[epub, mobi] Fix for a problem with epubs, mobi in drama ebooks without acts

[librarian.git] / librarian / pdf.py

diff --git a/librarian/pdf.py b/librarian/pdf.py
index b9ead15..2154985 100644 (file)
--- a/librarian/pdf.py
+++ b/librarian/pdf.py
@@ -3,41 +3,54 @@
 # This file is part of Librarian, licensed under GNU Affero GPLv3 or later.
 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
 #
 # This file is part of Librarian, licensed under GNU Affero GPLv3 or later.
 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
 #

+"""PDF creation library.
+
+Creates one big XML from the book and its children, converts it to LaTeX
+with TeXML, then runs it by XeLaTeX.
+
+"""

 from __future__ import with_statement
 import os
 import os.path
 import shutil
 from StringIO import StringIO
 from __future__ import with_statement
 import os
 import os.path
 import shutil
 from StringIO import StringIO

-from tempfile import mkdtemp
+from tempfile import mkdtemp, NamedTemporaryFile

 import re
 from copy import deepcopy
 import re
 from copy import deepcopy

-
-import sys
-sys.path.append('..') # for running from working copy
+from subprocess import call, PIPE

 
 from Texml.processor import process
 from lxml import etree
 from lxml.etree import XMLSyntaxError, XSLTApplyError
 
 
 from Texml.processor import process
 from lxml import etree
 from lxml.etree import XMLSyntaxError, XSLTApplyError
 

+from librarian.dcparser import Person

 from librarian.parser import WLDocument
 from librarian.parser import WLDocument

-from librarian import ParseError
+from librarian import ParseError, DCNS, get_resource, OutputFile

 from librarian import functions
 from librarian import functions

-
+from librarian.cover import DefaultEbookCover

 
 
 functions.reg_substitute_entities()
 
 
 functions.reg_substitute_entities()

-functions.reg_person_name()

 functions.reg_strip()
 functions.reg_starts_white()
 functions.reg_ends_white()
 functions.reg_strip()
 functions.reg_starts_white()
 functions.reg_ends_white()

+functions.reg_texcommand()

 
 STYLESHEETS = {
 
 STYLESHEETS = {

-    'wl2tex': 'xslt/wl2tex.xslt',
+    'wl2tex': 'pdf/wl2tex.xslt',

 }
 
 }
 

+#CUSTOMIZATIONS = [
+#    'nofootnotes',
+#    'nothemes',
+#    'defaultleading',
+#    'onehalfleading',
+#    'doubleleading',
+#    'nowlfont',
+#    ]

 
 

-def insert_tags(doc, split_re, tagname):
-    """ inserts <tagname> for every occurence of `split_re' in text nodes in the `doc' tree 
+def insert_tags(doc, split_re, tagname, exclude=None):
+    """ inserts <tagname> for every occurence of `split_re' in text nodes in the `doc' tree

 
     >>> t = etree.fromstring('<a><b>A-B-C</b>X-Y-Z</a>');
     >>> insert_tags(t, re.compile('-'), 'd');
 
     >>> t = etree.fromstring('<a><b>A-B-C</b>X-Y-Z</a>');
     >>> insert_tags(t, re.compile('-'), 'd');


@@ -45,160 +58,268 @@ def insert_tags(doc, split_re, tagname):
     <a><b>A<d/>B<d/>C</b>X<d/>Y<d/>Z</a>
     """
 
     <a><b>A<d/>B<d/>C</b>X<d/>Y<d/>Z</a>
     """
 

-    for elem in doc.iter():
+    for elem in doc.iter(tag=etree.Element):
+        if exclude and elem.tag in exclude:
+            continue

         if elem.text:
             chunks = split_re.split(elem.text)
         if elem.text:
             chunks = split_re.split(elem.text)

-            elem.text = chunks.pop(0)
-            while chunks:
+            while len(chunks) > 1:

                 ins = etree.Element(tagname)
                 ins.tail = chunks.pop()
                 elem.insert(0, ins)
                 ins = etree.Element(tagname)
                 ins.tail = chunks.pop()
                 elem.insert(0, ins)

+            elem.text = chunks.pop(0)

         if elem.tail:
             chunks = split_re.split(elem.tail)
             parent = elem.getparent()
             ins_index = parent.index(elem) + 1
         if elem.tail:
             chunks = split_re.split(elem.tail)
             parent = elem.getparent()
             ins_index = parent.index(elem) + 1

