Evolve dbt data diff output (#857)

* first draft * style fixes by ruff * past tense consistency * working draft of new table * style fixes by ruff * dbt diffs work, cloud broken for now * remove cached git repos * efficient naming * add type changed count * reorder for priority on prod changes * tabulate value diffs * style fixes by ruff * less horizontal space needed * leo's feedback * center align values * consistent formatting * shorter name same meaning * row counts and diff values working * deps impacts works now * default val * more readable * add primary key used * add model specific CI configs * consistency * conditional headers * style fixes by ruff * cleaner implementation * more cleaning * consistent format * fix unchanged calc * remove prints * default value * draft up tests * a couple more tests * new version * passing tests * style fixes by ruff * util unit test * add unit tests * test the templates * fix type hints * real test no mocking * update tests with all the new outputs * add more validations for mock * fix json bug --------- Co-authored-by: Sung Won Chung <[email protected]> Co-authored-by: sungchun12 <[email protected]> Co-authored-by: Dan Lawin <[email protected]>
datafold · Feb 6, 2024 · cac665b · cac665b
1 parent d4ca0e6
commit cac665b
Show file tree

Hide file tree

Showing 11 changed files with 304 additions and 61 deletions.
diff --git a/data_diff/cloud/datafold_api.py b/data_diff/cloud/datafold_api.py
@@ -144,18 +144,22 @@ class TSummaryResultSchemaStats(pydantic.BaseModel):
     exclusive_columns: Tuple[List[str], List[str]]
 
 
+class TSummaryResultDependencyDetails(pydantic.BaseModel):
+    deps: Dict[str, List[Dict]]
+
+
 class TCloudApiDataDiffSummaryResult(pydantic.BaseModel):
     status: str
     pks: Optional[TSummaryResultPrimaryKeyStats]
     values: Optional[TSummaryResultValueStats]
     schema_: Optional[TSummaryResultSchemaStats]
-    dependencies: Optional[Dict[str, Any]]
+    deps: Optional[TSummaryResultDependencyDetails]
 
     @classmethod
     def from_orm(cls, obj: Any) -> Self:
         pks = TSummaryResultPrimaryKeyStats(**obj["pks"]) if "pks" in obj else None
         values = TSummaryResultValueStats(**obj["values"]) if "values" in obj else None
-        deps = obj["deps"] if "deps" in obj else None
+        deps = TSummaryResultDependencyDetails(**obj["dependencies"]) if "dependencies" in obj else None
         schema = TSummaryResultSchemaStats(**obj["schema"]) if "schema" in obj else None
         return cls(
             status=obj["status"],

diff --git a/data_diff/dbt.py b/data_diff/dbt.py
@@ -306,12 +306,23 @@ def _local_diff(
         k for k, v in table2_columns.items() if k in table1_columns and v.data_type != table1_columns[k].data_type
     }
 
-    if columns_added:
-        diff_output_str += columns_added_template(columns_added)
+    diff_output_str += f"Primary Keys: {diff_vars.primary_keys} \n"
+
+    if diff_vars.where_filter:
+        diff_output_str += f"Where Filter: '{str(diff_vars.where_filter)}' \n"
+
+    if diff_vars.include_columns:
+        diff_output_str += f"Included Columns: {diff_vars.include_columns} \n"
+
+    if diff_vars.exclude_columns:
+        diff_output_str += f"Excluded Columns: {diff_vars.exclude_columns} \n"
 
     if columns_removed:
         diff_output_str += columns_removed_template(columns_removed)
 
+    if columns_added:
+        diff_output_str += columns_added_template(columns_added)
+
     if columns_type_changed:
         diff_output_str += columns_type_changed_template(columns_type_changed)
         column_set = column_set.difference(columns_type_changed)
@@ -349,13 +360,14 @@ def _local_diff(
             return
 
         dataset1_columns = [
-            (name, type_, table1.database.dialect.parse_type(table1.table_path, name, type_, *other))
-            for (name, type_, *other) in table1_columns.values()
+            (info.column_name, info.data_type, table1.database.dialect.parse_type(table1.table_path, info))
+            for info in table1_columns.values()
         ]
         dataset2_columns = [
-            (name, type_, table2.database.dialect.parse_type(table2.table_path, name, type_, *other))
-            for (name, type_, *other) in table2_columns.values()
+            (info.column_name, info.data_type, table2.database.dialect.parse_type(table2.table_path, info))
+            for info in table2_columns.values()
         ]
+
         print(
             json.dumps(
                 jsonify(
@@ -455,32 +467,57 @@ def _cloud_diff(
         rows_removed_count = diff_results.pks.exclusives[0]
 
         rows_updated = diff_results.values.rows_with_differences
-        total_rows = diff_results.values.total_rows
-        rows_unchanged = int(total_rows) - int(rows_updated)
+        total_rows_table1 = diff_results.pks.total_rows[0]
+        total_rows_table2 = diff_results.pks.total_rows[1]
+        total_rows_diff = total_rows_table2 - total_rows_table1
+
+        rows_unchanged = int(total_rows_table1) - int(rows_updated) - int(rows_removed_count)
         diff_percent_list = {
-            x.column_name: str(x.match) + "%" for x in diff_results.values.columns_diff_stats if x.match != 100.0
+            x.column_name: f"{str(round(100.00 - x.match, 2))}%"
+            for x in diff_results.values.columns_diff_stats
+            if x.match != 100.0
         }
-        columns_added = diff_results.schema_.exclusive_columns[1]
-        columns_removed = diff_results.schema_.exclusive_columns[0]
+        columns_added = set(diff_results.schema_.exclusive_columns[1])
+        columns_removed = set(diff_results.schema_.exclusive_columns[0])
         column_type_changes = diff_results.schema_.column_type_differs
 
-        if columns_added:
-            diff_output_str += columns_added_template(columns_added)
+        diff_output_str += f"Primary Keys: {diff_vars.primary_keys} \n"
+        if diff_vars.where_filter:
+            diff_output_str += f"Where Filter: '{str(diff_vars.where_filter)}' \n"
+
+        if diff_vars.include_columns:
+            diff_output_str += f"Included Columns: {diff_vars.include_columns} \n"
+
+        if diff_vars.exclude_columns:
+            diff_output_str += f"Excluded Columns: {diff_vars.exclude_columns} \n"
 
         if columns_removed:
             diff_output_str += columns_removed_template(columns_removed)
 
+        if columns_added:
+            diff_output_str += columns_added_template(columns_added)
+
         if column_type_changes:
             diff_output_str += columns_type_changed_template(column_type_changes)
 
+        deps_impacts = {
+            key: len(value) + sum(len(item.get("BiHtSync", [])) for item in value) if key == "hightouch" else len(value)
+            for key, value in diff_results.deps.deps.items()
+        }
+
         if any([rows_added_count, rows_removed_count, rows_updated]):
             diff_output = dbt_diff_string_template(
-                rows_added_count,
-                rows_removed_count,
-                rows_updated,
-                str(rows_unchanged),
-                diff_percent_list,
-                "Value Match Percent:",
+                total_rows_table1=total_rows_table1,
+                total_rows_table2=total_rows_table2,
+                total_rows_diff=total_rows_diff,
+                rows_added=rows_added_count,
+                rows_removed=rows_removed_count,
+                rows_updated=rows_updated,
+                rows_unchanged=str(rows_unchanged),
+                deps_impacts=deps_impacts,
+                is_cloud=True,
+                extra_info_dict=diff_percent_list,
+                extra_info_str="Value Changed:",
             )
             diff_output_str += f"\n{diff_url}\n {diff_output} \n"
             rich.print(diff_output_str)
@@ -524,7 +561,7 @@ def _cloud_diff(
 
 
 def _diff_output_base(dev_path: str, prod_path: str) -> str:
-    return f"\n[green]{prod_path} <> {dev_path}[/] \n"
+    return f"\n[blue]{prod_path}[/] <> [green]{dev_path}[/] \n"
 
 
 def _initialize_events(dbt_user_id: Optional[str], dbt_version: Optional[str], dbt_project_id: Optional[str]) -> None:

diff --git a/data_diff/diff_tables.py b/data_diff/diff_tables.py
@@ -138,14 +138,19 @@ def _get_stats(self, is_dbt: bool = False) -> DiffStats:
     def get_stats_string(self, is_dbt: bool = False):
         diff_stats = self._get_stats(is_dbt)
 
+        total_rows_diff = diff_stats.table2_count - diff_stats.table1_count
+
         if is_dbt:
             string_output = dbt_diff_string_template(
-                diff_stats.diff_by_sign["+"],
-                diff_stats.diff_by_sign["-"],
-                diff_stats.diff_by_sign["!"],
-                diff_stats.unchanged,
-                diff_stats.extra_column_diffs,
-                "Values Updated:",
+                total_rows_table1=diff_stats.table1_count,
+                total_rows_table2=diff_stats.table2_count,
+                total_rows_diff=total_rows_diff,
+                rows_added=diff_stats.diff_by_sign["+"],
+                rows_removed=diff_stats.diff_by_sign["-"],
+                rows_updated=diff_stats.diff_by_sign["!"],
+                rows_unchanged=diff_stats.unchanged,
+                extra_info_dict=diff_stats.extra_column_diffs,
+                extra_info_str="[u]Values Changed[/u]",
             )
 
         else:

