src/librarian/picture.py

   1 # -*- coding: utf-8 -*-
   2 from __future__ import unicode_literals
   3
   4 from operator import and_
   5
   6 from .dcparser import Field, WorkInfo, DCNS
   7 from librarian import (RDFNS, ValidationError, NoDublinCore, ParseError, WLURI)
   8 from xml.parsers.expat import ExpatError
   9 from os import path
  10 from lxml import etree
  11 from lxml.etree import (XMLSyntaxError, XSLTApplyError, Element)
  12 import re
  13 import six
  14
  15
  16 class WLPictureURI(WLURI):
  17     _re_wl_uri = re.compile(
  18         'http://wolnelektury.pl/katalog/obraz/(?P<slug>[-a-z0-9]+)/?$'
  19     )
  20
  21     @classmethod
  22     def from_slug(cls, slug):
  23         uri = 'http://wolnelektury.pl/katalog/obraz/%s/' % slug
  24         return cls(uri)
  25
  26
  27 def as_wlpictureuri_strict(text):
  28     return WLPictureURI.strict(text)
  29
  30
  31 class PictureInfo(WorkInfo):
  32     """
  33     Dublin core metadata for a picture
  34     """
  35     FIELDS = (
  36         Field(DCNS('language'), 'language', required=False),
  37         Field(DCNS('subject.period'), 'epochs', salias='epoch', multiple=True),
  38         Field(DCNS('subject.type'), 'kinds', salias='kind', multiple=True),
  39         Field(DCNS('subject.genre'), 'genres', salias='genre', multiple=True,
  40               required=False),
  41         Field(DCNS('subject.style'), 'styles', salias='style', multiple=True,
  42               required=False),
  43
  44         Field(DCNS('format.dimensions'), 'dimensions', required=False),
  45         Field(DCNS('format.checksum.sha1'), 'sha1', required=True),
  46         Field(DCNS('description.medium'), 'medium', required=False),
  47         Field(DCNS('description.dimensions'), 'original_dimensions',
  48               required=False),
  49         Field(DCNS('format'), 'mime_type', required=False),
  50         Field(DCNS('identifier.url'), 'url', WLPictureURI,
  51               strict=as_wlpictureuri_strict)
  52     )
  53
  54
  55 class ImageStore(object):
  56     EXT = ['gif', 'jpeg', 'png', 'swf', 'psd', 'bmp'
  57            'tiff', 'tiff', 'jpc', 'jp2', 'jpf', 'jb2', 'swc',
  58            'aiff', 'wbmp', 'xbm']
  59     MIME = ['image/gif', 'image/jpeg', 'image/png',
  60             'application/x-shockwave-flash', 'image/psd', 'image/bmp',
  61             'image/tiff', 'image/tiff', 'application/octet-stream',
  62             'image/jp2', 'application/octet-stream',
  63             'application/octet-stream', 'application/x-shockwave-flash',
  64             'image/iff', 'image/vnd.wap.wbmp', 'image/xbm']
  65
  66     def __init__(self, dir_):
  67         super(ImageStore, self).__init__()
  68         self.dir = dir_
  69
  70     def path(self, slug, mime_type):
  71         """
  72         Finds file by slug and mime type in our iamge store.
  73         Returns a file objects (perhaps should return a filename?)
  74         """
  75         try:
  76             i = self.MIME.index(mime_type)
  77         except ValueError:
  78             err = ValueError(
  79                 "Picture %s has unknown mime type: %s"
  80                 % (slug, mime_type)
  81             )
  82             err.slug = slug
  83             err.mime_type = mime_type
  84             raise err
  85         ext = self.EXT[i]
  86         # add some common extensions tiff->tif, jpeg->jpg
  87         return path.join(self.dir, slug + '.' + ext)
  88
  89
  90 class WLPicture(object):
  91     def __init__(self, edoc, parse_dublincore=True, image_store=None):
  92         self.edoc = edoc
  93         self.image_store = image_store
  94
  95         root_elem = edoc.getroot()
  96
  97         dc_path = './/' + RDFNS('RDF')
  98
  99         if root_elem.tag != 'picture':
 100             raise ValidationError(
 101                 "Invalid root element. Found '%s', should be 'picture'"
 102                 % root_elem.tag
 103             )
 104
 105         if parse_dublincore:
 106             self.rdf_elem = root_elem.find(dc_path)
 107
 108             if self.rdf_elem is None:
 109                 raise NoDublinCore(
 110                     "Document has no DublinCore - which is required."
