def _float_attrs(self, flt):
buf = list()
if flt.number != u"":
buf.append(u' number="%s"' % escape(flt.number))
if flt.pagenr != u"":
buf.append(u' pagenr="%s"' % escape(flt.pagenr))
return u"".join(buf)
def _float_attrs(self, flt):
buf = list()
if flt.number != u"":
buf.append(u' number="%s"' % escape(flt.number))
if flt.pagenr != u"":
buf.append(u' pagenr="%s"' % escape(flt.pagenr))
return u"".join(buf)
def _section_attrs(self, sec):
buf = list()
if sec.number != u"":
buf.append(u' number="%s"' % escape(sec.number))
buf.append(u' title="%s"' % escape(sec.title))
if sec.pagenr != u"":
buf.append(u' pagenr="%s"' % escape(sec.pagenr))
return u"".join(buf)
def _section_attrs(self, sec):
buf = list()
if sec.number != u"":
buf.append(u' number="%s"' % escape(sec.number))
buf.append(u' title="%s"' % escape(sec.title))
if sec.pagenr != u"":
buf.append(u' pagenr="%s"' % escape(sec.pagenr))
return u"".join(buf)
def _paragraph_attrs(self, para):
buf = list()
if para.pagenr != u"":
buf.append(u' pagenr="%s"' % escape(para.pagenr))
if para.font != u"":
buf.append(u' font="%s"' % escape(para.font))
if para.fontsize != u"":
buf.append(u' fontsize="%s"' % escape(para.fontsize))
if len(para.emph) > 0:
emph_str = EMPH_SEPARATOR.join(para.emph)
if emph_str != u"":
buf.append(u' emph="%s"' % escape(emph_str))
return u"".join(buf)
def _paragraph_attrs(self, para):
buf = list()
if para.pagenr != u"":
buf.append(u' pagenr="%s"' % escape(para.pagenr))
if para.font != u"":
buf.append(u' font="%s"' % escape(para.font))
if para.fontsize != u"":
buf.append(u' fontsize="%s"' % escape(para.fontsize))
if len(para.emph) > 0:
emph_str = EMPH_SEPARATOR.join(para.emph)
if emph_str != u"":
buf.append(u' emph="%s"' % escape(emph_str))
return u"".join(buf)
def to_XML(self, doc, pretty=False, linesep=u"", ident=u"", header=True):
"""Converts a Document to structure oriented XML.
Args:
doc: The Document or part of a document or list of Documents
to convert.
pretty: If true: returns a human readable XML string.
Takes care of parameters linesep and ident automatically.
linesep: Line seperator. Is overridden in case pretty is set.
ident: Ident for each line. Mainly used for recursion.
header: Place the XML header with version and encoding at the
beginning of the XML string.
Return:
Unicode string (XML markup).
"""
if doc == None:
return u""
buf = list()
if header:
buf.append(XML_HEADER)
if pretty:
linesep = u"\n"
if type(doc) == list:
buf.append(u"<documents>")
for d in doc:
buf.append(self.to_XML(d, linesep=linesep, ident=PRETTY_IDENT, header=False))
buf.append(u"</documents>")
elif type(doc) == Document:
buf.append(ident + u"<document>")
if doc.meta:
buf.append(self.to_XML(doc.meta, linesep=linesep, ident=ident + PRETTY_IDENT, header=False))
for c in doc.children():
buf.append(self.to_XML(c, linesep=linesep, ident=ident + PRETTY_IDENT, header=False))
buf.append(ident + u"</document>")
elif type(doc) == Meta:
buf.append(ident + u"<meta>")
if doc.title != u"":
title = u"%s%s<title>%s</title>" % (ident, PRETTY_IDENT, escape(doc.title))
buf.append(title)
for author in doc.authors:
author_xml = u"%s%s<author>%s</author>" % (ident, PRETTY_IDENT, escape(author))
buf.append(author_xml)
if doc.language != u"":
lang = u"%s%s<langauge>%s</language>" % (ident, PRETTY_IDENT, escape(doc.language))
buf.append(lang)
buf.append(ident + u"</meta>")
elif type(doc) == Section:
buf.append(u"%s<section%s>" % (ident, self._section_attrs(doc)))
for c in doc.children():
buf.append(self.to_XML(c, linesep=linesep, ident=ident + PRETTY_IDENT, header=False))
buf.append(ident + u"</section>")
elif type(doc) == Chapter:
buf.append(u"%s<chapter%s>" % (ident, self._section_attrs(doc)))
for c in doc.children():
buf.append(self.to_XML(c, linesep=linesep, ident=ident + PRETTY_IDENT, header=False))
buf.append(ident + u"</chapter>")
elif type(doc) == Paragraph:
buf.append(u"%s<paragraph%s>" % (ident, self._paragraph_attrs(doc)))
buf.append(escape(doc.text))
buf.append(ident + u"</paragraph>")
elif type(doc) == Float:
buf.append(ident + u"%s<float%s>" % (ident, self._float_attrs(doc)))
buf.append(escape(doc.text))
buf.append(ident + u"</float>")
elif type(doc) == Footnote:
buf.append(ident + u"%s<footnote%s>" % (ident, self._float_attrs(doc)))
buf.append(escape(doc.text))
buf.append(ident + u"</footnote>")
return linesep.join(buf)
def to_XML(self, doc, pretty=False, linesep=u"", ident=u"", header=True):
"""Converts a Document to structure oriented XML.
