feat: location in nomissingexternal models linting error

Author: benfdking

sqlmesh/core/linter/helpers.py

@@ -84,19 +84,8 @@ def to_range(self, read_file: t.Optional[t.List[str]]) -> Range:
         )
 
 
-def read_range_from_file(file: Path, text_range: Range) -> str:
-    """
-    Read the file and return the content within the specified range.
-
-    Args:
-        file: Path to the file to read
-        text_range: The range of text to extract
-
-    Returns:
-        The content within the specified range
-    """
-    with file.open("r", encoding="utf-8") as f:
-        lines = f.readlines()
+def read_range_from_string(content: str, text_range: Range) -> str:
+    lines = content.splitlines(keepends=False)
 
     # Ensure the range is within bounds
     start_line = max(0, text_range.start.line)
@@ -116,6 +105,23 @@ def read_range_from_file(file: Path, text_range: Range) -> str:
     return "".join(result)
 
 
+def read_range_from_file(file: Path, text_range: Range) -> str:
+    """
+    Read the file and return the content within the specified range.
+
+    Args:
+        file: Path to the file to read
+        text_range: The range of text to extract
+
+    Returns:
+        The content within the specified range
+    """
+    with file.open("r", encoding="utf-8") as f:
+        lines = f.readlines()
+
+    return read_range_from_string("".join(lines), text_range)
+
+
 def get_range_of_model_block(
     sql: str,
     dialect: str,

sqlmesh/core/linter/rules/builtin.py

@@ -7,10 +7,16 @@
 from sqlglot.expressions import Star
 from sqlglot.helper import subclasses
 
-from sqlmesh.core.linter.helpers import TokenPositionDetails, get_range_of_model_block
+from sqlmesh.core.dialect import normalize_model_name
+from sqlmesh.core.linter.helpers import (
+    TokenPositionDetails,
+    get_range_of_model_block,
+    read_range_from_string,
+)
 from sqlmesh.core.linter.rule import Rule, RuleViolation, Range, Fix, TextEdit
 from sqlmesh.core.linter.definition import RuleSet
 from sqlmesh.core.model import Model, SqlModel, ExternalModel
+from sqlmesh.utils.lineage import extract_references_from_query, ExternalModelReference
 
 
 class NoSelectStar(Rule):
@@ -113,7 +119,9 @@ def check_model(self, model: Model) -> t.Optional[RuleViolation]:
 class NoMissingExternalModels(Rule):
     """All external models must be registered in the external_models.yaml file"""
 
-    def check_model(self, model: Model) -> t.Optional[RuleViolation]:
+    def check_model(
+        self, model: Model
+    ) -> t.Optional[t.Union[RuleViolation, t.List[RuleViolation]]]:
         # Ignore external models themselves, because either they are registered,
         # and if they are not, they will be caught as referenced in another model.
         if isinstance(model, ExternalModel):
@@ -129,10 +137,74 @@ def check_model(self, model: Model) -> t.Optional[RuleViolation]:
         if not not_registered_external_models:
             return None
 
