X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/02e90dceef6fef2a804512d5df984c6f9b5af174..c87453a1da79443d74132398e7dd1aaa83140fee:/librarian/epub.py
diff --git a/librarian/epub.py b/librarian/epub.py
index 84a745f..95e65f1 100644
--- a/librarian/epub.py
+++ b/librarian/epub.py
@@ -1,21 +1,73 @@
# -*- coding: utf-8 -*-
#
# This file is part of Librarian, licensed under GNU Affero GPLv3 or later.
-# Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
+# Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
#
from __future__ import with_statement
import os
import os.path
+import re
+import subprocess
+from StringIO import StringIO
from copy import deepcopy
+from mimetypes import guess_type
+
from lxml import etree
import zipfile
+from tempfile import mkdtemp, NamedTemporaryFile
+from shutil import rmtree
+
+from librarian import RDFNS, WLNS, NCXNS, OPFNS, XHTMLNS, DCNS, OutputFile
+from librarian.cover import DefaultEbookCover
+
+from librarian import functions, get_resource
+
+from librarian.hyphenator import Hyphenator
-import sys
-sys.path.append('..') # for running from working copy
+functions.reg_person_name()
+functions.reg_lang_code_3to2()
-from librarian import XMLNamespace, RDFNS, DCNS, WLNS, NCXNS, OPFNS, NoDublinCore
-from librarian.dcparser import BookInfo
+
+def set_hyph_language(source_tree):
+ def get_short_lng_code(text):
+ result = ''
+ text = ''.join(text)
+ with open(get_resource('res/ISO-639-2_8859-1.txt'), 'rb') as f:
+ for line in f:
+ list = line.strip().split('|')
+ if list[0] == text:
+ result = list[2]
+ if result == '':
+ return text
+ else:
+ return result
+ bibl_lng = etree.XPath('//dc:language//text()',
+ namespaces={'dc': str(DCNS)})(source_tree)
+ short_lng = get_short_lng_code(bibl_lng[0])
+ try:
+ return Hyphenator(get_resource('res/hyph-dictionaries/hyph_' +
+ short_lng + '.dic'))
+ except:
+ pass
+
+
+def hyphenate_and_fix_conjunctions(source_tree, hyph):
+ texts = etree.XPath('/utwor/*[2]//text()')(source_tree)
+ for t in texts:
+ parent = t.getparent()
+ if hyph is not None:
+ newt = ''
+ wlist = re.compile(r'\w+|[^\w]', re.UNICODE).findall(t)
+ for w in wlist:
+ newt += hyph.inserted(w, u'\u00AD')
+ else:
+ newt = t
+ newt = re.sub(r'(?<=\s\w)\s+', u'\u00A0', newt)
+ if t.is_text:
+ parent.text = newt
+ elif t.is_tail:
+ parent.tail = newt
def inner_xml(node):
@@ -26,7 +78,8 @@ def inner_xml(node):
"""
nt = node.text if node.text is not None else ''
- return ''.join([nt] + [etree.tostring(child) for child in node])
+ return ''.join([nt] + [etree.tostring(child) for child in node])
+
def set_inner_xml(node, text):
""" sets node's text and children from a string
@@ -60,41 +113,40 @@ def node_name(node):
return tempnode.text
-def xslt(xml, sheet):
+def xslt(xml, sheet, **kwargs):
if isinstance(xml, etree._Element):
xml = etree.ElementTree(xml)
with open(sheet) as xsltf:
- return xml.xslt(etree.parse(xsltf))
-
-
-_resdir = os.path.join(os.path.dirname(os.path.realpath(__file__)), 'epub')
-def res(fname):
- return os.path.join(_resdir, fname)
+ transform = etree.XSLT(etree.parse(xsltf))
+ params = dict((key, transform.strparam(value)) for key, value in kwargs.iteritems())
+ return transform(xml, **params)
def replace_characters(node):
def replace_chars(text):
if text is None:
return None
- return text.replace("---", u"\u2014")\
+ return text.replace(u"\ufeff", u"")\
+ .replace("---", u"\u2014")\
.replace("--", u"\u2013")\
.replace(",,", u"\u201E")\
.replace('"', u"\u201D")\
.replace("'", u"\u2019")
- if node.tag == 'extra':
+ if node.tag in ('uwaga', 'extra'):
+ t = node.tail
node.clear()
- else:
- node.text = replace_chars(node.text)
- node.tail = replace_chars(node.tail)
- for child in node:
- replace_characters(child)
+ node.tail = t
+ node.text = replace_chars(node.text)
+ node.tail = replace_chars(node.tail)
+ for child in node:
+ replace_characters(child)
def find_annotations(annotations, source, part_no):
for child in source:
if child.tag in ('pe', 'pa', 'pt', 'pr'):
annotation = deepcopy(child)
- number = str(len(annotations)+1)
+ number = str(len(annotations) + 1)
annotation.set('number', number)
annotation.set('part', str(part_no))
annotation.tail = ''
@@ -103,70 +155,118 @@ def find_annotations(annotations, source, part_no):
child.clear()
child.tail = tail
child.text = number
- if child.tag not in ('extra', 'podtytul'):
+ if child.tag not in ('extra', 'uwaga'):
find_annotations(annotations, child, part_no)
+class Stanza(object):
+ """
+ Converts / verse endings into verse elements in a stanza.