Args:
doc: The Document or part of a document or list of Documents
to convert.
pretty: If true: returns a human readable XML string.
Takes care of parameters linesep and ident automatically.
linesep: Line seperator. Is overridden in case pretty is set.
ident: Ident for each line. Mainly used for recursion.
header: Place the XML header with version and encoding at the
beginning of the XML string.
Return:
Unicode string (XML markup).
"""
if doc == None:
return u""
buf = list()
if header:
buf.append(XML_HEADER)
if pretty:
linesep = u"\n"
if type(doc) == list:
buf.append(u"<documents>")
for d in doc:
buf.append(
self.to_XML(d,
linesep=linesep,
ident=PRETTY_IDENT,
header=False))
buf.append(u"</documents>")
elif type(doc) == Document:
buf.append(ident + u"<document>")
if doc.meta:
buf.append(
self.to_XML(doc.meta,
linesep=linesep,
ident=ident + PRETTY_IDENT,
header=False))
for c in doc.children():
buf.append(
self.to_XML(c,
linesep=linesep,
ident=ident + PRETTY_IDENT,
header=False))
buf.append(ident + u"</document>")
elif type(doc) == Meta:
buf.append(ident + u"<meta>")
if doc.title != u"":
title = u"%s%s<title>%s</title>" % (ident, PRETTY_IDENT,
escape(doc.title))
buf.append(title)
for author in doc.authors:
author_xml = u"%s%s<author>%s</author>" % (ident, PRETTY_IDENT,
escape(author))
buf.append(author_xml)
if doc.language != u"":
lang = u"%s%s<langauge>%s</language>" % (ident, PRETTY_IDENT,
escape(doc.language))
buf.append(lang)
buf.append(ident + u"</meta>")
elif type(doc) == Section:
buf.append(u"%s<section%s>" % (ident, self._section_attrs(doc)))
for c in doc.children():
buf.append(
self.to_XML(c,
linesep=linesep,
ident=ident + PRETTY_IDENT,
header=False))
buf.append(ident + u"</section>")
elif type(doc) == Chapter:
buf.append(u"%s<chapter%s>" % (ident, self._section_attrs(doc)))
for c in doc.children():
buf.append(
self.to_XML(c,
linesep=linesep,
ident=ident + PRETTY_IDENT,
header=False))
buf.append(ident + u"</chapter>")
elif type(doc) == Paragraph:
buf.append(u"%s<paragraph%s>" %
(ident, self._paragraph_attrs(doc)))
buf.append(escape(doc.text))
buf.append(ident + u"</paragraph>")
elif type(doc) == Float:
buf.append(ident + u"%s<float%s>" %
(ident, self._float_attrs(doc)))
buf.append(escape(doc.text))
buf.append(ident + u"</float>")
elif type(doc) == Footnote:
buf.append(ident + u"%s<footnote%s>" %
(ident, self._float_attrs(doc)))
buf.append(escape(doc.text))
buf.append(ident + u"</footnote>")
return linesep.join(buf)