explosion/spaCy
1
1
Fork 0
mirror of https://github.com/explosion/spaCy.git synced 2024-11-13 05:07:03 +03:00
Code Issues Packages Projects Releases Wiki Activity

* Tmp commit, while switch to new format that assumes alignment happens during training

Browse Source
This commit is contained in:
Matthew Honnibal 2015-05-23 17:39:04 +02:00
parent f35503018e
commit 983d954ef4
1 changed files with 6 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
bin/prepare_treebank.py
View File

@ -52,7 +52,7 @@ def format_doc(section, filename, raw_paras, ptb_loc, dep_loc):
word_idx = 0 word_idx = 0
i = 0 i = 0
doc = {'id': 'wsj_%s%s' % (section, filename), 'paragraphs': []} doc = {'id': filename, 'paragraphs': []}
for raw_sents in raw_paras: for raw_sents in raw_paras:
para = {'raw': ' '.join(sent.replace('<SEP>', '') for sent in raw_sents), para = {'raw': ' '.join(sent.replace('<SEP>', '') for sent in raw_sents),
'segmented': '<SENT>'.join(raw_sents), 'segmented': '<SENT>'.join(raw_sents),
@ -67,8 +67,8 @@ def format_doc(section, filename, raw_paras, ptb_loc, dep_loc):
_, annot = read_conll.parse(dep_sents[i], strip_bad_periods=True) _, annot = read_conll.parse(dep_sents[i], strip_bad_periods=True)
indices, word_idx, offset = _get_word_indices(raw_sent, 0, offset) indices, word_idx, offset = _get_word_indices(raw_sent, 0, offset)
for j, token in enumerate(annot): for j, token in enumerate(annot):
head = indices[token['head']] if token['head'] != -1 else -1
try: try:
head = indices[token['head']] if token['head'] != -1 else -1
para['tokens'].append({ para['tokens'].append({
'start': indices[token['id']], 'start': indices[token['id']],
'orth': words[j], 'orth': words[j],
@ -76,9 +76,6 @@ def format_doc(section, filename, raw_paras, ptb_loc, dep_loc):
'head': head, 'head': head,
'dep': token['dep']}) 'dep': token['dep']})
except: except:
print sorted(indices.items())
print token
print raw_sent
raise raise
for label, start, end in brackets: for label, start, end in brackets:
if start != end: if start != end:
@ -95,20 +92,18 @@ def main(onto_dir, raw_dir, out_dir):
section = str(i) if i >= 10 else ('0' + str(i)) section = str(i) if i >= 10 else ('0' + str(i))
raw_loc = path.join(raw_dir, 'wsj%s.json' % section) raw_loc = path.join(raw_dir, 'wsj%s.json' % section)
docs = [] docs = []
for j, raw_paras in enumerate(_iter_raw_files(raw_loc)): for j, (filename, raw_paras) in enumerate(_iter_raw_files(raw_loc)):
if section == '00': if section == '00':
j += 1 j += 1
filename = str(j) if j >= 9 else ('0' + str(j))
if section == '04' and filename == '55': if section == '04' and filename == '55':
continue continue
ptb_loc = path.join(onto_dir, section, 'wsj_%s%s.mrg' % (section, filename)) ptb_loc = path.join(onto_dir, section, '%s.parse' % filename)
dep_loc = ptb_loc + '.3.pa.gs.tab' dep_loc = ptb_loc + '.dep'
if path.exists(ptb_loc) and path.exists(dep_loc): if path.exists(ptb_loc) and path.exists(dep_loc):
print ptb_loc
doc = format_doc(section, filename, raw_paras, ptb_loc, dep_loc) doc = format_doc(section, filename, raw_paras, ptb_loc, dep_loc)
docs.append(doc) docs.append(doc)
with open(path.join(out_dir, '%s.json' % section), 'w') as file_: with open(path.join(out_dir, '%s.json' % section), 'w') as file_:
json.dump(docs, file_) json.dump(docs, file_, indent=4)
if __name__ == '__main__': if __name__ == '__main__':