X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/dd3e5f7c77134d034eb8592479288e1bb9fe28fd..665c88719c108626edee864affe07b085d28b7b4:/librarian/epub.py diff --git a/librarian/epub.py b/librarian/epub.py index c8a8668..be9488a 100644 --- a/librarian/epub.py +++ b/librarian/epub.py @@ -3,43 +3,94 @@ # This file is part of Librarian, licensed under GNU Affero GPLv3 or later. # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information. # -from __future__ import with_statement +from __future__ import print_function, unicode_literals import os import os.path import re import subprocess -from StringIO import StringIO +from six import BytesIO 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, OutputFile -from librarian.cover import DefaultEbookCover +from librarian import RDFNS, WLNS, NCXNS, OPFNS, XHTMLNS, DCNS, OutputFile +from librarian.cover import make_cover from librarian import functions, get_resource +from librarian.hyphenator import Hyphenator + functions.reg_person_name() +functions.reg_lang_code_3to2() + + +def squeeze_whitespace(s): + return re.sub(b'\\s+', b' ', s) + + +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.read().decode('latin1').split('\n'): + 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): """ returns node's text and children as a string - >>> print inner_xml(etree.fromstring('xyz')) + >>> print(inner_xml(etree.fromstring('xyz'))) xyz """ 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, encoding='unicode') for child in node]) + def set_inner_xml(node, text): """ sets node's text and children from a string >>> e = etree.fromstring('bxx') >>> set_inner_xml(e, 'xyz') - >>> print etree.tostring(e) + >>> print(etree.tostring(e, encoding='unicode')) xyz """ @@ -51,7 +102,7 @@ def set_inner_xml(node, text): def node_name(node): """ Find out a node's name - >>> print node_name(etree.fromstring('XYZ')) + >>> print(node_name(etree.fromstring('XYZ'))) XYZ """ @@ -66,11 +117,13 @@ 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)) + transform = etree.XSLT(etree.parse(xsltf)) + params = dict((key, transform.strparam(value)) for key, value in kwargs.items()) + return transform(xml, **params) def replace_characters(node): @@ -97,7 +150,7 @@ 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 = '' @@ -119,10 +172,10 @@ class Stanza(object): >>> s = etree.fromstring("a c c/\\nbx/\\nyc/ \\nd") >>> Stanza(s).versify() - >>> print etree.tostring(s) - a c cbx/ + >>> print(etree.tostring(s, encoding='unicode')) + a ccbx/ ycd - + """ def __init__(self, stanza_elem): self.stanza = stanza_elem @@ -137,7 +190,7 @@ class Stanza(object): tail = self.stanza.tail self.stanza.clear() self.stanza.tail = tail - self.stanza.extend(self.verses) + self.stanza.extend(verse for verse in self.verses if verse.text or len(verse) > 0) def open_normal_verse(self): self.open_verse = self.stanza.makeelement("wers_normalny") @@ -154,6 +207,8 @@ class Stanza(object): for i, verse_text in enumerate(re.split(r"/\s*\n", text)): if i: self.open_normal_verse() + if not verse_text.strip(): + continue verse = self.get_open_verse() if len(verse): verse[-1].tail = (verse[-1].tail or "") + verse_text @@ -184,18 +239,17 @@ 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) @@ -209,7 +263,7 @@ class TOC(object): 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_href, level-1, is_part) + return self.children[-1].add(name, part_href, level - 1, is_part) else: t = TOC(name) t.part_href = part_href @@ -247,7 +301,10 @@ 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) @@ -264,12 +321,12 @@ class TOC(object): texts.append( "
%s
" % (depth, child.href(), child.name)) - texts.append(child.html_part(depth+1)) + 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') + with open(get_resource('epub/toc.html'), 'rb') as f: + t = f.read().decode('utf-8') return t % self.html_part() @@ -287,21 +344,44 @@ 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 @@ -310,7 +390,9 @@ 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"): + 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), "part%d.html" % chunk_no, level=1, is_part=False) @@ -325,18 +407,22 @@ def transform_chunk(chunk_xml, chunk_no, annotations, empty=False, _empty_html_s 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, method="html", pretty_print=True) + 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): +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, with-full-fonts + flags: less-advertising, without-fonts, working-copy """ def transform_file(wldoc, chunk_counter=1, first=True, sample=None): @@ -344,16 +430,24 @@ def transform(wldoc, verbose=False, 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(wldoc.book_info.title, "part%d.html" % chunk_counter) chars = set() if first: # write book title page - html_tree = xslt(wldoc.edoc, get_resource('epub/xsltTitle.xsl')) + 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, method="html", pretty_print=True)) + html_string = etree.tostring( + html_tree, pretty_print=True, xml_declaration=True, + encoding="utf-8", + doctype='' + ) + zip.writestr('OPS/title.html', squeeze_whitespace(html_string)) # add a title page TOC entry toc.add(u"Strona tytułowa", "title.html") elif wldoc.book_info.parts: @@ -364,8 +458,13 @@ def