llm-join 0.2.7__tar.gz → 0.2.8__tar.gz

{llm_join-0.2.7 → llm_join-0.2.8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: llm-join
-Version: 0.2.7
+Version: 0.2.8
 Summary: Fuzzy join DataFrames using LLM scoring and embedding retrieval
 License: MIT
 Keywords: entity-resolution,fuzzy-join,llm,nlp,pandas

{llm_join-0.2.7 → llm_join-0.2.8}/llm_join/__init__.py

@@ -17,4 +17,4 @@ except ImportError as e:
 from llm_join.join import fuzzy_join
 
 __all__ = ["fuzzy_join"]
-__version__ = "0.2.7"
+__version__ = "0.2.8"

{llm_join-0.2.7 → llm_join-0.2.8}/llm_join/join.py

@@ -94,6 +94,10 @@ def fuzzy_join(
             break
 
         candidates = [c for c, _ in candidates_with_scores]
+        candidates_debug = [
+            {"candidate": c, "embed_score": round(s, 4)}
+            for c, s in candidates_with_scores
+        ]
         results = scorer.score(left_val, candidates, cfg.context_str, threshold=cfg.threshold)
         llm_call_count += 1
         if results is None:
@@ -106,11 +110,11 @@ def fuzzy_join(
                 reasoning="LLM failed — embed rank-0 fallback used",
                 embed_rank=0,
                 match_method="embed_fallback",
-                candidates=candidates_with_scores,
+                candidates=candidates_debug,
             ))
         elif results:
             for r in results:
-                r.candidates = candidates_with_scores
+                r.candidates = candidates_debug
             matches.extend(results)
 
     return merger.merge(df1, df2, left_col, right_col, matches, how=how, return_reasoning=return_reasoning)

{llm_join-0.2.7 → llm_join-0.2.8}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "llm-join"
-version = "0.2.7"
+version = "0.2.8"
 description = "Fuzzy join DataFrames using LLM scoring and embedding retrieval"
 readme = "README.md"
 requires-python = ">=3.9"