nicer mobi
[librarian.git] / librarian / epub.py
index cdb5aae..c5942a2 100644 (file)
@@ -1,44 +1,28 @@
 # -*- 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 subprocess
+from StringIO import StringIO
 from copy import deepcopy
 from lxml import etree
 import zipfile
+from tempfile import mkdtemp
+from shutil import rmtree
 
-from librarian import XMLNamespace, RDFNS, DCNS, WLNS, NoDublinCore
-from librarian.dcparser import BookInfo
-
-NCXNS = XMLNamespace("http://www.daisy.org/z3986/2005/ncx/")
-OPFNS = XMLNamespace("http://www.idpf.org/2007/opf")
-
-
-class DocProvider(object):
-    class DoesNotExist(Exception):
-        pass
-    
-    def by_slug(self, slug):
-        raise NotImplemented
-
-    def __getitem__(self, slug):
-        return self.by_slug(slug)
-
-    def by_uri(self, uri):
-        return self.by_slug(uri.rsplit('/', 1)[1])
+import sys
 
+from librarian import XMLNamespace, RDFNS, DCNS, WLNS, NCXNS, OPFNS, XHTMLNS, NoDublinCore
+from librarian.dcparser import BookInfo
 
-class DirDocProvider(DocProvider):
-    def __init__(self, dir):
-        self.dir = dir
-        self.files = {}
+from librarian import functions, get_resource
 
-    def by_slug(self, slug):
-        return open(os.path.join(self.dir, '%s.xml' % slug))
+functions.reg_person_name()
 
 
 def inner_xml(node):
@@ -49,7 +33,7 @@ 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
@@ -90,27 +74,24 @@ def xslt(xml, sheet):
         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)
-
-
 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):
@@ -126,7 +107,7 @@ 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)
 
 
@@ -177,19 +158,23 @@ 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%d.html" % 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
@@ -206,7 +191,13 @@ 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)
@@ -219,15 +210,34 @@ class TOC(object):
             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(
+                "<div style='margin-left:%dem;'><a href='%s'>%s</a></div>" %
+                (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 """
@@ -253,46 +263,74 @@ def chop(main_text):
     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)
+            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 <output_dir>/<author>/<slug>.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, method="html", pretty_print=True)
+    return output_html, toc, chars
+
+
+def transform(provider, slug=None, file_path=None, output_file=None, output_dir=None, make_dir=False, 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 <output_dir>/<author>/<slug>.epub instead of <output_dir>/<slug>.epub
+    sample=n: generate sample e-book (with at least n paragraphs)
+    cover: a cover.Cover object
+    flags: less-advertising, without-fonts
     """
 
-    def transform_file(input_xml, chunk_counter=1, first=True):
+    def transform_file(input_xml, 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'))]
 
         # every input file will have a TOC entry,
         # pointing to starting chunk
         toc = TOC(node_name(input_xml.find('.//'+DCNS('title'))), chunk_counter)
+        chars = set()
         if first:
             # write book title page
+            html_tree = xslt(input_xml, get_resource('epub/xsltTitle.xsl'))
+            chars = used_chars(html_tree.getroot())
             zip.writestr('OPS/title.html',
-                 etree.tostring(xslt(input_xml, res('xsltTitle.xsl')), pretty_print=True))
+                 etree.tostring(html_tree, method="html", pretty_print=True))
+            # add a title page TOC entry
+            toc.add(u"Strona tytułowa", "title.html")
         elif children:
             # 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(input_xml, 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)
             add_to_manifest(manifest, chunk_counter)
             add_to_spine(spine, chunk_counter)
             chunk_counter += 1
@@ -307,11 +345,17 @@ def transform(provider, slug, output_file=None, output_dir=None):
                 main_text = None
 
         if main_text is not None:
-            replace_characters(main_text)
+            for chunk_xml in chop(main_text):
+                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)
 
-            for chunk_no, chunk_xml in enumerate(chop(main_text), chunk_counter):
-                chunk_html, chunk_toc = transform_chunk(chunk_xml, chunk_counter, annotations)
                 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)
@@ -320,13 +364,28 @@ def transform(provider, slug, output_file=None, output_dir=None):
         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)
+                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)
 
-        return toc, chunk_counter
+        return toc, chunk_counter, chars, sample
 
     # read metadata from the first file
-    input_xml = etree.parse(provider[slug])
+    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])
+
+    if flags:
+        for flag in flags:
+            input_xml.getroot().set(flag, 'yes')
+
     metadata = input_xml.find('.//'+RDFNS('Description'))
     if metadata is None:
         raise NoDublinCore('Document has no DublinCore - which is required.')
@@ -335,14 +394,17 @@ def transform(provider, slug, output_file=None, output_dir=None):
 
     # 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')
-
+        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')
 
     zip = zipfile.ZipFile(output_file, 'w', zipfile.ZIP_DEFLATED)
 
@@ -357,41 +419,102 @@ def transform(provider, slug, output_file=None, output_dir=None):
                        '<rootfiles><rootfile full-path="OPS/content.opf" ' \
                        'media-type="application/oebps-package+xml" />' \
                        '</rootfiles></container>')
-    for fname in 'style.css', 'logo_wolnelektury.png':
-        zip.write(res(fname), os.path.join('OPS', fname))
+    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'))
 