+
+ Slashes may only occur directly in the stanza. Any slashes in subelements
+ will be ignored, and the subelements will be put inside verse elements.
+
+ >>> s = etree.fromstring("a c c/\\nbx/\\nyc/ \\nd")
+ >>> Stanza(s).versify()
+ >>> print etree.tostring(s)
+ a c cbx/
+ ycd
+
+ """
+ def __init__(self, stanza_elem):
+ self.stanza = stanza_elem
+ self.verses = []
+ self.open_verse = None
+
+ def versify(self):
+ self.push_text(self.stanza.text)
+ for elem in self.stanza:
+ self.push_elem(elem)
+ self.push_text(elem.tail)
+ tail = self.stanza.tail
+ self.stanza.clear()
+ self.stanza.tail = tail
+ self.stanza.extend(self.verses)
+
+ def open_normal_verse(self):
+ self.open_verse = self.stanza.makeelement("wers_normalny")
+ self.verses.append(self.open_verse)
+
+ def get_open_verse(self):
+ if self.open_verse is None:
+ self.open_normal_verse()
+ return self.open_verse
+
+ def push_text(self, text):
+ if not text:
+ return
+ for i, verse_text in enumerate(re.split(r"/\s*\n", text)):
+ if not verse_text.strip():
+ continue
+ if i:
+ self.open_normal_verse()
+ verse = self.get_open_verse()
+ if len(verse):
+ verse[-1].tail = (verse[-1].tail or "") + verse_text
+ else:
+ verse.text = (verse.text or "") + verse_text
+
+ def push_elem(self, elem):
+ if elem.tag.startswith("wers"):
+ verse = deepcopy(elem)
+ verse.tail = None
+ self.verses.append(verse)
+ self.open_verse = verse
+ else:
+ appended = deepcopy(elem)
+ appended.tail = None
+ self.get_open_verse().append(appended)
+
+
def replace_by_verse(tree):
""" Find stanzas and create new verses in place of a '/' character """
stanzas = tree.findall('.//' + WLNS('strofa'))
- for node in stanzas:
- for child_node in node:
- if child_node.tag in ('slowo_obce', 'wyroznienie'):
- foreign_verses = inner_xml(child_node).split('/\n')
- if len(foreign_verses) > 1:
- new_foreign = ''
- for foreign_verse in foreign_verses:
- if foreign_verse.startswith('', foreign_verse, ''))
- set_inner_xml(child_node, new_foreign)
- verses = inner_xml(node).split('/\n')
- if len(verses) > 1:
- modified_inner_xml = ''
- for verse in verses:
- if verse.startswith('', verse, ''))
- set_inner_xml(node, modified_inner_xml)
+ for stanza in stanzas:
+ Stanza(stanza).versify()
def add_to_manifest(manifest, partno):
""" Adds a node to the manifest section in content.opf file """
partstr = 'part%d' % partno
- e = manifest.makeelement(OPFNS('item'), attrib={
- 'id': partstr,
- 'href': partstr + '.html',
- 'media-type': 'application/xhtml+xml',
- })
+ e = manifest.makeelement(
+ OPFNS('item'), attrib={'id': partstr, 'href': partstr + '.html',
+ 'media-type': 'application/xhtml+xml'}
+ )
manifest.append(e)
def add_to_spine(spine, partno):
""" Adds a node to the spine section in content.opf file """
- e = spine.makeelement(OPFNS('itemref'), attrib={'idref': 'part%d' % partno});
+ e = spine.makeelement(OPFNS('itemref'), attrib={'idref': 'part%d' % partno})
spine.append(e)
class TOC(object):
- def __init__(self, name=None, part_number=None):
+ def __init__(self, name=None, part_href=None):
self.children = []
self.name = name
- self.part_number = part_number
+ self.part_href = part_href
self.sub_number = None
- def add(self, name, part_number, level=0, is_part=True):
+ def add(self, name, part_href, level=0, is_part=True, index=None):
+ assert level == 0 or index is None
if level > 0 and self.children:
- return self.children[-1].add(name, part_number, level-1, is_part)
+ return self.children[-1].add(name, part_href, level - 1, is_part)
else:
t = TOC(name)
- t.part_number = part_number
- self.children.append(t)
+ t.part_href = part_href
+ if index is not None:
+ self.children.insert(index, t)
