CT 1894 log partial parsing var changes and sort cli vars before hashing

.changes/unreleased/Fixes-20230124-141943.yaml

@@ -0,0 +1,6 @@
+kind: Fixes
+body: Sort cli vars before hashing for partial parsing
+time: 2023-01-24T14:19:43.333628-05:00
+custom:
+ Author: gshank
+ Issue: "6710"

core/dbt/events/types.proto

@@ -839,7 +839,21 @@ message UnableToPartialParseMsg {
  UnableToPartialParse data = 2;
 }
 
-// Skipped I025, I026, I027
+// I025
+message StateCheckVarsHash {
+ string checksum = 1;
+ string vars = 2;
+ string profile = 3;
+ string target = 4;
+ string version = 5;
+}
+
+message StateCheckVarsHashMsg {
+ EventInfo info = 1;
+ StateCheckVarsHash data = 2;
+}
+
+// Skipped I026, I027
 
 
 // I028

core/dbt/events/types.py

@@ -843,6 +843,15 @@ def message(self) -> str:
  return f"Unable to do partial parsing because {self.reason}"
 
 
+@dataclass
+class StateCheckVarsHash(DebugLevel, pt.StateCheckVarsHash):
+ def code(self):
+ return "I025"
+
+ def message(self) -> str:
+ return "checksum: {self.checksum}, vars: {self.vars}, profile: {self.profile}, target: {self.target}, version: {self.version}"
+
+
 # Skipped I025, I026, I026, I027
 
 

core/dbt/parser/manifest.py

@@ -8,6 +8,7 @@
 from itertools import chain
 import time
 from dbt.events.base_types import EventLevel
+import pprint
 
 import dbt.exceptions
 import dbt.tracking
@@ -29,6 +30,8 @@
  ParsedFileLoadFailed,
  InvalidDisabledTargetInTestNode,
  NodeNotFoundOrDisabled,
+ StateCheckVarsHash,
+ Note,
 )
 from dbt.logger import DbtProcessState
 from dbt.node_types import NodeType
@@ -569,6 +572,12 @@ def is_partial_parsable(self, manifest: Manifest) -> Tuple[bool, Optional[str]]:
  reason="config vars, config profile, or config target have changed"
  )
  )
+ fire_event(
+ Note(
+ msg=f"previous checksum: {self.manifest.state_check.vars_hash.checksum}, current checksum: {manifest.state_check.vars_hash.checksum}"
+ ),
+ level=EventLevel.DEBUG,
+ )
  valid = False
  reparse_reason = ReparseReason.vars_changed
  if self.manifest.state_check.profile_hash != manifest.state_check.profile_hash:
@@ -702,16 +711,28 @@ def build_manifest_state_check(self):
  # arg vars, but since any changes to that file will cause state_check
  # to not pass, it doesn't matter. If we move to more granular checking
  # of env_vars, that would need to change.
+ # We are using the parsed cli_vars instead of config.args.vars, in order
+ # to sort them and avoid reparsing because of ordering issues.
+ stringified_cli_vars = pprint.pformat(config.cli_vars)
  vars_hash = FileHash.from_contents(
  "\x00".join(
  [
- getattr(config.args, "vars", "{}") or "{}",
+ stringified_cli_vars,
  getattr(config.args, "profile", "") or "",
  getattr(config.args, "target", "") or "",
  __version__,
  ]
  )
  )
+ fire_event(
+ StateCheckVarsHash(
+ checksum=vars_hash.checksum,
+ vars=stringified_cli_vars,
+ profile=config.args.profile,
+ target=config.args.target,
+ version=__version__,
+ )
+ )
 
  # Create a FileHash of the env_vars in the project
  key_list = list(config.project_env_vars.keys())

tests/unit/test_events.py

@@ -185,6 +185,7 @@ def test_event_codes(self):
  PartialParsingError(exc_info={}),
  PartialParsingSkipParsing(),
  UnableToPartialParse(reason="something went wrong"),
+ StateCheckVarsHash(vars="testing", target="testing", profile="testing"),
  PartialParsingNotEnabled(),
  ParsedFileLoadFailed(path="", exc="", exc_info=""),
  StaticParserCausedJinjaRendering(path=""),