dbt-common 1.3.0__py3-none-any.whl → 1.5.0__py3-none-any.whl

dbt_common/__about__.py

@@ -1 +1 @@
-version = "1.3.0"
+version = "1.5.0"

dbt_common/clients/system.py

@@ -62,7 +62,9 @@ class FindMatchingParams:
         # Do not record or replay filesystem searches that were performed against
         # files which are actually part of dbt's implementation.
         return (
-            "dbt/include/global_project" not in self.root_path
+            "dbt/include"
+            not in self.root_path  # TODO: This actually obviates the next two checks but is probably too coarse?
+            and "dbt/include/global_project" not in self.root_path
             and "/plugins/postgres/dbt/include/" not in self.root_path
         )
 

dbt_common/record.py

@@ -2,15 +2,17 @@
 external systems during a command invocation, so that the command can be re-run
 later with the recording 'replayed' to dbt.
 
-The rationale for and architecture of this module is described in detail in the
+The rationale for and architecture of this module are described in detail in the
 docs/guides/record_replay.md document in this repository.
 """
 import functools
 import dataclasses
 import json
 import os
+
+from deepdiff import DeepDiff  # type: ignore
 from enum import Enum
-from typing import Any, Dict, List, Mapping, Optional, Type
+from typing import Any, Callable, Dict, List, Mapping, Optional, Type
 
 from dbt_common.context import get_invocation_context
 
@@ -51,15 +53,68 @@ class Record:
 
 
 class Diff:
-    """Marker class for diffs?"""
+    def __init__(self, current_recording_path: str, previous_recording_path: str) -> None:
+        self.current_recording_path = current_recording_path
+        self.previous_recording_path = previous_recording_path
+
+    def diff_query_records(self, current: List, previous: List) -> Dict[str, Any]:
+        # some of the table results are returned as a stringified list of dicts that don't
+        # diff because order isn't consistent. convert it into a list of dicts so it can
+        # be diffed ignoring order
+
+        for i in range(len(current)):
+            if current[i].get("result").get("table") is not None:
+                current[i]["result"]["table"] = json.loads(current[i]["result"]["table"])
+        for i in range(len(previous)):
+            if previous[i].get("result").get("table") is not None:
+                previous[i]["result"]["table"] = json.loads(previous[i]["result"]["table"])
+
+        return DeepDiff(previous, current, ignore_order=True, verbose_level=2)
+
+    def diff_env_records(self, current: List, previous: List) -> Dict[str, Any]:
+        # The mode and filepath may change.  Ignore them.
+
+        exclude_paths = [
+            "root[0]['result']['env']['DBT_RECORDER_FILE_PATH']",
+            "root[0]['result']['env']['DBT_RECORDER_MODE']",
+        ]
+
+        return DeepDiff(
+            previous, current, ignore_order=True, verbose_level=2, exclude_paths=exclude_paths
+        )
+
+    def diff_default(self, current: List, previous: List) -> Dict[str, Any]:
+        return DeepDiff(previous, current, ignore_order=True, verbose_level=2)
+
+    def calculate_diff(self) -> Dict[str, Any]:
+        with open(self.current_recording_path) as current_recording:
+            current_dct = json.load(current_recording)
 
-    pass
+        with open(self.previous_recording_path) as previous_recording:
+            previous_dct = json.load(previous_recording)
+
+        diff = {}
+        for record_type in current_dct:
+            if record_type == "QueryRecord":
+                diff[record_type] = self.diff_query_records(
+                    current_dct[record_type], previous_dct[record_type]
+                )
+            elif record_type == "GetEnvRecord":
+                diff[record_type] = self.diff_env_records(
+                    current_dct[record_type], previous_dct[record_type]
+                )
+            else:
+                diff[record_type] = self.diff_default(
+                    current_dct[record_type], previous_dct[record_type]
+                )
+
+        return diff
 
 
 class RecorderMode(Enum):
     RECORD = 1
     REPLAY = 2
-    RECORD_QUERIES = 3
+    DIFF = 3  # records and does diffing
 
 
 class Recorder:
@@ -67,15 +122,32 @@ class Recorder:
     _record_name_by_params_name: Dict[str, str] = {}
 
