batchalign 0.7.6a30__tar.gz → 0.7.6a32__tar.gz

{batchalign-0.7.6a30/batchalign.egg-info → batchalign-0.7.6a32}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: batchalign
-Version: 0.7.6a30
+Version: 0.7.6a32
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu

{batchalign-0.7.6a30 → batchalign-0.7.6a32}/batchalign/pipelines/fa/whisper_fa.py

@@ -49,7 +49,7 @@ class WhisperFAEngine(BatchalignEngine):
             if not isinstance(i, Utterance):
                 continue
             if i.alignment == None:
-                warnings.warn("We found at least one utterance without utterance-level alignment; this is usually not an issue, but if the entire transcript is unaligned, please run a pipeline with `Task.UTTERANCE_TIMING_RECOVERY` (\"bulletize\") before running forced-alignment.")
+                warnings.warn("We found at least one utterance without utterance-level alignment; this is usually not an issue, but if the entire transcript is unaligned, it means that utterance level timing recovery (which is fuzzy using ASR) failed due to the audio clarity. On this transcript, before running forced-alignment, please supply utterance-level links.")
                 continue
 
             # pop the previous group onto the stack

batchalign-0.7.6a32/batchalign/pipelines/morphosyntax/fr/apm.py

@@ -0,0 +1,15 @@
+"""
+auditory plural marking for french
+"""
+
+from .apmn import APMN
+
+apmn_lits = [
+    j.strip()
+    for i in APMN.strip().split("\n") if i.strip() != ""
+ for j in i.strip().split("\t") if j.strip() != ""
+]
+
+def is_apm_noun(n):
+    return bool(n.strip().lower() in apmn_lits)
+

batchalign-0.7.6a32/batchalign/pipelines/morphosyntax/fr/apmn.py

@@ -0,0 +1,81 @@
+APMN = """
+amiral	amiraux
+animal	animaux
+annal	annaux
+anormal	anormaux
+anticlérical	anticléricaux
+arsenal	arsenaux
+bocal	bocaux
+canal	canaux
+cantal	cantaux
+capital	capitaux
+caporal	caporaux
+cardinal	cardinaux
+central	centraux
+chenal	chenaux
+cheval	chevaux
+clérical	cléricaux
+collatéral	collatéraux
+colonial	coloniaux
+commensal	commensaux
+communal	communaux
+confessionnal	confessionnaux
+cordial	cordiaux
+corporal	corporaux
+cristal	cristaux
+cérébral	cérébraux
+fanal	fanaux
+frontal	frontaux
+fédéral	fédéraux
+féodal	féodaux
+gardénal	gardénaux
+général	généraux
+hôpital	hôpitaux
+idéal	idéaux
+international	internationaux
+journal	journaux
+libéral	libéraux
+local	locaux
+madrigal	madrigaux
+marsupial	marsupiaux
+maréchal	maréchaux
+mal	maux
+minéral	minéraux
+moral	moraux
+méridional	méridionaux
+métal	métaux
+nasal	nasaux
+national	nationaux
+normal	normaux
+numéral	numéraux
+occidental	occidentaux
+occipital	occipitaux
+oral	oraux
+ordinal	ordinaux
+oriental	orientaux
+original	originaux
+piédestal	piédestaux
+principal	principaux
+provincial	provinciaux
+quintal	quintaux
+radical	radicaux
+rival	rivaux
+rural	ruraux
+régional	régionaux
+sentimental	sentimentaux
+signal	signaux
+social	sociaux
+sénéchal	sénéchaux
+temporal	temporaux
+tergal	tergaux
+total	totaux
+tribunal	tribunaux
+urinal	urinaux
+vassal	vassaux
+val	vaux
+végétal	végétaux
+véronal	véronals
+éditorial	éditoriaux
+égal	égaux
+étal	étau
+"""

{batchalign-0.7.6a30 → batchalign-0.7.6a32}/batchalign/pipelines/morphosyntax/ud.py

@@ -216,6 +216,12 @@ def handler__NOUN(word, lang=None):
     case  = feats.get("Case", "").replace(",", "")
     type  = feats.get("PronType", "")
 
+    apm = ""
+    if lang == "fr":
+        from batchalign.pipelines.morphosyntax.fr.apm import is_apm_noun
+        apm = "apm" if is_apm_noun(word.text) else ""
+
+
     if word.deprel == "obj" and case.strip() == "":
         case = "Acc"
 
@@ -227,7 +233,7 @@ def handler__NOUN(word, lang=None):
     if gender_str == "-Com,Neut" or gender_str == "-Com" or gender_str == "-ComNeut": gender_str=""
     if number_str == "-Sing": number_str=""
 
-    return handler(word, lang)+gender_str+number_str+stringify_feats(case, type)+ger
+    return handler(word, lang)+gender_str+number_str+stringify_feats(case, type)+ger+stringify_feats(apm)
 
 def handler__PROPN(word, lang=None):
     # code as noun

batchalign-0.7.6a32/batchalign/version

@@ -0,0 +1,3 @@
+0.7.6-alpha.32
+November 26, 2024
+French APM

{batchalign-0.7.6a30 → batchalign-0.7.6a32/batchalign.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: batchalign
-Version: 0.7.6a30
+Version: 0.7.6a32
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu

{batchalign-0.7.6a30 → batchalign-0.7.6a32}/batchalign.egg-info/SOURCES.txt

@@ -73,6 +73,8 @@ batchalign/pipelines/morphosyntax/__init__.py
 batchalign/pipelines/morphosyntax/coref.py
 batchalign/pipelines/morphosyntax/ud.py
 batchalign/pipelines/morphosyntax/en/irr.py
+batchalign/pipelines/morphosyntax/fr/apm.py
+batchalign/pipelines/morphosyntax/fr/apmn.py
 batchalign/pipelines/morphosyntax/fr/case.py
 batchalign/pipelines/morphosyntax/ja/verbforms.py
 batchalign/pipelines/speaker/__init__.py

batchalign-0.7.6a30/batchalign/version

@@ -1,3 +0,0 @@
-0.7.6-alpha.30
-November 22, 2024
-French changes