7830528d120fa0fda020e88d6fafd98d2cae67b2
[librarian.git] / librarian / picture.py
1
2 from dcparser import (as_person, as_date, Field, WorkInfo, DCNS)
3 from librarian import (RDFNS, ValidationError, NoDublinCore, ParseError, WLURI)
4 from xml.parsers.expat import ExpatError
5 from os import path
6 from StringIO import StringIO
7 from lxml import etree
8 from lxml.etree import (XMLSyntaxError, XSLTApplyError)
9 import re
10
11
12 class WLPictureURI(WLURI):
13     _re_wl_uri = re.compile('http://wolnelektury.pl/katalog/obraz/'
14             '(?P<slug>[-a-z0-9]+)(/(?P<lang>[a-z]{3}))?/?$')
15
16     def __init__(self, *args, **kw):
17         super(WLPictureURI, self).__init__(*args, **kw)
18
19     @classmethod
20     def from_slug_and_lang(cls, slug, lang):
21         uri = 'http://wolnelektury.pl/katalog/obraz/%s/' % slug
22         return cls(uri)
23
24     def filename_stem(self):
25         return self.slug
26
27
28 class PictureInfo(WorkInfo):
29     """
30     Dublin core metadata for a picture
31     """
32     FIELDS = (
33         Field(DCNS('format.dimensions.digital'), 'dimensions', required=False),
34         Field(DCNS('format.dimensions.original'), 'dimensions_original', required=False),
35         Field(DCNS('format.physical'), 'physical', required=False),
36         Field(DCNS('format'), 'mime_type', required=False),
37         Field(DCNS('identifier.url'), 'url', WLPictureURI),
38         )
39
40     def validate(self):
41         """
42         WorkInfo has a language validation code only, which we do not need.
43         """
44         pass
45     
46
47 class ImageStore(object):
48     EXT = ['gif', 'jpeg', 'png', 'swf', 'psd', 'bmp'
49             'tiff', 'tiff', 'jpc', 'jp2', 'jpf', 'jb2', 'swc',
50             'aiff', 'wbmp', 'xbm']
51     MIME = ['image/gif', 'image/jpeg', 'image/png',
52             'application/x-shockwave-flash', 'image/psd', 'image/bmp',
53             'image/tiff', 'image/tiff', 'application/octet-stream',
54             'image/jp2', 'application/octet-stream', 'application/octet-stream',
55             'application/x-shockwave-flash', 'image/iff', 'image/vnd.wap.wbmp', 'image/xbm']
56
57     def __init__(self, dir_):
58         self.dir = dir_
59         return super(ImageStore, self).__init__()
60
61     def path(self, slug, mime_type):
62         """
63         Finds file by slug and mime type in our iamge store.
64         Returns a file objects (perhaps should return a filename?)
65         """
66         try:
67             i = self.MIME.index(mime_type)
68         except ValueError:
69             err = ValueError("Picture %s has unknown mime type: %s" % (slug, mime_type))
70             err.slug = slug
71             err.mime_type = mime_type
72             raise err
73         ext = self.EXT[i]
74         # add some common extensions tiff->tif, jpeg->jpg
75         return path.join(self.dir, slug + '.' + ext)
76
77
78 class WLPicture(object):
79     def __init__(self, edoc, parse_dublincore=True, image_store=None):
80         self.edoc = edoc
81         self.image_store = image_store
82
83         root_elem = edoc.getroot()
84
85         dc_path = './/' + RDFNS('RDF')
86
87         if root_elem.tag != 'picture':
88             raise ValidationError("Invalid root element. Found '%s', should be 'picture'" % root_elem.tag)
89
90         if parse_dublincore:
91             self.rdf_elem = root_elem.find(dc_path)
92
93             if self.rdf_elem is None:
94                 raise NoDublinCore('Document has no DublinCore - which is required.')
95
96             self.picture_info = PictureInfo.from_element(self.rdf_elem)
97         else:
98             self.picture_info = None
99
100     @classmethod
101     def from_string(cls, xml, *args, **kwargs):
102         return cls.from_file(StringIO(xml), *args, **kwargs)
103
104     @classmethod
105     def from_file(cls, xmlfile, parse_dublincore=True, image_store=None):
106
107         # first, prepare for parsing
108         if isinstance(xmlfile, basestring):
109             file = open(xmlfile, 'rb')
110             try:
111                 data = file.read()
112             finally:
113                 file.close()
114         else:
115             data = xmlfile.read()
116
117         if not isinstance(data, unicode):
118             data = data.decode('utf-8')
119
120         data = data.replace(u'\ufeff', '')
121
122         # assume images are in the same directory
123         if image_store is None and xmlfile.name is not None:
124             image_store = ImageStore(path.dirname(xmlfile.name))
125
126         try:
127             parser = etree.XMLParser(remove_blank_text=False)
128             tree = etree.parse(StringIO(data.encode('utf-8')), parser)
129
130             return cls(tree, parse_dublincore=parse_dublincore, image_store=image_store)
131         except (ExpatError, XMLSyntaxError, XSLTApplyError), e:
132             raise ParseError(e)
133
134     @property
135     def mime_type(self):
136         if self.picture_info is None:
137             raise ValueError('DC is not loaded, hence we don\'t know the image type')
138         return self.picture_info.mime_type
139
140     @property
141     def slug(self):
142         return self.picture_info.url.slug
143
144     @property
145     def image_path(self):
146         if self.image_store is None:
147             raise ValueError("No image store associated with whis WLPicture.")
148         return self.image_store.path(self.slug, self.mime_type)
149
150     def image_file(self, *args, **kwargs):
151         return open(self.image_path, *args, **kwargs)