Fix raw identifiers in collection schema (#76)

* Fix raw identifiers in collection schema

* Format

* to_string -> to_owned

Author: milesgranger

typesense/tests/derive/collection.rs

@@ -260,7 +260,66 @@ fn derived_document_handles_nested_and_flattened_fields() {
     assert_eq!(serde_json::to_value(schema).unwrap(), expected);
 }
 
-// Test 4: All Boolean Shorthand Attributes
+// Test 4: Raw Identifiers
+//
+// Rust raw identifiers (r#type, r#abstract, etc.) must have the r# prefix
+// stripped in all generated output: regular field names, flattened prefixes,
+// and default_sorting_field validation.
+
+#[allow(dead_code)]
+#[derive(Typesense, Serialize, Deserialize)]
+#[typesense(collection_name = "raw_ident_docs", default_sorting_field = "type")]
+struct RawIdentDoc {
+    id: String,
+    #[typesense(type = "int32")]
+    r#type: String,
+    r#abstract: Option<String>,
+    normal_field: i32,
+}
+
+#[derive(Typesense, Serialize, Deserialize)]
+struct RawIdentNested {
+    value: String,
+}
+
+#[allow(dead_code)]
+#[derive(Typesense, Serialize, Deserialize)]
+#[typesense(collection_name = "raw_ident_flat")]
+struct RawIdentFlat {
+    id: String,
+    #[typesense(flatten, skip)]
+    r#match: RawIdentNested,
+}
+
+#[test]
+fn derived_document_strips_raw_identifier_prefix() {
+    // Regular fields: r#type -> "type", r#abstract -> "abstract"
+    let schema = RawIdentDoc::collection_schema();
+    let expected = json!({
+      "name": "raw_ident_docs",
+      "default_sorting_field": "type",
+      "fields": [
+        { "name": "id", "type": "string" },
+        { "name": "type", "type": "int32" },
+        { "name": "abstract", "type": "string", "optional": true },
+        { "name": "normal_field", "type": "int32" }
+      ]
+    });
+    assert_eq!(serde_json::to_value(schema).unwrap(), expected);
+
+    // Flattened prefix: r#match -> "match.value"
+    let schema = RawIdentFlat::collection_schema();
+    let expected = json!({
+      "name": "raw_ident_flat",
+      "fields": [
+        { "name": "id", "type": "string" },
+        { "name": "match.value", "type": "string" }
+      ]
+    });
+    assert_eq!(serde_json::to_value(schema).unwrap(), expected);
+}
+
+// Test 5: All Boolean Shorthand Attributes
 
 #[allow(dead_code)]
 #[derive(Typesense, Serialize, Deserialize)]

typesense_derive/src/field_attributes.rs

@@ -1,4 +1,7 @@
-use crate::{bool_literal, get_inner_type, i32_literal, skip_eq, string_literal, ty_inner_type};
+use crate::{
+    bool_literal, get_inner_type, i32_literal, skip_eq, string_literal, strip_raw_prefix,
+    ty_inner_type,
+};
 use proc_macro2::TokenTree;
 use quote::quote;
 use syn::{Attribute, Field};
@@ -271,7 +274,7 @@ fn build_regular_field(field: &Field, field_attrs: &FieldAttributes) -> proc_mac
     let field_name = if let Some(rename) = &field_attrs.rename {
         quote! { #rename }
     } else {
-        let name_ident = field.ident.as_ref().unwrap().to_string();
+        let name_ident = strip_raw_prefix(&field.ident.as_ref().unwrap().to_string());
         quote! { #name_ident }
     };
 
@@ -322,7 +325,7 @@ pub(crate) fn process_field(
         let prefix = if let Some(rename_prefix) = &field_attrs.rename {
             quote! { #rename_prefix }
         } else {
-            let name_ident = field.ident.as_ref().unwrap().to_string();
+            let name_ident = strip_raw_prefix(&field.ident.as_ref().unwrap().to_string());
             quote! { #name_ident }
         };
 

typesense_derive/src/helpers.rs

@@ -133,6 +133,12 @@ pub(crate) fn ty_inner_type<'a>(ty: &'a syn::Type, wrapper: &'static str) -> Opt
     None
 }
 
+/// Strip the `r#` prefix from raw identifiers.
+/// ie, 'r#' in `r#type`should not appear in generated output like schema field names.
+pub(crate) fn strip_raw_prefix(s: &str) -> String {
+    s.strip_prefix("r#").unwrap_or(s).to_owned()
+}
+
 /// Helper to get the inner-most type from nested Option/Vec wrappers.
 pub(crate) fn get_inner_type(mut ty: &syn::Type) -> &syn::Type {
     while let Some(inner) = ty_inner_type(ty, "Option").or_else(|| ty_inner_type(ty, "Vec")) {

typesense_derive/src/lib.rs

@@ -56,9 +56,9 @@ fn impl_typesense_collection(item: ItemStruct) -> syn::Result<TokenStream> {
             .iter()
             .map(|field| {
                 extract_field_attrs(field).map(|attrs| {
-                    attrs
-                        .rename
-                        .unwrap_or_else(|| field.ident.as_ref().unwrap().to_string())
+                    attrs.rename.unwrap_or_else(|| {
+                        strip_raw_prefix(&field.ident.as_ref().unwrap().to_string())
+                    })
                 })
             })
             .collect::<syn::Result<Vec<String>>>()?;