tracepipe 0.3.2__py3-none-any.whl → 0.3.3__py3-none-any.whl

tracepipe/__init__.py

@@ -81,7 +81,7 @@ from .core import TracePipeConfig, TracePipeMode
 from .snapshot import DiffResult, Snapshot, diff, snapshot
 
 # === VERSION ===
-__version__ = "0.3.2"
+__version__ = "0.3.3"
 
 # === MINIMAL __all__ ===
 __all__ = [

tracepipe/convenience.py

@@ -361,8 +361,8 @@ def check(
     facts["rows_dropped"] = len(dropped)
     facts["total_steps"] = len(ctx.store.steps)
 
-    # Merge statistics
-    merge_stats_list = ctx.store.get_merge_stats() if hasattr(ctx.store, "get_merge_stats") else []
+    # Merge statistics - filter to df's lineage to avoid cross-contamination
+    merge_stats_list = _get_merge_stats_for_df(df, ctx)
 
     for i, (step_id, stats) in enumerate(merge_stats_list):
         facts[f"merge_{i}_expansion"] = stats.expansion_ratio
@@ -658,6 +658,53 @@ def find(
 # ============ HELPERS ============
 
 
+def _get_merge_stats_for_df(df: pd.DataFrame, ctx) -> list[tuple[int, Any]]:
+    """
+    Get merge stats relevant to df's lineage only.
+
+    This prevents cross-contamination where check(df) would show warnings
+    from merges that produced OTHER DataFrames in the same session.
+    """
+    if not hasattr(ctx.store, "get_merge_stats"):
+        return []
+
+    all_stats = ctx.store.get_merge_stats()
+    if not all_stats:
+        return []
+
+    # Get row IDs from df
+    rids = ctx.row_manager.get_ids_array(df)
+    if rids is None:
+        return []
+
+    # Find which merge steps produced rows in df
+    relevant_step_ids = set()
+
+    # Check merge mappings to find which merges produced df's rows
+    if hasattr(ctx.store, "merge_mappings"):
+        for mapping in ctx.store.merge_mappings:
+            # Check if any of df's row IDs are in this merge's output
+            for rid in rids:
+                # Binary search in sorted out_rids
+                i = np.searchsorted(mapping.out_rids, rid)
+                if i < len(mapping.out_rids) and mapping.out_rids[i] == rid:
+                    relevant_step_ids.add(mapping.step_id)
+                    break  # Found at least one match, this merge is relevant
+
+    # If no merge mappings found, fall back to checking if df was just merged
+    # by seeing if it has more columns than typical (heuristic)
+    if not relevant_step_ids and all_stats:
+        # Fallback: return only the most recent merge that could have produced df
+        # This handles the case where merge_mappings aren't available
+        for step_id, stats in reversed(all_stats):
+            if stats.result_rows == len(df):
+                relevant_step_ids.add(step_id)
+                break
+
+    # Filter stats to relevant merges only
+    return [(sid, stats) for sid, stats in all_stats if sid in relevant_step_ids]
+
+
 def _json_safe(val: Any) -> Any:
     """Convert value to JSON-serializable form."""
     if pd.isna(val):

tracepipe/instrumentation/series_capture.py

@@ -116,6 +116,10 @@ def wrap_series_assignment():
     """
     Wrap DataFrame.__setitem__ to capture diffs when assigning Series.
 
+    Note: For watched columns, _wrap_setitem (pandas_inst.py) already captures
+    the assignment. This wrapper only captures for NON-watched columns when
+    a TrackedSeries is assigned, to avoid double-logging.
+
     Handles:
     - df['col'] = series  (where series may have been modified)
     - df['col'] = scalar  (broadcast assignment)
@@ -127,28 +131,33 @@ def wrap_series_assignment():
     def tracked_setitem(self, key, value):
         ctx = get_context()
 
-        # Capture before state for watched columns
+        # For watched columns, _wrap_setitem already captures - skip to avoid double-logging
+        # We only capture here for NON-watched columns when a TrackedSeries is involved
+        should_capture_here = False
         before_values = None
+
         if (
             ctx.enabled
             and isinstance(key, str)
-            and key in ctx.watched_columns
             and key in self.columns
+            and key not in ctx.watched_columns  # Only capture NON-watched columns here
+            and isinstance(value, TrackedSeries)  # Only for TrackedSeries assignments
         ):
             rids = ctx.row_manager.get_ids_array(self)
             if rids is not None:
+                should_capture_here = True
                 before_values = {
                     "rids": rids.copy(),
                     "values": self[key].values.copy(),
                 }
 
-        # Always run original
+        # Always run original (which may be _wrap_setitem's wrapper)
         original_setitem(self, key, value)
 
         if not ctx.enabled:
             return
 
-        if before_values is None:
+        if not should_capture_here or before_values is None:
             return
 
         try:

{tracepipe-0.3.2.dist-info → tracepipe-0.3.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: tracepipe
-Version: 0.3.2
+Version: 0.3.3
 Summary: Row-level data lineage tracking for pandas pipelines
 Project-URL: Homepage, https://github.com/tracepipe/tracepipe
 Project-URL: Documentation, https://tracepipe.github.io/tracepipe/