+ else:
+ self.children.append(t)
if not is_part:
t.sub_number = len(self.children) + 1
return t.sub_number
@@ -183,7 +283,13 @@ class TOC(object):
else:
return 0
- def write_to_xml(self, nav_map, counter):
+ def href(self):
+ src = self.part_href
+ if self.sub_number is not None:
+ src += '#sub%d' % self.sub_number
+ return src
+
+ def write_to_xml(self, nav_map, counter=1):
for child in self.children:
nav_point = nav_map.makeelement(NCXNS('navPoint'))
nav_point.set('id', 'NavPoint-%d' % counter)
@@ -191,20 +297,42 @@ class TOC(object):
nav_label = nav_map.makeelement(NCXNS('navLabel'))
text = nav_map.makeelement(NCXNS('text'))
- text.text = child.name
+ if child.name is not None:
+ text.text = re.sub(r'\n', ' ', child.name)
+ else:
+ text.text = child.name
nav_label.append(text)
nav_point.append(nav_label)
content = nav_map.makeelement(NCXNS('content'))
- src = 'part%d.html' % child.part_number
- if child.sub_number is not None:
- src += '#sub%d' % child.sub_number
- content.set('src', src)
+ content.set('src', child.href())
nav_point.append(content)
nav_map.append(nav_point)
counter = child.write_to_xml(nav_point, counter + 1)
return counter
+ def html_part(self, depth=0):
+ texts = []
+ for child in self.children:
+ texts.append(
+ "" %
+ (depth, child.href(), child.name))
+ texts.append(child.html_part(depth + 1))
+ return "\n".join(texts)
+
+ def html(self):
+ with open(get_resource('epub/toc.html')) as f:
+ t = unicode(f.read(), 'utf-8')
+ return t % self.html_part()
+
+
+def used_chars(element):
+ """ Lists characters used in an ETree Element """
+ chars = set((element.text or '') + (element.tail or ''))
+ for child in element:
+ chars = chars.union(used_chars(child))
+ return chars
+
def chop(main_text):
""" divide main content of the XML file into chunks """
@@ -212,190 +340,373 @@ def chop(main_text):
# prepare a container for each chunk
part_xml = etree.Element('utwor')
etree.SubElement(part_xml, 'master')
- main_xml_part = part_xml[0] # master
+ main_xml_part = part_xml[0] # master
last_node_part = False
+
+ # the below loop are workaround for a problem with epubs in drama ebooks without acts
+ is_scene = False
+ is_act = False
+ for one_part in main_text:
+ name = one_part.tag
+ if name == 'naglowek_scena':
+ is_scene = True
+ elif name == 'naglowek_akt':
+ is_act = True
+
for one_part in main_text:
name = one_part.tag
- if name == 'naglowek_czesc':
- yield part_xml
- last_node_part = True
- main_xml_part[:] = [deepcopy(one_part)]
- elif not last_node_part and name in ("naglowek_rozdzial", "naglowek_akt", "srodtytul"):
- yield part_xml
- main_xml_part[:] = [deepcopy(one_part)]
+ if is_act is False and is_scene is True:
+ if name == 'naglowek_czesc':
+ yield part_xml
+ last_node_part = True
+ main_xml_part[:] = [deepcopy(one_part)]
+ elif not last_node_part and name == "naglowek_scena":
+ yield part_xml
+ main_xml_part[:] = [deepcopy(one_part)]
+ else:
+ main_xml_part.append(deepcopy(one_part))
+ last_node_part = False
else:
- main_xml_part.append(deepcopy(one_part))
- last_node_part = False
+ if name == 'naglowek_czesc':
+ yield part_xml
+ last_node_part = True
+ main_xml_part[:] = [deepcopy(one_part)]
+ elif not last_node_part and name in ("naglowek_rozdzial", "naglowek_akt", "srodtytul"):
+ yield part_xml
+ main_xml_part[:] = [deepcopy(one_part)]
+ else:
+ main_xml_part.append(deepcopy(one_part))
+ last_node_part = False
yield part_xml
-def transform_chunk(chunk_xml, chunk_no, annotations):
- """ transforms one chunk, returns a HTML string and a TOC object """
+def transform_chunk(chunk_xml, chunk_no, annotations, empty=False, _empty_html_static=[]):