transform(wldoc, verbose=False, else: 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) + html_string = etree.tostring( + html_tree, pretty_print=True, xml_declaration=True, + encoding="utf-8", + doctype='' + ) + zip.writestr('OPS/part%d.html' % chunk_counter, squeeze_whitespace(html_string)) add_to_manifest(manifest, chunk_counter) add_to_spine(spine, chunk_counter) chunk_counter += 1 @@ -391,7 +490,7 @@ def transform(wldoc, verbose=False, toc.extend(chunk_toc) chars = chars.union(chunk_chars) - zip.writestr('OPS/part%d.html' % chunk_counter, chunk_html) + zip.writestr('OPS/part%d.html' % chunk_counter, squeeze_whitespace(chunk_html)) add_to_manifest(manifest, chunk_counter) add_to_spine(spine, chunk_counter) chunk_counter += 1 @@ -404,7 +503,6 @@ def transform(wldoc, verbose=False, return toc, chunk_counter, chars, sample - document = deepcopy(wldoc) del wldoc @@ -412,9 +510,14 @@ def transform(wldoc, verbose=False, for flag in flags: document.edoc.getroot().set(flag, 'yes') + document.clean_ed_note() + document.clean_ed_note('abstrakt') + # add editors info - document.edoc.getroot().set('editors', u', '.join(sorted( - editor.readable() for editor in document.editors()))) + 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)) @@ -429,28 +532,47 @@ def transform(wldoc, verbose=False, 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): + ilustr_elements = set(ilustr.get('src') for ilustr in document.edoc.findall('//ilustr')) + for i, filename in enumerate(os.listdir(ilustr_path)): + if filename not in ilustr_elements: + continue + 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( + '' % (image_id, filename, guess_type(file_path)[0]))) + # write static elements mime = zipfile.ZipInfo() mime.filename = 'mimetype' mime.compress_type = zipfile.ZIP_STORED - mime.extra = '' - zip.writestr(mime, 'application/epub+zip') - zip.writestr('META-INF/container.xml', '' \ - '' \ - '') - 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')) + mime.extra = b'' + zip.writestr(mime, b'application/epub+zip') + zip.writestr( + 'META-INF/container.xml', + b'' + b'' + b'' + b'' + ) + 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 = make_cover - cover_file = StringIO() + cover_file = BytesIO() bound_cover = cover(document.book_info) bound_cover.save(cover_file) cover_name = 'cover.%s' % bound_cover.ext() @@ -460,7 +582,11 @@ def transform(wldoc, verbose=False, 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)) + cover_tree, pretty_print=True, xml_declaration=True, + encoding="utf-8", + doctype='' + )) if bound_cover.uses_dc_cover: if document.book_info.cover_by: @@ -472,18 +598,20 @@ def transform(wldoc, verbose=False, '')) manifest.append(etree.fromstring( '' % (cover_name, bound_cover.mime_type()))) - spine.insert(0, etree.fromstring('')) + spine.insert(0, etree.fromstring('')) opf.getroot()[0].append(etree.fromstring('')) guide.append(etree.fromstring('')) - annotations = etree.Element('annotations') - toc_file = etree.fromstring('' \ - '' \ - '') + toc_file = etree.fromstring( + b'' + b'' + b'' + ) nav_map = toc_file[-1] if html_toc: @@ -509,28 +637,36 @@ def transform(wldoc, verbose=False, html_tree = xslt(annotations, get_resource('epub/xsltAnnotations.xsl')) chars = chars.union(used_chars(html_tree.getroot())) zip.writestr('OPS/annotations.html', etree.tostring( - html_tree, method="html", pretty_print=True)) + html_tree, pretty_print=True, xml_declaration=True, + encoding="utf-8", + doctype='' + )) toc.add("Wesprzyj Wolne Lektury", "support.html") manifest.append(etree.fromstring( '')) spine.append(etree.fromstring( '')) - html_string = open(get_resource('epub/support.html')).read() + html_string = open(get_resource('epub/support.html'), 'rb').read() chars.update(used_chars(etree.fromstring(html_string))) - zip.writestr('OPS/support.html', html_string) + zip.writestr('OPS/support.html', squeeze_whitespace(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')) + html_tree = xslt(document.edoc, get_resource('epub/xsltLast.xsl'), outputtype=output_type) 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: + zip.writestr('OPS/last.html', squeeze_whitespace(etree.tostring( + html_tree, pretty_print=True, xml_declaration=True, + encoding="utf-8", + doctype='' + ))) + + if not flags or 'without-fonts' not in flags: # strip fonts tmpdir = mkdtemp('-librarian-epub') try: @@ -540,24 +676,25 @@ def transform(wldoc, verbose=False, 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)) + optimizer_call = ['perl', 'subset.pl', '--chars', + ''.join(chars).encode('utf-8'), + get_resource('fonts/' + fname), + os.path.join(tmpdir, fname)] + env = {"PERL_USE_UNSAFE_INC": "1"} + if verbose: + print("Running font-optimizer") + subprocess.check_call(optimizer_call, env=env) else: - zip.write(get_resource('fonts/' + fname), os.path.join('OPS', fname)) + dev_null = open(os.devnull, 'w') + subprocess.check_call(optimizer_call, stdout=dev_null, stderr=dev_null, env=env) + zip.write(os.path.join(tmpdir, 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)) + 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: @@ -574,7 +711,8 @@ def transform(wldoc, verbose=False, 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)) + 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)