added m3u8 parser library

build/plugin/src/engine/client.py

@@ -18,6 +18,7 @@
 from Plugins.Extensions.archivCZSK.engine.tools.util import toString, toUnicode
 from Plugins.Extensions.archivCZSK.gui.captcha import Captcha
 from Plugins.Extensions.archivCZSK.resources.libraries import simplejson as json
+from Plugins.Extensions.archivCZSK.resources.libraries import m3u8
 
 
 GItem_lst = VideoAddonContentProvider.get_shared_itemlist()

build/plugin/src/resources/libraries/iso8601/__init__.py

@@ -0,0 +1 @@
+from .iso8601 import *

build/plugin/src/resources/libraries/iso8601/iso8601.py

@@ -0,0 +1,209 @@
+"""ISO 8601 date time string parsing
+
+Basic usage:
+>>> import iso8601
+>>> iso8601.parse_date("2007-01-25T12:00:00Z")
+datetime.datetime(2007, 1, 25, 12, 0, tzinfo=<iso8601.iso8601.Utc ...>)
+>>>
+
+"""
+
+from datetime import (
+    datetime,
+    timedelta,
+    tzinfo
+)
+from decimal import Decimal
+import sys
+import re
+
+__all__ = ["parse_date", "ParseError", "UTC"]
+
+if sys.version_info >= (3, 0, 0):
+    _basestring = str
+else:
+    _basestring = basestring
+
+
+# Adapted from http://delete.me.uk/2005/03/iso8601.html
+ISO8601_REGEX = re.compile(
+    r"""
+    (?P<year>[0-9]{4})
+    (
+        (
+            (-(?P<monthdash>[0-9]{1,2}))
+            |
+            (?P<month>[0-9]{2})
+            (?!$)  # Don't allow YYYYMM
+        )
+        (
+            (
+                (-(?P<daydash>[0-9]{1,2}))
+                |
+                (?P<day>[0-9]{2})
+            )
+            (
+                (
+                    (?P<separator>[ T])
+                    (?P<hour>[0-9]{2})
+                    (:{0,1}(?P<minute>[0-9]{2})){0,1}
+                    (
+                        :{0,1}(?P<second>[0-9]{1,2})
+                        ([.,](?P<second_fraction>[0-9]+)){0,1}
+                    ){0,1}
+                    (?P<timezone>
+                        Z
+                        |
+                        (
+                            (?P<tz_sign>[-+])
+                            (?P<tz_hour>[0-9]{2})
+                            :{0,1}
+                            (?P<tz_minute>[0-9]{2}){0,1}
+                        )
+                    ){0,1}
+                ){0,1}
+            )
+        ){0,1}  # YYYY-MM
+    ){0,1}  # YYYY only
+    $
+    """,
+    re.VERBOSE
+)
+
+class ParseError(Exception):
+    """Raised when there is a problem parsing a date string"""
+
+# Yoinked from python docs
+ZERO = timedelta(0)
+class Utc(tzinfo):
+    """UTC Timezone
+
+    """
+    def utcoffset(self, dt):
+        return ZERO
+
+    def tzname(self, dt):
+        return "UTC"
+
+    def dst(self, dt):
+        return ZERO
+
+    def __repr__(self):
+        return "<iso8601.Utc>"
+
+UTC = Utc()
+
+class FixedOffset(tzinfo):
+    """Fixed offset in hours and minutes from UTC
+
+    """
+    def __init__(self, offset_hours, offset_minutes, name):
+        self.__offset_hours = offset_hours  # Keep for later __getinitargs__
+        self.__offset_minutes = offset_minutes  # Keep for later __getinitargs__
+        self.__offset = timedelta(hours=offset_hours, minutes=offset_minutes)
+        self.__name = name
+
+    def __eq__(self, other):
+        if isinstance(other, FixedOffset):
+            return (
+                (other.__offset == self.__offset)
+                and
+                (other.__name == self.__name)
+            )
+        if isinstance(other, tzinfo):
+            return other == self
+        return False
+
+    def __getinitargs__(self):