+ """ transforms one chunk, returns a HTML string, a TOC object and a set of used characters """
toc = TOC()
for element in chunk_xml[0]:
- if element.tag in ("naglowek_czesc", "naglowek_rozdzial", "naglowek_akt", "srodtytul"):
- toc.add(node_name(element), chunk_no)
+ if element.tag == "naglowek_czesc":
+ toc.add(node_name(element), "part%d.html#book-text" % chunk_no)
+ elif element.tag in ("naglowek_rozdzial", "naglowek_akt", "srodtytul"):
+ toc.add(node_name(element), "part%d.html" % chunk_no)
elif element.tag in ('naglowek_podrozdzial', 'naglowek_scena'):
- subnumber = toc.add(node_name(element), chunk_no, level=1, is_part=False)
+ subnumber = toc.add(node_name(element), "part%d.html" % chunk_no, level=1, is_part=False)
element.set('sub', str(subnumber))
- find_annotations(annotations, chunk_xml, chunk_no)
- replace_by_verse(chunk_xml)
- output_html = etree.tostring(xslt(chunk_xml, res('xsltScheme.xsl')), pretty_print=True)
- return output_html, toc
-
-
-def transform(provider, slug, output_file=None, output_dir=None):
- """ produces an epub
-
- provider is a DocProvider
- either output_file (a file-like object) or output_dir (path to file/dir) should be specified
- if output_dir is specified, file will be written to //.epub
+ if empty:
+ if not _empty_html_static:
+ _empty_html_static.append(open(get_resource('epub/emptyChunk.html')).read())
+ chars = set()
+ output_html = _empty_html_static[0]
+ else:
+ find_annotations(annotations, chunk_xml, chunk_no)
+ replace_by_verse(chunk_xml)
+ html_tree = xslt(chunk_xml, get_resource('epub/xsltScheme.xsl'))
+ chars = used_chars(html_tree.getroot())
+ output_html = etree.tostring(
+ html_tree, pretty_print=True, xml_declaration=True,
+ encoding="utf-8",
+ doctype=''
+ )
+ return output_html, toc, chars
+
+
+def transform(wldoc, verbose=False, style=None, html_toc=False,
+ sample=None, cover=None, flags=None, hyphenate=False, ilustr_path='', output_type='epub'):
+ """ produces a EPUB file
+
+ sample=n: generate sample e-book (with at least n paragraphs)
+ cover: a cover.Cover factory or True for default
+ flags: less-advertising, without-fonts, working-copy
"""
- def transform_file(input_xml, chunk_counter=1, first=True):
+ def transform_file(wldoc, chunk_counter=1, first=True, sample=None):
""" processes one input file and proceeds to its children """
- children = [child.text for child in input_xml.findall('.//'+DCNS('relation.hasPart'))]
+ replace_characters(wldoc.edoc.getroot())
+
+ hyphenator = set_hyph_language(wldoc.edoc.getroot()) if hyphenate else None
+ hyphenate_and_fix_conjunctions(wldoc.edoc.getroot(), hyphenator)
# every input file will have a TOC entry,
# pointing to starting chunk
- toc = TOC(node_name(input_xml.find('.//'+DCNS('title'))), chunk_counter)
+ toc = TOC(wldoc.book_info.title, "part%d.html" % chunk_counter)
+ chars = set()
if first:
# write book title page
- zip.writestr('OPS/title.html',
- etree.tostring(xslt(input_xml, res('xsltTitle.xsl')), pretty_print=True))
- elif children:
+ html_tree = xslt(wldoc.edoc, get_resource('epub/xsltTitle.xsl'), outputtype=output_type)
+ chars = used_chars(html_tree.getroot())
+ zip.writestr(
+ 'OPS/title.html',
+ etree.tostring(
+ html_tree, pretty_print=True, xml_declaration=True,
+ encoding="utf-8",
+ doctype=''
+ )
+ )
+ # add a title page TOC entry
+ toc.add(u"Strona tytuÅowa", "title.html")
+ elif wldoc.book_info.parts:
# write title page for every parent
- zip.writestr('OPS/part%d.html' % chunk_counter,
- etree.tostring(xslt(input_xml, res('xsltChunkTitle.xsl')), pretty_print=True))
+ if sample is not None and sample <= 0:
+ chars = set()
+ html_string = open(get_resource('epub/emptyChunk.html')).read()
+ else:
