def discourse_rels(request):
db = request.corpus
words = db.words
text_ids = db.corpus.attribute(
corpus_d_sattr.get(db.corpus_name, 'text_id'), 's')
results = db.db.discourse.find({'_user': request.user})
docs = {}
rel_counts = defaultdict(int)
rel_occurrences = defaultdict(list)
sum_all = 0
for r in results:
try:
docid = int(r['_docno'])
except KeyError:
pass
else:
txt0 = text_ids[docid]
txt = "%s: %s" % (txt0[2], ' '.join(words[txt0[0]:txt0[0] + 5]))
txt = txt.decode('ISO-8859-15')
rels = isolate_relations(r['relations'])
for k in rels:
sum_all += len(rels[k])
rel_counts[k] += len(rels[k])
rel_occurrences[k].append((docid, txt, rels[k]))
result = []
for rel in sorted(rel_counts.keys(), key=lambda x: -rel_counts[x]):
result.append((rel, rel_counts[rel], rel_occurrences[rel]))
return render_template('discourse_rels.html',
corpus_name=db.corpus_name,
results=result,
sum_all=sum_all)
def __init__(self, ctx, db):
self.ctx = ctx
self.texts = db.corpus.attribute(
corpus_d_sattr.get(db.corpus_name, 'text_id'), 's')
self.sentences = db.corpus.attribute("s", 's')
self.words = db.words
self.postags = db.corpus.attribute("pos", 'p')
def list_discourse(request):
db = request.corpus
words = db.words
text_ids = db.corpus.attribute(
corpus_d_sattr.get(db.corpus_name, 'text_id'), 's')
docids = sorted(
set([
r['_docno'] for r in db.db.discourse.find(
{'_user': {
'$in': [request.user, '*gold*']
}}) if '_docno' in r
]))
doc_lst = []
for docid in docids:
txt0 = text_ids[docid]
txt = "%s: %s" % (txt0[2], ' '.join(words[txt0[0]:txt0[0] + 5]))
if request.user in ADMINS:
users = [
doc['_user'] for doc in db.db.discourse.find({'_docno': docid})
]
else:
users = [
doc['_user'] for doc in db.db.discourse.find({'_docno': docid})
if (doc['_user'] in ['*gold*', request.user]
or request.user is not None
and doc['_user'].startswith(request.user + '*'))
]
doc_lst.append((request.user, docid, txt.decode('ISO-8859-15'), users))
return render_template('discourse_list.html',
corpus_name=db.corpus_name,
user=request.user,
results=doc_lst)
def render_discourse(request, disc_no):
db = request.corpus
corpus = db.corpus
t_id = int(disc_no)
doc = db.get_discourse(t_id, request.user)
texts = corpus.attribute(corpus_d_sattr.get(db.corpus_name, 'text_id'),
's')
sents = corpus.attribute("s", 's')
start, end, text_attrs = texts[t_id]
sent_id = sents.cpos2struc(start)
response = render_template_nocache(
'discourse.html',
corpus_name=json.dumps(request.corpus.corpus_name),
disc_id=disc_no,
sent_id=sent_id,
sentences=json.dumps(doc['sentences']),
edus=json.dumps(doc['edus']),
tokens=json.dumps(doc['tokens']),
indent=json.dumps(doc['indent']),
relations=json.dumps(doc.get('relations', '')),
nonedu=json.dumps(doc.get('nonedu', {})),
uedus=json.dumps(doc.get('uedus', {})),
topics=json.dumps(doc.get('topics', [])))
request.set_corpus_cookie(response)
return response
def __init__(self, ctx, db):
self.ctx = ctx
self.texts = db.corpus.attribute(
corpus_d_sattr.get(db.corpus_name, 'text_id'), 's')
self.sentences = db.corpus.attribute("s", 's')
self.words = db.words
self.postags = db.corpus.attribute("pos", 'p')
self.morph = db.corpus.attribute("morph", 'p')
self.deprel = db.corpus.attribute("deprel", "p")
self.attach = db.corpus.attribute("attach", "p")
self.lemma = db.corpus.attribute("lemma", "p")
self.db = db
def render_sentence(request, sent_no):
db = request.corpus
tueba_corpus = db.corpus
sno = int(sent_no) - 1
words = db.words
sents = db.sentences
texts = tueba_corpus.attribute(corpus_sattr.get(db.corpus_name, 'text_id'),
's')
texts_d = tueba_corpus.attribute(
corpus_d_sattr.get(db.corpus_name, 'text_id'), 's')
max_sent = len(sents)
start, end = sents[sno][:2]
tokens = []
for i in xrange(start, end + 1):
tokens.append(words[i].decode('ISO-8859-1'))
t_id = texts.cpos2struc(end - 1)
t_id_d = texts_d.cpos2struc(end - 1)
unused_start, unused_end, t_attrs = texts[t_id]
if db.corpus_name in corpus_urls:
text_url = corpus_urls[db.corpus_name](t_attrs, db.corpus_name)
else:
text_url = '#'
parses = db.get_parses(sno)
alignments = db.get_alignments(sno)
trees_out = StringIO()
parses_html = ''
response = render_template('sentence.tmpl',
sent_id=sno + 1,
sent_text=' '.join(tokens),
parses_html=parses_html,
text_id=t_attrs,
text_url=text_url,
prev_sent='/pycwb/sentence/%d' % (sno, ),