diff --git a/data_diff/utils.py b/data_diff/utils.py
@@ -459,19 +459,59 @@ def __repr__(self) -> str:
 
 
 def dbt_diff_string_template(
-    rows_added: str, rows_removed: str, rows_updated: str, rows_unchanged: str, extra_info_dict: Dict, extra_info_str
+    total_rows_table1: int,
+    total_rows_table2: int,
+    total_rows_diff: int,
+    rows_added: int,
+    rows_removed: int,
+    rows_updated: int,
+    rows_unchanged: int,
+    extra_info_dict: Dict,
+    extra_info_str: str,
+    is_cloud: Optional[bool] = False,
+    deps_impacts: Optional[Dict] = None,
 ) -> str:
-    string_output = f"\n{tabulate([[rows_added, rows_removed]], headers=['Rows Added', 'Rows Removed'])}"
+    # main table
+    main_rows = [
+        ["Total", total_rows_table1, "", f"{total_rows_table2} [{diff_int_dynamic_color_template(total_rows_diff)}]"],
+        ["Added", "", diff_int_dynamic_color_template(rows_added), ""],
+        ["Removed", "", diff_int_dynamic_color_template(-rows_removed), ""],
+        ["Different", "", rows_updated, ""],
+        ["Unchanged", "", rows_unchanged, ""],
+    ]
+
+    main_headers = ["rows", "PROD", "<>", "DEV"]
+    main_table = tabulate(main_rows, headers=main_headers)
+
+    # diffs table
+    diffs_rows = sorted(list(extra_info_dict.items()))
+
+    diffs_headers = ["columns", "% diff values" if is_cloud else "# diff values"]
+    diffs_table = tabulate(diffs_rows, headers=diffs_headers)
+
+    # deps impacts table
+    deps_impacts_table = ""
+    if deps_impacts:
+        deps_impacts_rows = list(deps_impacts.items())
+        deps_impacts_headers = ["deps", "# data assets"]
+        deps_impacts_table = f"\n\n{tabulate(deps_impacts_rows, headers=deps_impacts_headers)}"
+
+    # combine all tables
+    string_output = f"\n{main_table}\n\n{diffs_table}{deps_impacts_table}"
 
-    string_output += f"\n\nUpdated Rows: {rows_updated}\n"
-    string_output += f"Unchanged Rows: {rows_unchanged}\n\n"
+    return string_output
 
-    string_output += extra_info_str
 
-    for k, v in extra_info_dict.items():
-        string_output += f"\n{k}: {v}"
+def diff_int_dynamic_color_template(diff_value: int) -> str:
+    if not isinstance(diff_value, int):
+        return diff_value
 
-    return string_output
+    if diff_value > 0:
+        return f"[green]+{diff_value}[/]"
+    elif diff_value < 0:
+        return f"[red]{diff_value}[/]"
+    else:
+        return "0"
 
 
 def _jsons_equiv(a: str, b: str):
@@ -498,18 +538,18 @@ def diffs_are_equiv_jsons(diff: list, json_cols: dict):
     return match, overriden_diff_cols
 
 
-def columns_removed_template(columns_removed) -> str:
-    columns_removed_str = f"Column(s) removed: {columns_removed}\n"
+def columns_removed_template(columns_removed: set) -> str:
+    columns_removed_str = f"[red]Columns removed [-{len(columns_removed)}]:[/] [blue]{columns_removed}[/]\n"
     return columns_removed_str
 
 
-def columns_added_template(columns_added) -> str:
-    columns_added_str = f"Column(s) added: {columns_added}\n"
+def columns_added_template(columns_added: set) -> str:
+    columns_added_str = f"[green]Columns added [+{len(columns_added)}]: {columns_added}[/]\n"
     return columns_added_str
 
 
 def columns_type_changed_template(columns_type_changed) -> str:
-    columns_type_changed_str = f"Type change: {columns_type_changed}\n"
+    columns_type_changed_str = f"Type changed [{len(columns_type_changed)}]: [green]{columns_type_changed}[/]\n"
     return columns_type_changed_str
 
 

diff --git a/data_diff/version.py b/data_diff/version.py
@@ -1 +1 @@
-__version__ = "0.10.1"
+__version__ = "0.11.0"
diff --git a/pyproject.toml b/pyproject.toml
@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "data-diff"
-version = "0.10.1"
+version = "0.11.0"
 description = "Command-line tool and Python library to efficiently diff rows across two different databases."
 authors = ["Datafold <[email protected]>"]
 license = "MIT"

diff --git a/tests/dbt_artifacts/jaffle_shop.duckdb b/tests/dbt_artifacts/jaffle_shop.duckdb
diff --git a/tests/dbt_artifacts/target/manifest.json b/tests/dbt_artifacts/target/manifest.json