+        # If the model is anything other than a sql model that and has a path
+        # that ends with .sql, we cannot extract the references from the query.
+        path = model._path
+        if not isinstance(model, SqlModel) or not path or not str(path).endswith(".sql"):
+            return self._standard_error_message(
+                model_name=model.fqn,
+                external_models=not_registered_external_models,
+            )
+
+        with open(path, "r", encoding="utf-8") as file:
+            read_file = file.read()
+        split_read_file = read_file.splitlines()
+
+        # If there are any unregistered external models, return a violation find
+        # the ranges for them.
+        references = extract_references_from_query(
+            query=model.query,
+            context=self.context,
+            document_path=path,
+            read_file=split_read_file,
+            depends_on=model.depends_on,
+            dialect=model.dialect,
+        )
+        external_references = {
+            normalize_model_name(
+                table=read_range_from_string(read_file, ref.range),
+                default_catalog=model.default_catalog,
+                dialect=model.dialect,
+            ): ref
+            for ref in references
+            if isinstance(ref, ExternalModelReference) and ref.path is None
+        }
+
+        # Ensure that depends_on and external references match.
+        if not_registered_external_models != set(external_references.keys()):
+            return self._standard_error_message(
+                model_name=model.fqn,
+                external_models=not_registered_external_models,
+            )
+
+        # Return a violation for each unregistered external model with its range.
+        violations = []
+        for ref_name, ref in external_references.items():
+            if ref_name in not_registered_external_models:
+                violations.append(
+                    RuleViolation(
+                        rule=self,
+                        violation_msg=f"Model '{model.fqn}' depends on unregistered external model '{ref_name}'. "
+                        "Please register it in the external models file. This can be done by running 'sqlmesh create_external_models'.",
+                        violation_range=ref.range,
+                    )
+                )
+
+        if len(violations) < len(not_registered_external_models):
+            return self._standard_error_message(
+                model_name=model.fqn,
+                external_models=not_registered_external_models,
+            )
+
+        return violations
+
+    def _standard_error_message(
+        self, model_name: str, external_models: t.Set[str]
+    ) -> RuleViolation:
         return RuleViolation(
             rule=self,
-            violation_msg=f"Model '{model.name}' depends on unregistered external models: "
-            f"{', '.join(m for m in not_registered_external_models)}. "
+            violation_msg=f"Model '{model_name}' depends on unregistered external models: "
+            f"{', '.join(m for m in external_models)}. "
             "Please register them in the external models file. This can be done by running 'sqlmesh create_external_models'.",
         )
 

sqlmesh/lsp/completions.py

@@ -8,8 +8,8 @@
 from sqlmesh import macro
 import typing as t
 from sqlmesh.lsp.context import AuditTarget, LSPContext, ModelTarget
-from sqlmesh.lsp.description import generate_markdown_description
 from sqlmesh.lsp.uri import URI
+from sqlmesh.utils.lineage import generate_markdown_description
 
 
 def get_sql_completions(

sqlmesh/lsp/description.py

@@ -23,6 +23,10 @@
     ApiResponseGetLineage,
     ApiResponseGetModels,
 )
+
+# Define the command constant
+EXTERNAL_MODEL_UPDATE_COLUMNS = "sqlmesh.external_model_update_columns"
+
 from sqlmesh.lsp.completions import get_sql_completions
 from sqlmesh.lsp.context import (
     LSPContext,
@@ -60,15 +64,15 @@
 from sqlmesh.lsp.helpers import to_lsp_range, to_sqlmesh_position
 from sqlmesh.lsp.hints import get_hints
 from sqlmesh.lsp.reference import (
-    LSPCteReference,
-    LSPModelReference,
-    LSPExternalModelReference,
+    CTEReference,
+    ModelReference,
     get_references,
     get_all_references,
 )
 from sqlmesh.lsp.rename import prepare_rename, rename_symbol, get_document_highlights
 from sqlmesh.lsp.uri import URI
 from sqlmesh.utils.errors import ConfigError
+from sqlmesh.utils.lineage import ExternalModelReference
 from web.server.api.endpoints.lineage import column_lineage, model_lineage
 from web.server.api.endpoints.models import get_models
 from typing import Union
@@ -479,7 +483,7 @@ def hover(ls: LanguageServer, params: types.HoverParams) -> t.Optional[types.Hov
                 if not references:
                     return None
                 reference = references[0]
-                if isinstance(reference, LSPCteReference) or not reference.markdown_description:
+                if isinstance(reference, CTEReference) or not reference.markdown_description:
                     return None
                 return types.Hover(
                     contents=types.MarkupContent(
@@ -525,7 +529,7 @@ def goto_definition(
                 location_links = []
                 for reference in references:
                     # Use target_range if available (CTEs, Macros, and external models in YAML)
-                    if isinstance(reference, LSPModelReference):
+                    if isinstance(reference, ModelReference):
                         # Regular SQL models - default to start of file
                         target_range = types.Range(
                             start=types.Position(line=0, character=0),
@@ -535,7 +539,7 @@ def goto_definition(
                             start=types.Position(line=0, character=0),
                             end=types.Position(line=0, character=0),
                         )
-                    elif isinstance(reference, LSPExternalModelReference):
+                    elif isinstance(reference, ExternalModelReference):
                         # External models may have target_range set for YAML files
                         target_range = types.Range(
                             start=types.Position(line=0, character=0),