag_div__(name, *a, **b)

    def __call__(self, html):
        return web2pyHTMLParser(decoder.decoder(html)).tree

TAG = __TAG__()


class HTML(DIV):
    """
    There are four predefined document type definitions.
    They can be specified in the 'doctype' parameter:

    - 'strict' enables strict doctype
    - 'transitional' enables transitional doctype (default)
    - 'frameset' enables frameset doctype
    - 'html5' enables HTML 5 doctype
    - any other string will be treated as user's own doctype

    'lang' parameter specifies the language of the document.
    Defaults to 'en'.

    See also `DIV`
    """

    tag = b'html'

    strict = b'<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">\n'
    transitional = \
        b'<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">\n'
    frameset = \
        b'<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Frameset//EN" "http://www.w3.org/TR/html4/frameset.dtd">\n'
    html5 = b'<!DOCTYPE HTML>\n'

    def xml(self):
        lang = self['lang']
        if not lang:
            lang = 'en'
        self.attributes['_lang'] = lang
        doctype = self['doctype']
        if doctype is None:
            doctype = self.transitional
        elif doctype == 'strict':
            doctype = self.strict
        elif doctype == 'transitional':
            doctype = self.transitional
        elif doctype == 'frameset':
            doctype = self.frameset
        elif doctype == 'html5':
            doctype = self.html5
        elif doctype == '':
            doctype = b''
        else:
            doctype = b'%s\n' % to_bytes(doctype)
        (fa, co) = self._xml()

        return b'%s<%s%s>%s</%s>' % (doctype, self.tag, fa, co, self.tag)


class XHTML(DIV):
    """
    This is XHTML version of the HTML helper.

    There are three predefined document type definitions.
    They can be specified in the 'doctype' parameter:

    - 'strict' enables strict doctype
    - 'transitional' enables transitional doctype (default)
    - 'frameset' enables frameset doctype
    - any other string will be treated as user's own doctype

    'lang' parameter specifies the language of the document and the xml document.
    Defaults to 'en'.

    'xmlns' parameter specifies the xml namespace.
    Defaults to 'http://www.w3.org/1999/xhtml'.

    See also `DIV`
    """

    tag = b'html'

    strict = b'<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">\n'
    transitional = b'<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">\n'
    frameset = b'<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Frameset//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-frameset.dtd">\n'
    xmlns = 'http://www.w3.org/1999/xhtml'

    def xml(self):
        xmlns = self['xmlns']
        if xmlns:
            self.attributes['_xmlns'] = xmlns
        else:
            self.attributes['_xmlns'] = self.xmlns
        lang = self['lang']
        if not lang:
            lang = 'en'
        self.attributes['_lang'] = lang
        self.attributes['_xml:lang'] = lang
        doctype = self['doctype']
        if doctype:
            if doctype == 'strict':
                doctype = self.strict
            elif doctype == 'transitional':
                doctype = self.transitional
            elif doctype == 'frameset':
                doctype = self.frameset
            else:
                doctype = b'%s\n' % to_bytes(doctype)
        else:
            doctype = self.transitional
        (fa, co) = self._xml()
        return b'%s<%s%s>%s</%s>' % (doctype, self.tag, fa, co, self.tag)


class HEAD(DIV):

    tag = 'head'


class TITLE(DIV):

    tag = 'title'


class META(DIV):

    tag = 'meta/'


class LINK(DIV):

    tag = 'link/'


class SCRIPT(DIV):

    tag = b'script'
    tagname = to_bytes(tag)

    def xml(self):
        (fa, co) = self._xml()
        fa = to_bytes(fa)
        # no escaping of subcomponents
        co = b'\n'.join(map(to_bytes,
            # allow xml components (i.e. ASSIGNJS)
            map(lambda c: c.xml() if hasattr(c, 'xml') and callable(c.xml) else c,
            self.components)))
        if co:
            # <script [attributes]><!--//--><![CDATA[//><!--
            # script body
            # //--><!]]></script>
            # return '<%s%s><!--//--><![CDATA[//><!--\n%s\n//--><!]]></%s>' % (self.tag, fa, co, self.tag)
            return b'<%s%s><!--\n%s\n//--></%s>' % (self.tagname, fa, co, self.tagname)
        else:
            return DIV.xml(self)


class STYLE(DIV):

    tag = 'style'
    tagname = to_bytes(tag)

    def xml(self):
        (fa, co) = self._xml()
        fa = to_bytes(fa)
        # no escaping of subcomponents
        co = b'\n'.join([to_bytes(component) for component in
                       self.components])
        if co:
            # <style [attributes]><!--/*--><![CDATA[/*><!--*/
            # style body
            # /*]]>*/--></style>
            return b'<%s%s><!--/*--><![CDATA[/*><!--*/\n%s\n/*]]>*/--></%s>' % (self.tagname, fa, co, self.tagname)
        else:
            return DIV.xml(self)


class IMG(DIV):

    tag = 'img/'


class SPAN(DIV):

    tag = 'span'


class BODY(DIV):

    tag = 'body'


class H1(DIV):

    tag = 'h1'


class H2(DIV):

    tag = 'h2'


class H3(DIV):

    tag = 'h3'


class H4(DIV):

    tag = 'h4'


class H5(DIV):

    tag = 'h5'


class H6(DIV):

    tag = 'h6'


class P(DIV):
    """
    Will replace ``\\n`` by ``<br />`` if the `cr2br` attribute is provided.

