X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/cba76114f031d47c93e1af947a350230cbef0a1f..6b75d94412c089b10bd7637b1008bdaf9057a440:/librarian/pdf.py diff --git a/librarian/pdf.py b/librarian/pdf.py index 9fb92b1..246fd0c 100644 --- a/librarian/pdf.py +++ b/librarian/pdf.py @@ -27,7 +27,8 @@ from librarian.dcparser import Person from librarian.parser import WLDocument from librarian import ParseError, DCNS, get_resource, OutputFile from librarian import functions -from librarian.cover import WLCover +from librarian.cover import DefaultEbookCover +from .sponsor import sponsor_logo functions.reg_substitute_entities() @@ -40,20 +41,21 @@ STYLESHEETS = { 'wl2tex': 'pdf/wl2tex.xslt', } -#CUSTOMIZATIONS = [ -# 'nofootnotes', -# 'nothemes', -# 'defaultleading', -# 'onehalfleading', -# 'doubleleading', -# 'nowlfont', -# ] +# CUSTOMIZATIONS = [ +# 'nofootnotes', +# 'nothemes', +# 'defaultleading', +# 'onehalfleading', +# 'doubleleading', +# 'nowlfont', +# ] + def insert_tags(doc, split_re, tagname, exclude=None): """ inserts for every occurence of `split_re' in text nodes in the `doc' tree - >>> t = etree.fromstring('A-B-CX-Y-Z'); - >>> insert_tags(t, re.compile('-'), 'd'); + >>> t = etree.fromstring('A-B-CX-Y-Z') + >>> insert_tags(t, re.compile('-'), 'd') >>> print etree.tostring(t) ABCXYZ """ @@ -95,12 +97,26 @@ def fix_hanging(doc): ) +def fix_tables(doc): + for kol in doc.iter(tag='kol'): + if kol.tail is not None: + if not kol.tail.strip(): + kol.tail = None + for table in doc.iter(tag='tabela'): + if table.get('ramka') == '1' or table.get('ramki') == '1': + table.set('_format', '|' + 'X|' * len(table[0])) + else: + table.set('_format', 'X' * len(table[0])) + + 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 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'): + 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) @@ -146,9 +162,8 @@ def parse_creator(doc): 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]: + 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) @@ -184,7 +199,7 @@ def package_available(package, args='', verbose=False): def transform(wldoc, verbose=False, save_tex=None, morefloats=None, - cover=None, flags=None, customizations=None): + cover=None, flags=None, customizations=None, ilustr_path=''): """ produces a PDF file with XeLaTeX wldoc: a WLDocument @@ -204,16 +219,15 @@ def transform(wldoc, verbose=False, save_tex=None, morefloats=None, if cover: if cover is True: - cover = WLCover - bound_cover = cover(book_info) + cover = DefaultEbookCover + bound_cover = cover(book_info, width=1200) 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) + root.set('data-cover-source', book_info.cover_source) if flags: for flag in flags: root.set('flag-' + flag, 'yes') @@ -229,8 +243,14 @@ def transform(wldoc, verbose=False, save_tex=None, morefloats=None, root.set('customizations', u','.join(customizations)) # add editors info - root.set('editors', u', '.join(sorted( - editor.readable() for editor in document.editors()))) + editors = document.editors() + if editors: + root.set('editors', u', '.join(sorted( + editor.readable() for editor in 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) @@ -238,21 +258,38 @@ def transform(wldoc, verbose=False, save_tex=None, morefloats=None, parse_creator(document.edoc) substitute_hyphens(document.edoc) fix_hanging(document.edoc) + fix_tables(document.edoc) # wl -> TeXML style_filename = get_stylesheet("wl2tex") style = etree.parse(style_filename) - - texml = document.transform(style) + functions.reg_mathml_latex() # TeXML -> LaTeX temp = mkdtemp('-wl2pdf') + for ilustr in document.edoc.findall("//ilustr"): + shutil.copy(os.path.join(ilustr_path, ilustr.get("src")), temp) + + for sponsor in book_info.sponsors: + ins = etree.Element("data-sponsor", name=sponsor) + logo = sponsor_logo(sponsor) + if logo: + fname = 'sponsor-%s' % os.path.basename(logo) + shutil.copy(logo, os.path.join(temp, fname)) + ins.set('src', fname) + root.insert(0, ins) + + if book_info.sponsor_note: + root.set("sponsor-note", book_info.sponsor_note) + + texml = document.transform(style) + if cover: with open(os.path.join(temp, 'cover.png'), 'w') as f: - bound_cover.save(f) + bound_cover.save(f, quality=80) - del document # no longer needed large object :) + del document # no longer needed large object :) tex_path = os.path.join(temp, 'doc.tex') fout = open(tex_path, 'w') @@ -311,8 +348,7 @@ def load_including_children(wldoc=None, provider=None, uri=None): text = re.sub(ur"([\u0400-\u04ff]+)", ur"\1", text) - document = WLDocument.from_string(text, - parse_dublincore=True, provider=provider) + document = WLDocument.from_string(text, parse_dublincore=True, provider=provider) document.swap_endlines() for child_uri in document.book_info.parts: