From: Marcin Koziej Date: Wed, 14 Nov 2012 16:10:57 +0000 (+0100) Subject: fixes for autotagger X-Git-Url: https://git.mdrn.pl/redakcja.git/commitdiff_plain/7499a88c62e78c49d7f9e393b8e67a315acc16ba fixes for autotagger --- diff --git a/apps/catalogue/management/edumed.py b/apps/catalogue/management/edumed.py index 250eccb7..c384dac2 100644 --- a/apps/catalogue/management/edumed.py +++ b/apps/catalogue/management/edumed.py @@ -4,7 +4,7 @@ import re from slughifi import slughifi -class Tagger: +class Tagger(object): def __init__(self, state, lines): self.state = state self.lines = lines @@ -17,7 +17,7 @@ class Tagger: ignore = [re.compile(r"^[\[][PA][\]] - [^ ]+$")] empty_line = re.compile(r"^\s+$") - + def skip_empty(self, position): while self.line(position) == "" or \ self.empty_line.match(self.line(position)) or \ @@ -38,12 +38,15 @@ Return None -- means that we can't tag it in any way @staticmethod def anymatches(regex): return lambda x: regex.match(x) - class Section(Tagger): looks_like = re.compile(r"^[IVX]+[.]\s+(.*)$") + def __init__(self, *a): + super(Section, self).__init__(*a) + self.is_podrozdzial = False + def tag(self, pos): pos2 = self.skip_empty(pos) pos = pos2 @@ -53,7 +56,8 @@ class Section(Tagger): return pos + 1 def __unicode__(self): - return self.wrap("naglowek_rozdzial", self.title) + return self.wrap(self.is_podrozdzial and "naglowek_podrozdzial" or "naglowek_rozdzial", + self.title) class Meta(Tagger): @@ -91,25 +95,41 @@ class Informacje(Tagger): class List(Tagger): - point = re.compile(r"^[\s]*([-*])") - - def tag(self, pos): + point = re.compile(r"^[\s]*[-*·]{1,2}(.*)") + num = re.compile(r"^[\s]*[0-9a-z]{1,2}[.]\s+(.*)") + + def __init__(self, *args): + + super(List, self).__init__(*args) self.items = [] + self.type = 'punkt' + + def tag(self, pos): while True: l = self.line(pos) m = self.point.match(l) + if not m: + m = self.num.match(l) + if m: self.type = 'num' if l and m: - self.items.append(l[1:].strip()) + self.items.append(m.groups()[0].lstrip()) pos += 1 else: break if self.items: return pos + def append(self, tagger): + self.items.append(tagger) + def __unicode__(self): - s = '' + s = '' % self.type for i in self.items: - s += "\n%s" % i + if isinstance(i, list): + x = "\n".join(map(lambda elem: unicode(elem), i)) + else: + x = unicode(i) + s += "\n%s" % x s += "\n\n" return s @@ -120,21 +140,22 @@ class Paragraph(Tagger): re.compile(r"^[\s]*$") ] podrozdzial = [ - re.compile(r"[\s]*(przebieg zajęć|opcje dodatkowe)[\s]*", re.I), + re.compile(r"[\s]*(przebieg zaj..|opcje dodatkowe)[\s]*", re.I), ] + def tag(self, pos): self.line = self.lines[pos] self.ignore = False self.is_podrozdzial = False - + for x in self.remove_this: if x.match(self.line): self.ignore = True - + for x in self.podrozdzial: if x.match(self.line): self.is_podrozdzial = True - + return pos + 1 def __unicode__(self): @@ -196,7 +217,7 @@ def eatseq(pos, *taggers): return (tuple(good), pos) -def tagger(text): +def tagger(text, pretty_print=False): """ tagger(text) function name and signature is a contract. returns auto-tagged text @@ -208,7 +229,7 @@ returns auto-tagged text content = [] state = {} info = Informacje(state, lines) - + ((info,), pos) = eatseq(pos, info) # print "[i] %d. %s" % (pos, lines[pos]) @@ -226,22 +247,61 @@ returns auto-tagged text pos += 1 if pos >= len(lines): break - - return toxml(content) + + return toxml(content, pretty_print=pretty_print) dc_fixed = { 'description': u'Publikacja zrealizowana w ramach projektu Cyfrowa Przyszłość (http://cyfrowaprzyszlosc.pl).', - 'relation': u'moduły powiązane linki', + 'relation': u'moduły powiązane linki', 'description.material': u'linki do załączników', 'rights': u'Creative Commons Uznanie autorstwa - Na tych samych warunkach 