    see also `DIV`
    """

    tag = 'p'

    def xml(self):
        text = DIV.xml(self)
        if self['cr2br']:
            text = text.replace(b'\n', b'<br />')
        return text


class STRONG(DIV):

    tag = 'strong'


class B(DIV):

    tag = 'b'


class BR(DIV):

    tag = 'br/'


class HR(DIV):

    tag = 'hr/'


class A(DIV):
    """
    Generates an A() link.
    A() in web2py is really important and with the included web2py.js
    allows lots of Ajax interactions in the page

    On top of "usual" `_attributes`, it takes

    Args:
        callback: an url to call but not redirect to
        cid: if you want to load the _href into an element of the page (component)
            pass its id (without the #) here
        delete: element to delete after calling callback
        target: same thing as cid
        confirm: text to display upon a callback with a delete
        noconfirm: don't display alert upon a callback with delete

    """

    tag = 'a'

    def xml(self):
        if not self.components and self['_href']:
            self.append(self['_href'])
        disable_needed = ['callback', 'cid', 'delete', 'component', 'target']
        disable_needed = any((self[attr] for attr in disable_needed))
        if disable_needed:
            self['_data-w2p_disable_with'] = self['_disable_with'] or 'default'
            self['_disable_with'] = None
        if self['callback'] and not self['_id']:
            self['_id'] = web2py_uuid()
        if self['delete']:
            self['_data-w2p_remove'] = self['delete']
        if self['target']:
            if self['target'] == '<self>':
                self['target'] = self['_id']
            self['_data-w2p_target'] = self['target']
        if self['component']:
            self['_data-w2p_method'] = 'GET'
            self['_href'] = self['component']
        elif self['callback']:
            self['_data-w2p_method'] = 'POST'
            self['_href'] = self['callback']
            if self['delete'] and not self['noconfirm']:
                if not self['confirm']:
                    self['_data-w2p_confirm'] = 'default'
                else:
                    self['_data-w2p_confirm'] = self['confirm']
        elif self['cid']:
            self['_data-w2p_method'] = 'GET'
            self['_data-w2p_target'] = self['cid']
            if self['pre_call']:
                self['_data-w2p_pre_call'] = self['pre_call']
        return DIV.xml(self)


class BUTTON(DIV):

    tag = 'button'


class EM(DIV):

    tag = 'em'


class EMBED(DIV):

    tag = 'embed/'


class TT(DIV):

    tag = 'tt'


class PRE(DIV):

    tag = 'pre'


class CENTER(DIV):

    tag = 'center'


class CODE(DIV):

    """
    Displays code in HTML with syntax highlighting.

    Args:
        language: indicates the language, otherwise PYTHON is assumed
        link: can provide a link
        styles: for styles

    Examples:

    {{=CODE(\"print('hello world')\", language='python', link=None,
        counter=1, styles={}, highlight_line=None)}}


    supported languages are

        "python", "html_plain", "c", "cpp", "web2py", "html"

    The "html" language interprets {{ and }} tags as "web2py" code,
    "html_plain" doesn't.

    if a link='/examples/global/vars/' is provided web2py keywords are linked to
    the online docs.

    the counter is used for line numbering, counter can be None or a prompt
    string.
    """

    def xml(self):
        language = self['language'] or 'PYTHON'
        link = self['link']
        counter = self.attributes.get('counter', 1)
        highlight_line = self.attributes.get('highlight_line', None)
        context_lines = self.attributes.get('context_lines', None)
        styles = self['styles'] or {}
        return highlight(
            join(self.components),
            language=language,
            link=link,
            counter=counter,
            styles=styles,
            attributes=self.attributes,
            highlight_line=highlight_line,
            context_lines=context_lines,
        )


class LABEL(DIV):

    tag = 'label'


class LI(DIV):

    tag = 'li'


class UL(DIV):
    """
    UL Component.

    If subcomponents are not LI-components they will be wrapped in a LI

    """

    tag = 'ul'

    def _fixup(self):
        self._wrap_components(LI, LI)


class OL(UL):

    tag = 'ol'


class TD(DIV):

    tag = 'td'


class TH(DIV):

    tag = 'th'


class TR(DIV):
    """
    TR Component.

    If subcomponents are not TD/TH-components they will be wrapped in a TD

    """

    tag = 'tr'

    def _fixup(self):
        self._wrap_components((TD, TH), TD)


class __TRHEAD__(DIV):
    """
    __TRHEAD__ Component, internal only

    If subcomponents are not TD/TH-components they will be wrapped in a TH

    """

    tag = 'tr'

    def _fixup(self):
        self._wrap_components((TD, TH), TH)


class THEAD(DIV):

    tag = 'thead'

    def _fixup(self):
        self._wrap_components((__TRHEAD__, TR), __TRHEAD__)


class TBODY(DIV):

    tag = 'tbody'

    def _fixup(self):
        self._wrap_components(TR, TR)


class TFOOT(DIV):

    tag = 'tfoot'

    def _fixup(self):
        self._wrap_components(TR, TR)


class COL(DIV):

    tag = 'col/'


class COLGROUP(DIV):

    tag = 'colgroup'


class TABLE(DIV):
    """
    TABLE Component.

    If subcomponents are not TR/TBODY/THEAD/TFOOT-components
    they will be wrapped in a TR

    """

    tag = 'table'

    def _fixup(self):
        self._wrap_components((TR, TBODY, THEAD, TFOOT, COL, COLGROUP), TR)


class I(DIV):

    tag = 'i'


class IFRAME(DIV):

    tag = 'iframe'


class INPUT(DIV):

    """
    INPUT Component

    Takes two special attributes value= and requires=.

    Args:
        value: used to pass the initial value for the input field.
            value differs from _value because it works for checkboxes, radio,
            textarea and select/option too.
            For a checkbox value should be '' or 'on'.
            For a radio or select/option value should be the _value
            of the checked/selected item.

        requires: should be None, or a validator or a list of validators
            for the value of the field.

