batchalign 0.8.2.post2__tar.gz → 0.8.2.post3__tar.gz

{batchalign-0.8.2.post2/batchalign.egg-info → batchalign-0.8.2.post3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: batchalign
-Version: 0.8.2.post2
+Version: 0.8.2.post3
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu

{batchalign-0.8.2.post2 → batchalign-0.8.2.post3}/batchalign/cli/cli.py

@@ -415,10 +415,11 @@ def benchmark(ctx, in_dir, out_dir, lang, num_speakers, whisper, whisper_oai, **
 def compare(ctx, in_dir, out_dir, lang, **kwargs):
     """Compare transcripts against gold-standard references.
 
-    For each FILE.cha in IN_DIR, expects a companion FILE.gold.cha in the
-    same directory.  Runs morphosyntax analysis on the main transcript, then
-    produces a word-level diff stored as %%xsrep / %%xsmor tiers and writes
-    error metrics to a .compare.csv file in OUT_DIR.
+    For each FILE.cha in IN_DIR, looks for a companion FILE.gold.cha in the
+    same directory.  If not found, falls back to template.gold.cha as the
+    default gold for all files.  Runs morphosyntax analysis on the main
+    transcript, then produces a word-level diff stored as %%xsrep / %%xsmor
+    tiers and writes error metrics to a .compare.csv file in OUT_DIR.
     """
 
     _dispatch("compare", lang, 1, ["cha"], ctx,

{batchalign-0.8.2.post2 → batchalign-0.8.2.post3}/batchalign/cli/dispatch.py

@@ -204,13 +204,17 @@ def _run_pipeline_for_file(command, pipeline, file, output, loader_info, writer_
         if file.endswith(".gold.cha"):
             return
 
-        # Find companion gold file
+        # Find companion gold file: FILE.gold.cha overrides, then
+        # template.gold.cha serves as the default for the whole directory
         p = P(file)
         gold_path = p.parent / (p.stem + ".gold.cha")
+        if not gold_path.exists():
+            gold_path = p.parent / "template.gold.cha"
         if not gold_path.exists():
             raise FileNotFoundError(
                 f"No gold .cha file found for comparison. "
-                f"main: {p.name}, expected: {gold_path.name}, looked in: {str(gold_path)}"
+                f"main: {p.name}, expected: {p.stem}.gold.cha or template.gold.cha, "
+                f"looked in: {str(p.parent)}"
             )
 
         main_doc = CHATFile(path=str(p)).doc

{batchalign-0.8.2.post2 → batchalign-0.8.2.post3}/batchalign/formats/chat/generator.py

@@ -131,6 +131,9 @@ def generate_chat_utterance(utterance: Utterance, special_mor=False, write_wor=T
             prefix = "+" if tok.status == "extra_main" else ("-" if tok.status == "extra_gold" else "")
             xsrep_parts.append(f"{prefix}{tok.text}")
             xsmor_parts.append(f"{prefix}{tok.pos or '?'}")
+        # Strip PUNCT POS from final element, keep the delimiter char (same as %mor)
+        if xsmor_parts and xsrep_parts and xsmor_parts[-1] == "PUNCT":
+            xsmor_parts[-1] = xsrep_parts[-1]
         result.append(f"%xsrep:\t" + " ".join(xsrep_parts))
         result.append(f"%xsmor:\t" + " ".join(xsmor_parts))
 

{batchalign-0.8.2.post2 → batchalign-0.8.2.post3}/batchalign/pipelines/analysis/compare.py

@@ -395,6 +395,12 @@ class CompareEngine(BatchalignEngine):
                     pos="PUNCT",
                     status="match"
                 )))
+                # Ensure punctuation forms have morphology so %mor includes
+                # the final delimiter (matches generator's PUNCT|<char> logic)
+                if form.morphology is None:
+                    form.morphology = [Morphology(
+                        pos="PUNCT", lemma=form.text.strip(), feats=""
+                    )]
             utt_positioned[utt_idx].sort(key=lambda x: x[0])
 
         # --- 7. Set comparison on each gold utterance ---

batchalign-0.8.2.post3/batchalign/version

@@ -0,0 +1,3 @@
+0.8.2-post.3
+March 28 2026
+fix bugs

{batchalign-0.8.2.post2 → batchalign-0.8.2.post3/batchalign.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: batchalign
-Version: 0.8.2.post2
+Version: 0.8.2.post3
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu

batchalign-0.8.2.post2/batchalign/version

@@ -1,3 +0,0 @@
-0.8.2-post.2
-Feb 27th 2026
-adds new feature regarding compare