     def __init__(
-        self, mode: RecorderMode, types: Optional[List], recording_path: Optional[str] = None
+        self,
+        mode: RecorderMode,
+        types: Optional[List],
+        current_recording_path: str = "recording.json",
+        previous_recording_path: Optional[str] = None,
     ) -> None:
         self.mode = mode
-        self.types = types
+        self.recorded_types = types
         self._records_by_type: Dict[str, List[Record]] = {}
+        self._unprocessed_records_by_type: Dict[str, List[Dict[str, Any]]] = {}
         self._replay_diffs: List["Diff"] = []
+        self.diff: Optional[Diff] = None
+        self.previous_recording_path = previous_recording_path
+        self.current_recording_path = current_recording_path
+
+        if self.previous_recording_path is not None and self.mode in (
+            RecorderMode.REPLAY,
+            RecorderMode.DIFF,
+        ):
+            self.diff = Diff(
+                current_recording_path=self.current_recording_path,
+                previous_recording_path=self.previous_recording_path,
+            )
 
-        if recording_path is not None:
-            self._records_by_type = self.load(recording_path)
+            if self.mode == RecorderMode.REPLAY:
+                self._unprocessed_records_by_type = self.load(self.previous_recording_path)
 
     @classmethod
     def register_record_type(cls, rec_type) -> Any:
@@ -90,7 +162,14 @@ class Recorder:
         self._records_by_type[rec_cls_name].append(record)
 
     def pop_matching_record(self, params: Any) -> Optional[Record]:
-        rec_type_name = self._record_name_by_params_name[type(params).__name__]
+        rec_type_name = self._record_name_by_params_name.get(type(params).__name__)
+
+        if rec_type_name is None:
+            raise Exception(
+                f"A record of type {type(params).__name__} was requested, but no such type has been registered."
+            )
+
+        self._ensure_records_processed(rec_type_name)
         records = self._records_by_type[rec_type_name]
         match: Optional[Record] = None
         for rec in records:
@@ -101,8 +180,8 @@ class Recorder:
 
         return match
 
-    def write(self, file_name) -> None:
-        with open(file_name, "w") as file:
+    def write(self) -> None:
+        with open(self.current_recording_path, "w") as file:
             json.dump(self._to_dict(), file)
 
     def _to_dict(self) -> Dict:
@@ -115,22 +194,20 @@ class Recorder:
         return dct
 
     @classmethod
-    def load(cls, file_name: str) -> Dict[str, List[Record]]:
+    def load(cls, file_name: str) -> Dict[str, List[Dict[str, Any]]]:
         with open(file_name) as file:
-            loaded_dct = json.load(file)
+            return json.load(file)
 
-        records_by_type: Dict[str, List[Record]] = {}
+    def _ensure_records_processed(self, record_type_name: str) -> None:
+        if record_type_name in self._records_by_type:
+            return
 
-        for record_type_name in loaded_dct:
-            # TODO: this breaks with QueryRecord on replay since it's
-            # not in common so isn't part of cls._record_cls_by_name yet
-            record_cls = cls._record_cls_by_name[record_type_name]
-            rec_list = []
-            for record_dct in loaded_dct[record_type_name]:
-                rec = record_cls.from_dict(record_dct)
-                rec_list.append(rec)  # type: ignore
-            records_by_type[record_type_name] = rec_list
-        return records_by_type
+        rec_list = []
+        record_cls = self._record_cls_by_name[record_type_name]
+        for record_dct in self._unprocessed_records_by_type[record_type_name]:
+            rec = record_cls.from_dict(record_dct)
+            rec_list.append(rec)  # type: ignore
+        self._records_by_type[record_type_name] = rec_list
 
     def expect_record(self, params: Any) -> Any:
         record = self.pop_matching_record(params)
@@ -138,24 +215,27 @@ class Recorder:
         if record is None:
             raise Exception()
 
+        if record.result is None:
+            return None
+
         result_tuple = dataclasses.astuple(record.result)
         return result_tuple[0] if len(result_tuple) == 1 else result_tuple
 
     def write_diffs(self, diff_file_name) -> None:
-        json.dump(
-            self._replay_diffs,
-            open(diff_file_name, "w"),
-        )
+        assert self.diff is not None
+        with open(diff_file_name, "w") as f:
+            json.dump(self.diff.calculate_diff(), f)
 
     def print_diffs(self) -> None:
-        print(repr(self._replay_diffs))
+        assert self.diff is not None
+        print(repr(self.diff.calculate_diff()))
 
 
 def get_record_mode_from_env() -> Optional[RecorderMode]:
     """
     Get the record mode from the environment variables.
 
