From 5867c7992131569190cda33e03bf027db77a10bf Mon Sep 17 00:00:00 2001
From: ceramisch <carlos.ramisch@lis-lab.fr>
Date: Mon, 16 Dec 2024 00:00:38 +0100
Subject: [PATCH] Minimal update un Sequoia simplification script
---
cm-code/bert-minimal.py | 4 ++--
cm-code/petits.conllu | 3 ++-
lib/conllulib.py | 4 ++--
sequoia/bin/simplify_sequoia.py | 18 +++++++++---------
4 files changed, 15 insertions(+), 14 deletions(-)
diff --git a/cm-code/bert-minimal.py b/cm-code/bert-minimal.py
index 477e1e4..114b2e5 100755
--- a/cm-code/bert-minimal.py
+++ b/cm-code/bert-minimal.py
@@ -6,8 +6,8 @@ from transformers import AutoModel, AutoTokenizer
name = 'almanach/camembert-base'
#sent = "Des poids lourds et engins en feu \
# dans une entreprise en Vendée ."
-#sent = "La gare routière attend toujours ses illuminations ."
-sent = "Quelle surprise ! Arturo a la covid"
+sent = "La gare routière attend toujours ses illuminations ."
+#sent = "Quelle surprise ! Arturo a la covid"
tok = AutoTokenizer.from_pretrained(name)
model = AutoModel.from_pretrained(name)
diff --git a/cm-code/petits.conllu b/cm-code/petits.conllu
index ba3517c..be0ba52 100644
--- a/cm-code/petits.conllu
+++ b/cm-code/petits.conllu
@@ -1,4 +1,5 @@
-# text = Les petits ruisseaux font les grandes rivières .
+# global.columns = ID FORM LEMMA UPOS XPOS FEATS HEAD DEPREL DEPS MISC
+# text = Les petits ruisseaux font les grandes rivières.
1 Les le DET _ Definite=Def|Number=Plur|PronType=Art 3 det _ _
2 petits petit ADJ _ Gender=Masc|Number=Plur 3 amod _ _
3 ruisseaux ruisseau NOUN _ Gender=Masc|Number=Plur 4 nsubj _ _
diff --git a/lib/conllulib.py b/lib/conllulib.py
index b587782..fa707f4 100644
--- a/lib/conllulib.py
+++ b/lib/conllulib.py
@@ -538,8 +538,8 @@ class TransBasedConfig(object):
`next_act` is a string among "SHIFT", "RIGHT-ARC-X" or "LEFT-ARC-X" where
"X" is the name of any valid syntactic relation label (deprel).
Returns a new syntactic relation added by the action, or None for "SHIFT"
- Returned relation is a triple (mod, head, deprel) with modifier, head, and
- deprel label if `add_deprel=True` (default), or a pair (mod, head) if
+ Returned relation is a triple (dep, head, deprel) with dependent, head, and
+ deprel label if `add_deprel=True` (default), or a pair (dep, head) if
`add_deprel=False`.
"""
if next_act == "SHIFT":
diff --git a/sequoia/bin/simplify_sequoia.py b/sequoia/bin/simplify_sequoia.py
index 3d4afda..70cd719 100755
--- a/sequoia/bin/simplify_sequoia.py
+++ b/sequoia/bin/simplify_sequoia.py
@@ -134,10 +134,10 @@ def is_projective(sent):
if dep_id > head_id :
start = head_id
end = dep_id
- for token_i in range(start,end-1): # sent is 0-indexed, ID is 1-indexed
- if sent[token_i]["head"] < start or sent[token_i]["head"] > end :
- return False
- return True
+ for token_i in range(start, end - 1): # sent is 0-indexed, ID is 1-indexed
+ if sent[token_i]["head"] < start or sent[token_i]["head"] > end :
+ return False
+ return True
#########################################
@@ -150,14 +150,14 @@ def remove_subrelations(sent):
#########################################
if len(sys.argv) != 2:
- print('Usage: {} <input_corpus.conllu>'.format(sys.argv[0]), file=sys.stderr)
+ print('Usage: {} <input_corpus.conllu>'.format(sys.argv[0]), file=sys.stderr)
exit(-1)
with open(sys.argv[1], "r", encoding="UTF=8") as f:
np_counter = range_counter = del_ne_counter = 0
del_ssense_counter = mod_ssense_counter = 0 #subrel_counter = 0
- np_ids = []
- for sent in conllu.parse_incr(f):
+ np_ids = []
+ for sent in conllu.parse_incr(f):
range_counter = range_counter + remove_range_tokens(sent)
del_ssense_ci, mod_ssense_ci = simplify_supersense(sent)
del_ssense_counter = del_ssense_counter + del_ssense_ci
@@ -171,7 +171,7 @@ with open(sys.argv[1], "r", encoding="UTF=8") as f:
print(sent.serialize(), end="")
else:
np_counter += 1
- np_ids.append(sent.metadata["sent_id"])
+ np_ids.append((sent.metadata["sent_id"],len(sent)))
print( "{} range tokens removed.\n".format(range_counter), file=sys.stderr)
@@ -182,4 +182,4 @@ print( "{} supersense tags modified (complex operators).\n".format(mod_ssense_co
#print( "{} subrelations removed from deprel.".format(subrel_counter), file=sys.stderr)
print( "{} non-projective sentences removed:".format(np_counter), file=sys.stderr)
-print(", ".join(np_ids), file=sys.stderr)
+print("\n".join([f"{np_id} -> {lgth}" for (np_id, lgth) in np_ids]), file=sys.stderr)
--
GitLab