Merge pull request #484 from guessit-io/feature/size-output

Enhance size output value / Detect bit rate

HISTORY.rst

@@ -97,6 +97,10 @@ History
 - Added default and configurable list of allowed languages and countries
 - Added `VC-1` as new `video_codec` possible value
 - Enhanced dash-separated `release_group` detection.
+- Changed `size` output to return `guessit.Quantity` object.
+- Changed `size` output to return `guessit.Size` object.
+- Added `audio_video_rate` as new possible property.
+- Added `video_video_rate` as new possible property.
 
 2.1.4 (2017-06-01)
 ------------------

docs/properties.rst

@@ -180,6 +180,13 @@ Video properties
   - ``DXVA``
 
 
+- **video_bit_rate**
+
+  Video bit rate (Mbps). Examples: ``25Mbps`` (``<BitRate [25Mbps]>``), ``40Mbps`` (``<BitRate [40Mbps]>``).
+
+  - ``[<guessit.BitRate>]`` (object has ``magnitude`` and ``units``)
+
+
 Audio properties
 ----------------
 
@@ -206,6 +213,13 @@ Audio properties
     ``Master Audio``
 
 
+- **audio_bit_rate**
+
+  Audio bit rate (Kbps, Mbps). Examples: ``448Kbps`` (``<BitRate [448Kbps]>``), ``1.5Mbps`` (``<BitRate [1.5Mbps]>``).
+
+  - ``[<guessit.BitRate>]`` (object has ``magnitude`` and ``units``)
+
+
 Localization properties
 -----------------------
 
@@ -265,7 +279,9 @@ Other properties
 
 - **size**
 
-  Size (MB, GB, TB).
+  Size (MB, GB, TB). Examples: ``1.2GB`` (``<Size [1.2GB]>``), ``430MB`` (``<Size [430MB]>``).
+
+  - ``[<guessit.Size>]`` (object has ``magnitude`` and ``units``)
 
 
 - **edition**

guessit/__init__.py

@@ -5,5 +5,6 @@ Extracts as much information as possible from a video file.
 """
 from .api import guessit, GuessItApi
 from .options import ConfigurationException
+from .rules.common.quantity import Size
 
 from .__version__ import __version__

guessit/rules/__init__.py

@@ -24,6 +24,7 @@ from .properties.release_group import release_group
 from .properties.streaming_service import streaming_service
 from .properties.other import other
 from .properties.size import size
+from .properties.bit_rate import bit_rate
 from .properties.edition import edition
 from .properties.cds import cds
 from .properties.bonus import bonus
@@ -63,6 +64,7 @@ def rebulk_builder():
     rebulk.rebulk(streaming_service())
     rebulk.rebulk(other())
     rebulk.rebulk(size())
+    rebulk.rebulk(bit_rate())
     rebulk.rebulk(edition())
     rebulk.rebulk(cds())
     rebulk.rebulk(bonus())

guessit/rules/common/quantity.py

@@ -0,0 +1,90 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+"""
+Quantities: Size
+"""
+import re
+from abc import abstractmethod
+
+import six
+
+from ..common import seps
+
+
+class Quantity(object):
+    """
+    Represent a quantity object with magnitude and units.
+    """
+
+    parser_re = re.compile(r'(?P<magnitude>\d+(?:[.]\d+)?)(?P<units>[^\d]+)')
+
+    def __init__(self, magnitude, units):
+        self.magnitude = magnitude
+        self.units = units
+
+    @classmethod
+    @abstractmethod
+    def parse_units(cls, value):
+        """
+        Parse a string to a proper unit notation.
+        """
+        raise NotImplementedError
+
+    @classmethod
+    def fromstring(cls, string):
+        """
+        Parse the string into a quantity object.
+        :param string:
+        :return:
+        """
+        values = cls.parser_re.match(string).groupdict()
+        try:
+            magnitude = int(values['magnitude'])
+        except ValueError:
+            magnitude = float(values['magnitude'])
+        units = cls.parse_units(values['units'])
+
+        return cls(magnitude, units)
+
+    def __hash__(self):
+        return hash(str(self))
+
+    def __eq__(self, other):
+        if isinstance(other, six.string_types):
+            return str(self) == other
+        if not isinstance(other, self.__class__):
+            return NotImplemented
+        return self.magnitude == other.magnitude and self.units == other.units
+
+    def __ne__(self, other):
+        return not self == other
+
+    def __repr__(self):
+        return '<{0} [{1}]>'.format(self.__class__.__name__, self)
+
+    def __str__(self):
+        return '{0}{1}'.format(self.magnitude, self.units)
+
+
+class Size(Quantity):
+    """
+    Represent size.
+
+    e.g.: 1.1GB, 300MB
+    """
+
+    @classmethod
+    def parse_units(cls, value):
+        return value.strip(seps).upper()
+
+
+class BitRate(Quantity):
+    """
+    Represent bit rate.
+
+    e.g.: 320Kbps, 1.5Mbps
+    """
+
+    @classmethod
+    def parse_units(cls, value):
+        return value.strip(seps).capitalize()