3.0', } +def find_block(content, title_re, begin=-1, end=-1): + title_re = re.compile(title_re, re.I | re.UNICODE) + print "looking for %s" % title_re.pattern + if title_re.pattern[0:6] == 'pomoce': + import pdb; pdb.set_trace() + + rb = -1 + if begin < 0: begin = 0 + if end < 0: end = len(content) + + for i in range(begin, end): + elem = content[i] + if isinstance(elem, Paragraph): + if title_re.match(elem.line): + rb = i + continue + if isinstance(elem, Section): + if title_re.match(elem.title): + rb = i + continue + if rb >= 0: + if isinstance(elem, List): + continue + if isinstance(elem, Paragraph) and elem.line: + continue + break + if rb >= 0: + return rb, i + + +def remove_block(content, title_re, removed=None): + rb, re = find_block(content, title_re) + + if removed is not None and isinstance(removed, list): + removed += content[rb:re][:] + content[rb:re] = [] + return content + + def mark_activities(content): i = 0 tl = len(content) is_przebieg = re.compile(r"[\s]*przebieg zaj..[\s]*", re.I) - # import pdb; pdb.set_trace() + is_next_section = re.compile(r"^[IVX]+[.]? ") is_activity = re.compile(r"^[0-9]+[.]? (.+)") @@ -306,7 +366,8 @@ def mark_dictionary(content): i = 0 is_dictionary = re.compile(r"[\s]*s.owniczek[\s]*", re.I) is_dictentry = re.compile(r"([^-]+) - (.+)") - slowniczek = [] + slowniczek = content[0].spawn(List) + slowniczek.type = 'slowniczek' while i < len(content): e = content[i] if isinstance(e, Section): @@ -314,14 +375,15 @@ def mark_dictionary(content): db = i + 1 elif db >= 1: de = i - content[db:de] = [Container('slowniczek', *slowniczek)] + content[db:de] = [slowniczek] break elif db >= 0: if isinstance(e, Paragraph): m = is_dictentry.match(e.line) if m: - slowniczek.append(Container('definiendum', m.groups()[0])) - slowniczek.append(Container('definiens', m.groups()[1])) + slowniczek.append([Container('definiendum', m.groups()[0]), + Container('definiens', m.groups()[1])]) + else: slowniczek.append(e) i += 1 @@ -329,9 +391,36 @@ def mark_dictionary(content): return content -def toxml(content): +def move_evaluation(content): + evaluation = [] + + content = remove_block(content, r"ewaluacja[+ PA\[\].]*", evaluation) + if evaluation: + # print "found evaluation %s" % (evaluation,) + evaluation[0].is_podrozdzial = True + # evaluation place + opcje_dodatkowe = find_block(content, r"opcje dodatkowe\s*") + if opcje_dodatkowe: + # print "putting evaluation just before opcje dodatkowe @ %s" % (opcje_dodatkowe, ) + content[opcje_dodatkowe[0]:opcje_dodatkowe[0]] = evaluation + else: + materialy = find_block(content, r"materia.y[+ AP\[\].]*") + if materialy: + # print "putting evaluation just before materialy @ %s" % (materialy, ) + content[materialy[0]:materialy[0]] = evaluation + else: + print "er.. no idea where to place evaluation" + return content + + +def toxml(content, pretty_print=False): + # some transformations content = mark_activities(content) content = mark_dictionary(content) + content = remove_block(content, r"wykorzyst(yw)?ane metody[+ PA\[\].]*") + content = remove_block(content, r"(pomoce|potrzebne materia.y)[+ PA\[\]]*") + content = move_evaluation(content) + info = content.pop(0) state = info.state @@ -387,9 +476,9 @@ def toxml(content): p(u'') t(u'nazwa_utworu', meta.get(u'Tytuł modułu', u'')) - p(u'') - a(u'Numer porządkowy: %s' % meta.get(u'Numer porządkowy', u'')) - p(u'') + # p(u'') + a(u'' % meta.get(u'Numer porządkowy', u'')) + # p(u'') p(unicode(info.title)) for elm in content: @@ -401,6 +490,12 @@ def toxml(content): p(u'') p(u'') + if pretty_print: + from lxml import etree + from StringIO import StringIO + xml = etree.parse(StringIO(holder['xml'])) + holder['xml'] = etree.tostring(xml, pretty_print=pretty_print, encoding=unicode) + return holder['xml']