+        return (self.__offset_hours, self.__offset_minutes, self.__name)
+
+    def utcoffset(self, dt):
+        return self.__offset
+
+    def tzname(self, dt):
+        return self.__name
+
+    def dst(self, dt):
+        return ZERO
+
+    def __repr__(self):
+        return "<FixedOffset %r %r>" % (self.__name, self.__offset)
+
+def to_int(d, key, default_to_zero=False, default=None, required=True):
+    """Pull a value from the dict and convert to int
+
+    :param default_to_zero: If the value is None or empty, treat it as zero
+    :param default: If the value is missing in the dict use this default
+
+    """
+    value = d.get(key) or default
+    if (value in ["", None]) and default_to_zero:
+        return 0
+    if value is None:
+        if required:
+            raise ParseError("Unable to read %s from %s" % (key, d))
+    else:
+        return int(value)
+
+def parse_timezone(matches, default_timezone=UTC):
+    """Parses ISO 8601 time zone specs into tzinfo offsets
+
+    """
+
+    if matches["timezone"] == "Z":
+        return UTC
+    # This isn't strictly correct, but it's common to encounter dates without
+    # timezones so I'll assume the default (which defaults to UTC).
+    # Addresses issue 4.
+    if matches["timezone"] is None:
+        return default_timezone
+    sign = matches["tz_sign"]
+    hours = to_int(matches, "tz_hour")
+    minutes = to_int(matches, "tz_minute", default_to_zero=True)
+    description = "%s%02d:%02d" % (sign, hours, minutes)
+    if sign == "-":
+        hours = -hours
+        minutes = -minutes
+    return FixedOffset(hours, minutes, description)
+
+def parse_date(datestring, default_timezone=UTC):
+    """Parses ISO 8601 dates into datetime objects
+
+    The timezone is parsed from the date string. However it is quite common to
+    have dates without a timezone (not strictly correct). In this case the
+    default timezone specified in default_timezone is used. This is UTC by
+    default.
+
+    :param datestring: The date to parse as a string
+    :param default_timezone: A datetime tzinfo instance to use when no timezone
+                             is specified in the datestring. If this is set to
+                             None then a naive datetime object is returned.
+    :returns: A datetime.datetime instance
+    :raises: ParseError when there is a problem parsing the date or
+             constructing the datetime instance.
+
+    """
+    if not isinstance(datestring, _basestring):
+        raise ParseError("Expecting a string %r" % datestring)
+    m = ISO8601_REGEX.match(datestring)
+    if not m:
+        raise ParseError("Unable to parse date string %r" % datestring)
+    groups = m.groupdict()
+
+    tz = parse_timezone(groups, default_timezone=default_timezone)
+
+    groups["second_fraction"] = int(Decimal("0.%s" % (groups["second_fraction"] or 0)) * Decimal("1000000.0"))
+
+    try:
+        return datetime(
+            year=to_int(groups, "year"),
+            month=to_int(groups, "month", default=to_int(groups, "monthdash", required=False, default=1)),
+            day=to_int(groups, "day", default=to_int(groups, "daydash", required=False, default=1)),
+            hour=to_int(groups, "hour", default_to_zero=True),
+            minute=to_int(groups, "minute", default_to_zero=True),
+            second=to_int(groups, "second", default_to_zero=True),
+            microsecond=groups["second_fraction"],
+            tzinfo=tz,
+        )
+    except Exception as e:
+        raise ParseError(e)

build/plugin/src/resources/libraries/iso8601/test_iso8601.py

@@ -0,0 +1,98 @@
+# coding=UTF-8
+from __future__ import absolute_import
+
+import copy
+import datetime
+import pickle
+
+import pytest
+
+from iso8601 import iso8601
+
+def test_iso8601_regex():
+    assert iso8601.ISO8601_REGEX.match("2006-10-11T00:14:33Z")
+
+def test_parse_no_timezone_different_default():
+    tz = iso8601.FixedOffset(2, 0, "test offset")
+    d = iso8601.parse_date("2007-01-01T08:00:00", default_timezone=tz)
+    assert d == datetime.datetime(2007, 1, 1, 8, 0, 0, 0, tz)
+    assert d.tzinfo == tz
+
+def test_parse_utc_different_default():
+    """Z should mean 'UTC', not 'default'.
+
+    """
+    tz = iso8601.FixedOffset(2, 0, "test offset")
+    d = iso8601.parse_date("2007-01-01T08:00:00Z", default_timezone=tz)
+    assert d == datetime.datetime(2007, 1, 1, 8, 0, 0, 0, iso8601.UTC)
+
+@pytest.mark.parametrize("invalid_date, error_string", [
+    ("2013-10-", "Unable to parse date string"),
+    ("2013-", "Unable to parse date string"),
+    ("", "Unable to parse date string"),
+    (None, "Expecting a string"),
+    ("wibble", "Unable to parse date string"),
+    ("23", "Unable to parse date string"),
+    ("131015T142533Z", "Unable to parse date string"),
+    ("131015", "Unable to parse date string"),
+    ("20141", "Unable to parse date string"),
+    ("201402", "Unable to parse date string"),
+    ("2007-06-23X06:40:34.00Z", "Unable to parse date string"),  # https://code.google.com/p/pyiso8601/issues/detail?id=14
+    ("2007-06-23 06:40:34.00Zrubbish", "Unable to parse date string"),  # https://code.google.com/p/pyiso8601/issues/detail?id=14
+    ("20114-01-03T01:45:49", "Unable to parse date string"),
+])
+def test_parse_invalid_date(invalid_date, error_string):
+    assert isinstance(invalid_date, str) or invalid_date is None  # Why? 'cos I've screwed up the parametrize before :)
+    with pytest.raises(iso8601.ParseError) as exc:
+        iso8601.parse_date(invalid_date)
+    assert exc.errisinstance(iso8601.ParseError)
+    assert str(exc.value).startswith(error_string)
+
+@pytest.mark.parametrize("valid_date,expected_datetime,isoformat", [
+    ("2007-06-23 06:40:34.00Z", datetime.datetime(2007, 6, 23, 6, 40, 34, 0, iso8601.UTC), "2007-06-23T06:40:34+00:00"),  # Handle a separator other than T
+    ("1997-07-16T19:20+01:00", datetime.datetime(1997, 7, 16, 19, 20, 0, 0, iso8601.FixedOffset(1, 0, "+01:00")), "1997-07-16T19:20:00+01:00"),  # Parse with no seconds
+    ("2007-01-01T08:00:00", datetime.datetime(2007, 1, 1, 8, 0, 0, 0, iso8601.UTC), "2007-01-01T08:00:00+00:00"),  # Handle timezone-less dates. Assumes UTC. http://code.google.com/p/pyiso8601/issues/detail?id=4
+    ("2006-10-20T15:34:56.123+02:30", datetime.datetime(2006, 10, 20, 15, 34, 56, 123000, iso8601.FixedOffset(2, 30, "+02:30")), None),
+    ("2006-10-20T15:34:56Z", datetime.datetime(2006, 10, 20, 15, 34, 56, 0, iso8601.UTC), "2006-10-20T15:34:56+00:00"),
+    ("2007-5-7T11:43:55.328Z", datetime.datetime(2007, 5, 7, 11, 43, 55, 328000, iso8601.UTC), "2007-05-07T11:43:55.328000+00:00"),  # http://code.google.com/p/pyiso8601/issues/detail?id=6
+    ("2006-10-20T15:34:56.123Z", datetime.datetime(2006, 10, 20, 15, 34, 56, 123000, iso8601.UTC), "2006-10-20T15:34:56.123000+00:00"),
+    ("2013-10-15T18:30Z", datetime.datetime(2013, 10, 15, 18, 30, 0, 0, iso8601.UTC), "2013-10-15T18:30:00+00:00"),
+    ("2013-10-15T22:30+04", datetime.datetime(2013, 10, 15, 22, 30, 0, 0, iso8601.FixedOffset(4, 0, "+04:00")), "2013-10-15T22:30:00+04:00"),  # <time>±hh:mm
+    ("2013-10-15T1130-0700", datetime.datetime(2013, 10, 15, 11, 30, 0, 0, iso8601.FixedOffset(-7, 0, "-07:00")), "2013-10-15T11:30:00-07:00"),  # <time>±hhmm
+    ("2013-10-15T1130+0700", datetime.datetime(2013, 10, 15, 11, 30, 0, 0, iso8601.FixedOffset(+7, 0, "+07:00")), "2013-10-15T11:30:00+07:00"),  # <time>±hhmm