-    opf = xslt(metadata, res('xsltContent.xsl'))
+    opf = xslt(metadata, get_resource('epub/xsltContent.xsl'))
     manifest = opf.find('.//' + OPFNS('manifest'))
+    guide = opf.find('.//' + OPFNS('guide'))
     spine = opf.find('.//' + OPFNS('spine'))
 
+    if cover:
+        cover_file = StringIO()
+        c = cover(book_info.author.readable(), book_info.title)
+        c.save(cover_file)
+        c_name = 'cover.%s' % c.ext()
+        zip.writestr(os.path.join('OPS', c_name), cover_file.getvalue())
+        del cover_file
+
+        cover_tree = etree.parse(get_resource('epub/cover.html'))
+        cover_tree.find('//' + XHTMLNS('img')).set('src', c_name)
+        zip.writestr('OPS/cover.html', etree.tostring(
+                        cover_tree, method="html", pretty_print=True))
+
+        manifest.append(etree.fromstring(
+            '<item id="cover" href="cover.html" media-type="application/xhtml+xml" />'))
+        manifest.append(etree.fromstring(
+            '<item id="cover-image" href="%s" media-type="%s" />' % (c_name, c.mime_type())))
+        spine.insert(0, etree.fromstring('<itemref idref="cover" linear="no" />'))
+        opf.getroot()[0].append(etree.fromstring('<meta name="cover" content="cover-image"/>'))
+        guide.append(etree.fromstring('<reference href="cover.html" type="cover" title="Okładka"/>'))
+
+
     annotations = etree.Element('annotations')
 
     toc_file = etree.fromstring('<?xml version="1.0" encoding="utf-8"?><!DOCTYPE ncx PUBLIC ' \
                                '"-//NISO//DTD ncx 2005-1//EN" "http://www.daisy.org/z3986/2005/ncx-2005-1.dtd">' \
                                '<ncx xmlns="http://www.daisy.org/z3986/2005/ncx/" xml:lang="pl" ' \
                                'version="2005-1"><head></head><docTitle></docTitle><navMap>' \
-                               '<navPoint id="NavPoint-1" playOrder="1"><navLabel>' \
-                               '<text>Strona tytułowa</text></navLabel><content src="title.html" />' \
-                               '</navPoint><navPoint id="NavPoint-2" playOrder="2"><navLabel>' \
-                               '<text>Początek utworu</text></navLabel><content src="part1.html" />' \
-                               '</navPoint></navMap></ncx>')
+                               '</navMap></ncx>')
     nav_map = toc_file[-1]
 
-    toc, chunk_counter = transform_file(input_xml)
-    toc_counter = toc.write_to_xml(nav_map, 3) # we already have 2 navpoints
+    if html_toc:
+        manifest.append(etree.fromstring(
+            '<item id="html_toc" href="toc.html" media-type="application/xhtml+xml" />'))
+        spine.append(etree.fromstring(
+            '<itemref idref="html_toc" />'))
+        guide.append(etree.fromstring('<reference href="toc.html" type="toc" title="Spis treści"/>'))
+
+    toc, chunk_counter, chars, sample = transform_file(input_xml, sample=sample)
+
+    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(
-            '<navPoint id="NavPoint-%(i)d" playOrder="%(i)d" ><navLabel><text>Przypisy</text>'\
-            '</navLabel><content src="annotations.html" /></navPoint>' % {'i': toc_counter}))
+        toc.add("Przypisy", "annotations.html")
         manifest.append(etree.fromstring(
             '<item id="annotations" href="annotations.html" media-type="application/xhtml+xml" />'))
         spine.append(etree.fromstring(
             '<itemref idref="annotations" />'))
         replace_by_verse(annotations)
+        html_tree = xslt(annotations, get_resource('epub/xsltAnnotations.xsl'))
+        chars = chars.union(used_chars(html_tree.getroot()))
         zip.writestr('OPS/annotations.html', etree.tostring(
-                            xslt(annotations, res("xsltAnnotations.xsl")), pretty_print=True))
+                            html_tree, method="html", pretty_print=True))
+
+    toc.add("Strona redakcyjna", "last.html")
+    manifest.append(etree.fromstring(
+        '<item id="last" href="last.html" media-type="application/xhtml+xml" />'))
+    spine.append(etree.fromstring(
+        '<itemref idref="last" />'))
+    html_tree = xslt(input_xml, 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')
+        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))
+            manifest.append(etree.fromstring(
+                '<item id="%s" href="%s" media-type="font/ttf" />' % (fname, fname)))
+        rmtree(tmpdir)
+        os.chdir(cwd)
 
     zip.writestr('OPS/content.opf', etree.tostring(opf, pretty_print=True))
     contents = []
@@ -405,20 +528,11 @@ def transform(provider, slug, output_file=None, output_dir=None):
     toc_file[0][0].set('content', ''.join((title, 'WolneLektury.pl')))
     toc_file[0][1].set('content', str(toc.depth()))
     set_inner_xml(toc_file[1], ''.join(('<text>', title, '</text>')))
+
+    # 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))
     zip.close()
-
-
-if __name__ == '__main__':
-    import sys
-
-    if len(sys.argv) < 2:
-        print >> sys.stderr, 'Usage: python epub.py <input file>'
-        sys.exit(1)
-
-    main_input = sys.argv[1]
-    basepath, ext = os.path.splitext(main_input)
-    path, slug = os.path.realpath(basepath).rsplit('/', 1)
-    provider = DirDocProvider(path)
-    transform(provider, slug, output_dir=path)
-