-            elem.tail = chunks.pop(0)
-            while chunks:
+            while len(chunks) > 1:

                 ins = etree.Element(tagname)
                 ins.tail = chunks.pop()
                 parent.insert(ins_index, ins)
                 ins = etree.Element(tagname)
                 ins.tail = chunks.pop()
                 parent.insert(ins_index, ins)

+            elem.tail = chunks.pop(0)

 
 
 def substitute_hyphens(doc):
 
 
 def substitute_hyphens(doc):

-    insert_tags(doc, 
+    insert_tags(doc,

                 re.compile("(?<=[^-\s])-(?=[^-\s])"),
                 re.compile("(?<=[^-\s])-(?=[^-\s])"),

-                "dywiz")
+                "dywiz",
+                exclude=[DCNS("identifier.url"), DCNS("rights.license")]
+                )

 
 
 def fix_hanging(doc):
 
 
 def fix_hanging(doc):

-    insert_tags(doc, 
+    insert_tags(doc,

                 re.compile("(?<=\s\w)\s+"),
                 re.compile("(?<=\s\w)\s+"),

-                "nbsp")
+                "nbsp",
+                exclude=[DCNS("identifier.url"), DCNS("rights.license")]
+                )
+
+
+def move_motifs_inside(doc):
+    """ moves motifs to be into block elements """
+    for master in doc.xpath('//powiesc|//opowiadanie|//liryka_l|//liryka_lp|//dramat_wierszowany_l|//dramat_wierszowany_lp|//dramat_wspolczesny'):
+        for motif in master.xpath('motyw'):
+            for sib in motif.itersiblings():
+                if sib.tag not in ('sekcja_swiatlo', 'sekcja_asterysk', 'separator_linia', 'begin', 'end', 'motyw', 'extra', 'uwaga'):
+                    # motif shouldn't have a tail - it would be untagged text
+                    motif.tail = None
+                    motif.getparent().remove(motif)
+                    sib.insert(0, motif)
+                    break

 
 
 
 

-def get_resource(path):
-    return os.path.join(os.path.dirname(__file__), path)
+def hack_motifs(doc):
+    """ dirty hack for the marginpar-creates-orphans LaTeX problem
+    see http://www.latex-project.org/cgi-bin/ltxbugs2html?pr=latex/2304
+
+    moves motifs in stanzas from first verse to second
+    and from next to last to last, then inserts negative vspace before them
+    """
+    for motif in doc.findall('//strofa//motyw'):
+        # find relevant verse-level tag
+        verse, stanza = motif, motif.getparent()
+        while stanza is not None and stanza.tag != 'strofa':
+            verse, stanza = stanza, stanza.getparent()
+        breaks_before = sum(1 for i in verse.itersiblings('br', preceding=True))
+        breaks_after = sum(1 for i in verse.itersiblings('br'))
+        if (breaks_before == 0 and breaks_after > 0) or breaks_after == 1:
+            move_by = 1
+            if breaks_after == 2:
+                move_by += 1
+            moved_motif = deepcopy(motif)
+            motif.tag = 'span'
+            motif.text = None
+            moved_motif.tail = None
+            moved_motif.set('moved', str(move_by))
+
+            for br in verse.itersiblings('br'):
+                if move_by > 1:
+                    move_by -= 1
+                    continue
+                br.addnext(moved_motif)
+                break
+
+
+def parse_creator(doc):
+    """Generates readable versions of creator and translator tags.
+
+    Finds all dc:creator and dc.contributor.translator tags
+    and adds *_parsed versions with forenames first.
+    """
+    for person in doc.xpath("|".join('//dc:'+(tag) for tag in (
+                    'creator', 'contributor.translator')),
+                    namespaces = {'dc': str(DCNS)})[::-1]:
+        if not person.text:
+            continue
+        p = Person.from_text(person.text)
+        person_parsed = deepcopy(person)
+        person_parsed.tag = person.tag + '_parsed'
+        person_parsed.set('sortkey', person.text)
+        person_parsed.text = p.readable()
+        person.getparent().insert(0, person_parsed)
+

 
 def get_stylesheet(name):
     return get_resource(STYLESHEETS[name])
 
 
 def get_stylesheet(name):
     return get_resource(STYLESHEETS[name])
 

-def transform(provider, slug, output_file=None, output_dir=None):
-    """ produces a pdf file

 
 