 111                 )
 112
 113             self.picture_info = PictureInfo.from_element(self.rdf_elem)
 114         else:
 115             self.picture_info = None
 116         self.frame = None
 117
 118     @classmethod
 119     def from_bytes(cls, xml, *args, **kwargs):
 120         return cls.from_file(six.BytesIO(xml), *args, **kwargs)
 121
 122     @classmethod
 123     def from_file(cls, xmlfile, parse_dublincore=True, image_store=None):
 124
 125         # first, prepare for parsing
 126         if isinstance(xmlfile, six.text_type):
 127             file = open(xmlfile, 'rb')
 128             try:
 129                 data = file.read()
 130             finally:
 131                 file.close()
 132         else:
 133             data = xmlfile.read()
 134
 135         if not isinstance(data, six.text_type):
 136             data = data.decode('utf-8')
 137
 138         data = data.replace(u'\ufeff', '')
 139
 140         # assume images are in the same directory
 141         if image_store is None and getattr(xmlfile, 'name', None):
 142             image_store = ImageStore(path.dirname(xmlfile.name))
 143
 144         try:
 145             parser = etree.XMLParser(remove_blank_text=False)
 146             tree = etree.parse(six.BytesIO(data.encode('utf-8')), parser)
 147
 148             me = cls(tree, parse_dublincore=parse_dublincore,
 149                      image_store=image_store)
 150             me.load_frame_info()
 151             return me
 152         except (ExpatError, XMLSyntaxError, XSLTApplyError) as e:
 153             raise ParseError(e)
 154
 155     @property
 156     def mime_type(self):
 157         if self.picture_info is None:
 158             raise ValueError(
 159                 "DC is not loaded, hence we don't know the image type."
 160             )
 161         return self.picture_info.mime_type
 162
 163     @property
 164     def slug(self):
 165         return self.picture_info.url.slug
 166
 167     @property
 168     def image_path(self):
 169         if self.image_store is None:
 170             raise ValueError("No image store associated with whis WLPicture.")
 171
 172         return self.image_store.path(self.slug, self.mime_type)
 173
 174     def image_file(self, *args, **kwargs):
 175         return open(self.image_path, 'rb', *args, **kwargs)
 176
 177     def get_sem_coords(self, sem):
 178         area = sem.find("div[@type='rect']")
 179         if area is None:
 180             area = sem.find("div[@type='whole']")
 181             return [[0, 0], [-1, -1]]
 182
 183         def has_all_props(node, props):
 184             return reduce(and_, map(lambda prop: prop in node.attrib, props))
 185
 186         if not has_all_props(area, ['x1', 'x2', 'y1', 'y2']):
 187             return None
 188
 189         def n(prop): return int(area.get(prop))
 190         return [[n('x1'), n('y1')], [n('x2'), n('y2')]]
 191
 192     def partiter(self):
 193         """
 194         Iterates the parts of this picture and returns them
 195         and their metadata.
 196         """
 197         # omg no support for //sem[(@type='theme') or (@type='object')] ?
 198         for part in list(self.edoc.iterfind("//sem[@type='theme']")) +\
 199                 list(self.edoc.iterfind("//sem[@type='object']")):
 200             pd = {'type': part.get('type')}
 201
 202             coords = self.get_sem_coords(part)
 203             if coords is None:
 204                 continue
 205             pd['coords'] = coords
 206
 207             def want_unicode(x):
 208                 if not isinstance(x, six.text_type):
 209                     return x.decode('utf-8')
 210                 else:
 211                     return x
 212             pd['object'] = (
 213                 part.attrib['type'] == 'object'
 214                 and want_unicode(part.attrib.get('object', u''))
 215                 or None
 216             )
 217             pd['themes'] = (
 218                 part.attrib['type'] == 'theme'
 219                 and [part.attrib.get('theme', u'')]
 220                 or []
 221             )
 222             yield pd
 223
 224     def load_frame_info(self):
 225         k = self.edoc.find("//sem[@object='kadr']")
 226
 227         if k is not None:
 228             clip = self.get_sem_coords(k)
 229             self.frame = clip
 230             frm = Element("sem", {"type": "frame"})
 231             frm.append(k.iter("div").next())
 232             self.edoc.getroot().append(frm)
 233             k.getparent().remove(k)
 234         else:
 235             frm = self.edoc.find("//sem[@type='frame']")
 236             if frm:
 237                 self.frame = self.get_sem_coords(frm)
 238             else:
 239                 self.frame = None
 240         return self