+ html_tree = xslt(wldoc.edoc, get_resource('epub/xsltChunkTitle.xsl'))
+ chars = used_chars(html_tree.getroot())
+ html_string = etree.tostring(
+ html_tree, pretty_print=True, xml_declaration=True,
+ encoding="utf-8",
+ doctype=''
+ )
+ zip.writestr('OPS/part%d.html' % chunk_counter, html_string)
add_to_manifest(manifest, chunk_counter)
add_to_spine(spine, chunk_counter)
chunk_counter += 1
- if len(input_xml.getroot()) > 1:
+ if len(wldoc.edoc.getroot()) > 1:
# rdf before style master
- main_text = input_xml.getroot()[1]
+ main_text = wldoc.edoc.getroot()[1]
else:
# rdf in style master
- main_text = input_xml.getroot()[0]
+ main_text = wldoc.edoc.getroot()[0]
if main_text.tag == RDFNS('RDF'):
main_text = None
if main_text is not None:
- replace_characters(main_text)
-
for chunk_xml in chop(main_text):
- chunk_html, chunk_toc = transform_chunk(chunk_xml, chunk_counter, annotations)
+ empty = False
+ if sample is not None:
+ if sample <= 0:
+ empty = True
+ else:
+ sample -= len(chunk_xml.xpath('//strofa|//akap|//akap_cd|//akap_dialog'))
+ chunk_html, chunk_toc, chunk_chars = transform_chunk(chunk_xml, chunk_counter, annotations, empty)
+
toc.extend(chunk_toc)
+ chars = chars.union(chunk_chars)
zip.writestr('OPS/part%d.html' % chunk_counter, chunk_html)
add_to_manifest(manifest, chunk_counter)
add_to_spine(spine, chunk_counter)
chunk_counter += 1
- if children:
- for child in children:
- child_xml = etree.parse(provider.by_uri(child))
- child_toc, chunk_counter = transform_file(child_xml, chunk_counter, first=False)
- toc.append(child_toc)
-
- return toc, chunk_counter
-
- # read metadata from the first file
- input_xml = etree.parse(provider[slug])
- metadata = input_xml.find('.//'+RDFNS('Description'))
- if metadata is None:
- raise NoDublinCore('Document has no DublinCore - which is required.')
- book_info = BookInfo.from_element(input_xml)
- metadata = etree.ElementTree(metadata)
-
- # if output to dir, create the file
- if output_dir is not None:
- author = unicode(book_info.author)
- author_dir = os.path.join(output_dir, author)
- try:
- os.makedirs(author_dir)
- except OSError:
- pass
- output_file = open(os.path.join(author_dir, '%s.epub' % slug), 'w')
+ for child in wldoc.parts():
+ child_toc, chunk_counter, chunk_chars, sample = transform_file(
+ child, chunk_counter, first=False, sample=sample)
+ toc.append(child_toc)
+ chars = chars.union(chunk_chars)
+
+ return toc, chunk_counter, chars, sample
+
+ document = deepcopy(wldoc)
+ del wldoc
+ if flags:
+ for flag in flags:
+ document.edoc.getroot().set(flag, 'yes')
+ document.clean_ed_note()
+ document.clean_ed_note('abstrakt')
+
+ # add editors info
+ editors = document.editors()
+ if editors:
+ document.edoc.getroot().set('editors', u', '.join(sorted(
+ editor.readable() for editor in editors)))
+ if document.book_info.funders:
+ document.edoc.getroot().set('funders', u', '.join(
+ document.book_info.funders))
+ if document.book_info.thanks:
+ document.edoc.getroot().set('thanks', document.book_info.thanks)
+
+ opf = xslt(document.book_info.to_etree(), get_resource('epub/xsltContent.xsl'))
+ manifest = opf.find('.//' + OPFNS('manifest'))
+ guide = opf.find('.//' + OPFNS('guide'))
+ spine = opf.find('.//' + OPFNS('spine'))
+
+ output_file = NamedTemporaryFile(prefix='librarian', suffix='.epub', delete=False)
zip = zipfile.ZipFile(output_file, 'w', zipfile.ZIP_DEFLATED)
+ functions.reg_mathml_epub(zip)
+
+ if os.path.isdir(ilustr_path):
+ for i, filename in enumerate(os.listdir(ilustr_path)):
+ file_path = os.path.join(ilustr_path, filename)
+ zip.write(file_path, os.path.join('OPS', filename))
+ image_id = 'image%s' % i
+ manifest.append(etree.fromstring(
+ '