-    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>.pdf
+def package_available(package, args='', verbose=False):
+    """ check if a verion of a latex package accepting given args is available """
+    tempdir = mkdtemp('-wl2pdf-test')
+    fpath = os.path.join(tempdir, 'test.tex')
+    f = open(fpath, 'w')
+    f.write(r"""
+        \documentclass{wl}
+        \usepackage[%s]{%s}
+        \begin{document}
+        \end{document}
+        """ % (args, package))
+    f.close()
+    if verbose:
+        p = call(['xelatex', '-output-directory', tempdir, fpath])
+    else:
+        p = call(['xelatex', '-interaction=batchmode', '-output-directory', tempdir, fpath], stdout=PIPE, stderr=PIPE)
+    shutil.rmtree(tempdir)
+    return p == 0
+
+
+def transform(wldoc, verbose=False, save_tex=None, morefloats=None,
+              cover=None, flags=None, customizations=None):
+    """ produces a PDF file with XeLaTeX
+
+    wldoc: a WLDocument
+    verbose: prints all output from LaTeX
+    save_tex: path to save the intermediary LaTeX file to
+    morefloats (old/new/none): force specific morefloats
+    cover: a cover.Cover factory or True for default
+    flags: less-advertising,
+    customizations: user requested customizations regarding various formatting parameters (passed to wl LaTeX class)

     """
 
     # Parse XSLT
     try:
     """
 
     # Parse XSLT
     try:

+        book_info = wldoc.book_info
+        document = load_including_children(wldoc)
+        root = document.edoc.getroot()
+
+        if cover:
+            if cover is True:
+                cover = DefaultEbookCover
+            bound_cover = cover(book_info)
+            root.set('data-cover-width', str(bound_cover.width))
+            root.set('data-cover-height', str(bound_cover.height))
+            if bound_cover.uses_dc_cover:
+                if book_info.cover_by:
+                    root.set('data-cover-by', book_info.cover_by)
+                if book_info.cover_source:
+                    root.set('data-cover-source',
+                            book_info.cover_source)
+        if flags:
+            for flag in flags:
+                root.set('flag-' + flag, 'yes')
+
+        # check for LaTeX packages
+        if morefloats:
+            root.set('morefloats', morefloats.lower())
+        elif package_available('morefloats', 'maxfloats=19'):
+            root.set('morefloats', 'new')
+
+        # add customizations
+        if customizations is not None:
+            root.set('customizations', u','.join(customizations))
+
+        # add editors info
+        root.set('editors', u', '.join(sorted(
+            editor.readable() for editor in document.editors())))
+        if document.book_info.funders:
+            root.set('funders', u', '.join(document.book_info.funders))
+        if document.book_info.thanks:
+            root.set('thanks', document.book_info.thanks)
+
+        # hack the tree
+        move_motifs_inside(document.edoc)
+        hack_motifs(document.edoc)
+        parse_creator(document.edoc)
+        substitute_hyphens(document.edoc)
+        fix_hanging(document.edoc)
+
+        # wl -> TeXML

         style_filename = get_stylesheet("wl2tex")
         style = etree.parse(style_filename)
 
         style_filename = get_stylesheet("wl2tex")
         style = etree.parse(style_filename)
 

-        document = load_including_children(provider, slug)
-
-        # dirty hack for the marginpar-creates-orphans LaTeX problem
-        # see http://www.latex-project.org/cgi-bin/ltxbugs2html?pr=latex/2304
-        for motif in document.edoc.findall('//strofa//motyw'):
-            # find relevant verse-level tag
-            verse, stanza = motif, motif.getparent()
-            while stanza is not None and stanza.tag != 'strofa':
-                verse, stanza = stanza, stanza.getparent()
-            breaks_before = sum(1 for i in verse.itersiblings('br', preceding=True))
-            breaks_after = sum(1 for i in verse.itersiblings('br'))
-            if (breaks_before == 0 and breaks_after > 0) or breaks_after == 1:
-                move_by = 1
-                if breaks_after == 2:
-                    move_by += 1
-                moved_motif = deepcopy(motif)
-                motif.tag = 'span'
-                motif.text = None
-                moved_motif.tail = None
-                moved_motif.set('moved', str(move_by))
-
-                for br in verse.itersiblings(tag='br'):
-                    if move_by > 1:
-                        move_by -= 1
-                        continue
-                    br.addnext(moved_motif)
-                    break
+        texml = document.transform(style)

 
 

-        substitute_hyphens(document.edoc)
-        fix_hanging(document.edoc)
+        # TeXML -> LaTeX
+        temp = mkdtemp('-wl2pdf')

 
 

-        # if output to dir, create the file
-        if output_dir is not None:
-            author = unicode(document.book_info.author)
-            output_dir = os.path.join(output_dir, author)
+        if cover:
+            with open(os.path.join(temp, 'cover.png'), 'w') as f:
+                bound_cover.save(f)

