Feat(lsp): Add support for find and go to references for Model usages (#4680)

Author: themisvaltinos

sqlmesh/lsp/main.py

@@ -44,7 +44,12 @@
     CustomMethod,
 )
 from sqlmesh.lsp.hints import get_hints
-from sqlmesh.lsp.reference import get_references, get_cte_references
+from sqlmesh.lsp.reference import (
+    LSPCteReference,
+    LSPModelReference,
+    get_references,
+    get_all_references,
+)
 from sqlmesh.lsp.uri import URI
 from web.server.api.endpoints.lineage import column_lineage, model_lineage
 from web.server.api.endpoints.models import get_models
@@ -368,7 +373,7 @@ def hover(ls: LanguageServer, params: types.HoverParams) -> t.Optional[types.Hov
                 if not references:
                     return None
                 reference = references[0]
-                if not reference.markdown_description:
+                if isinstance(reference, LSPCteReference) or not reference.markdown_description:
                     return None
                 return types.Hover(
                     contents=types.MarkupContent(
@@ -418,8 +423,8 @@ def goto_definition(
                 references = get_references(self.lsp_context, uri, params.position)
                 location_links = []
                 for reference in references:
-                    # Use target_range if available (for CTEs), otherwise default to start of file
-                    if reference.target_range:
+                    # Use target_range if available (CTEs, Macros), otherwise default to start of file
+                    if not isinstance(reference, LSPModelReference):
                         target_range = reference.target_range
                         target_selection_range = reference.target_range
                     else:
@@ -449,18 +454,18 @@ def goto_definition(
         def find_references(
             ls: LanguageServer, params: types.ReferenceParams
         ) -> t.Optional[t.List[types.Location]]:
-            """Find all references of a symbol (currently supporting CTEs)"""
+            """Find all references of a symbol (supporting CTEs, models for now)"""
             try:
                 uri = URI(params.text_document.uri)
                 self._ensure_context_for_document(uri)
                 document = ls.workspace.get_text_document(params.text_document.uri)
                 if self.lsp_context is None:
                     raise RuntimeError(f"No context found for document: {document.path}")
 
-                cte_references = get_cte_references(self.lsp_context, uri, params.position)
+                all_references = get_all_references(self.lsp_context, uri, params.position)
 
                 # Convert references to Location objects
-                locations = [types.Location(uri=ref.uri, range=ref.range) for ref in cte_references]
+                locations = [types.Location(uri=ref.uri, range=ref.range) for ref in all_references]
 
                 return locations if locations else None
             except Exception as e:

sqlmesh/lsp/reference.py

@@ -1,6 +1,7 @@
 from lsprotocol.types import Range, Position
 import typing as t
 from pathlib import Path
+from pydantic import Field
 
 from sqlmesh.core.audit import StandaloneAudit
 from sqlmesh.core.dialect import normalize_model_name
@@ -23,21 +24,37 @@
 import inspect
 
 
-class Reference(PydanticModel):
-    """
-    A reference to a model or CTE.
+class LSPModelReference(PydanticModel):
+    """A LSP reference to a model."""
+
+    type: t.Literal["model"] = "model"
+    uri: str
+    range: Range
+    markdown_description: t.Optional[str] = None
 
-    Attributes:
-        range: The range of the reference in the source file
-        uri: The uri of the referenced model
-        markdown_description: The markdown description of the referenced model
-        target_range: The range of the definition for go-to-definition (optional, used for CTEs)
-    """
 
+class LSPCteReference(PydanticModel):
+    """A LSP reference to a CTE."""
+
+    type: t.Literal["cte"] = "cte"
+    uri: str
     range: Range
+    target_range: Range
+
+
+class LSPMacroReference(PydanticModel):
+    """A LSP reference to a macro."""
+
+    type: t.Literal["macro"] = "macro"
     uri: str
+    range: Range
+    target_range: Range
     markdown_description: t.Optional[str] = None
-    target_range: t.Optional[Range] = None
+
+
+Reference = t.Annotated[
+    t.Union[LSPModelReference, LSPCteReference, LSPMacroReference], Field(discriminator="type")
+]
 
 
 def by_position(position: Position) -> t.Callable[[Reference], bool]:
@@ -136,7 +153,7 @@ def get_model_definitions_for_a_path(
         return []
 
