Added --fields and --ignorefields options allowing specification of which fields will be compared

Author: datsom1

db_diff/__init__.py

@@ -52,7 +52,7 @@ def _simplify_json_row(r, common_keys):
     return r
 
 
-def compare(previous, current, show_unchanged=False):
+def compare(previous, current, show_unchanged=False, fields=None, ignorefields=None):
     result = {
         "added": [],
         "removed": [],
@@ -64,45 +64,60 @@ def compare(previous, current, show_unchanged=False):
     previous_columns = set(next(iter(previous.values())).keys())
     current_columns = set(next(iter(current.values())).keys())
     ignore_columns = None
-    if previous_columns != current_columns:
-        result["columns_added"] = [
-            c for c in current_columns if c not in previous_columns
-        ]
-        result["columns_removed"] = [
-            c for c in previous_columns if c not in current_columns
-        ]
-        ignore_columns = current_columns.symmetric_difference(previous_columns)
+
+    # Apply fields/ignorefields filtering
+    if fields:
+        compare_columns = set(fields)
+    elif ignorefields:
+        compare_columns = previous_columns | current_columns
+        compare_columns = compare_columns - set(ignorefields)
+    else:
+        compare_columns = previous_columns | current_columns
+
+    # Adjust columns_added/removed based on compare_columns
+    result["columns_added"] = [c for c in current_columns if c not in previous_columns and c in compare_columns]
+    result["columns_removed"] = [c for c in previous_columns if c not in current_columns and c in compare_columns]
+    ignore_columns = (previous_columns | current_columns) - compare_columns
+
     # Have any rows been removed or added?
     removed = [id for id in previous if id not in current]
     added = [id for id in current if id not in previous]
-    # How about changed?
     removed_or_added = set(removed) | set(added)
     potential_changes = [id for id in current if id not in removed_or_added]
-    changed = [id for id in potential_changes if current[id] != previous[id]]
+    changed = [
+        id for id in potential_changes
+        if any(
+            (k in compare_columns) and (current[id].get(k) != previous[id].get(k))
+            for k in compare_columns
+        )
+    ]
     if added:
         result["added"] = [current[id] for id in added]
     if removed:
         result["removed"] = [previous[id] for id in removed]
     if changed:
         for id in changed:
-            diffs = list(diff(previous[id], current[id], ignore=ignore_columns))
+            diffs = list(diff(
+                previous[id], current[id],
+                ignore=ignore_columns if ignore_columns else None
+            ))
             if diffs:
                 changes = {
                     "key": id,
                     "changes": {
-                        # field can be a list if id contained '.' - #7
                         field[0] if isinstance(field, list) else field: [
                             prev_value,
                             current_value,
                         ]
                         for _, field, (prev_value, current_value) in diffs
+                        if (field[0] if isinstance(field, list) else field) in compare_columns
                     },
                 }
                 if show_unchanged:
                     changes["unchanged"] = {
                         field: value
                         for field, value in previous[id].items()
-                        if field not in changes["changes"] and field != "id"
+                        if field not in changes["changes"] and field != "id" and field in compare_columns
                     }
                 result["changed"].append(changes)
     return result

db_diff/cli.py

@@ -69,10 +69,23 @@
     show_default=True,
     help="Input File Encoding. Available: (utf-8|utf-16|utf-16le|utf-16be|latin1|cp1252|ascii|...).",
 )
+@click.option(
+    "--fields",
+    type=str,
+    default=None,
+    help="Comma-separated list of fields to compare (all others ignored). Mutually exclusive with --ignorefields.",
+)
+@click.option(
+    "--ignorefields",
+    type=str,
+    default=None,
+    help="Comma-separated list of fields to ignore during comparison. Mutually exclusive with --fields.",
+)
 @click.version_option()
 def cli(
     previous, current, key, input_format,
-    show_unchanged, encoding, show_time, output, output_file, output_path
+    show_unchanged, encoding, show_time, output, output_file, output_path,
+    fields, ignorefields
 ):
     """Compare the differences between two CSV or JSON files."""
     dialect = {
@@ -127,9 +140,23 @@ def load(filename, key):
         if not key:
             raise click.ClickException("Could not determine key column from the previous file.")
 
+    # Enforce mutual exclusivity
+    if fields and ignorefields:
+        raise click.ClickException("You cannot use --fields and --ignorefields at the same time.")
+
+    # Parse fields/ignorefields into sets
+    fields_set = set(f.strip() for f in fields.split(",")) if fields else None
+    ignorefields_set = set(f.strip() for f in ignorefields.split(",")) if ignorefields else None
+
     previous_data = load(previous, key)
     current_data = load(current, key)
-    diff = compare(previous_data, current_data, show_unchanged)
+    diff = compare(
+        previous_data,
+        current_data,
+        show_unchanged,
+        fields=fields_set,
+        ignorefields=ignorefields_set,
+    )
 
     if output == "json":
         print(std_json.dumps(diff, indent=4))