 
 

-        texml = document.transform(style)

         del document # no longer needed large object :)
 
         del document # no longer needed large object :)
 

-        temp = mkdtemp('wl2pdf-')

         tex_path = os.path.join(temp, 'doc.tex')
         fout = open(tex_path, 'w')
         tex_path = os.path.join(temp, 'doc.tex')
         fout = open(tex_path, 'w')

-        process(StringIO(texml), fout, 'utf8', 255, 0, 0)
+        process(StringIO(texml), fout, 'utf-8')

         fout.close()
         del texml
 
         fout.close()
         del texml
 

-        shutil.copy(get_resource('pdf/wl.sty'), temp)
-        shutil.copy(get_resource('pdf/wl-logo.png'), temp)
-        print "pdflatex -output-directory %s %s" % (temp, os.path.join(temp, 'doc.tex'))
-        if os.system("pdflatex -output-directory %s %s" % (temp, os.path.join(temp, 'doc.tex'))):
+        if save_tex:
+            shutil.copy(tex_path, save_tex)
+
+        # LaTeX -> PDF
+        shutil.copy(get_resource('pdf/wl.cls'), temp)
+        shutil.copy(get_resource('res/wl-logo.png'), temp)
+
+        try:
+            cwd = os.getcwd()
+        except OSError:
+            cwd = None
+        os.chdir(temp)
+
+        if verbose:
+            p = call(['xelatex', tex_path])
+        else:
+            p = call(['xelatex', '-interaction=batchmode', tex_path], stdout=PIPE, stderr=PIPE)
+        if p:

             raise ParseError("Error parsing .tex file")
 
             raise ParseError("Error parsing .tex file")
 

+        if cwd is not None:
+            os.chdir(cwd)
+
+        output_file = NamedTemporaryFile(prefix='librarian', suffix='.pdf', delete=False)

         pdf_path = os.path.join(temp, 'doc.pdf')
         pdf_path = os.path.join(temp, 'doc.pdf')

-        if output_dir is not None:
-            try:
-                os.makedirs(output_dir)
-            except OSError:
-                pass
-            output_path = os.path.join(output_dir, '%s.pdf' % slug)
-            shutil.move(pdf_path, output_path)
-        else:
-            with open(pdf_path) as f:
-                output_file.write(f.read())
-            output_file.close()
+        shutil.move(pdf_path, output_file.name)
+        shutil.rmtree(temp)
+        return OutputFile.from_filename(output_file.name)

 
 

-        return True

     except (XMLSyntaxError, XSLTApplyError), e:
         raise ParseError(e)
 
 
     except (XMLSyntaxError, XSLTApplyError), e:
         raise ParseError(e)
 
 

-def load_including_children(provider, slug=None, uri=None):
-    """ makes one big xml file with children inserted at end 
-    either slug or uri must be provided
+def load_including_children(wldoc=None, provider=None, uri=None):
+    """ Makes one big xml file with children inserted at end.
+    
+    Either wldoc or provider and URI must be provided.

     """
 
     """
 

-    if uri:
+    if uri and provider:

         f = provider.by_uri(uri)
         f = provider.by_uri(uri)

-    elif slug:
-        f = provider[slug]
+        text = f.read().decode('utf-8')
+        f.close()
+    elif wldoc is not None:
+        text = etree.tostring(wldoc.edoc, encoding=unicode)
+        provider = wldoc.provider

     else:
     else:

-        raise ValueError('Neither slug nor URI provided for a book.')
+        raise ValueError('Neither a WLDocument, nor provider and URI were provided.')
+
+    text = re.sub(ur"([\u0400-\u04ff]+)", ur"<alien>\1</alien>", text)

 
 

-    document = WLDocument.from_file(f, True,
-        parse_dublincore=True,
-        preserve_lines=False)
+    document = WLDocument.from_string(text,
+                parse_dublincore=True, provider=provider)
+    document.swap_endlines()

 
     for child_uri in document.book_info.parts:
 
     for child_uri in document.book_info.parts:

-        child = load_including_children(provider, uri=child_uri)
+        child = load_including_children(provider=provider, uri=child_uri)

         document.edoc.getroot().append(child.edoc.getroot())
         document.edoc.getroot().append(child.edoc.getroot())

-

     return document
     return document

-
-
-if __name__ == '__main__':
-    import sys
-    from librarian import DirDocProvider
-
-    if len(sys.argv) < 2:
-        print >> sys.stderr, 'Usage: python pdf.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)
-