Fixes #2631: losing whitespace in epub/mobi stanzas.
[librarian.git] / librarian / pdf.py
index 659390d..9fb92b1 100644 (file)
@@ -3,40 +3,54 @@
 # 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 tempfile import mkdtemp
+from tempfile import mkdtemp, NamedTemporaryFile
 import re
-
-import sys
-sys.path.append('..') # for running from working copy
+from copy import deepcopy
+from subprocess import call, PIPE
 
 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 import ParseError
+from librarian import ParseError, DCNS, get_resource, OutputFile
 from librarian import functions
-
+from librarian.cover import WLCover
 
 
 functions.reg_substitute_entities()
-functions.reg_person_name()
 functions.reg_strip()
 functions.reg_starts_white()
 functions.reg_ends_white()
+functions.reg_texcommand()
 
 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');
@@ -44,134 +58,264 @@ def insert_tags(doc, split_re, tagname):
     <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)
-            elem.text = chunks.pop(0)
-            while chunks:
+            while len(chunks) > 1:
                 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
-            elem.tail = chunks.pop(0)
-            while chunks:
+            while len(chunks) > 1:
                 ins = etree.Element(tagname)
                 ins.tail = chunks.pop()
                 parent.insert(ins_index, ins)
+            elem.tail = chunks.pop(0)
 
 
 def substitute_hyphens(doc):
-    insert_tags(doc, 
+    insert_tags(doc,
                 re.compile("(?<=[^-\s])-(?=[^-\s])"),
-                "dywiz")
+                "dywiz",
+                exclude=[DCNS("identifier.url"), DCNS("rights.license")]
+                )
 
 
 def fix_hanging(doc):
-    insert_tags(doc, 
+    insert_tags(doc,
                 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 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_resource(path):
-    return os.path.join(os.path.dirname(__file__), path)
 
 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:
+        book_info = wldoc.book_info
+        document = load_including_children(wldoc)
+        root = document.edoc.getroot()
+
+        if cover:
+            if cover is True:
+                cover = WLCover
+            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())))
+
+        # 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)
 
-        document = load_including_children(provider, slug)
+        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 :)
 
-        temp = mkdtemp('wl2pdf-')
         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
 
-        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")
 
+        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')
-        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)
 
 
-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)
-    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:
-        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:
-        child = load_including_children(provider, uri=child_uri)
+        child = load_including_children(provider=provider, uri=child_uri)
         document.edoc.getroot().append(child.edoc.getroot())
-
     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)
-