-    If the mode is not set to 'RECORD' or 'REPLAY', return None.
+    If the mode is not set to 'RECORD', 'DIFF' or 'REPLAY', return None.
     Expected format: 'DBT_RECORDER_MODE=RECORD'
     """
     record_mode = os.environ.get("DBT_RECORDER_MODE")
@@ -165,6 +245,9 @@ def get_record_mode_from_env() -> Optional[RecorderMode]:
 
     if record_mode.lower() == "record":
         return RecorderMode.RECORD
+    # diffing requires a file path, otherwise treat as noop
+    elif record_mode.lower() == "diff" and os.environ.get("DBT_RECORDER_FILE_PATH") is not None:
+        return RecorderMode.DIFF
     # replaying requires a file path, otherwise treat as noop
     elif record_mode.lower() == "replay" and os.environ.get("DBT_RECORDER_FILE_PATH") is not None:
         return RecorderMode.REPLAY
@@ -190,7 +273,21 @@ def get_record_types_from_env() -> Optional[List]:
     return record_types_str.split(",")
 
 
-def record_function(record_type, method=False, tuple_result=False):
+def get_record_types_from_dict(fp: str) -> List:
+    """
+    Get the record subset from the dict.
+    """
+    with open(fp) as file:
+        loaded_dct = json.load(file)
+    return list(loaded_dct.keys())
+
+
+def record_function(
+    record_type,
+    method: bool = False,
+    tuple_result: bool = False,
+    id_field_name: Optional[str] = None,
+) -> Callable:
     def record_function_inner(func_to_record):
         # To avoid runtime overhead and other unpleasantness, we only apply the
         # record/replay decorator if a relevant env var is set.
@@ -208,12 +305,17 @@ def record_function(record_type, method=False, tuple_result=False):
             if recorder is None:
                 return func_to_record(*args, **kwargs)
 
-            if recorder.types is not None and record_type.__name__ not in recorder.types:
+            if (
+                recorder.recorded_types is not None
+                and record_type.__name__ not in recorder.recorded_types
+            ):
                 return func_to_record(*args, **kwargs)
 
             # For methods, peel off the 'self' argument before calling the
             # params constructor.
             param_args = args[1:] if method else args
+            if method and id_field_name is not None:
+                param_args = (getattr(args[0], id_field_name),) + param_args
 
             params = record_type.params_cls(*param_args, **kwargs)
 
@@ -230,7 +332,7 @@ def record_function(record_type, method=False, tuple_result=False):
             r = func_to_record(*args, **kwargs)
             result = (
                 None
-                if r is None or record_type.result_cls is None
+                if record_type.result_cls is None
                 else record_type.result_cls(*r)
                 if tuple_result
                 else record_type.result_cls(r)

{dbt_common-1.3.0.dist-info → dbt_common-1.5.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: dbt-common
-Version: 1.3.0
+Version: 1.5.0
 Summary: The shared common utilities that dbt-core and adapter implementations use
 Project-URL: Homepage, https://github.com/dbt-labs/dbt-common
 Project-URL: Repository, https://github.com/dbt-labs/dbt-common.git
@@ -26,6 +26,7 @@ Classifier: Programming Language :: Python :: Implementation :: PyPy
 Requires-Python: >=3.8
 Requires-Dist: agate<1.10,>=1.7.0
 Requires-Dist: colorama<0.5,>=0.3.9
+Requires-Dist: deepdiff<8.0,>=7.0
 Requires-Dist: isodate<0.7,>=0.6
 Requires-Dist: jinja2<4,>=3.1.3
 Requires-Dist: jsonschema<5.0,>=4.0