    Examples:

    >>> INPUT(_type='text', _name='name', value='Max').xml()
    '<input name=\"name\" type=\"text\" value=\"Max\" />'

    >>> INPUT(_type='checkbox', _name='checkbox', value='on').xml()
    '<input checked=\"checked\" name=\"checkbox\" type=\"checkbox\" value=\"on\" />'

    >>> INPUT(_type='radio', _name='radio', _value='yes', value='yes').xml()
    '<input checked=\"checked\" name=\"radio\" type=\"radio\" value=\"yes\" />'

    >>> INPUT(_type='radio', _name='radio', _value='no', value='yes').xml()
    '<input name=\"radio\" type=\"radio\" value=\"no\" />'


        """

    tag = 'input/'

    def _validate(self):

        # # this only changes value, not _value

        name = self['_name']
        if name is None or name == '':
            return True
        name = str(name)
        request_vars_get = self.request_vars.get
        if self['_type'] != 'checkbox':
            self['old_value'] = self['value'] or self['_value'] or ''
            value = request_vars_get(name, '')
            self['value'] = value if not hasattr(value, 'file') else None
        else:
            self['old_value'] = self['value'] or False
            value = request_vars_get(name)
            if isinstance(value, (tuple, list)):
                self['value'] = self['_value'] in value
            else:
                self['value'] = self['_value'] == value
        requires = self['requires']
        if requires:
            if not isinstance(requires, (list, tuple)):
                requires = [requires]
            for k, validator in enumerate(requires):
                try:
                    (value, errors) = validator(value)
                except:
                    import traceback
                    print(traceback.format_exc())
                    msg = "Validation error, field:%s %s" % (name, validator)
                    raise Exception(msg)
                if errors is not None:
                    self.vars[name] = value
                    self.errors[name] = errors
                    break
        if name not in self.errors:
            self.vars[name] = value
            return True
        return False

    def _postprocessing(self):
        t = self['_type']
        if not t:
            t = self['_type'] = 'text'
        t = t.lower()
        value = self['value']
        if self['_value'] is None or isinstance(self['_value'], cgi.FieldStorage):
            _value = None
        else:
            _value = str(self['_value'])
        if '_checked' in self.attributes and 'value' not in self.attributes:
            pass
        elif t == 'checkbox':
            if not _value:
                _value = self['_value'] = 'on'
            if not value:
                value = []
            elif value is True:
                value = [_value]
            elif not isinstance(value, (list, tuple)):
                value = str(value).split('|')
            self['_checked'] = _value in value and 'checked' or None
        elif t == 'radio':
            if str(value) == str(_value):
                self['_checked'] = 'checked'
            else:
                self['_checked'] = None
        elif t == 'password' and value != DEFAULT_PASSWORD_DISPLAY:
            self['value'] = ''
        elif not t == 'submit':
            if value is None:
                self['value'] = _value
            elif not isinstance(value, list):
                self['_value'] = value

    def xml(self):
        name = self.attributes.get('_name', None)
        if name and hasattr(self, 'errors') \
                and self.errors.get(name, None) \
                and self['hideerror'] is not True:
            self['_class'] = (self['_class'] and self['_class'] + ' ' or '') + 'invalidinput'
            return DIV.xml(self) + DIV(
                DIV(
                    self.errors[name], _class='error',
                    errors=None, _id='%s__error' % name),
                _class='error_wrapper').xml()
        else:
            if self['_class'] and self['_class'].endswith('invalidinput'):
                self['_class'] = self['_class'][:-12]
                if self['_class'] == '':
                    self['_class'] = None
            return DIV.xml(self)


class TEXTAREA(INPUT):

    """
    Examples::

        TEXTAREA(_name='sometext', value='blah ' * 100, requires=IS_NOT_EMPTY())

    'blah blah blah ...' will be the content of the textarea field.

    """

    tag = 'textarea'

    def _postprocessing(self):
        if '_rows' not in self.attributes:
            self['_rows'] = 10
        if '_cols' not in self.attributes:
            self['_cols'] = 40
        if self['value'] is not None:
            self.components = [self['value']]
        elif self.components:
            self['value'] = self.components[0]


class OPTION(DIV):

    tag = 'option'

    def _fixup(self):
        if '_value' not in self.attributes:
            self.attributes['_value'] = str(self.components[0])


class OBJECT(DIV):

    tag = 'object'


class OPTGROUP(DIV):

    tag = 'optgroup'

    def _fixup(self):
        components = []
        for c in self.components:
            if isinstance(c, OPTION):
                components.append(c)
            else:
                components.append(OPTION(c, _value=str(c)))
        self.components = components


class SELECT(INPUT):
    """
    Examples:

    >>> from validators import IS_IN_SET
    >>> SELECT('yes', 'no', _name='selector', value='yes',
    ...    requires=IS_IN_SET(['yes', 'no'])).xml()
    '<select name=\"selector\"><option selected=\"selected\" value=\"yes\">yes</option><option value=\"no\">no</option></select>'

    """

    tag = 'select'

    def _fixup(self):
        components = []
        for c in self.components:
            if isinstance(c, (OPTION, OPTGROUP)):
                components.append(c)
            else:
                components.append(OPTION(c, _value=str(c)))
        self.components = components

    def _postprocessing(self):
        component_list = []
        for c in self.components:
            if isinstance(c, OPTGROUP):
                component_list.append(c.components)
            else:
                component_list.append([c])
        options = itertools.chain(*component_list)

        value = self['value']
        if value is not None:
            if not self['_multiple']:
                for c in options:  # my patch
                    if (value is not None) and (str(c['_value']) == str(value)):
                        c['_selected'] = 'selected'
                    else:
                        c['_selected'] = None
            else:
                if isinstance(value, (list, tuple)):
                    values = [str(item) for item in value]
                else:
                    values = [str(value)]
                for c in options:  # my patch
                    if (value is not None) and (str(c['_value']) in values):
                        c['_selected'] = 'selected'
                    else:
                        c['_selected'] = None


class FIELDSET(DIV):

    tag = 'fieldset'


class LEGEND(DIV):

    tag = 'legend'


class FORM(DIV):

    """
    Examples:

    >>> from validators import IS_NOT_EMPTY
    >>> form=FORM(INPUT(_name="test", requires=IS_NOT_EMPTY()))
    >>> form.xml()
    '<form action=\"#\" enctype=\"multipart/form-data\" method=\"post\"><input name=\"test\" type=\"text\" /></form>'


    a FORM is container for INPUT, TEXTAREA, SELECT and other helpers

    form has one important method::

        form.accepts(request.vars, session)

    if form is accepted (and all validators pass) form.vars contains the
    accepted vars, otherwise form.errors contains the errors.
    in case of errors the form is modified to present the errors to the user.
    """

    tag = 'form'

    def __init__(self, *components, **attributes):
        DIV.__init__(self, *components, **attributes)
        self.vars = Storage()
        self.errors = Storage()
        self.latest = Storage()
        self.accepted = None  # none for not submitted

    def assert_status(self, status, request_vars):
        return status

    def accepts(self,
                request_vars,
                session=None,
                formname='default',
                keepvalues=False,
                onvalidation=None,
                hideerror=False,
                **kwargs
                ):
        """
        kwargs is not used but allows to specify the same interface for FORM and SQLFORM
        """
        if request_vars.__class__.__name__ == 'Request':
            request_vars = request_vars.post_vars
        self.errors.clear()
        self.request_vars = Storage()
        self.request_vars.update(request_vars)
        self.session = session
        self.formname = formname
        self.keepvalues = keepvalues

        # if this tag is a form and we are in accepting mode (status=True)
        # check formname and formkey

        status = True
        changed = False
        request_vars = self.request_vars
        if session is not None:
            formkey = request_vars._formkey
            keyname = '_formkey[%s]' % formname
            formkeys = list(session.get(keyname, []))
            # check if user tampering with form and void CSRF
            if not (formkey and formkeys and formkey in formkeys):
                status = False
            else:
                session[keyname].remove(formkey)
        if formname != request_vars._formname:
            status = False
        if status and session:
            # check if editing a record that has been modified by the server
            if hasattr(self, 'record_hash') and self.record_hash != formkey.split(':')[0]:
                status = False
                self.record_changed = changed = True
        status = self._traverse(status, hideerror)
        status = self.assert_status(status, request_vars)
        if onvalidation:
            if isinstance(onvalidation, dict):
                onsuccess = onvalidation.get('onsuccess', None)
                onfailure = onvalidation.get('onfailure', None)
                onchange = onvalidation.get('onchange', None)
                if [k for k in onvalidation if k not in ('onsuccess', 'onfailure', 'onchange')]:
                    raise RuntimeError('Invalid key in onvalidate dict')
                if onsuccess and status:
                    call_as_list(onsuccess, self)
                if onfailure and request_vars and not status:
                    call_as_list(onfailure, self)
                    status = len(self.errors) == 0
                if changed:
                    if onchange and self.record_changed and \
                            self.detect_record_change:
                        call_as_list(onchange, self)
            elif status:
                call_as_list(onvalidation, self)
        if self.errors:
            status = False
        if session is not None:
            if hasattr(self, 'record_hash'):
                formkey = self.record_hash + ':' + web2py_uuid()
            else:
                formkey = web2py_uuid()
            self.formkey = formkey
            keyname = '_formkey[%s]' % formname
            session[keyname] = list(session.get(keyname, []))[-9:] + [formkey]
        if status and not keepvalues:
            self._traverse(False, hideerror)
        self.accepted = status
        return status

    def _postprocessing(self):
        if '_action' not in self.attributes:
            self['_action'] = '#'
        if '_method' not in self.attributes:
            self['_method'] = 'post'
        if '_enctype' not in self.attributes:
            self['_enctype'] = 'multipart/form-data'

    def hidden_fields(self):
        c = []
        attr = self.attributes.get('hidden', {})
        if 'hidden' in self.attributes:
            c = [INPUT(_type='hidden', _name=key, _value=value) for (key, value) in iteritems(attr)]
        if hasattr(self, 'formkey') and self.formkey:
            c.append(INPUT(_type='hidden', _name='_formkey', _value=self.formkey))
        if hasattr(self, 'formname') and self.formname:
            c.append(INPUT(_type='hidden', _name='_formname', _value=self.formname))
        return DIV(c, _style="display:none;")

    def xml(self):
        newform = FORM(*self.components, **self.attributes)
        hidden_fields = self.hidden_fields()
        if hidden_fields.components:
            newform.append(hidden_fields)
        return DIV.xml(newform)

    def validate(self, **kwargs):
        """
        This function validates the form,
        you can use it instead of directly form.accepts.

        Usage:
        In controller::

            def action():
                form=FORM(INPUT(_name=\"test\", requires=IS_NOT_EMPTY()))
                form.validate() #you can pass some args here - see below
                return dict(form=form)

        This can receive a bunch of arguments

        onsuccess = 'flash' - will show message_onsuccess in response.flash
                    None - will do nothing
                    can be a function (lambda form: pass)
        onfailure = 'flash' - will show message_onfailure in response.flash
                    None - will do nothing
                    can be a function (lambda form: pass)
        onchange = 'flash' - will show message_onchange in response.flash
                    None - will do nothing
                    can be a function (lambda form: pass)

        message_onsuccess
        message_onfailure
        message_onchange
        next      = where to redirect in case of success
        any other kwargs will be passed for form.accepts(...)
        """
        from gluon.globals import current
        from gluon.http import redirect
        kwargs['request_vars'] = kwargs.get(
            'request_vars', current.request.post_vars)
        kwargs['session'] = kwargs.get('session', current.session)
        kwargs['dbio'] = kwargs.get('dbio', False)  # necessary for SQLHTML forms

        onsuccess = kwargs.get('onsuccess', 'flash')
        onfailure = kwargs.get('onfailure', 'flash')
        onchange = kwargs.get('onchange', 'flash')
        message_onsuccess = kwargs.get('message_onsuccess',
                                       current.T("Success!"))
        message_onfailure = kwargs.get('message_onfailure',
                                       current.T("Errors in form, please check it out."))
        message_onchange = kwargs.get('message_onchange',
                                      current.T("Form consecutive submissions not allowed. " +
                                                "Try re-submitting or refreshing the form page."))
        next = kwargs.get('next', None)
        for key in ('message_onsuccess', 'message_onfailure', 'onsuccess',
                    'onfailure', 'next', 'message_onchange', 'onchange'):
            if key in kwargs:
                del kwargs[key]

        if self.accepts(**kwargs):
            if onsuccess == 'flash':
                if next:
                    current.session.flash = message_onsuccess
                else:
                    current.response.flash = message_onsuccess
            elif callable(onsuccess):
                onsuccess(self)
            if next:
                if self.vars:
                    for key, value in iteritems(self.vars):
                        next = next.replace('[%s]' % key,
                                            urllib_quote(str(value)))
                    if not next.startswith('/'):
                        next = URL(next)
                redirect(next)
            return True
        elif self.errors:
            if onfailure == 'flash':
                current.response.flash = message_onfailure
            elif callable(onfailure):
                onfailure(self)
            return False
        elif hasattr(self, "record_changed"):
            if self.record_changed and self.detect_record_change:
                if onchange == 'flash':
                    current.response.flash = message_onchange
                elif callable(onchange):
                    onchange(self)
            return False

    def process(self, **kwargs):
        """
        Perform the .validate() method but returns the form

        Usage in controllers::

            # directly on return
            def action():
                #some code here
                return dict(form=FORM(...).process(...))

        You can use it with FORM, SQLFORM or FORM based plugins::

            # response.flash messages
            def action():
                form = SQLFORM(db.table).process(message_onsuccess='Sucess!')
                return dict(form=form)

            # callback function
            # callback receives True or False as first arg, and a list of args.
            def my_callback(status, msg):
                response.flash = "Success! "+msg if status else "Errors occured"

            # after argument can be 'flash' to response.flash messages
            # or a function name to use as callback or None to do nothing.
            def action():
                return dict(form=SQLFORM(db.table).process(onsuccess=my_callback)


        """
        kwargs['dbio'] = kwargs.get('dbio', True)  # necessary for SQLHTML forms
        self.validate(**kwargs)
        return self

    REDIRECT_JS = "window.location='%s';return false"

    def add_button(self, value, url, _class=None):
        submit = self.element(_type='submit')
        _class = "%s w2p-form-button" % _class if _class else "w2p-form-button"
        submit.parent.append(
            TAG['button'](value, _class=_class,
                          _onclick=url if url.startswith('javascript:') else
                          self.REDIRECT_JS % url))

    @staticmethod
    def confirm(text='OK', buttons=None, hidden=None):
        if not buttons:
            buttons = {}
        if not hidden:
            hidden = {}
        inputs = [INPUT(_type='button',
                        _value=name,
                        _onclick=FORM.REDIRECT_JS % link)
                  for name, link in iteritems(buttons)]
        inputs += [INPUT(_type='hidden',
                         _name=name,
                         _value=value)
                   for name, value in iteritems(hidden)]
        form = FORM(INPUT(_type='submit', _value=text), *inputs)
        form.process()
        return form

    def as_dict(self, flat=False, sanitize=True):
        """EXPERIMENTAL

        Sanitize is naive. It should catch any unsafe value
        for client retrieval.
        """
        SERIALIZABLE = (int, float, bool, basestring, long,
                        set, list, dict, tuple, Storage, type(None))
        UNSAFE = ("PASSWORD", "CRYPT")
        d = self.__dict__

        def sanitizer(obj):
            if isinstance(obj, dict):
                for k in obj.keys():
                    if any([unsafe in str(k).upper() for unsafe in UNSAFE]):
                        # erease unsafe pair
                        obj.pop(k)
            else:
                # not implemented
                pass
            return obj

        def flatten(obj):
            if isinstance(obj, (dict, Storage)):
                newobj = obj.copy()
            else:
                newobj = obj
            if sanitize:
                newobj = sanitizer(newobj)
            if flat:
                if type(obj) in SERIALIZABLE:
                    if isinstance(newobj, (dict, Storage)):
                        for k in newobj:
                            newk = flatten(k)
                            newobj[newk] = flatten(newobj[k])
                            if k != newk:
                                newobj.pop(k)
                        return newobj
                    elif isinstance(newobj, (list, tuple, set)):
                        return [flatten(item) for item in newobj]
                    else:
                        return newobj
                else:
                    return str(newobj)
            else:
                return newobj
        return flatten(d)

    def as_json(self, sanitize=True):
        d = self.as_dict(flat=True, sanitize=sanitize)
        from gluon.serializers import json
        return json(d)

    def as_yaml(self, sanitize=True):
        d = self.as_dict(flat=True, sanitize=sanitize)
        from gluon.serializers import yaml
        return yaml(d)

    def as_xml(self, sanitize=True):
        d = self.as_dict(flat=True, sanitize=sanitize)
        from gluon.serializers import xml
        return xml(d)


class BEAUTIFY(DIV):
    """
    Turns any list, dictionary, etc into decent looking html.