     # Find all possible references
-    references = []
+    references: t.List[Reference] = []
 
     with open(file_path, "r", encoding="utf-8") as file:
         read_file = file.readlines()
@@ -173,7 +190,7 @@ def get_model_definitions_for_a_path(
                             table_range = to_lsp_range(table_range_sqlmesh)
 
                             references.append(
-                                Reference(
+                                LSPCteReference(
                                     uri=document_uri.value,  # Same file
                                     range=table_range,
                                     target_range=target_range,
@@ -227,7 +244,7 @@ def get_model_definitions_for_a_path(
                 description = generate_markdown_description(referenced_model)
 
                 references.append(
-                    Reference(
+                    LSPModelReference(
                         uri=referenced_model_uri.value,
                         range=Range(
                             start=to_lsp_position(start_pos_sqlmesh),
@@ -286,7 +303,7 @@ def get_macro_definitions_for_a_path(
         return []
 
     references = []
-    config_for_model, config_path = lsp_context.context.config_for_path(
+    _, config_path = lsp_context.context.config_for_path(
         file_path,
     )
 
@@ -372,7 +389,7 @@ def get_macro_reference(
         # Create a reference to the macro definition
         macro_uri = URI.from_path(path)
 
-        return Reference(
+        return LSPMacroReference(
             uri=macro_uri.value,
             range=to_lsp_range(macro_range),
             target_range=Range(
@@ -405,7 +422,7 @@ def get_built_in_macro_reference(macro_name: str, macro_range: Range) -> t.Optio
     # Calculate the end line number by counting the number of source lines
     end_line_number = line_number + len(source_lines) - 1
 
-    return Reference(
+    return LSPMacroReference(
         uri=URI.from_path(Path(filename)).value,
         range=macro_range,
         target_range=Range(
@@ -416,9 +433,99 @@ def get_built_in_macro_reference(macro_name: str, macro_range: Range) -> t.Optio
     )
 
 
+def get_model_find_all_references(
+    lint_context: LSPContext, document_uri: URI, position: Position
+) -> t.List[LSPModelReference]:
+    """
+    Get all references to a model across the entire project.
+
+    This function finds all usages of a model in other files by searching through
+    all models in the project and checking their dependencies.
+
+    Args:
+        lint_context: The LSP context
+        document_uri: The URI of the document
+        position: The position to check for model references
+
+    Returns:
+        A list of references to the model across all files
+    """
+    # First, get the references in the current file to determine what model we're looking for
+    current_file_references = [
+        ref
+        for ref in get_model_definitions_for_a_path(lint_context, document_uri)
+        if isinstance(ref, LSPModelReference)
+    ]
+
+    # Find the model reference at the cursor position
+    target_model_uri: t.Optional[str] = None
+    for ref in current_file_references:
+        if _position_within_range(position, ref.range):
+            # This is a model reference, get the target model URI
+            target_model_uri = ref.uri
+            break
+
+    if target_model_uri is None:
+        return []
+
+    # Start with the model definition
+    all_references: t.List[LSPModelReference] = [
+        LSPModelReference(
+            uri=ref.uri,
+            range=Range(
+                start=Position(line=0, character=0),
+                end=Position(line=0, character=0),
+            ),
+            markdown_description=ref.markdown_description,
+        )
+    ]
+
+    # Then add the original reference
+    for ref in current_file_references:
+        if ref.uri == target_model_uri and isinstance(ref, LSPModelReference):
+            all_references.append(
+                LSPModelReference(
+                    uri=document_uri.value,
+                    range=ref.range,
+                    markdown_description=ref.markdown_description,
+                )
+            )
+
+    # Search through the models in the project
+    for path, target in lint_context.map.items():
+        if not isinstance(target, (ModelTarget, AuditTarget)):
+            continue
+
+        file_uri = URI.from_path(path)
+
+        # Skip current file, already processed
+        if file_uri.value == document_uri.value:
+            continue
+
+        # Get model references for this file
+        file_references = [
+            ref
+            for ref in get_model_definitions_for_a_path(lint_context, file_uri)
+            if isinstance(ref, LSPModelReference)
+        ]
+
+        # Add references that point to the target model file
+        for ref in file_references:
+            if ref.uri == target_model_uri and isinstance(ref, LSPModelReference):
+                all_references.append(
+                    LSPModelReference(
+                        uri=file_uri.value,
+                        range=ref.range,
+                        markdown_description=ref.markdown_description,
+                    )
+                )
+
+    return all_references
+
+
 def get_cte_references(
     lint_context: LSPContext, document_uri: URI, position: Position
-) -> t.List[Reference]:
+) -> t.List[LSPCteReference]:
     """
     Get all references to a CTE at a specific position in a document.
 
