Merge pull request #2510 from andrewdotn/master

Fix parsing of escaped backslashes

Author: Robert Mosolgo

lib/graphql/language/lexer.rb

@@ -35,11 +35,17 @@
   RBRACKET =      ']';
   COLON =         ':';
   QUOTE =         '"';
+  BACKSLASH = '\\';
+  # Could limit to hex here, but “bad unicode escape” on 0XXF is probably a
+  # more helpful error than “unknown char”
+  UNICODE_ESCAPE = '\\u' [0-9A-Za-z]{4};
+  # https://graphql.github.io/graphql-spec/June2018/#sec-String-Value
+  STRING_ESCAPE = '\\' [\\/bfnrt];
   BLOCK_QUOTE =   '"""';
   ESCAPED_BLOCK_QUOTE = '\\"""';
   BLOCK_STRING_CHAR = (ESCAPED_BLOCK_QUOTE | ^QUOTE | QUOTE{1,2} ^QUOTE);
   ESCAPED_QUOTE = '\\"';
-  STRING_CHAR =   (ESCAPED_QUOTE | ^QUOTE);
+  STRING_CHAR =   ((ESCAPED_QUOTE | ^QUOTE) - BACKSLASH) | UNICODE_ESCAPE | STRING_ESCAPE;
   VAR_SIGN =      '$';
   DIR_SIGN =      '@';
   ELLIPSIS =      '...';
@@ -53,6 +59,11 @@
   # catch-all for anything else. must be at the bottom for precedence.
   UNKNOWN_CHAR =         /./;
 
+  # Used with ragel -V for graphviz visualization
+  str := |*
+      QUOTED_STRING => { emit_string(ts, te, meta, block: false) };
+  *|;
+
   main := |*
     INT           => { emit(:INT, ts, te, meta) };
     FLOAT         => { emit(:FLOAT, ts, te, meta) };

lib/graphql/language/lexer.rl

@@ -32,6 +32,11 @@
       assert_equal 'c', tokens[1].value
     end
 
+    it "handles escaped backslashes before escaped quotes" do
+      tokens = subject.tokenize('text: "b\\\\", otherText: "a"')
+      assert_equal ['text', ':', 'b\\', 'otherText', ':', 'a',], tokens.map(&:value)
+    end
+
     describe "block strings" do
       let(:query_string) { %|{ a(b: """\nc\n \\""" d\n""" """""e""""")}|}
 
@@ -60,6 +65,11 @@
 
         assert_equal '', tokens[0].value
       end
+
+      it "tokenizes escaped backslashes at the end of blocks" do
+        tokens = subject.tokenize('text: """b\\\\""", otherText: "a"')
+        assert_equal ['text', ':', 'b\\', 'otherText', ':', 'a',], tokens.map(&:value)
+      end
     end
 
     it "unescapes escaped characters" do

spec/graphql/language/lexer_spec.rb

@@ -122,6 +122,30 @@
     end
   end
 
+  it "parses backslashes in arguments" do
+    document = subject.parse <<-GRAPHQL
+      query {
+        item(text: "a", otherText: "b\\\\") {
+          text
+          otherText
+        }
+      }
+    GRAPHQL
+    assert_equal "b\\", document.definitions[0].selections[0].arguments[1].value
+  end
+
+  it "parses backslases in non-last arguments" do
+    document = subject.parse <<-GRAPHQL
+      query {
+        item(text: "b\\\\", otherText: "a") {
+          text
+          otherText
+        }
+      }
+    GRAPHQL
+    assert_equal "b\\", document.definitions[0].selections[0].arguments[0].value
+  end
+
   it "parses the test schema" do
     schema = Dummy::Schema
     schema_string = GraphQL::Schema::Printer.print_schema(schema)