Merge pull request #631 from datafold/json-values-diff-stats-fix

include only rows with equal PK to values diff stats in --json output

Author: dlawin

data_diff/diff_tables.py

@@ -165,8 +165,8 @@ def get_stats_string(self, is_dbt: bool = False):
 
         return string_output
 
-    def get_stats_dict(self):
-        diff_stats = self._get_stats()
+    def get_stats_dict(self, is_dbt: bool = False):
+        diff_stats = self._get_stats(is_dbt)
         json_output = {
             "rows_A": diff_stats.table1_count,
             "rows_B": diff_stats.table2_count,
@@ -177,6 +177,8 @@ def get_stats_dict(self):
             "total": sum(diff_stats.diff_by_sign.values()),
             "stats": self.stats,
         }
+        if diff_stats.extra_column_diffs:
+            json_output["values"] = diff_stats.extra_column_diffs
 
         return json_output
 

data_diff/format.py

@@ -51,7 +51,7 @@ def jsonify(
 
     summary = None
     if with_summary:
-        summary = _jsonify_diff_summary(diff.get_stats_dict())
+        summary = _jsonify_diff_summary(diff.get_stats_dict(is_dbt=True))
 
     columns = None
     if with_columns:
@@ -258,7 +258,7 @@ def _jsonify_diff_summary(stats_dict: dict) -> JsonDiffSummary:
             updated=stats_dict["updated"],
             unchanged=stats_dict["unchanged"],
         ),
-        stats=Stats(diffCounts=stats_dict["stats"]["diff_counts"]),
+        stats=Stats(diffCounts=stats_dict["values"]),
     )