@@ -432,12 +539,12 @@ def get_cte_references(
     Returns:
         A list of references to the CTE (including its definition and all usages)
     """
-    references = get_model_definitions_for_a_path(lint_context, document_uri)
 
-    # Filter for CTE references (those with target_range set and same URI)
-    # TODO: Consider extending Reference class to explicitly indicate reference type instead
-    cte_references = [
-        ref for ref in references if ref.target_range is not None and ref.uri == document_uri.value
+    # Filter to get the CTE references
+    cte_references: t.List[LSPCteReference] = [
+        ref
+        for ref in get_model_definitions_for_a_path(lint_context, document_uri)
+        if isinstance(ref, LSPCteReference)
     ]
 
     if not cte_references:
@@ -450,7 +557,7 @@ def get_cte_references(
             target_cte_definition_range = ref.target_range
             break
         # Check if cursor is on the CTE definition
-        elif ref.target_range and _position_within_range(position, ref.target_range):
+        elif _position_within_range(position, ref.target_range):
             target_cte_definition_range = ref.target_range
             break
 
@@ -459,27 +566,55 @@ def get_cte_references(
 
     # Add the CTE definition
     matching_references = [
-        Reference(
+        LSPCteReference(
             uri=document_uri.value,
             range=target_cte_definition_range,
-            markdown_description="CTE definition",
+            target_range=target_cte_definition_range,
         )
     ]
 
     # Add all usages
     for ref in cte_references:
         if ref.target_range == target_cte_definition_range:
             matching_references.append(
-                Reference(
+                LSPCteReference(
                     uri=document_uri.value,
                     range=ref.range,
-                    markdown_description="CTE usage",
+                    target_range=ref.target_range,
                 )
             )
 
     return matching_references
 
 
+def get_all_references(
+    lint_context: LSPContext, document_uri: URI, position: Position
+) -> t.Sequence[Reference]:
+    """
+    Get all references of a symbol at a specific position in a document.
+
+    This function determines the type of reference (CTE, model for now) at the cursor
+    position and returns all references to that symbol across the project.
+
+    Args:
+        lint_context: The LSP context
+        document_uri: The URI of the document
+        position: The position to check for references
+
+    Returns:
+        A list of references to the symbol at the given position
+    """
+    # First try CTE references (within same file)
+    if cte_references := get_cte_references(lint_context, document_uri, position):
+        return cte_references
+
+    # Then try model references (across files)
+    if model_references := get_model_find_all_references(lint_context, document_uri, position):
+        return model_references
+
+    return []
+
+
 def _position_within_range(position: Position, range: Range) -> bool:
     """Check if a position is within a given range."""
     return (

tests/lsp/test_reference.py

@@ -1,7 +1,7 @@
 from lsprotocol.types import Position
 from sqlmesh.core.context import Context
 from sqlmesh.lsp.context import LSPContext, ModelTarget, AuditTarget
-from sqlmesh.lsp.reference import get_model_definitions_for_a_path, by_position
+from sqlmesh.lsp.reference import LSPModelReference, get_model_definitions_for_a_path, by_position
 from sqlmesh.lsp.uri import URI
 
 
@@ -47,9 +47,13 @@ def test_reference_with_alias() -> None:
         if isinstance(info, ModelTarget) and "sushi.waiter_revenue_by_day" in info.names
     )
 
-    references = get_model_definitions_for_a_path(
-        lsp_context, URI.from_path(waiter_revenue_by_day_path)
-    )
+    references = [
+        ref
+        for ref in get_model_definitions_for_a_path(
+            lsp_context, URI.from_path(waiter_revenue_by_day_path)
+        )
+        if isinstance(ref, LSPModelReference)
+    ]
     assert len(references) == 3
 
     with open(waiter_revenue_by_day_path, "r") as file: