- return self.slug, self.language == other.slug, other.language
-
- def filename_stem(self):
- stem = self.slug
- if self.language != self.DEFAULT_LANGUAGE:
- stem += '_' + self.language
- return stem
-
- def validate_language(self, language):
- if language != self.language:
- raise ValidationError("Incorrect language definition in URI")
-
-
-class DocProvider(object):
- """Base class for a repository of XML files.
-
- Used for generating joined files, like EPUBs.
- """
-
- def by_slug_and_lang(self, slug, lang=None):
- """Should return a file-like object with a WL document XML."""
- raise NotImplementedError
-
- def by_slug(self, slug):
- """Should return a file-like object with a WL document XML."""
- return self.by_slug_and_lang(slug)
-
- def by_uri(self, uri):
- """Should return a file-like object with a WL document XML."""
- wluri = WLURI(uri)
- return self.by_slug_and_lang(wluri.slug, wluri.language)
-
-
-class DirDocProvider(DocProvider):
- """ Serve docs from a directory of files in form <slug>.xml """
-
- def __init__(self, dir_):
- self.dir = dir_
- self.files = {}
-
- def by_slug_and_lang(self, slug, lang=None):
- fname = WLURI.from_slug_and_lang(slug, lang).filename_stem() + '.xml'
- return open(os.path.join(self.dir, fname))
-
-
-import lxml.etree as etree
-import dcparser
-
-DEFAULT_BOOKINFO = dcparser.BookInfo(
- { RDFNS('about'): u'http://wiki.wolnepodreczniki.pl/Lektury:Template'},
- { DCNS('creator'): [u'Some, Author'],
- DCNS('title'): [u'Some Title'],
- DCNS('subject.period'): [u'Unknown'],
- DCNS('subject.type'): [u'Unknown'],
- DCNS('subject.genre'): [u'Unknown'],
- DCNS('date'): ['1970-01-01'],
- DCNS('language'): [WLURI.DEFAULT_LANGUAGE],
- # DCNS('date'): [creation_date],
- DCNS('publisher'): [u"Fundacja Nowoczesna Polska"],
- DCNS('description'):
- [u"""Publikacja zrealizowana w ramach projektu
- Wolne Lektury (http://wolnelektury.pl). Reprodukcja cyfrowa
- wykonana przez Bibliotekę Narodową z egzemplarza
- pochodzącego ze zbiorów BN."""],
- DCNS('identifier.url'): [WLURI.example],
- DCNS('rights'):
- [u"Domena publiczna - zm. [OPIS STANU PRAWNEGO TEKSTU]"] })
-
-def xinclude_forURI(uri):
- e = etree.Element(XINS("include"))
- e.set("href", uri)
- return etree.tostring(e, encoding=unicode)
-
-def wrap_text(ocrtext, creation_date, bookinfo=DEFAULT_BOOKINFO):
- """Wrap the text within the minimal XML structure with a DC template."""
- bookinfo.created_at = creation_date
-
- dcstring = etree.tostring(bookinfo.to_etree(), \
- method='xml', encoding=unicode, pretty_print=True)
-
- return u'<utwor>\n' + dcstring + u'\n<plain-text>\n' + ocrtext + \
- u'\n</plain-text>\n</utwor>'
-
-
-def serialize_raw(element):
- b = u'' + (element.text or '')
-
- for child in element.iterchildren():
- e = etree.tostring(child, method='xml', encoding=unicode,
- pretty_print=True)
- b += e
-
- return b
-
-SERIALIZERS = {
- 'raw': serialize_raw,
-}
-
-def serialize_children(element, format='raw'):
- return SERIALIZERS[format](element)
-
-def get_resource(path):
- return os.path.join(os.path.dirname(__file__), path)
-
-
-class OutputFile(object):
- """Represents a file returned by one of the converters."""
-
- _string = None
- _filename = None
-
- def __del__(self):
- if self._filename:
- os.unlink(self._filename)
-
- def __nonzero__(self):
- return self._string is not None or self._filename is not None
-
- @classmethod
- def from_string(cls, string):
- """Converter returns contents of a file as a string."""
-
- instance = cls()
- instance._string = string
- return instance
-
- @classmethod
- def from_filename(cls, filename):
- """Converter returns contents of a file as a named file."""
-
- instance = cls()
- instance._filename = filename
- return instance
-
- def get_string(self):
- """Get file's contents as a string."""
-
- if self._filename is not None:
- with open(self._filename) as f:
- return f.read()
- else:
- return self._string
-
- def get_file(self):
- """Get file as a file-like object."""
-
- if self._string is not None:
- from StringIO import StringIO
- return StringIO(self._string)
- elif self._filename is not None:
- return open(self._filename)
-
- def get_filename(self):
- """Get file as a fs path."""
-
- if self._filename is not None:
- return self._filename
- elif self._string is not None:
- from tempfile import NamedTemporaryFile
- temp = NamedTemporaryFile(prefix='librarian-', delete=False)
- temp.write(self._string)
- temp.close()
- self._filename = temp.name
- return self._filename
- else:
- return None
fnp / librarian.git / blobdiff