lkrsnik/STARK
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Sorted input + fixed certain errors

Browse Source
This commit is contained in:
Luka 2019-12-27 17:36:54 +01:00
parent f1bba0e58f
commit 36223d3a98
1 changed files with 11 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
dependency-parsetree.py
View File

@ -186,6 +186,10 @@ def create_trees(input_path, internal_saves, feats_detailed_dict={}, save=True):
corpus_size += 1 corpus_size += 1
for token_id, token in enumerate(token_nodes): for token_id, token in enumerate(token_nodes):
if isinstance(token.parent, int) or token.parent == '':
root = None
print('No parent: ' + sentence.id)
break
if int(token.parent) == 0: if int(token.parent) == 0:
token.set_parent(None) token.set_parent(None)
else: else:
@ -380,7 +384,7 @@ def count_trees(cpu_cores, all_trees, query_tree, create_output_string_functs, f
# else: # else:
# result_dict[r_k] = r_v # result_dict[r_k] = r_v
# 1.02 s (16 cores) # 1.02 s (16 cores)
if cpu_cores > 0: if cpu_cores > 1:
# input_data = (tree, query_tree, create_output_string_functs, filters) # input_data = (tree, query_tree, create_output_string_functs, filters)
all_unigrams = p.map(get_unigrams, [(tree, query_tree, create_output_string_functs, filters) for tree in all_trees]) all_unigrams = p.map(get_unigrams, [(tree, query_tree, create_output_string_functs, filters) for tree in all_trees])
for unigrams in all_unigrams: for unigrams in all_unigrams:
@ -559,16 +563,17 @@ def main():
already_processed, result_dict, unigrams_dict, corpus_size, feats_detailed_list = load_zipped_pickle( already_processed, result_dict, unigrams_dict, corpus_size, feats_detailed_list = load_zipped_pickle(
checkpoint_path) checkpoint_path)
for path in os.listdir(input_path): for path in sorted(os.listdir(input_path)):
path_obj = Path(input_path, path) path_obj = Path(input_path, path)
pathlist = path_obj.glob('**/*.conllu') pathlist = path_obj.glob('**/*.conllu')
if path_obj.name in already_processed: if path_obj.name in already_processed:
continue continue
start_exe_time = time.time() start_exe_time = time.time()
for path in pathlist: for path in sorted(pathlist):
# because path is object not string # because path is object not string
path_str = str(path) path_str = str(path)
# if Path(path_str).name == 'GF0003946-dedup.conllu':
# break
# print(path_in_str) # print(path_in_str)
(all_trees, form_dict, lemma_dict, upos_dict, xpos_dict, deprel_dict, sub_corpus_size, (all_trees, form_dict, lemma_dict, upos_dict, xpos_dict, deprel_dict, sub_corpus_size,
@ -585,9 +590,10 @@ def main():
already_processed.add(path_obj.name) already_processed.add(path_obj.name)
# 15.26
print("Execution time:") print("Execution time:")
print("--- %s seconds ---" % (time.time() - start_exe_time)) print("--- %s seconds ---" % (time.time() - start_exe_time))
# print(1 + 'asd')
save_zipped_pickle( save_zipped_pickle(
(already_processed, result_dict, unigrams_dict, corpus_size, feats_detailed_list), (already_processed, result_dict, unigrams_dict, corpus_size, feats_detailed_list),
checkpoint_path, protocol=2) checkpoint_path, protocol=2)