X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/e4836157dee23cebc8e69e9dafab3cac41f20035..ab4401272d1b8507effd51da00f40bafe81f2b44:/librarian/pdf.py diff --git a/librarian/pdf.py b/librarian/pdf.py index 2d45372..1bfd949 100644 --- a/librarian/pdf.py +++ b/librarian/pdf.py @@ -21,7 +21,7 @@ from lxml.etree import XMLSyntaxError, XSLTApplyError from librarian.dcparser import Person from librarian.parser import WLDocument -from librarian import ParseError, DCNS +from librarian import ParseError, DCNS, get_resource from librarian import functions @@ -32,12 +32,19 @@ functions.reg_ends_white() functions.reg_texcommand() STYLESHEETS = { - 'wl2tex': 'xslt/wl2tex.xslt', + 'wl2tex': 'pdf/wl2tex.xslt', } +CUSTOMIZATIONS = [ + 'nofootnotes', + 'nothemes', + '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 + """ 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'); @@ -67,7 +74,7 @@ def insert_tags(doc, split_re, tagname, exclude=None): def substitute_hyphens(doc): - insert_tags(doc, + insert_tags(doc, re.compile("(?<=[^-\s])-(?=[^-\s])"), "dywiz", exclude=[DCNS("identifier.url"), DCNS("rights.license")] @@ -75,7 +82,7 @@ def substitute_hyphens(doc): def fix_hanging(doc): - insert_tags(doc, + insert_tags(doc, re.compile("(?<=\s\w)\s+"), "nbsp", exclude=[DCNS("identifier.url"), DCNS("rights.license")] @@ -128,29 +135,31 @@ def hack_motifs(doc): def parse_creator(doc): - """ find all dc:creator tags and add dc:creator_parsed with forenames first """ - for creator in doc.findall('//'+DCNS('creator')): - p = Person.from_text(creator.text) - creator_parsed = deepcopy(creator) - creator_parsed.tag = DCNS('creator_parsed') - creator_parsed.text = ' '.join(p.first_names + (p.last_name,)) - creator.getparent().insert(0, creator_parsed) - + """ find all dc:creator and dc.contributor tags and add *_parsed versions with forenames first """ + for person in doc.xpath("|".join('//dc:'+(tag) for tag in ( + 'creator', 'contributor.translator', 'contributor.editor', 'contributor.technical_editor')), + 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 package_available(package, args='', verbose=False): - """ check if a verion of a latex package accepting given args is available """ + """ 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{book} + \documentclass{wl} \usepackage[%s]{%s} \begin{document} \end{document} @@ -164,8 +173,9 @@ def package_available(package, args='', verbose=False): return p == 0 -def transform(provider, slug=None, file_path=None, - output_file=None, output_dir=None, make_dir=False, verbose=False, save_tex=None, morefloats=None): +def transform(provider, slug=None, file_path=None, + output_file=None, output_dir=None, make_dir=False, verbose=False, save_tex=None, morefloats=None, + cover=None, flags=None, customizations=None): """ produces a PDF file with XeLaTeX provider: a DocProvider @@ -177,6 +187,9 @@ def transform(provider, slug=None, file_path=None, 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 object + flags: less-advertising, + customizations: user requested customizations regarding various formatting parameters (passed to wl LaTeX class) """ # Parse XSLT @@ -190,12 +203,23 @@ def transform(provider, slug=None, file_path=None, raise ValueError('either slug or file_path should be specified') document = load_including_children(provider, slug=slug) + if cover: + document.edoc.getroot().set('data-cover-width', str(cover.width)) + document.edoc.getroot().set('data-cover-height', str(cover.height)) + if flags: + for flag in flags: + document.edoc.getroot().set('flag-' + flag, 'yes') + # check for LaTeX packages if morefloats: document.edoc.getroot().set('morefloats', morefloats.lower()) elif package_available('morefloats', 'maxfloats=19'): document.edoc.getroot().set('morefloats', 'new') + # add customizations + if customizations is not None: + document.edoc.getroot().set('customizations', u','.join(customizations)) + # hack the tree move_motifs_inside(document.edoc) hack_motifs(document.edoc) @@ -211,11 +235,19 @@ def transform(provider, slug=None, file_path=None, # wl -> TeXML style_filename = get_stylesheet("wl2tex") style = etree.parse(style_filename) + texml = document.transform(style) - del document # no longer needed large object :) # TeXML -> LaTeX temp = mkdtemp('-wl2pdf') + + if cover: + c = cover(document.book_info.author.readable(), document.book_info.title) + with open(os.path.join(temp, 'cover.png'), 'w') as f: + c.save(f) + + del document # no longer needed large object :) + tex_path = os.path.join(temp, 'doc.tex') fout = open(tex_path, 'w') process(StringIO(texml), fout, 'utf-8') @@ -226,8 +258,8 @@ def transform(provider, slug=None, file_path=None, shutil.copy(tex_path, save_tex) # LaTeX -> PDF - shutil.copy(get_resource('pdf/wl.sty'), temp) - shutil.copy(get_resource('pdf/wl-logo.png'), temp) + shutil.copy(get_resource('pdf/wl.cls'), temp) + shutil.copy(get_resource('res/wl-logo.png'), temp) cwd = os.getcwd() os.chdir(temp) @@ -269,7 +301,7 @@ def transform(provider, slug=None, file_path=None, def load_including_children(provider, slug=None, uri=None, file_path=None): - """ makes one big xml file with children inserted at end + """ makes one big xml file with children inserted at end either slug or uri must be provided """ @@ -282,13 +314,15 @@ def load_including_children(provider, slug=None, uri=None, file_path=None): else: raise ValueError('Neither slug, URI nor file path provided for a book.') - document = WLDocument.from_file(f, True, + text = f.read().decode('utf-8') + text = re.sub(ur"([\u0400-\u04ff]+)", ur"\1", text) + + document = WLDocument.from_string(text, True, parse_dublincore=True) f.close() - for child_uri in document.book_info.parts: + print child_uri child = load_including_children(provider, uri=child_uri) document.edoc.getroot().append(child.edoc.getroot()) - return document