X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/18a72965fc7b90155130e87d4f1a4dd4aa354549..5782cd6d2ee17eb613055e6b641f01dcd58af80e:/librarian/epub.py diff --git a/librarian/epub.py b/librarian/epub.py index 666bcc9..8dba5ac 100644 --- a/librarian/epub.py +++ b/librarian/epub.py @@ -1,29 +1,66 @@ # -*- 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 lxml import etree import zipfile -from tempfile import mkdtemp +from tempfile import mkdtemp, NamedTemporaryFile from shutil import rmtree -import sys - -from librarian import XMLNamespace, RDFNS, DCNS, WLNS, NCXNS, OPFNS, NoDublinCore -from librarian.dcparser import BookInfo +from librarian import RDFNS, WLNS, NCXNS, OPFNS, XHTMLNS, DCNS, OutputFile +from librarian.cover import DefaultEbookCover from librarian import functions, get_resource -functions.reg_person_name() - +from librarian.hyphenator import Hyphenator +functions.reg_person_name() +functions.reg_lang_code_3to2() + +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): + if hyph is not None: + texts = etree.XPath('/utwor/*[2]//text()')(source_tree) + for t in texts: + parent = t.getparent() + newt = '' + wlist = re.compile(r'\w+|[^\w]', re.UNICODE).findall(t) + for w in wlist: + newt += hyph.inserted(w, u'\u00AD') + 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): """ returns node's text and children as a string @@ -32,7 +69,7 @@ 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 @@ -83,13 +120,14 @@ def replace_characters(node): .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): @@ -105,35 +143,78 @@ 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 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): @@ -156,19 +237,23 @@ def add_to_spine(spine, partno): 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 @@ -185,7 +270,13 @@ 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) @@ -193,20 +284,34 @@ 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( + "
%s
" % + (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 """ @@ -225,18 +330,41 @@ def chop(main_text): 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 @@ -245,10 +373,12 @@ def transform_chunk(chunk_xml, chunk_no, annotations, empty=False, _empty_html_s 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)) if empty: if not _empty_html_static: @@ -264,41 +394,44 @@ def transform_chunk(chunk_xml, chunk_no, annotations, empty=False, _empty_html_s return output_html, toc, chars -def transform(provider, slug=None, file_path=None, output_file=None, output_dir=None, make_dir=False, verbose=False, sample=None): +def transform(wldoc, verbose=False, + style=None, html_toc=False, + sample=None, cover=None, flags=None): """ produces a EPUB file - provider: a DocProvider - slug: slug of file to process, available by provider - output_file: file-like object or path to output file - output_dir: path to directory to save output file to; either this or output_file must be present - make_dir: writes output to //.epub instead of /.epub 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, with-full-fonts """ - def transform_file(input_xml, chunk_counter=1, first=True, sample=None): + def transform_file(wldoc, chunk_counter=1, first=True, sample=None): """ processes one input file and proceeds to its children """ - replace_characters(input_xml.getroot()) - - children = [child.text for child in input_xml.findall('.//'+DCNS('relation.hasPart'))] - + replace_characters(wldoc.edoc.getroot()) + + hyphenator = set_hyph_language(wldoc.edoc.getroot()) + 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 - html_tree = xslt(input_xml, get_resource('epub/xsltTitle.xsl')) + html_tree = xslt(wldoc.edoc, get_resource('epub/xsltTitle.xsl')) chars = used_chars(html_tree.getroot()) zip.writestr('OPS/title.html', etree.tostring(html_tree, method="html", pretty_print=True)) - elif children: + # add a title page TOC entry + toc.add(u"Strona tytułowa", "title.html") + elif wldoc.book_info.parts: # write title page for every parent if sample is not None and sample <= 0: chars = set() html_string = open(get_resource('epub/emptyChunk.html')).read() else: - html_tree = xslt(input_xml, get_resource('epub/xsltChunkTitle.xsl')) + html_tree = xslt(wldoc.edoc, get_resource('epub/xsltChunkTitle.xsl')) chars = used_chars(html_tree.getroot()) html_string = etree.tostring(html_tree, method="html", pretty_print=True) zip.writestr('OPS/part%d.html' % chunk_counter, html_string) @@ -306,12 +439,12 @@ def transform(provider, slug=None, file_path=None, output_file=None, output_dir= 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 @@ -332,47 +465,37 @@ def transform(provider, slug=None, file_path=None, output_file=None, output_dir= 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, chunk_chars, sample = transform_file(child_xml, chunk_counter, first=False, sample=sample) - toc.append(child_toc) - chars = chars.union(chunk_chars) + 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 - # read metadata from the first file - if file_path: - if slug: - raise ValueError('slug or file_path should be specified, not both') - f = open(file_path, 'r') - input_xml = etree.parse(f) - f.close() - else: - if not slug: - raise ValueError('either slug or file_path should be specified') - 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: - if make_dir: - author = unicode(book_info.author) - output_dir = os.path.join(output_dir, author) - try: - os.makedirs(output_dir) - except OSError: - pass - if slug: - output_file = open(os.path.join(output_dir, '%s.epub' % slug), 'w') - else: - output_file = open(os.path.join(output_dir, os.path.splitext(os.path.basename(file_path))[0] + '.epub'), 'w') + document = deepcopy(wldoc) + del wldoc + + if flags: + for flag in flags: + document.edoc.getroot().set(flag, 'yes') + + # add editors info + document.edoc.getroot().set('editors', u', '.join(sorted( + editor.readable() for editor in document.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) # write static elements @@ -386,12 +509,42 @@ def transform(provider, slug=None, file_path=None, output_file=None, output_dir= '' \ '') - zip.write(get_resource('epub/style.css'), os.path.join('OPS', 'style.css')) zip.write(get_resource('res/wl-logo-small.png'), os.path.join('OPS', 'logo_wolnelektury.png')) + zip.write(get_resource('res/jedenprocent.png'), os.path.join('OPS', 'jedenprocent.png')) + if not style: + style = get_resource('epub/style.css') + zip.write(style, os.path.join('OPS', 'style.css')) + + if cover: + if cover is True: + cover = DefaultEbookCover + + cover_file = StringIO() + bound_cover = cover(document.book_info) + bound_cover.save(cover_file) + cover_name = 'cover.%s' % bound_cover.ext() + zip.writestr(os.path.join('OPS', cover_name), cover_file.getvalue()) + del cover_file + + cover_tree = etree.parse(get_resource('epub/cover.html')) + cover_tree.find('//' + XHTMLNS('img')).set('src', cover_name) + zip.writestr('OPS/cover.html', etree.tostring( + cover_tree, method="html", pretty_print=True)) + + if bound_cover.uses_dc_cover: + if document.book_info.cover_by: + document.edoc.getroot().set('data-cover-by', document.book_info.cover_by) + if document.book_info.cover_source: + document.edoc.getroot().set('data-cover-source', document.book_info.cover_source) + + manifest.append(etree.fromstring( + '')) + manifest.append(etree.fromstring( + '' % (cover_name, bound_cover.mime_type()))) + spine.insert(0, etree.fromstring('')) + opf.getroot()[0].append(etree.fromstring('')) + guide.append(etree.fromstring('')) - opf = xslt(metadata, get_resource('epub/xsltContent.xsl')) - manifest = opf.find('.//' + OPFNS('manifest')) - spine = opf.find('.//' + OPFNS('spine')) annotations = etree.Element('annotations') @@ -399,22 +552,24 @@ def transform(provider, slug=None, file_path=None, output_file=None, output_dir= '"-//NISO//DTD ncx 2005-1//EN" "http://www.daisy.org/z3986/2005/ncx-2005-1.dtd">' \ '' \ - '' \ - 'Strona tytułowa' \ - '') + '') nav_map = toc_file[-1] - toc, chunk_counter, chars, sample = transform_file(input_xml, sample=sample) + if html_toc: + manifest.append(etree.fromstring( + '')) + spine.append(etree.fromstring( + '')) + guide.append(etree.fromstring('')) + + toc, chunk_counter, chars, sample = transform_file(document, sample=sample) - if not toc.children: - toc.add(u"Początek