{tracepipe-0.3.2.dist-info → tracepipe-0.3.3.dist-info}/RECORD

@@ -1,8 +1,8 @@
-tracepipe/__init__.py,sha256=MuwxV2mU4XxHqab62vQxaDAlhMvRCgUCmr_YU9R16ss,3342
+tracepipe/__init__.py,sha256=VhfrIqYHgzY7mDQwcsmH_Gw8_j7U_libclBKXTFMwE0,3342
 tracepipe/api.py,sha256=WdcKvvzI3voDt6fxZWa8vjyZQU8lfRshx7T78oj7oFE,13351
 tracepipe/context.py,sha256=_povLpqa5wd_ESHt5hbSmWTSMTF3nUfeutEQo4RMK2E,3856
 tracepipe/contracts.py,sha256=m-rjPrgnCiAgKEkweOS7P95jrjDptt5UPdvUlqaV_rU,16226
-tracepipe/convenience.py,sha256=SZGcSOKPjAeJ9udPP_Fa_zTZY5GeDX61W6uftMwafjc,26563
+tracepipe/convenience.py,sha256=KuDz_ZzNivVG1SS8Srr3plu4CTwFmNhYL4rk3vV6cbE,28421
 tracepipe/core.py,sha256=kAXks694rR0Z4tD7Gyty0TyJGWx2whsSdteYYpHuazo,8010
 tracepipe/debug.py,sha256=6t2GKVZLwn7SJLhrStE9qsmTiVIHATTE3jJPQ2DYtnc,10140
 tracepipe/safety.py,sha256=jTBZv4QGDJfnZETsSZeMKbdOUtGXk-_XkmllhnGWM-M,5537
@@ -14,7 +14,7 @@ tracepipe/instrumentation/filter_capture.py,sha256=onlYLU5bBZSM3WmxM2AFHfktnlx7R
 tracepipe/instrumentation/indexer_capture.py,sha256=1ATCeJ-uNA1uGiSbgnUx0wdVsIlZGHeUBaFJPXgFQNg,28440
 tracepipe/instrumentation/merge_capture.py,sha256=Eze-PTrn7IXxZRZBYX9R13mOY3diWKAkjp4z-wa1tEk,13349
 tracepipe/instrumentation/pandas_inst.py,sha256=2YSoju9ml2PjLOYzsx8MHH1iqhjgnXHbIidnF0JDpaY,29546
-tracepipe/instrumentation/series_capture.py,sha256=N1Cf-pQDh23qQLLd8DNsxbcaD-91sTJkRd5AnccKZGE,10649
+tracepipe/instrumentation/series_capture.py,sha256=i7FiA2ndEzS6duIj5y-a7SDfIMl2cCY_jGC1tmG7TGU,11271
 tracepipe/storage/__init__.py,sha256=pGFMfbIgIi2kofVPwYDqe2HTYMYJoabiGjTq77pYi-g,348
 tracepipe/storage/base.py,sha256=7DV_-rp37DjBMr9B1w85hLVYhC8OQShk2PcEhT-n4tE,4894
 tracepipe/storage/lineage_store.py,sha256=swMMf59isoCQZHaezCmquA-0R5iGNH3eGWjc9d9LGmo,27392
@@ -23,7 +23,7 @@ tracepipe/utils/__init__.py,sha256=CI_GXViCjdMbu1j6HuzZhoQZEW0sIB6WAve6j5pfOC0,1
 tracepipe/utils/value_capture.py,sha256=wGgegQmJnVHxHbwHSH9di7JAOBChzD3ERJrabZNiayk,4092
 tracepipe/visualization/__init__.py,sha256=M3s44ZTUNEToyghjhQW0FgbmWHKPr4Xc-7iNF6DpI_E,132
 tracepipe/visualization/html_export.py,sha256=G0hfZTJctUCfpun17zXX1NIXhvJZbca6hKmP3rcIjbg,42282
-tracepipe-0.3.2.dist-info/METADATA,sha256=ik5FLmADKLqj25TprTnJPi21SW4EJ88mBTG-aQ4p-gc,9152
-tracepipe-0.3.2.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
-tracepipe-0.3.2.dist-info/licenses/LICENSE,sha256=HMOAFHBClL79POwWL-2_aDcx42DJAq7Ce-nwJPvMB9U,1075
-tracepipe-0.3.2.dist-info/RECORD,,
+tracepipe-0.3.3.dist-info/METADATA,sha256=OnHkiCjZ0-fUERI4PoUh3VmIWMZ_sePk8-pFBwo_8cI,9152
+tracepipe-0.3.3.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
+tracepipe-0.3.3.dist-info/licenses/LICENSE,sha256=HMOAFHBClL79POwWL-2_aDcx42DJAq7Ce-nwJPvMB9U,1075
+tracepipe-0.3.3.dist-info/RECORD,,