X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/d329e7e5c88a9445e10083536ae1549e6fc1a1be..565aa32ff31bdbe6e2aaf9f47b124ac1f883a990:/librarian/pypdf.py?ds=sidebyside diff --git a/librarian/pypdf.py b/librarian/pypdf.py index 0793951..9851cb1 100644 --- a/librarian/pypdf.py +++ b/librarian/pypdf.py @@ -18,8 +18,7 @@ from urllib2 import urlopen from lxml import etree -from xmlutils import Xmill, tag, tagged, ifoption, tag_open_close -from librarian.dcparser import Person +from xmlutils import Xmill, ifoption, tag_open_close from librarian import DCNS, get_resource, IOFile from librarian import functions from pdf import PDFFormat, substitute_hyphens, fix_hanging @@ -33,7 +32,8 @@ def escape(really): prefix = (u'' % (really and 1 or 0)) postfix = u'' if isinstance(value, list): - import pdb; pdb.set_trace() + import pdb + pdb.set_trace() if isinstance(value, tuple): return prefix + value[0], value[1] + postfix else: @@ -69,6 +69,7 @@ class EduModule(Xmill): def __init__(self, options=None, state=None): super(EduModule, self).__init__(options, state) self.activity_counter = 0 + self.activity_last = None self.exercise_counter = 0 def swap_endlines(txt): @@ -86,16 +87,15 @@ class EduModule(Xmill): return values def handle_rdf__RDF(self, _): - "skip metadata in generation" + """skip metadata in generation""" return @escape(True) def get_rightsinfo(self, element): rights_lic = self.get_dc(element, 'rights.license', True) - return u'' + \ - (rights_lic and u'%s' % rights_lic or '') +\ - u'%s' % self.get_dc(element, 'rights', True) +\ - u'' + return u'' + (rights_lic and u'%s' % rights_lic or '') + \ + u'%s' % self.get_dc(element, 'rights', True) + \ + u'' @escape(True) def get_authors(self, element, which=None): @@ -106,7 +106,7 @@ class EduModule(Xmill): dc.authors_expert else: authors = getattr(dc, "authors_%s" % which) - return u', '.join(author.readable() for author in authors) + return u', '.join(author.readable() for author in authors if author) @escape(1) def get_title(self, element): @@ -115,31 +115,31 @@ class EduModule(Xmill): def handle_utwor(self, element): lines = [ u''' - - - \\documentclass[%s]{wl} - \\usepackage{style}''' % self.options['customization_str'], - self.options['has_cover'] and '\usepackage{makecover}', - (self.options['morefloats'] == 'new' and '\usepackage[maxfloats=64]{morefloats}') or - (self.options['morefloats'] == 'old' and '\usepackage{morefloats}') or - (self.options['morefloats'] == 'none' and - u'''\\IfFileExists{morefloats.sty}{ - \\usepackage{morefloats} - }{}'''), - u'''\\def\\authors{%s}''' % self.get_authors(element), - u'''\\def\\authorsexpert{%s}''' % self.get_authors(element, 'expert'), - u'''\\def\\authorsscenario{%s}''' % self.get_authors(element, 'scenario'), - u'''\\def\\authorstextbook{%s}''' % self.get_authors(element, 'textbook'), - - u'''\\author{\\authors}''', - u'''\\title{%s}''' % self.get_title(element), - u'''\\def\\bookurl{%s}''' % self.options['wldoc'].book_info.url.canonical(), - u'''\\def\\rightsinfo{%s}''' % self.get_rightsinfo(element), - u''] + + + \\documentclass[%s]{wl} + \\usepackage{style}''' % self.options['customization_str'], + self.options['has_cover'] and '\usepackage{makecover}', + (self.options['morefloats'] == 'new' and '\usepackage[maxfloats=64]{morefloats}') or + (self.options['morefloats'] == 'old' and '\usepackage{morefloats}') or + (self.options['morefloats'] == 'none' and + u'''\\IfFileExists{morefloats.sty}{ + \\usepackage{morefloats} + }{}'''), + u'''\\def\\authors{%s}''' % self.get_authors(element), + u'''\\def\\authorsexpert{%s}''' % self.get_authors(element, 'expert'), + u'''\\def\\authorsscenario{%s}''' % self.get_authors(element, 'scenario'), + u'''\\def\\authorstextbook{%s}''' % self.get_authors(element, 'textbook'), + + u'''\\author{\\authors}''', + u'''\\title{%s}''' % self.get_title(element), + u'''\\def\\bookurl{%s}''' % self.options['wldoc'].book_info.url.canonical(), + u'''\\def\\rightsinfo{%s}''' % self.get_rightsinfo(element), + u'' + ] return u"".join(filter(None, lines)), u'' - @escape(1) def handle_powiesc(self, element): return u""" @@ -153,50 +153,46 @@ class EduModule(Xmill): return u'' % cmd, u'' handle_akap = \ - handle_akap = \ - handle_akap_cd = \ - handle_akap_cd = \ - handle_akap_dialog = \ - handle_akap_dialog = \ - handle_autor_utworu = \ - handle_dedykacja = \ - handle_didaskalia = \ - handle_didask_tekst = \ - handle_dlugi_cytat = \ - handle_dzielo_nadrzedne = \ - handle_lista_osoba = \ - handle_mat = \ - handle_miejsce_czas = \ - handle_motto = \ - handle_motto_podpis = \ - handle_naglowek_akt = \ - handle_naglowek_czesc = \ - handle_naglowek_listy = \ - handle_naglowek_osoba = \ - handle_naglowek_podrozdzial = \ - handle_naglowek_scena = \ - handle_nazwa_utworu = \ - handle_nota = \ - handle_osoba = \ - handle_pa = \ - handle_pe = \ - handle_podtytul = \ - handle_poezja_cyt = \ - handle_pr = \ - handle_pt = \ - handle_sekcja_asterysk = \ - handle_sekcja_swiatlo = \ - handle_separator_linia = \ - handle_slowo_obce = \ - handle_srodtytul = \ - handle_tytul_dziela = \ - handle_wyroznienie = \ - handle_dywiz = \ - handle_texcommand + handle_akap_cd = \ + handle_akap_dialog = \ + handle_autor_utworu = \ + handle_dedykacja = \ + handle_didaskalia = \ + handle_didask_tekst = \ + handle_dlugi_cytat = \ + handle_dzielo_nadrzedne = \ + handle_lista_osoba = \ + handle_mat = \ + handle_miejsce_czas = \ + handle_motto = \ + handle_motto_podpis = \ + handle_naglowek_akt = \ + handle_naglowek_czesc = \ + handle_naglowek_listy = \ + handle_naglowek_osoba = \ + handle_naglowek_scena = \ + handle_nazwa_utworu = \ + handle_nota = \ + handle_osoba = \ + handle_pa = \ + handle_pe = \ + handle_podtytul = \ + handle_poezja_cyt = \ + handle_pr = \ + handle_pt = \ + handle_sekcja_asterysk = \ + handle_sekcja_swiatlo = \ + handle_separator_linia = \ + handle_slowo_obce = \ + handle_srodtytul = \ + handle_tytul_dziela = \ + handle_wyroznienie = \ + handle_dywiz = \ + handle_texcommand def handle_naglowek_rozdzial(self, element): if not self.options['teacher']: - if element.text.startswith((u'Wiedza', u'Zadania', u'Słowniczek')): + if element.text.startswith((u'Wiedza', u'Zadania', u'Słowniczek', u'Dla ucznia')): self.state['mute'] = False else: self.state['mute'] = True @@ -204,9 +200,23 @@ class EduModule(Xmill): return self.handle_texcommand(element) handle_naglowek_rozdzial.unmuter = True + def handle_naglowek_podrozdzial(self, element): + self.activity_counter = 0 + if not self.options['teacher']: + if element.text.startswith(u'Dla ucznia'): + self.state['mute'] = False + return None + elif element.text.startswith(u'Dla nauczyciela'): + self.state['mute'] = True + return None + elif self.state['mute']: + return None + return self.handle_texcommand(element) + handle_naglowek_podrozdzial.unmuter = True def handle_uwaga(self, _e): return None + def handle_extra(self, _e): return None @@ -234,13 +244,16 @@ class EduModule(Xmill): opis = '' n = element.xpath('wskazowki') - if n: wskazowki = submill.generate(n[0]) - - else: wskazowki = '' + if n: + wskazowki = submill.generate(n[0]) + else: + wskazowki = '' n = element.xpath('pomoce') - if n: pomoce = submill.generate(n[0]) - else: pomoce = '' + if n: + pomoce = submill.generate(n[0]) + else: + pomoce = '' forma = ''.join(element.xpath('forma/text()')) @@ -248,9 +261,16 @@ class EduModule(Xmill): counter = self.activity_counter + if element.getnext().tag == 'aktywnosc' or (self.activity_last and self.activity_last.getnext() == element): + counter_tex = """%(counter)d.""" % locals() + else: + counter_tex = '' + + self.activity_last = element + return u""" -%(counter)d. +%(counter_tex)s %(czas)s %(forma)s @@ -276,7 +296,7 @@ class EduModule(Xmill): def handle_forma(self, *_): return - def handle_lista(self, element, attrs={}): + def handle_lista(self, element, attrs=None): ltype = element.attrib.get('typ', 'punkt') if not element.findall("punkt"): if ltype == 'czytelnia': @@ -288,16 +308,16 @@ class EduModule(Xmill): if surl is None: # print '** missing src on , setting default' surl = 'http://edukacjamedialna.edu.pl/lekcje/slowniczek/' - sxml = None - if surl: - sxml = etree.fromstring(self.options['wldoc'].provider.by_uri(surl).get_string()) - self.options = {'slowniczek': True, 'slowniczek_xml': sxml } + sxml = etree.fromstring(self.options['wldoc'].provider.by_uri(surl).get_string()) + self.options = {'slowniczek': True, 'slowniczek_xml': sxml} - listcmd = {'num': 'enumerate', - 'punkt': 'itemize', - 'alfa': 'itemize', - 'slowniczek': 'itemize', - 'czytelnia': 'itemize'}[ltype] + listcmd = { + 'num': 'enumerate', + 'punkt': 'itemize', + 'alfa': 'itemize', + 'slowniczek': 'itemize', + 'czytelnia': 'itemize' + }[ltype] return u'' % listcmd, u'' @@ -316,7 +336,7 @@ class EduModule(Xmill): typ = element.attrib['typ'] self.exercise_counter += 1 - if not typ in exercise_handlers: + if typ not in exercise_handlers: return '(no handler)' self.options = {'exercise_counter': self.exercise_counter} handler = exercise_handlers[typ](self.options, self.state) @@ -332,7 +352,10 @@ class EduModule(Xmill): if self.options['slowniczek_xml'] is not None and (nxt is None or nxt.tag != 'definiens'): sxml = self.options['slowniczek_xml'] assert element.text != '' - defloc = sxml.xpath("//definiendum[text()='%s']" % element.text) + if "'" in (element.text or ''): + defloc = sxml.xpath("//definiendum[text()=\"%s\"]" % (element.text or '').strip()) + else: + defloc = sxml.xpath("//definiendum[text()='%s']" % (element.text or '').strip()) if defloc: definiens = defloc[0].getnext() if definiens.tag == 'definiens': @@ -355,14 +378,13 @@ class EduModule(Xmill): max_col = len(ks) self.options = {'columnts': max_col} # styling: - # has_frames = int(element.attrib.get("ramki", "0")) - # if has_frames: frames_c = "framed" - # else: frames_c = "" - # return u"""""" % frames_c, u"
" + # has_frames = int(element.attrib.get("ramki", "0")) + # if has_frames: frames_c = "framed" + # else: frames_c = "" + # return u"""""" % frames_c, u"
" return u''' tabular%s - ''' % ('l' * max_col), \ - u'''tabular''' + ''' % ('l' * max_col), u'''tabular''' @escape(1) def handle_wiersz(self, element): @@ -388,6 +410,7 @@ class EduModule(Xmill): frmt = self.options['format'] name = element.attrib.get('nazwa', '').strip() image = frmt.get_image(name.strip()) + name = image.get_filename().rsplit('/', 1)[-1] img_path = "obraz/%s" % name.replace("_", "") frmt.attachments[img_path] = image return cmd("obraz", parms=[img_path])(self) @@ -402,8 +425,7 @@ class EduModule(Xmill): print '!! unknown