# -*- coding: utf-8 -*-

# copyright 2003-2011 LOGILAB S.A. (Paris, FRANCE), all rights reserved.

# contact http://www.logilab.fr/ -- mailto:contact@logilab.fr

#

# This file is part of CubicWeb.

#

# CubicWeb is free software: you can redistribute it and/or modify it under the

# terms of the GNU Lesser General Public License as published by the Free

# Software Foundation, either version 2.1 of the License, or (at your option)

# any later version.

#

# CubicWeb is distributed in the hope that it will be useful, but WITHOUT

# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS

# FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public License for more

# details.

#

# You should have received a copy of the GNU Lesser General Public License along

# with CubicWeb.  If not, see <http://www.gnu.org/licenses/>.

"""user interface libraries

contains some functions designed to help implementation of cubicweb user

interface.

"""

__docformat__ = "restructuredtext en"

import csv

import re

from StringIO import StringIO

from logilab.mtconverter import xml_escape, html_unescape

from logilab.common.date import ustrftime

from logilab.common.deprecation import deprecated

from cubicweb.utils import JSString, json_dumps

def rql_for_eid(eid):

    """return the rql query necessary to fetch entity with the given eid.  This

    function should only be used to generate link with rql inside, not to give

    to cursor.execute (in which case you won't benefit from rql cache).

    :Parameters:

      - `eid`: the eid of the entity we should search

    :rtype: str

    :return: the rql query

    """

    return 'Any X WHERE X eid %s' % eid

def eid_param(name, eid):

    assert name is not None

    assert eid is not None

    return '%s:%s' % (name, eid)

def print_bytes(value, req, props, displaytime=True):

    return u''

def print_string(value, req, props, displaytime=True):

    # don't translate empty value if you don't want strange results

    if props is not None and value and props.get('internationalizable'):

        return req._(value)

    return value

def print_int(value, req, props, displaytime=True):

    return unicode(value)

def print_date(value, req, props, displaytime=True):

    return ustrftime(value, req.property_value('ui.date-format'))

def print_time(value, req, props, displaytime=True):

    return ustrftime(value, req.property_value('ui.time-format'))

def print_tztime(value, req, props, displaytime=True):

    return ustrftime(value, req.property_value('ui.time-format')) + u' UTC'

def print_datetime(value, req, props, displaytime=True):

    if displaytime:

        return ustrftime(value, req.property_value('ui.datetime-format'))

    return ustrftime(value, req.property_value('ui.date-format'))

def print_tzdatetime(value, req, props, displaytime=True):

    if displaytime:

        return ustrftime(value, req.property_value('ui.datetime-format')) + u' UTC'

    return ustrftime(value, req.property_value('ui.date-format'))

_('%d years')

_('%d months')

_('%d weeks')

_('%d days')

_('%d hours')

_('%d minutes')

_('%d seconds')

def print_timedelta(value, req, props, displaytime=True):

    if isinstance(value, (int, long)):

        # `date - date`, unlike `datetime - datetime` gives an int

        # (number of days), not a timedelta

        # XXX should rql be fixed to return Int instead of Interval in

        #     that case? that would be probably the proper fix but we

        #     loose information on the way...

        value = timedelta(days=value)

    if value.days > 730 or value.days < -730: # 2 years

        return req._('%d years') % (value.days // 365)

    elif value.days > 60 or value.days < -60: # 2 months

        return req._('%d months') % (value.days // 30)

    elif value.days > 14 or value.days < -14: # 2 weeks

        return req._('%d weeks') % (value.days // 7)

    elif value.days > 2 or value.days < -2:

        return req._('%d days') % int(value.days)

    else:

        minus = 1 if value.days > 0 else -1

        if value.seconds > 3600:

            return req._('%d hours') % (int(value.seconds // 3600) * minus)

        elif value.seconds >= 120:

            return req._('%d minutes') % (int(value.seconds // 60) * minus)

        else:

            return req._('%d seconds') % (int(value.seconds) * minus)

def print_boolean(value, req, props, displaytime=True):

    if value:

        return req._('yes')

    return req._('no')

def print_float(value, req, props, displaytime=True):

    return unicode(req.property_value('ui.float-format') % value)

