SERVER-75138 Add shapification for DocumentSources which use IDL for serialization

7 files changed, 66 insertions, 27 deletions

diff --git a/buildscripts/idl/idl/ast.py b/buildscripts/idl/idl/ast.py
index a07d547248e..a3808963213 100644
--- a/buildscripts/idl/idl/ast.py
+++ b/buildscripts/idl/idl/ast.py
@@ -109,6 +109,9 @@ class Type(common.SourceLocation):
         self.first_element_field_name = None  # type: str
         self.deserialize_with_tenant = False  # type: bool
         self.internal_only = False  # type: bool
+        # Marks whether this type is a query shape component.
+        # Can only be true if is_struct is true.
+        self.is_query_shape_component = False  # type: bool
         super(Type, self).__init__(file_name, line, column)
 
 
@@ -251,14 +254,14 @@ class Field(common.SourceLocation):
         # See WRITING-13831 for details on query shape.
         self.query_shape_literal = None  # type: Optional[bool]
         # Determines whether or not this field represents a fieldpath that should be anonymized.
-        self.query_shape_fieldpath = None  # type: Optional[bool]
+        self.query_shape_anonymize = None  # type: Optional[bool]
 
         super(Field, self).__init__(file_name, line, column)
 
     @property
     def should_serialize_query_shape(self):
         # type: () -> bool
-        return self.query_shape_fieldpath or self.query_shape_literal
+        return self.query_shape_anonymize or self.query_shape_literal
 
 
 class Privilege(common.SourceLocation):
diff --git a/buildscripts/idl/idl/binder.py b/buildscripts/idl/idl/binder.py
index da4191d83fc..179768e6df1 100644
--- a/buildscripts/idl/idl/binder.py
+++ b/buildscripts/idl/idl/binder.py
@@ -327,17 +327,17 @@ def _bind_struct_common(ctxt, parsed_spec, struct, ast_struct):
 
             # Verify that each field on the struct defines a query shape type on the field if and only if
             # query_shape_component is defined on the struct.
-            defined_query_shape_type = ast_field.query_shape_literal is not None or ast_field.query_shape_fieldpath is not None
+            defined_query_shape_type = ast_field.query_shape_literal is not None or ast_field.query_shape_anonymize is not None
             if not field.hidden and struct.query_shape_component and not defined_query_shape_type:
                 ctxt.add_must_declare_shape_type(ast_field, ast_struct.name, ast_field.name)
 
             if not struct.query_shape_component and defined_query_shape_type:
                 ctxt.add_must_be_query_shape_component(ast_field, ast_struct.name, ast_field.name)
 
-            if ast_field.query_shape_fieldpath and ast_field.type.cpp_type not in [
+            if ast_field.query_shape_anonymize and ast_field.type.cpp_type not in [
                     "std::string", "std::vector<std::string>"
             ]:
-                ctxt.add_query_shape_fieldpath_must_be_string(ast_field, ast_field.name,
+                ctxt.add_query_shape_anonymize_must_be_string(ast_field, ast_field.name,
                                                               ast_field.type.cpp_type)
 
     # Fill out the field comparison_order property as needed
@@ -455,6 +455,7 @@ def _bind_struct_type(struct):
     ast_type.cpp_type = _get_struct_qualified_cpp_name(struct)
     ast_type.bson_serialization_type = ["object"]
     ast_type.first_element_field_name = struct.fields[0].name if struct.fields else None
+    ast_type.is_query_shape_component = struct.query_shape_component
     return ast_type
 
 
@@ -1019,6 +1020,7 @@ def _bind_type(idltype):
     ast_type.deserializer = _normalize_method_name(idltype.cpp_type, idltype.deserializer)
     ast_type.deserialize_with_tenant = idltype.deserialize_with_tenant
     ast_type.internal_only = idltype.internal_only
+    ast_type.is_query_shape_component = True
     return ast_type
 
 
@@ -1045,7 +1047,7 @@ def _bind_field(ctxt, parsed_spec, field):
     ast_field.always_serialize = field.always_serialize
     ast_field.preparse = field.preparse
     ast_field.query_shape_literal = field.query_shape_literal
-    ast_field.query_shape_fieldpath = field.query_shape_fieldpath
+    ast_field.query_shape_anonymize = field.query_shape_anonymize
 
     ast_field.cpp_name = field.name
     if field.cpp_name:
@@ -1056,11 +1058,11 @@ def _bind_field(ctxt, parsed_spec, field):
         ctxt.add_array_not_valid_error(ast_field, "field", ast_field.name)
 
     # Validate that 'field' is not both a query shape literal and query shape fieldpath. The two are mutually exclusive.
-    if ast_field.query_shape_literal is not None and ast_field.query_shape_fieldpath is not None:
+    if ast_field.query_shape_literal is not None and ast_field.query_shape_anonymize is not None:
         ctxt.add_field_cannot_be_literal_and_fieldpath(ast_field, ast_field.name)
 