utworu", 1) - toc_counter = toc.write_to_xml(nav_map, 2) + if len(toc.children) < 2: + toc.add(u"Początek utworu", "part1.html") # Last modifications in container files and EPUB creation if len(annotations) > 0: - nav_map.append(etree.fromstring( - 'Przypisy'\ - '' % {'i': toc_counter})) + toc.add("Przypisy", "annotations.html") manifest.append(etree.fromstring( '')) spine.append(etree.fromstring( @@ -425,34 +580,69 @@ def transform(provider, slug=None, file_path=None, output_file=None, output_dir= zip.writestr('OPS/annotations.html', etree.tostring( html_tree, method="html", pretty_print=True)) - # strip fonts - tmpdir = mkdtemp('-librarian-epub') - cwd = os.getcwd() - - os.chdir(os.path.join(os.path.dirname(os.path.realpath(__file__)), 'font-optimizer')) - for fname in 'DejaVuSerif.ttf', 'DejaVuSerif-Bold.ttf', 'DejaVuSerif-Italic.ttf', 'DejaVuSerif-BoldItalic.ttf': - optimizer_call = ['perl', 'subset.pl', '--chars', ''.join(chars).encode('utf-8'), - get_resource('fonts/' + fname), os.path.join(tmpdir, fname)] - if verbose: - print "Running font-optimizer" - subprocess.check_call(optimizer_call) - else: - subprocess.check_call(optimizer_call, stdout=subprocess.PIPE, stderr=subprocess.PIPE) - zip.write(os.path.join(tmpdir, fname), os.path.join('OPS', fname)) - rmtree(tmpdir) - os.chdir(cwd) - - zip.writestr('OPS/content.opf', etree.tostring(opf, pretty_print=True)) - contents = [] - title = node_name(etree.ETXPath('.//'+DCNS('title'))(input_xml)[0]) + toc.add("Wesprzyj Wolne Lektury", "support.html") + manifest.append(etree.fromstring( + '')) + spine.append(etree.fromstring( + '')) + html_string = open(get_resource('epub/support.html')).read() + chars.update(used_chars(etree.fromstring(html_string))) + zip.writestr('OPS/support.html', html_string) + + toc.add("Strona redakcyjna", "last.html") + manifest.append(etree.fromstring( + '')) + spine.append(etree.fromstring( + '')) + html_tree = xslt(document.edoc, get_resource('epub/xsltLast.xsl')) + chars.update(used_chars(html_tree.getroot())) + zip.writestr('OPS/last.html', etree.tostring( + html_tree, method="html", pretty_print=True)) + + if not flags or not 'without-fonts' in flags: + # strip fonts + tmpdir = mkdtemp('-librarian-epub') + try: + cwd = os.getcwd() + except OSError: + cwd = None + + os.chdir(os.path.join(os.path.dirname(os.path.realpath(__file__)), 'font-optimizer')) + for fname in 'DejaVuSerif.ttf', 'DejaVuSerif-Bold.ttf', 'DejaVuSerif-Italic.ttf', 'DejaVuSerif-BoldItalic.ttf': + if not flags or not 'with-full-fonts' in flags: + optimizer_call = ['perl', 'subset.pl', '--chars', ''.join(chars).encode('utf-8'), + get_resource('fonts/' + fname), os.path.join(tmpdir, fname)] + if verbose: + print "Running font-optimizer" + subprocess.check_call(optimizer_call) + else: + subprocess.check_call(optimizer_call, stdout=subprocess.PIPE, stderr=subprocess.PIPE) + zip.write(os.path.join(tmpdir, fname), os.path.join('OPS', fname)) + else: + zip.write(get_resource('fonts/' + fname), os.path.join('OPS', fname)) + manifest.append(etree.fromstring( + '' % (fname, fname))) + rmtree(tmpdir) + if cwd is not None: + os.chdir(cwd) + zip.writestr('OPS/content.opf', etree.tostring(opf, pretty_print=True, xml_declaration = True, encoding='UTF-8')) + title = document.book_info.title attributes = "dtb:uid", "dtb:depth", "dtb:totalPageCount", "dtb:maxPageNumber" for st in attributes: meta = toc_file.makeelement(NCXNS('meta')) meta.set('name', st) meta.set('content', '0') toc_file[0].append(meta) - toc_file[0][0].set('content', ''.join((title, 'WolneLektury.pl'))) + toc_file[0][0].set('content', str(document.book_info.url)) toc_file[0][1].set('content', str(toc.depth())) set_inner_xml(toc_file[1], ''.join(('', title, ''))) - zip.writestr('OPS/toc.ncx', etree.tostring(toc_file, pretty_print=True)) + + # write TOC + if html_toc: + toc.add(u"Spis treści", "toc.html", index=1) + zip.writestr('OPS/toc.html', toc.html().encode('utf-8')) + toc.write_to_xml(nav_map) + zip.writestr('OPS/toc.ncx', etree.tostring(toc_file, pretty_print=True, xml_declaration = True, encoding='UTF-8')) zip.close() + + return OutputFile.from_filename(output_file.name)