PRINTERS = {

    'Bytes': print_bytes,

    'String': print_string,

    'Int': print_int,

    'BigInt': print_int,

    'Date': print_date,

    'Time': print_time,

    'TZTime': print_tztime,

    'Datetime': print_datetime,

    'TZDatetime': print_tzdatetime,

    'Boolean': print_boolean,

    'Float': print_float,

    'Decimal': print_float,

    'Interval': print_timedelta,

    }

@deprecated('[3.14] use req.printable_value(attrtype, value, ...)')

def printable_value(req, attrtype, value, props=None, displaytime=True):

    return req.printable_value(attrtype, value, props, displaytime)

# text publishing #############################################################

try:

    from cubicweb.ext.rest import rest_publish # pylint: disable=W0611

except ImportError:

    def rest_publish(entity, data):

        """default behaviour if docutils was not found"""

        return xml_escape(data)

TAG_PROG = re.compile(r'</?.*?>', re.U)

def remove_html_tags(text):

    """Removes HTML tags from text

    >>> remove_html_tags('<td>hi <a href="http://www.google.fr">world</a></td>')

    'hi world'

    >>>

    """

    return TAG_PROG.sub('', text)

REF_PROG = re.compile(r"<ref\s+rql=([\'\"])([^\1]*?)\1\s*>([^<]*)</ref>", re.U)

def _subst_rql(view, obj):

    delim, rql, descr = obj.groups()

    return u'<a href="%s">%s</a>' % (view._cw.build_url(rql=rql), descr)

def html_publish(view, text):

    """replace <ref rql=''> links by <a href="...">"""

    if not text:

        return u''

    return REF_PROG.sub(lambda obj, view=view:_subst_rql(view, obj), text)

# fallback implementation, nicer one defined below if lxml> 2.0 is available

def safe_cut(text, length):

    """returns a string of length <length> based on <text>, removing any html

    tags from given text if cut is necessary."""

    if text is None:

        return u''

    noenttext = html_unescape(text)

    text_nohtml = remove_html_tags(noenttext)

    # try to keep html tags if text is short enough

    if len(text_nohtml) <= length:

        return text

    # else if un-tagged text is too long, cut it

    return xml_escape(text_nohtml[:length] + u'...')

fallback_safe_cut = safe_cut

REM_ROOT_HTML_TAGS = re.compile('</(body|html)>', re.U)

from lxml import etree, html

from lxml.html import clean, defs

ALLOWED_TAGS = (defs.general_block_tags | defs.list_tags | defs.table_tags |

                defs.phrase_tags | defs.font_style_tags |

                set(('span', 'a', 'br', 'img', 'map', 'area', 'sub', 'sup'))

                )

CLEANER = clean.Cleaner(allow_tags=ALLOWED_TAGS, remove_unknown_tags=False,

                        style=True, safe_attrs_only=True,

                        add_nofollow=False,

                        )

def soup2xhtml(data, encoding):

    """tidy html soup by allowing some element tags and return the result

    """

    # remove spurious </body> and </html> tags, then normalize line break

    # (see http://www.w3.org/Protocols/rfc2616/rfc2616-sec3.html#sec3.7.1)

    data = REM_ROOT_HTML_TAGS.sub('', u'\n'.join(data.splitlines()))

    xmltree = etree.HTML(CLEANER.clean_html('<div>%s</div>' % data))

    # NOTE: lxml 2.0 does support encoding='unicode', but last time I (syt)

    # tried I got weird results (lxml 2.2.8)

    body = etree.tostring(xmltree[0], encoding=encoding)

    # remove <body> and </body> and decode to unicode

    snippet = body[6:-7].decode(encoding)

    # take care to bad xhtml (for instance starting with </div>) which

    # may mess with the <div> we added below. Only remove it if it's

    # still there...

    if snippet.startswith('<div>') and snippet.endswith('</div>'):

        snippet = snippet[5:-6]

    return snippet

    # lxml.Cleaner envelops text elements by internal logic (not accessible)

    # see http://www.w3.org/Protocols/rfc2616/rfc2616-sec3.html#sec3.7.1

    # TODO drop attributes in elements

    # TODO add policy configuration (content only, embedded content, ...)

    # XXX this is buggy for "<p>text1</p><p>text2</p>"...

    # XXX drop these two snippets action and follow the lxml behaviour

    # XXX (tests need to be updated)

    # if snippet.startswith('<div>') and snippet.endswith('</div>'):

    #     snippet = snippet[5:-6]

    # if snippet.startswith('<p>') and snippet.endswith('</p>'):

    #     snippet = snippet[3:-4]

    return snippet.decode(encoding)

if hasattr(etree.HTML('<div>test</div>'), 'iter'): # XXX still necessary?

    # pylint: disable=E0102

    def safe_cut(text, length):

        """returns an html document of length <length> based on <text>,

        and cut is necessary.

        """

        if text is None:

            return u''

        dom = etree.HTML(text)

        curlength = 0

        add_ellipsis = False

        for element in dom.iter():

            if curlength >= length:

                parent = element.getparent()

                parent.remove(element)

                if curlength == length and (element.text or element.tail):

                    add_ellipsis = True

            else:

                if element.text is not None:

                    element.text = cut(element.text, length - curlength)

                    curlength += len(element.text)

                if element.tail is not None:

                    if curlength < length:

                        element.tail = cut(element.tail, length - curlength)

                        curlength += len(element.tail)

                    elif curlength == length:

                        element.tail = '...'

                    else:

                        element.tail = ''

        text = etree.tounicode(dom[0])[6:-7] # remove wrapping <body></body>

        if add_ellipsis:

            return text + u'...'

        return text

def text_cut(text, nbwords=30, gotoperiod=True):

    """from the given plain text, return a text with at least <nbwords> words,

    trying to go to the end of the current sentence.

    :param nbwords: the minimum number of words required

    :param gotoperiod: specifies if the function should try to go to

                       the first period after the cut (i.e. finish

                       the sentence if possible)

    Note that spaces are normalized.

    """

    if text is None:

        return u''

    words = text.split()

    text = u' '.join(words) # normalize spaces

    textlength = minlength = len(' '.join(words[:nbwords]))

    if gotoperiod:

        textlength = text.find('.', minlength) + 1

        if textlength == 0: # no period found

            textlength = minlength

    return text[:textlength]

def cut(text, length):

    """returns a string of a maximum length <length> based on <text>

    (approximatively, since if text has been  cut, '...' is added to the end of the string,

    resulting in a string of len <length> + 3)

    """

    if text is None:

        return u''

    if len(text) <= length:

        return text

    # else if un-tagged text is too long, cut it

    return text[:length] + u'...'

# HTML generation helper functions ############################################

class _JSId(object):

    def __init__(self, id, parent=None):

        self.id = id

        self.parent = parent

    def __unicode__(self):

        if self.parent:

            return u'%s.%s' % (self.parent, self.id)

        return unicode(self.id)

    def __str__(self):

        return unicode(self).encode('utf8')

    def __getattr__(self, attr):

        return _JSId(attr, self)

    def __call__(self, *args):

        return _JSCallArgs(args, self)

class _JSCallArgs(_JSId):

    def __init__(self, args, parent=None):

        assert isinstance(args, tuple)

        self.args = args

        self.parent = parent

    def __unicode__(self):

        args = []

        for arg in self.args:

            if isinstance(arg, JSString):

                args.append(arg)

            else:

                args.append(json_dumps(arg))

        if self.parent:

            return u'%s(%s)' % (self.parent, ','.join(args))

        return ','.join(args)

class _JS(object):

    def __getattr__(self, attr):

        return _JSId(attr)

js = _JS()

js.__doc__ = """\

magic object to return strings suitable to call some javascript function with

the given arguments (which should be correctly typed).

>>> str(js.pouet(1, "2"))

'pouet(1,"2")'

>>> str(js.cw.pouet(1, "2"))

'cw.pouet(1,"2")'

>>> str(js.cw.pouet(1, "2").pouet(None))

'cw.pouet(1,"2").pouet(null)'

>>> str(js.cw.pouet(1, JSString("$")).pouet(None))

'cw.pouet(1,$).pouet(null)'

"""

def domid(string):

    """return a valid DOM id from a string (should also be usable in jQuery

    search expression...)

    """

    return string.replace('.', '_').replace('-', '_')

HTML4_EMPTY_TAGS = frozenset(('base', 'meta', 'link', 'hr', 'br', 'param',

                              'img', 'area', 'input', 'col'))

def sgml_attributes(attrs):

    return u' '.join(u'%s="%s"' % (attr, xml_escape(unicode(value)))

                     for attr, value in sorted(attrs.items())

                     if value is not None)