-    if ast_field.query_shape_fieldpath is False:
-        ctxt.add_field_cannot_have_query_shape_fieldpath_false(ast_field)
+    if ast_field.query_shape_anonymize is False:
+        ctxt.add_field_cannot_have_query_shape_anonymize_false(ast_field)
 
     if field.ignore:
         ast_field.ignore = field.ignore
@@ -1134,6 +1136,8 @@ def _bind_field(ctxt, parsed_spec, field):
         if ast_field.validator is None:
             return None
 
+    if ast_field.should_serialize_query_shape and not ast_field.type.is_query_shape_component:
+        ctxt.add_must_be_query_shape_component(ast_field, ast_field.type.name, ast_field.name)
     return ast_field
 
 
diff --git a/buildscripts/idl/idl/errors.py b/buildscripts/idl/idl/errors.py
index e9386535ffe..5837cb11648 100644
--- a/buildscripts/idl/idl/errors.py
+++ b/buildscripts/idl/idl/errors.py
@@ -950,10 +950,10 @@ class ParserContext(object):
 
     def add_must_declare_shape_type(self, location, struct_name, field_name):
         # type: (common.SourceLocation, str, str) -> None
-        """Add an error about a field not specifying either query_shape_literal or query_shape_fieldpath if the struct is query_shape_component."""
+        """Add an error about a field not specifying either query_shape_literal or query_shape_anonymize if the struct is query_shape_component."""
         self._add_error(
             location, ERROR_ID_FIELD_MUST_DECLARE_SHAPE_LITERAL,
-            f"Field '{field_name}' must specify either 'query_shape_literal' or 'query_shape_fieldpath' since struct '{struct_name}' is a query shape component."
+            f"Field '{field_name}' must specify either 'query_shape_literal' or 'query_shape_anonymize' since struct '{struct_name}' is a query shape component."
         )
 
     def add_must_be_query_shape_component(self, location, struct_name, field_name):
@@ -963,7 +963,7 @@ class ParserContext(object):
             f"Field '{field_name}' cannot specify 'query_shape_literal' property since struct '{struct_name}' is not a query shape component."
         )
 
-    def add_query_shape_fieldpath_must_be_string(self, location, field_name, field_type):
+    def add_query_shape_anonymize_must_be_string(self, location, field_name, field_type):
         self._add_error(
             location, ERROR_ID_INVALID_TYPE_FOR_SHAPIFY,
             f"In order for {field_name} to be marked as a query shape fieldpath, it must have a string type, not {field_type}."
@@ -975,9 +975,9 @@ class ParserContext(object):
             f"{field_name} cannot be marked as both a query shape literal and query shape fieldpath."
         )
 
-    def add_field_cannot_have_query_shape_fieldpath_false(self, location):
+    def add_field_cannot_have_query_shape_anonymize_false(self, location):
         self._add_error(location, ERROR_ID_QUERY_SHAPE_FIELDPATH_CANNOT_BE_FALSE,
-                        "'query_shape_fieldpath' cannot be defined as false if it is set.")
+                        "'query_shape_anonymize' cannot be defined as false if it is set.")
 
 
 def _assert_unique_error_messages():
diff --git a/buildscripts/idl/idl/generator.py b/buildscripts/idl/idl/generator.py
index b472ffe17a6..26792609546 100644
--- a/buildscripts/idl/idl/generator.py
+++ b/buildscripts/idl/idl/generator.py
@@ -2166,7 +2166,7 @@ class _CppSourceFileWriter(_CppFileWriterBase):
                         self._writer.write_template(
                             'options.serializeLiteralValue(${expression}).serializeForIDL(${field_name}, builder);'
                         )
-                    elif field.query_shape_fieldpath:
+                    elif field.query_shape_anonymize:
                         self._writer.write_template(
                             'builder->append(${field_name}, options.serializeFieldPathFromString(${expression}));'
                         )
@@ -2321,7 +2321,7 @@ class _CppSourceFileWriter(_CppFileWriterBase):
                             self._writer.write_template(
                                 'options.serializeLiteralValue(${expression}).serializeForIDL(${field_name}, builder);'
                             )
-                        elif field.query_shape_fieldpath:
+                        elif field.query_shape_anonymize:
                             self._writer.write_template(
                                 'builder->append(${field_name}, options.serializeFieldPathFromString(${expression}));'
                             )
@@ -2335,7 +2335,7 @@ class _CppSourceFileWriter(_CppFileWriterBase):
                             self._writer.write_template(
                                 'options.serializeLiteralValue(value).serializeForIDL(${field_name}, builder);'
                             )
-                        elif field.query_shape_fieldpath:
+                        elif field.query_shape_anonymize:
                             self._writer.write_template(
                                 'idl::idlSerialize(builder, ${field_name}, options.serializeFieldPathFromString(value));'
                             )
@@ -2382,7 +2382,7 @@ class _CppSourceFileWriter(_CppFileWriterBase):
                         self._writer.write_line(
                             'options.serializeLiteralValue(%s).serializeForIDL(%s, builder);' %
                             (_access_member(field), _get_field_constant_name(field)))
-                    elif field.query_shape_fieldpath:
+                    elif field.query_shape_anonymize:
                         self._writer.write_line(
                             'builder->append(%s, options.serializeFieldPathFromString(%s));' %
                             (_get_field_constant_name(field), _access_member(field)))
diff --git a/buildscripts/idl/idl/parser.py b/buildscripts/idl/idl/parser.py
index cc4f047231e..022ad0dd3f9 100644
--- a/buildscripts/idl/idl/parser.py
+++ b/buildscripts/idl/idl/parser.py
@@ -408,7 +408,7 @@ def _parse_field(ctxt, name, node):
                 _RuleDesc('bool_scalar'),
             "query_shape_literal":
                 _RuleDesc('required_bool_scalar'),
-            "query_shape_fieldpath":
+            "query_shape_anonymize":
                 _RuleDesc('required_bool_scalar'),
         })
 