{dbt_common-1.3.0.dist-info → dbt_common-1.5.0.dist-info}/RECORD

@@ -1,4 +1,4 @@
-dbt_common/__about__.py,sha256=qg6ZAU0TpLWOHtkFC272q_j3aVEzW76H0lONbKcgFA8,18
+dbt_common/__about__.py,sha256=Wgohrh7g_peUNIz0Cro4w7u88rZU94fNypIcm9oMWgQ,18
 dbt_common/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 dbt_common/constants.py,sha256=-Y5DIL1SDPQWtlCNizXRYxFgbx1D7LaLs1ysamvGMRk,278
 dbt_common/context.py,sha256=BhgT7IgyvpZHEtIdFVVuBBBX5LuU7obXT7NvIPeuD2g,1760
@@ -6,7 +6,7 @@ dbt_common/dataclass_schema.py,sha256=t3HGD0oXTSjitctuCVHv3iyq5BT3jxoSxv_VGkrJlE
 dbt_common/helper_types.py,sha256=NoxqGFAq9bOjh7rqtz_eepXAxk20n3mmW_gUVpnMyYU,3901
 dbt_common/invocation.py,sha256=Zw8jRPn75oi2VrUD6qGvaCDtSyIfqm5pJlPpRjs3s1E,202
 dbt_common/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-dbt_common/record.py,sha256=fwp8Q2x0UiB9fHbMFkmOlAgKLst1dhDxQpSSsd04aDw,8204
+dbt_common/record.py,sha256=F-FbFt6Js_U4r5b5hXGp8DY-MTEKNb-dLMaBT80UWHw,12415
 dbt_common/semver.py,sha256=2zoZYCQ7PfswqslT2NHuMGgPGMuMuX-yRThVoqfDWQU,13954
 dbt_common/tests.py,sha256=6lC_JuRtoYO6cbAF8-R5aTM4HtQiM_EH8X5m_97duGY,315
 dbt_common/ui.py,sha256=rc2TEM29raBFc_LXcg901pMDD07C2ohwp9qzkE-7pBY,2567
@@ -14,7 +14,7 @@ dbt_common/clients/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuF
 dbt_common/clients/_jinja_blocks.py,sha256=xoJK9Y0F93U2PKfT_3SJbBopCGYCtl7LiwKuylXnrEE,12947
 dbt_common/clients/agate_helper.py,sha256=n5Q0_gJPbBhFvjd286NGYGlcTtdEExYmIT3968lppyg,9124
 dbt_common/clients/jinja.py,sha256=i6VQ94FU4F6ZCQLHTxNSeGHmvyYSIe34nDhNkH6wO08,18502
-dbt_common/clients/system.py,sha256=OOhRDWR5t0Ns3OhkqjPTNTtyl_RMRWPDHWCzDoFtgkA,23014
+dbt_common/clients/system.py,sha256=K-b9Lx8ZhgR5tKq1vnqbX8sEVaVRQHqkEkLYQIjFc64,23158
 dbt_common/contracts/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 dbt_common/contracts/constraints.py,sha256=hyqTW2oPB1dfXWW388LWnL-EFdqTpQciKISH3CeLkro,1267
 dbt_common/contracts/metadata.py,sha256=K_M06Rue0wmrQhFP_mq3uvQszq10CIt93oGiAVgbRfE,1293
@@ -56,7 +56,7 @@ dbt_common/utils/encoding.py,sha256=6_kSY2FvGNYMg7oX7PrbvVioieydih3Kl7Ii802LaHI,
 dbt_common/utils/executor.py,sha256=Zyzd1wML3aN-iYn9ZG2Gc_jj5vknmvQNyH-c0RaPIpo,2446
 dbt_common/utils/formatting.py,sha256=JUn5rzJ-uajs9wPCN0-f2iRFY1pOJF5YjTD9dERuLoc,165
 dbt_common/utils/jinja.py,sha256=XNfZHuZhLM_R_yPmzYojPm6bF7QOoxIjSWrkJRw6wks,965
-dbt_common-1.3.0.dist-info/METADATA,sha256=WLyv25ll2nMlvoBBwbM8rrEIyB6MwbUTpT69REVxOg4,5264
-dbt_common-1.3.0.dist-info/WHEEL,sha256=zEMcRr9Kr03x1ozGwg5v9NQBKn3kndp6LSoSlVg-jhU,87
-dbt_common-1.3.0.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-dbt_common-1.3.0.dist-info/RECORD,,
+dbt_common-1.5.0.dist-info/METADATA,sha256=phlsK2cVVw1AWNpJa2_uNaM2KG5HroA0RZw4gUsMCFQ,5298
+dbt_common-1.5.0.dist-info/WHEEL,sha256=1yFddiXMmvYK7QYTqtRNtX66WJ0Mz8PYEiEUoOUUxRY,87
+dbt_common-1.5.0.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+dbt_common-1.5.0.dist-info/RECORD,,

{dbt_common-1.3.0.dist-info → dbt_common-1.5.0.dist-info}/WHEEL

@@ -1,4 +1,4 @@
 Wheel-Version: 1.0
-Generator: hatchling 1.24.2
+Generator: hatchling 1.25.0
 Root-Is-Purelib: true
 Tag: py3-none-any