guessit/rules/properties/bit_rate.py

@@ -0,0 +1,48 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+"""
+video_bit_rate and audio_bit_rate properties
+"""
+import re
+
+from rebulk import Rebulk
+from rebulk.rules import Rule, RenameMatch
+
+from ..common import dash, seps
+from ..common.quantity import BitRate
+from ..common.validators import seps_surround
+
+
+def bit_rate():
+    """
+    Builder for rebulk object.
+    :return: Created Rebulk object
+    :rtype: Rebulk
+    """
+    rebulk = Rebulk().regex_defaults(flags=re.IGNORECASE, abbreviations=[dash])
+    rebulk.defaults(name='audio_bit_rate', validator=seps_surround)
+    rebulk.regex(r'\d+-?[kmg]bps', r'\d+\.\d+-?[kmg]bps',
+                 conflict_solver=(
+                     lambda match, other: match
+                     if other.name == 'audio_channels' and 'weak-audio_channels' not in other.tags
+                     else other
+                 ),
+                 formatter=BitRate.fromstring, tags=['release-group-prefix'])
+
+    rebulk.rules(BitRateTypeRule)
+
+    return rebulk
+
+
+class BitRateTypeRule(Rule):
+    """
+    Convert audio bit rate guess into video bit rate.
+    """
+    consequence = RenameMatch('video_bit_rate')
+
+    def when(self, matches, context):
+        for match in matches.named('audio_bit_rate'):
+            previous = matches.previous(match, index=0,
+                                        predicate=lambda m: m.name in ('source', 'screen_size', 'video_codec'))
+            if previous and not matches.holes(previous.end, match.start, predicate=lambda m: m.value.strip(seps)):
+                yield match

guessit/rules/properties/size.py

@@ -7,8 +7,9 @@ import re
 
 from rebulk import Rebulk
 
-from ..common.validators import seps_surround
 from ..common import dash
+from ..common.quantity import Size
+from ..common.validators import seps_surround
 
 
 def size():
@@ -17,13 +18,8 @@ def size():
     :return: Created Rebulk object
     :rtype: Rebulk
     """
-
-    def format_size(value):
-        """Format size using uppercase and no space."""
-        return re.sub(r'(?<=\d)[.](?=[^\d])', '', value.upper())
-
     rebulk = Rebulk().regex_defaults(flags=re.IGNORECASE, abbreviations=[dash])
     rebulk.defaults(name='size', validator=seps_surround)
-    rebulk.regex(r'\d+\.?[mgt]b', r'\d+\.\d+[mgt]b', formatter=format_size, tags=['release-group-prefix'])
+    rebulk.regex(r'\d+-?[mgt]b', r'\d+\.\d+-?[mgt]b', formatter=Size.fromstring, tags=['release-group-prefix'])
 
     return rebulk

guessit/test/episodes.yml

@@ -4338,3 +4338,14 @@
   episode_title: AKA Sin Bin
   container: mkv
   type: episode
+
+? Hotel.Hell.S01E01.720p.DD5.1.448kbps-ALANiS
+: title: Hotel Hell
+  season: 1
+  episode: 1
+  screen_size: 720p
+  audio_codec: Dolby Digital
+  audio_channels: '5.1'
+  audio_bit_rate: 448Kbps
+  release_group: ALANiS
+  type: episode

guessit/test/various.yml

@@ -864,3 +864,41 @@
   video_codec: H.264
   release_group: SDH
   type: episode
+
+? Katy Perry - Pepsi & Billboard Summer Beats Concert Series 2012 1080i HDTV 20 Mbps DD2.0 MPEG2-TrollHD.ts
+: title: Katy Perry
+  alternative_title: Pepsi & Billboard Summer Beats Concert
+  year: 2012
+  screen_size: 1080i
+  source: HDTV
+  video_bit_rate: 20Mbps
+  audio_codec: Dolby Digital
+  audio_channels: '2.0'
+  video_codec: MPEG-2
+  release_group: TrollHD
+  container: ts
+
+? Justin Timberlake - MTV Video Music Awards 2013 1080i 32 Mbps DTS-HD 5.1.ts
+: title: Justin Timberlake
+  alternative_title: MTV Video Music Awards
+  year: 2013
+  screen_size: 1080i
+  video_bit_rate: 32Mbps
+  audio_codec: DTS-HD
+  audio_channels: '5.1'
+  container: ts
+  type: movie
+
+? Chuck Berry The Very Best Of Chuck Berry(2010)[320 Kbps]
+: title: Chuck Berry The Very Best Of Chuck Berry
+  year: 2010
+  audio_bit_rate: 320Kbps
+  type: movie
+
+? Title Name [480p][1.5Mbps][.mp4]
+: title: Title Name
+  screen_size: 480p
+  video_bit_rate: 1.5Mbps
+  container: mp4
+  type: movie
+