acryl-datahub 1.0.0.2rc5__py3-none-any.whl → 1.0.0.3rc2__py3-none-any.whl

datahub/sql_parsing/sqlglot_utils.py

@@ -257,7 +257,10 @@ def generate_hash(text: str) -> str:
 
 
 def get_query_fingerprint_debug(
-    expression: sqlglot.exp.ExpOrStr, platform: DialectOrStr, fast: bool = False
+    expression: sqlglot.exp.ExpOrStr,
+    platform: DialectOrStr,
+    fast: bool = False,
+    secondary_id: Optional[str] = None,
 ) -> Tuple[str, Optional[str]]:
     try:
         if not fast:
@@ -272,16 +275,18 @@ def get_query_fingerprint_debug(
         logger.debug("Failed to generalize query for fingerprinting: %s", e)
         expression_sql = None
 
-    fingerprint = generate_hash(
-        expression_sql
-        if expression_sql is not None
-        else _expression_to_string(expression, platform=platform)
-    )
+    text = expression_sql or _expression_to_string(expression, platform=platform)
+    if secondary_id:
+        text = text + " -- " + secondary_id
+    fingerprint = generate_hash(text=text)
     return fingerprint, expression_sql
 
 
 def get_query_fingerprint(
-    expression: sqlglot.exp.ExpOrStr, platform: DialectOrStr, fast: bool = False
+    expression: sqlglot.exp.ExpOrStr,
+    platform: DialectOrStr,
+    fast: bool = False,
+    secondary_id: Optional[str] = None,
 ) -> str:
     """Get a fingerprint for a SQL query.
 
@@ -298,12 +303,15 @@ def get_query_fingerprint(
     Args:
         expression: The SQL query to fingerprint.
         platform: The SQL dialect to use.
+        secondary_id: An optional additional id string to included in the final fingerprint.
 
     Returns:
         The fingerprint for the SQL query.
     """
 
-    return get_query_fingerprint_debug(expression, platform, fast=fast)[0]
+    return get_query_fingerprint_debug(
+        expression=expression, platform=platform, fast=fast, secondary_id=secondary_id
+    )[0]
 
 
 @functools.lru_cache(maxsize=FORMAT_QUERY_CACHE_SIZE)

datahub/testing/mcp_diff.py

@@ -2,7 +2,7 @@ import dataclasses
 import json
 import re
 from collections import defaultdict
-from typing import Any, Dict, List, Sequence, Set, Tuple, Union
+from typing import Any, Dict, List, Optional, Sequence, Set, Tuple, Union
 
 import deepdiff.serialization
 import yaml
@@ -34,6 +34,7 @@ class AspectForDiff:
     aspect_name: str
     aspect: Dict[str, Any] = dataclasses.field(hash=False)
     delta_info: "DeltaInfo" = dataclasses.field(hash=False, repr=False)
+    headers: Optional[Dict[str, str]] = dataclasses.field(default=None, hash=False)
 
     @classmethod
     def create_from_mcp(cls, idx: int, obj: Dict[str, Any]) -> "AspectForDiff":
@@ -44,6 +45,7 @@ class AspectForDiff:
             aspect_name=obj["aspectName"],
             aspect=aspect.get("json", aspect),
             delta_info=DeltaInfo(idx=idx, original=obj),
+            headers=obj.get("headers"),
         )
 
     def __repr__(self):
@@ -240,9 +242,12 @@ class MCPDiff:
                         s.append(serialize_aspect(ga.aspect))
                 for (i, old, new), diffs in aspect_diffs.aspects_changed.items():
                     s.append(self.report_aspect(old, i, "changed") + ":")
+
+                    print_aspects = False
                     for diff_level in diffs:
                         s.append(self.report_diff_level(diff_level, i))
-                    if verbose:
+                        print_aspects |= self.is_diff_level_on_aspect(diff_level)
+                    if verbose and print_aspects:
                         s.append(f"Old aspect:\n{serialize_aspect(old.aspect)}")
                         s.append(f"New aspect:\n{serialize_aspect(new.aspect)}")
 
@@ -271,6 +276,14 @@ class MCPDiff:
             f"root[{idx}].", ""
         )
 
+    @staticmethod
+    def is_diff_level_on_aspect(diff: DiffLevel) -> bool:
+        skip_print_fields = ["changeType", "headers"]
+        try:
+            return diff.path(output_format="list")[1] not in skip_print_fields
+        except IndexError:
+            return True
+
 
 def serialize_aspect(aspect: Union[AspectForDiff, Dict[str, Any]]) -> str:
     if isinstance(aspect, AspectForDiff):  # Unpack aspect