diff --git a/buildscripts/idl/idl/syntax.py b/buildscripts/idl/idl/syntax.py
index c8b3573294f..f161488062e 100644
--- a/buildscripts/idl/idl/syntax.py
+++ b/buildscripts/idl/idl/syntax.py
@@ -494,7 +494,7 @@ class Field(common.SourceLocation):
         self.constructed = False  # type: bool
 
         self.query_shape_literal = None  # type: Optional[bool]
-        self.query_shape_fieldpath = None  # type: Optional[bool]
+        self.query_shape_anonymize = None  # type: Optional[bool]
 
         self.hidden = False  # type: bool
 
diff --git a/buildscripts/idl/tests/test_binder.py b/buildscripts/idl/tests/test_binder.py
index a891c6f15b2..c4e3c806d71 100644
--- a/buildscripts/idl/tests/test_binder.py
+++ b/buildscripts/idl/tests/test_binder.py
@@ -2704,7 +2704,7 @@ class TestBinder(testcase.IDLTestcase):
                             query_shape_literal: false
         """), idl.errors.ERROR_ID_CANNOT_DECLARE_SHAPE_LITERAL)
 
-        # Validating query_shape_fieldpath relies on std::string
+        # Validating query_shape_anonymize relies on std::string
         basic_types = textwrap.dedent("""
             types:
                 string:
@@ -2731,7 +2731,7 @@ class TestBinder(testcase.IDLTestcase):
                     description: ""
                     fields:
                         field1:
-                            query_shape_fieldpath: true
+                            query_shape_anonymize: true
                             type: string
                         field2:
                             query_shape_literal: false
@@ -2746,7 +2746,7 @@ class TestBinder(testcase.IDLTestcase):
                     description: ""
                     fields:
                         field1:
-                            query_shape_fieldpath: true
+                            query_shape_anonymize: true
                             type: array<string>
                         field2:
                             query_shape_literal: false
@@ -2762,7 +2762,7 @@ class TestBinder(testcase.IDLTestcase):
                     description: ""
                     fields:
                         field1:
-                            query_shape_fieldpath: false
+                            query_shape_anonymize: false
                             type: string
                         field2:
                             query_shape_literal: false
@@ -2778,7 +2778,7 @@ class TestBinder(testcase.IDLTestcase):
                     description: ""
                     fields:
                         field1:
-                            query_shape_fieldpath: true
+                            query_shape_anonymize: true
                             type: bool
                         field2:
                             query_shape_literal: false
@@ -2794,7 +2794,7 @@ class TestBinder(testcase.IDLTestcase):
                     description: ""
                     fields:
                         field1:
-                            query_shape_fieldpath: true
+                            query_shape_anonymize: true
                             type: array<bool>
                         field2:
                             query_shape_literal: false
@@ -2810,7 +2810,7 @@ class TestBinder(testcase.IDLTestcase):
                     description: ""
                     fields:
                         field1:
-                            query_shape_fieldpath: true
+                            query_shape_anonymize: true
                             query_shape_literal: true
                             type: string
                         field2:
@@ -2818,6 +2818,38 @@ class TestBinder(testcase.IDLTestcase):
                             type: bool
         """), idl.errors.ERROR_ID_CANNOT_BE_LITERAL_AND_FIELDPATH)
 
+        self.assert_bind_fail(
+            basic_types + textwrap.dedent("""
+            structs:
+                StructZero:
+                    strict: true
+                    description: ""
+                    fields:
+                        field1:
+                            query_shape_literal: true
+                            type: string
+            """), idl.errors.ERROR_ID_CANNOT_DECLARE_SHAPE_LITERAL)
+
+        self.assert_bind_fail(
+            basic_types + textwrap.dedent("""
+            structs:
+                StructZero:
+                    strict: true
+                    description: ""
+                    fields:
+                        field1:
+                            type: string
+                struct1:
+                    query_shape_component: true
+                    strict: true
+                    description: ""
+                    fields:
+                        field2:
+                            type: StructZero
+                            description: ""
+                            query_shape_literal: true
+            """), idl.errors.ERROR_ID_CANNOT_DECLARE_SHAPE_LITERAL)
+
 
 if __name__ == '__main__':