    Two special attributes are

    - sorted: a function that takes the dict and returned sorted keys
    - keyfilter: a function that takes a key and returns its representation or
      None if the key is to be skipped.
      By default key[:1]=='_' is skipped.

    Examples:

    >>> BEAUTIFY(['a', 'b', {'hello': 'world'}]).xml()
    '<div><table><tr><td><div>a</div></td></tr><tr><td><div>b</div></td></tr><tr><td><div><table><tr><td style="font-weight:bold;vertical-align:top;">hello</td><td style="vertical-align:top;">:</td><td><div>world</div></td></tr></table></div></td></tr></table></div>'

    """

    tag = 'div'

    @staticmethod
    def no_underscore(key):
        if key[:1] == '_':
            return None
        return key

    def __init__(self, component, **attributes):
        self.components = [component]
        self.attributes = attributes
        sorter = attributes.get('sorted', sorted)
        keyfilter = attributes.get('keyfilter', BEAUTIFY.no_underscore)
        components = []
        attributes = copy.copy(self.attributes)
        level = attributes['level'] = attributes.get('level', 6) - 1
        if '_class' in attributes:
            attributes['_class'] += 'i'
        if level == 0:
            return
        for c in self.components:
            if hasattr(c, 'value') and not callable(c.value) and not isinstance(c, cgi.FieldStorage):
                if c.value:
                    components.append(c.value)
            if hasattr(c, 'xml') and callable(c.xml):
                components.append(c)
                continue
            elif hasattr(c, 'keys') and callable(c.keys):
                rows = []
                try:
                    keys = (sorter and sorter(c)) or c
                    for key in keys:
                        if isinstance(key, (str, unicodeT)) and keyfilter:
                            filtered_key = keyfilter(key)
                        else:
                            filtered_key = str(key)
                        if filtered_key is None:
                            continue
                        value = c[key]
                        if isinstance(value, types.LambdaType):
                            continue
                        rows.append(
                            TR(
                                TD(filtered_key, _style='font-weight:bold;vertical-align:top;'),
                                TD(':', _style='vertical-align:top;'),
                                TD(BEAUTIFY(value, **attributes))))
                    components.append(TABLE(*rows, **attributes))
                    continue
                except:
                    pass
            if isinstance(c, str):
                components.append(str(c))
            elif isinstance(c, unicodeT):
                components.append(c.encode('utf8'))
            elif isinstance(c, (list, tuple)):
                items = [TR(TD(BEAUTIFY(item, **attributes)))
                         for item in c]
                components.append(TABLE(*items, **attributes))
            elif isinstance(c, cgi.FieldStorage):
                components.append('FieldStorage object')
            else:
                components.append(repr(c))
        self.components = components


class MENU(DIV):
    """
    Used to build menus

    Args:
        _class: defaults to 'web2py-menu web2py-menu-vertical'
        ul_class: defaults to 'web2py-menu-vertical'
        li_class: defaults to 'web2py-menu-expand'
        li_first: defaults to 'web2py-menu-first'
        li_last: defaults to 'web2py-menu-last'

    Use like::

        menu = MENU([['name', False, URL(...), [submenu]], ...])
        {{=menu}}

    """

    tag = 'ul'

    def __init__(self, data, **args):
        self.data = data
        self.attributes = args
        self.components = []
        if '_class' not in self.attributes:
            self['_class'] = 'web2py-menu web2py-menu-vertical'
        if 'ul_class' not in self.attributes:
            self['ul_class'] = 'web2py-menu-vertical'
        if 'li_class' not in self.attributes:
            self['li_class'] = 'web2py-menu-expand'
        if 'li_first' not in self.attributes:
            self['li_first'] = 'web2py-menu-first'
        if 'li_last' not in self.attributes:
            self['li_last'] = 'web2py-menu-last'
        if 'li_active' not in self.attributes:
            self['li_active'] = 'web2py-menu-active'
        if 'mobile' not in self.attributes:
            self['mobile'] = False

    def serialize(self, data, level=0):
        if level == 0:
            ul = UL(**self.attributes)
        else:
            ul = UL(_class=self['ul_class'])
        for item in data:
            if isinstance(item, LI):
                ul.append(item)
            else:
                (name, active, link) = item[:3]
                if isinstance(link, DIV):
                    li = LI(link)
                elif 'no_link_url' in self.attributes and self['no_link_url'] == link:
                    li = LI(DIV(name))
                elif isinstance(link, dict):
                    li = LI(A(name, **link))
                elif link:
                    li = LI(A(name, _href=link))
                elif not link and isinstance(name, A):
                    li = LI(name)
                else:
                    li = LI(A(name, _href='#',
                              _onclick='javascript:void(0);return false;'))
                if level == 0 and item == data[0]:
                    li['_class'] = self['li_first']
                elif level == 0 and item == data[-1]:
                    li['_class'] = self['li_last']
                if len(item) > 3 and item[3]:
                    li['_class'] = self['li_class']
                    li.append(self.serialize(item[3], level + 1))
                if active or ('active_url' in self.attributes and self['active_url'] == link):
                    if li['_class']:
                        li['_class'] = li['_class'] + ' ' + self['li_active']
                    else:
                        li['_class'] = self['li_active']
                if len(item) <= 4 or item[4] is True:
                    ul.append(li)
        return ul