+    ("2013-10-15T1130+07", datetime.datetime(2013, 10, 15, 11, 30, 0, 0, iso8601.FixedOffset(+7, 0, "+07:00")), "2013-10-15T11:30:00+07:00"),  # <time>±hh
+    ("2013-10-15T1130-07", datetime.datetime(2013, 10, 15, 11, 30, 0, 0, iso8601.FixedOffset(-7, 0, "-07:00")), "2013-10-15T11:30:00-07:00"),  # <time>±hh
+    ("2013-10-15T15:00-03:30", datetime.datetime(2013, 10, 15, 15, 0, 0, 0, iso8601.FixedOffset(-3, -30, "-03:30")), "2013-10-15T15:00:00-03:30"),
+    ("2013-10-15T183123Z", datetime.datetime(2013, 10, 15, 18, 31, 23, 0, iso8601.UTC), "2013-10-15T18:31:23+00:00"),  # hhmmss
+    ("2013-10-15T1831Z", datetime.datetime(2013, 10, 15, 18, 31, 0, 0, iso8601.UTC), "2013-10-15T18:31:00+00:00"),  # hhmm
+    ("2013-10-15T18Z", datetime.datetime(2013, 10, 15, 18, 0, 0, 0, iso8601.UTC), "2013-10-15T18:00:00+00:00"),  # hh
+    ("2013-10-15", datetime.datetime(2013, 10, 15, 0, 0, 0, 0, iso8601.UTC), "2013-10-15T00:00:00+00:00"),  # YYYY-MM-DD
+    ("20131015T18:30Z", datetime.datetime(2013, 10, 15, 18, 30, 0, 0, iso8601.UTC), "2013-10-15T18:30:00+00:00"),  # YYYYMMDD
+    ("2012-12-19T23:21:28.512400+00:00", datetime.datetime(2012, 12, 19, 23, 21, 28, 512400, iso8601.FixedOffset(0, 0, "+00:00")), "2012-12-19T23:21:28.512400+00:00"),  # https://code.google.com/p/pyiso8601/issues/detail?id=21
+    ("2006-10-20T15:34:56.123+0230", datetime.datetime(2006, 10, 20, 15, 34, 56, 123000, iso8601.FixedOffset(2, 30, "+02:30")), "2006-10-20T15:34:56.123000+02:30"),  # https://code.google.com/p/pyiso8601/issues/detail?id=18
+    ("19950204", datetime.datetime(1995, 2, 4, tzinfo=iso8601.UTC), "1995-02-04T00:00:00+00:00"),  # https://code.google.com/p/pyiso8601/issues/detail?id=1
+    ("2010-07-20 15:25:52.520701+00:00", datetime.datetime(2010, 7, 20, 15, 25, 52, 520701, iso8601.FixedOffset(0, 0, "+00:00")), "2010-07-20T15:25:52.520701+00:00"),  # https://code.google.com/p/pyiso8601/issues/detail?id=17
+    ("2010-06-12", datetime.datetime(2010, 6, 12, tzinfo=iso8601.UTC), "2010-06-12T00:00:00+00:00"),  # https://code.google.com/p/pyiso8601/issues/detail?id=16
+    ("1985-04-12T23:20:50.52-05:30", datetime.datetime(1985, 4, 12, 23, 20, 50, 520000, iso8601.FixedOffset(-5, -30, "-05:30")), "1985-04-12T23:20:50.520000-05:30"),  # https://bitbucket.org/micktwomey/pyiso8601/issue/8/015-parses-negative-timezones-incorrectly
+    ("1997-08-29T06:14:00.000123Z", datetime.datetime(1997, 8, 29, 6, 14, 0, 123, iso8601.UTC), "1997-08-29T06:14:00.000123+00:00"),  # https://bitbucket.org/micktwomey/pyiso8601/issue/9/regression-parsing-microseconds
+    ("2014-02", datetime.datetime(2014, 2, 1, 0, 0, 0, 0, iso8601.UTC), "2014-02-01T00:00:00+00:00"),  # https://bitbucket.org/micktwomey/pyiso8601/issue/14/regression-yyyy-mm-no-longer-parses
+    ("2014", datetime.datetime(2014, 1, 1, 0, 0, 0, 0, iso8601.UTC), "2014-01-01T00:00:00+00:00"),  # YYYY
+    ("1997-08-29T06:14:00,000123Z", datetime.datetime(1997, 8, 29, 6, 14, 0, 123, iso8601.UTC), "1997-08-29T06:14:00.000123+00:00"),  # Use , as decimal separator
+])
+def test_parse_valid_date(valid_date, expected_datetime, isoformat):
+    parsed = iso8601.parse_date(valid_date)
+    assert parsed.year == expected_datetime.year
+    assert parsed.month == expected_datetime.month
+    assert parsed.day == expected_datetime.day
+    assert parsed.hour == expected_datetime.hour
+    assert parsed.minute == expected_datetime.minute
+    assert parsed.second == expected_datetime.second
+    assert parsed.microsecond == expected_datetime.microsecond
+    assert parsed.tzinfo == expected_datetime.tzinfo
+    assert parsed == expected_datetime
+    assert parsed.isoformat() == expected_datetime.isoformat()
+    copy.deepcopy(parsed)  # ensure it's deep copy-able
+    pickle.dumps(parsed)  # ensure it pickles
+    if isoformat:
+        assert parsed.isoformat() == isoformat
+    assert iso8601.parse_date(parsed.isoformat()) == parsed  # Test round trip