next_sent='/pycwb/sentence/%d' % (sno + 2, ),
disc_id=t_id_d,
corpus_name=request.corpus.corpus_name,
has_gold=False)
request.set_corpus_cookie(response)
return response
edu_markable = Edu()
edu_markable.span = (start + ctx_start, end + ctx_start)
edu_markable.xml_id = 'edu_%s_%d_%d' % (t_id, next_sent,
sub_edu)
text_markable.edus['%d.%d' %
(next_sent, sub_edu)] = edu_markable
edu_markable.edu_idx = len(text_markable.edu_list)
text_markable.edu_list.append(edu_markable)
ctx.register_object(edu_markable)
parse_relations(doc['relations'], text_markable, ctx)
if __name__ == '__main__':
db = database.get_corpus('TUEBA4')
text_ids = db.corpus.attribute(
corpus_d_sattr.get(db.corpus_name, 'text_id'), 's')
if len(sys.argv) > 1:
annotator = sys.argv[1]
else:
annotator = '*gold*'
results = db.db.discourse.find({'_user': annotator})
doc = make_implicit_doc()
print '<?xml version="1.0" encoding="ISO-8859-15"?>'
print '<exml-doc>'
reader = DiscourseReader(doc, db)
# do the actual conversion
for r in results:
try:
docid = int(r['_docno'])
except KeyError:
pass
def render_sentence(request, sent_no):
db = request.corpus
tueba_corpus = db.corpus
sno = int(sent_no) - 1
words = db.words
sents = db.sentences
texts = tueba_corpus.attribute(corpus_sattr.get(db.corpus_name, 'text_id'),
's')
texts_d = tueba_corpus.attribute(
corpus_d_sattr.get(db.corpus_name, 'text_id'), 's')
max_sent = len(sents)
start, end = sents[sno][:2]
tokens = []
for i in xrange(start, end + 1):
tokens.append(words[i].decode('ISO-8859-1'))
t_id = texts.cpos2struc(end - 1)
coref = db.db.referential.find_one({'_id': t_id})
discourse = db.db.discourse.find_one({'_id': '%s~*gold*' % (t_id, )})
t_id_d = texts_d.cpos2struc(end - 1)
unused_start, unused_end, t_attrs = texts[t_id]
if db.corpus_name in corpus_urls:
text_url = corpus_urls[db.corpus_name](t_attrs, db.corpus_name)
else:
text_url = '#'
parses = db.get_parses(sno)
alignments = db.get_alignments(sno)
trees_out = StringIO()
names_parses = sorted([k for k in parses.iterkeys() if k != '_id'])
names_alignments = sorted([k for k in alignments.iterkeys() if k != '_id'])
if coref is not None:
names_coref = sorted([k for k in coref.iterkeys() if k != '_id'])
else:
names_coref = []
annotations = db.find_annotations([start, end], '*gold*')
if names_parses or names_alignments or annotations:
print >> trees_out, '<div id="parses-tabs">'
print >> trees_out, '<ul class="nav nav-tabs">'
for k in names_parses:
print >> trees_out, '<li><a href="#parses-%s">%s (parse)</a></li>' % (
k, k)
for k in names_alignments:
print >> trees_out, '<li><a href="#alignments-%s">%s (align)</a></li>' % (
k, k)
for k in names_coref:
print >> trees_out, '<li><a href="#coref-%s">%s (coref)</a></li>' % (
k, k)
levels = defaultdict(StringIO)
for anno in annotations:
level = anno['level']
schema.schemas[level].make_display(anno, db, levels[level], None)
names = sorted(levels.iterkeys())
for k in names:
print >> trees_out, '<li><a href="#level-tabs-%s">%s</a></li>' % (
k, k)
print >> trees_out, '</ul>'
for k in names_parses:
v = parses[k]
print >> trees_out, '<div id="parses-%s">' % (k, )
#trees_out.write('<b>%s</b> <a href="javascript:$(\'tree:%s\').toggle()">[show]</a><br/>\n'%(k,k))
t = export.from_json(v)
csstree.write_html(t, trees_out, _id='tree-' + k)
print >> trees_out, '</div>'
for k in names_alignments:
v = alignments[k]
print >> trees_out, '<div id="alignments-%s">' % (k, )
write_alignment(v, trees_out)
print >> trees_out, '</div>'
for k in names_coref:
v = coref[k]
print >> trees_out, '<div id="coref-%s">' % (k, )
write_coref(db, v, trees_out, start, end + 1)
print >> trees_out, '</div>'
for k in names:
print >> trees_out, '<div id="level-tabs-%s">' % (k, )
trees_out.write(''.join(levels[k].getvalue()))
print >> trees_out, "</div>"
print >> trees_out, '</div>'
parses_html = trees_out.getvalue().decode('ISO-8859-15')
else:
parses_html = ''
response = render_template('sentence.tmpl',
sent_id=sno + 1,
sent_text=' '.join(tokens),
parses_html=parses_html,
text_id=t_attrs,
text_url=text_url,
prev_sent='/pycwb/sentence/%d' % (sno, ),
next_sent='/pycwb/sentence/%d' % (sno + 2, ),
disc_id=t_id_d,
corpus_name=request.corpus.corpus_name,
has_gold=(discourse is not None))
request.set_corpus_cookie(response)
return response