    def serialize_mobile(self, data, select=None, prefix=''):
        if not select:
            select = SELECT(**self.attributes)
        custom_items = []
        for item in data:
            # Custom item aren't serialized as mobile
            if len(item) >= 3 and (not item[0]) or (isinstance(item[0], DIV) and not (item[2])):
                # ex: ('', False, A('title', _href=URL(...), _title="title"))
                # ex: (A('title', _href=URL(...), _title="title"), False, None)
                custom_items.append(item)
            elif len(item) <= 4 or item[4] is True:
                select.append(OPTION(CAT(prefix, item[0]),
                                     _value=item[2], _selected=item[1]))
                if len(item) > 3 and len(item[3]):
                    self.serialize_mobile(
                        item[3], select, prefix=CAT(prefix, item[0], '/'))
        select['_onchange'] = 'window.location=this.value'
        # avoid to wrap the select if no custom items are present
        html = DIV(select, self.serialize(custom_items)) if len(custom_items) else select
        return html

    def xml(self):
        if self['mobile']:
            return self.serialize_mobile(self.data, 0).xml()
        else:
            return self.serialize(self.data, 0).xml()


def embed64(filename=None,
            file=None,
            data=None,
            extension='image/gif'
            ):
    """
    helper to encode the provided (binary) data into base64.

    Args:
        filename: if provided, opens and reads this file in 'rb' mode
        file: if provided, reads this file
        data: if provided, uses the provided data
    """

    if filename and os.path.exists(file):
        fp = open(filename, 'rb')
        data = fp.read()
        fp.close()
    data = base64.b64encode(data)
    return 'data:%s;base64,%s' % (extension, data)


# TODO: Check if this test() is still relevant now that we have gluon/tests/test_html.py
def test():
    """
    Example:

    >>> from validators import *
    >>> print(DIV(A('click me', _href=URL(a='a', c='b', f='c')), BR(), HR(), DIV(SPAN("World"), _class='unknown')).xml())
    <div><a href="/a/b/c">click me</a><br /><hr /><div class=\"unknown\"><span>World</span></div></div>
    >>> print(DIV(UL("doc","cat","mouse")).xml())
    <div><ul><li>doc</li><li>cat</li><li>mouse</li></ul></div>
    >>> print(DIV(UL("doc", LI("cat", _class='feline'), 18)).xml())
    <div><ul><li>doc</li><li class=\"feline\">cat</li><li>18</li></ul></div>
    >>> print(TABLE(['a', 'b', 'c'], TR('d', 'e', 'f'), TR(TD(1), TD(2), TD(3))).xml())
    <table><tr><td>a</td><td>b</td><td>c</td></tr><tr><td>d</td><td>e</td><td>f</td></tr><tr><td>1</td><td>2</td><td>3</td></tr></table>
    >>> form=FORM(INPUT(_type='text', _name='myvar', requires=IS_EXPR('int(value)<10')))
    >>> print(form.xml())
    <form action=\"#\" enctype=\"multipart/form-data\" method=\"post\"><input name=\"myvar\" type=\"text\" /></form>
    >>> print(form.accepts({'myvar':'34'}, formname=None))
    False
    >>> print(form.xml())
    <form action="#" enctype="multipart/form-data" method="post"><input class="invalidinput" name="myvar" type="text" value="34" /><div class="error_wrapper"><div class="error" id="myvar__error">Invalid expression</div></div></form>
    >>> print(form.accepts({'myvar':'4'}, formname=None, keepvalues=True))
    True
    >>> print(form.xml())
    <form action=\"#\" enctype=\"multipart/form-data\" method=\"post\"><input name=\"myvar\" type=\"text\" value=\"4\" /></form>
    >>> form=FORM(SELECT('cat', 'dog', _name='myvar'))
    >>> print(form.accepts({'myvar':'dog'}, formname=None, keepvalues=True))
    True
    >>> print(form.xml())
    <form action=\"#\" enctype=\"multipart/form-data\" method=\"post\"><select name=\"myvar\"><option value=\"cat\">cat</option><option selected=\"selected\" value=\"dog\">dog</option></select></form>
    >>> form=FORM(INPUT(_type='text', _name='myvar', requires=IS_MATCH('^\w+$', 'only alphanumeric!')))
    >>> print(form.accepts({'myvar':'as df'}, formname=None))
    False
    >>> print(form.xml())
    <form action="#" enctype="multipart/form-data" method="post"><input class="invalidinput" name="myvar" type="text" value="as df" /><div class="error_wrapper"><div class="error" id="myvar__error">only alphanumeric!</div></div></form>
    >>> session={}
    >>> form=FORM(INPUT(value="Hello World", _name="var", requires=IS_MATCH('^\w+$')))
    >>> isinstance(form.as_dict(), dict)
    True
    >>> "vars" in form.as_dict(flat=True)
    True
    >>> isinstance(form.as_json(), basestring) and len(form.as_json(sanitize=False)) > 0
    True
    >>> if form.accepts({}, session,formname=None): print('passed')
    >>> if form.accepts({'var':'test ', '_formkey': session['_formkey[None]']}, session, formname=None): print('passed')
    """
    pass


class web2pyHTMLParser(HTMLParser):
    """
    obj = web2pyHTMLParser(text) parses and html/xml text into web2py helpers.
    obj.tree contains the root of the tree, and tree can be manipulated
    """

    def __init__(self, text, closed=('input', 'link')):
        HTMLParser.__init__(self)
        self.tree = self.parent = TAG['']()
        self.closed = closed
        self.last = None
        self.feed(text)