build/plugin/src/resources/libraries/m3u8/__init__.py

@@ -0,0 +1,75 @@
+# coding: utf-8
+# Copyright 2014 Globo.com Player authors. All rights reserved.
+# Use of this source code is governed by a MIT License
+# license that can be found in the LICENSE file.
+
+import sys
+PYTHON_MAJOR_VERSION = sys.version_info
+
+import os
+import posixpath
+
+try:
+    import urlparse as url_parser
+    import urllib2
+    urlopen = urllib2.urlopen
+except ImportError:
+    import urllib.parse as url_parser
+    from urllib.request import urlopen as url_opener
+    urlopen = url_opener
+
+
+from m3u8.model import M3U8, Playlist, IFramePlaylist, Media, Segment
+from m3u8.parser import parse, is_url
+
+__all__ = ('M3U8', 'Playlist', 'IFramePlaylist', 'Media',
+           'Segment', 'loads', 'load', 'parse')
+
+def loads(content):
+    '''
+    Given a string with a m3u8 content, returns a M3U8 object.
+    Raises ValueError if invalid content
+    '''
+    return M3U8(content)
+
+def load(uri):
+    '''
+    Retrieves the content from a given URI and returns a M3U8 object.
+    Raises ValueError if invalid content or IOError if request fails.
+    '''
+    if is_url(uri):
+        return _load_from_uri(uri)
+    else:
+        return _load_from_file(uri)
+
+# Support for python3 inspired by https://github.com/szemtiv/m3u8/
+def _load_from_uri(uri):
+    resource = urlopen(uri)
+    base_uri = _parsed_url(_url_for(uri))
+    if PYTHON_MAJOR_VERSION < (3,):
+        content = _read_python2x(resource)
+    else:
+        content = _read_python3x(resource)
+    return M3U8(content, base_uri=base_uri)
+
+def _url_for(uri):
+    return urlopen(uri).geturl()
+
+def _parsed_url(url):
+    parsed_url = url_parser.urlparse(url)
+    prefix = parsed_url.scheme + '://' + parsed_url.netloc
+    base_path = posixpath.normpath(parsed_url.path + '/..')
+    return url_parser.urljoin(prefix, base_path)
+
+def _read_python2x(resource):
+    return resource.read().strip()
+
+def _read_python3x(resource):
+    return  resource.read().decode(resource.headers.get_content_charset(failobj="utf-8"))
+
+def _load_from_file(uri):
+    with open(uri) as fileobj:
+        raw_content = fileobj.read().strip()
+    base_uri = os.path.dirname(uri)
+    return M3U8(raw_content, base_uri=base_uri)
+