fix: add medium-level validation guards for facets and tokenizer args (#28)

Co-authored-by: Ankit <573824+ankitml@users.noreply.github.com>

Author: philippemnoel

lib/parade_db/migration_helpers.rb

@@ -213,10 +213,11 @@ def tokenizer_args(options)
       positional = Array(opts.delete(:__positional)).map { |value| tokenizer_positional_arg_sql(value) }
 
       if opts.key?(:min) || opts.key?(:max)
-        if opts.key?(:min) && opts.key?(:max)
-          positional << Integer(opts.delete(:min)).to_s
-          positional << Integer(opts.delete(:max)).to_s
+        unless opts.key?(:min) && opts.key?(:max)
+          raise ArgumentError, "tokenizer named args :min and :max must be provided together"
         end
+        positional << Integer(opts.delete(:min)).to_s
+        positional << Integer(opts.delete(:max)).to_s
       end
 
       named = opts.map do |k, v|

lib/parade_db/search_methods.rb

@@ -599,9 +599,24 @@ def normalize_facet_fields(fields, agg:)
         raise ArgumentError, "Facet field names must be unique."
       end
 
+      validate_facet_fields_indexed!(normalized)
       normalized
     end
 
+    def validate_facet_fields_indexed!(fields)
+      return unless klass.respond_to?(:paradedb_indexed_fields)
+
+      indexed_fields = klass.paradedb_indexed_fields
+      return if indexed_fields.empty?
+
+      unknown = fields.reject { |field| indexed_fields.include?(field) }
+      return if unknown.empty?
+
+      raise ParadeDB::FieldNotIndexed,
+            "#{klass.name}.facets contains non-indexed fields #{unknown.join(', ')}. " \
+            "Indexed fields: #{indexed_fields.join(', ')}"
+    end
+
     def normalize_facet_size(size)
       return nil if size.nil?
 

spec/index_dsl_unit_spec.rb

@@ -141,6 +141,38 @@
     assert_includes sql, "pdb.ngram(2, 5)"
   end
 
+  it "rejects partial ngram bounds in named_args" do
+    klass = Class.new(ParadeDB::Index) do
+      self.table_name = :products
+      self.key_field = :id
+      self.fields = {
+        id: {},
+        description: { tokenizer: :ngram, named_args: { min: 2 } }
+      }
+    end
+
+    error = assert_raises(ArgumentError) do
+      ActiveRecord::Base.connection.send(:build_create_sql, klass.compiled_definition, if_not_exists: false)
+    end
+    assert_includes error.message, ":min and :max must be provided together"
+  end
+
+  it "rejects max-only ngram bounds in named_args" do
+    klass = Class.new(ParadeDB::Index) do
+      self.table_name = :products
+      self.key_field = :id
+      self.fields = {
+        id: {},
+        description: { tokenizer: :ngram, named_args: { max: 5 } }
+      }
+    end
+
+    error = assert_raises(ArgumentError) do
+      ActiveRecord::Base.connection.send(:build_create_sql, klass.compiled_definition, if_not_exists: false)
+    end
+    assert_includes error.message, ":min and :max must be provided together"
+  end
+
   it "renders qualified and inline tokenizer forms" do
     klass = Class.new(ParadeDB::Index) do
       self.table_name = :products

spec/index_runtime_features_unit_spec.rb

@@ -126,6 +126,26 @@
     assert_includes sql, "::pdb.alias('description_simple')"
   end
 
+  it "facets raises FieldNotIndexed for non-indexed facet fields" do
+    Object.const_set("RuntimeProduct", Class.new(ActiveRecord::Base) do
+      self.table_name = :products
+      include ParadeDB::Model
+    end)
+    Object.const_set("RuntimeProductIndex", Class.new(ParadeDB::Index) do
+      self.table_name = :products
+      self.key_field = :id
+      self.fields = { id: {}, description: {} }
+    end)
+
+    conn = ActiveRecord::Base.connection
+    conn.create_paradedb_index(RuntimeProductIndex, if_not_exists: true)
+
+    error = assert_raises(ParadeDB::FieldNotIndexed) do
+      RuntimeProduct.search(:description).matching_all("shoe").build_facet_query(fields: [:price]).sql
+    end
+    assert_includes error.message, "non-indexed fields"
+  end
+
   it "raise mode detects missing catalog index drift" do
     Object.const_set("DriftProduct", Class.new(ActiveRecord::Base) do
       self.table_name = :products