    def handle_starttag(self, tagname, attrs):
        if tagname in self.closed:
            tagname += '/'
        tag = TAG[tagname]()
        for key, value in attrs:
            tag['_' + key] = value
        tag.parent = self.parent
        self.parent.append(tag)
        if not tag.tag.endswith('/'):
            self.parent = tag
        else:
            self.last = tag.tag[:-1]

    def handle_data(self, data):
        if not isinstance(data, unicodeT):
            try:
                data = data.decode('utf8')
            except:
                data = data.decode('latin1')
        self.parent.append(data.encode('utf8', 'xmlcharref'))

    def handle_charref(self, name):
        if name.startswith('x'):
            self.parent.append(unichr(int(name[1:], 16)).encode('utf8'))
        else:
            self.parent.append(unichr(int(name)).encode('utf8'))

    def handle_entityref(self, name):
        self.parent.append(entitydefs[name])

    def handle_endtag(self, tagname):
        # this deals with unbalanced tags
        if tagname == self.last:
            return
        while True:
            try:
                parent_tagname = self.parent.tag
                self.parent = self.parent.parent
            except:
                raise RuntimeError("unable to balance tag %s" % tagname)
            if parent_tagname[:len(tagname)] == tagname:
                break


def markdown_serializer(text, tag=None, attr=None):
    attr = attr or {}
    if tag is None:
        return re.sub('\s+', ' ', text)
    if tag == 'br':
        return '\n\n'
    if tag == 'h1':
        return '#' + text + '\n\n'
    if tag == 'h2':
        return '#' * 2 + text + '\n\n'
    if tag == 'h3':
        return '#' * 3 + text + '\n\n'
    if tag == 'h4':
        return '#' * 4 + text + '\n\n'
    if tag == 'p':
        return text + '\n\n'
    if tag == 'b' or tag == 'strong':
        return '**%s**' % text
    if tag == 'em' or tag == 'i':
        return '*%s*' % text
    if tag == 'tt' or tag == 'code':
        return '`%s`' % text
    if tag == 'a':
        return '[%s](%s)' % (text, attr.get('_href', ''))
    if tag == 'img':
        return '![%s](%s)' % (attr.get('_alt', ''), attr.get('_src', ''))
    return text


def markmin_serializer(text, tag=None, attr=None):
    attr = attr or {}
    # if tag is None: return re.sub('\s+',' ',text)
    if tag == 'br':
        return '\n\n'
    if tag == 'h1':
        return '# ' + text + '\n\n'
    if tag == 'h2':
        return '#' * 2 + ' ' + text + '\n\n'
    if tag == 'h3':
        return '#' * 3 + ' ' + text + '\n\n'
    if tag == 'h4':
        return '#' * 4 + ' ' + text + '\n\n'
    if tag == 'p':
        return text + '\n\n'
    if tag == 'li':
        return '\n- ' + text.replace('\n', ' ')
    if tag == 'tr':
        return text[3:].replace('\n', ' ') + '\n'
    if tag in ['table', 'blockquote']:
        return '\n-----\n' + text + '\n------\n'
    if tag in ['td', 'th']:
        return ' | ' + text
    if tag in ['b', 'strong', 'label']:
        return '**%s**' % text
    if tag in ['em', 'i']:
        return "''%s''" % text
    if tag in ['tt']:
        return '``%s``' % text.strip()
    if tag in ['code']:
        return '``\n%s``' % text
    if tag == 'a':
        return '[[%s %s]]' % (text, attr.get('_href', ''))
    if tag == 'img':
        return '[[%s %s left]]' % (attr.get('_alt', 'no title'), attr.get('_src', ''))
    return text


class MARKMIN(XmlComponent):
    """
    For documentation: http://web2py.com/examples/static/markmin.html
    """
    def __init__(self,
                 text, extra=None, allowed=None, sep='p',
                 url=None, environment=None, latex='google',
                 autolinks='default',
                 protolinks='default',
                 class_prefix='',
                 id_prefix='markmin_',
                 **kwargs):
        self.text = to_bytes(text)
        self.extra = extra or {}
        self.allowed = allowed or {}
        self.sep = sep
        self.url = URL if url is True else url
        self.environment = environment
        self.latex = latex
        self.autolinks = autolinks
        self.protolinks = protolinks
        self.class_prefix = class_prefix
        self.id_prefix = id_prefix
        self.kwargs = kwargs

    def flatten(self):
        return self.text

    def xml(self):
        from gluon.contrib.markmin.markmin2html import render
        html = render(self.text, extra=self.extra,
                      allowed=self.allowed, sep=self.sep, latex=self.latex,
                      URL=self.url, environment=self.environment,
                      autolinks=self.autolinks, protolinks=self.protolinks,
                      class_prefix=self.class_prefix, id_prefix=self.id_prefix)
        return to_bytes(html) if not self.kwargs else to_bytes(DIV(XML(html), **self.kwargs).xml())

    def __str__(self):
        # In PY3 __str__ cannot return bytes (TypeError: __str__ returned non-string (type bytes))
        return to_native(self.xml())


def ASSIGNJS(**kargs):
    """
    Example:
        ASSIGNJS(var1='1', var2='2') will return the following javascript variables assignations :

            var var1 = "1";
            var var2 = "2";

    Args:
        **kargs: Any keywords arguments and assigned values.

    Returns:
        Javascript vars assignations for the key/value passed.

    """
    from gluon.serializers import json
    s = ""
    for key, value in kargs.items():
        s += 'var %s = %s;\n' % (key, json(value))
    return XML(s)


if __name__ == '__main__':
    import doctest
    doctest.testmod()
