graphql 0.16.0 → 2.0.15

data/lib/graphql/language/lexer.rl

@@ -17,52 +17,97 @@
   QUERY =         'query';
   MUTATION =      'mutation';
   SUBSCRIPTION =  'subscription';
-  RCURLY =        '{';
-  LCURLY =        '}';
-  RPAREN =        '(';
-  LPAREN =        ')';
-  RBRACKET =      '[';
-  LBRACKET =      ']';
+  SCHEMA =        'schema';
+  SCALAR =        'scalar';
+  TYPE =          'type';
+  EXTEND =        'extend';
+  IMPLEMENTS =    'implements';
+  INTERFACE =     'interface';
+  UNION =         'union';
+  ENUM =          'enum';
+  INPUT =         'input';
+  DIRECTIVE =     'directive';
+  REPEATABLE =    'repeatable';
+  LCURLY =        '{';
+  RCURLY =        '}';
+  LPAREN =        '(';
+  RPAREN =        ')';
+  LBRACKET =      '[';
+  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_DIGIT = [0-9A-Za-z];
+  FOUR_DIGIT_UNICODE = UNICODE_DIGIT{4};
+  N_DIGIT_UNICODE = LCURLY UNICODE_DIGIT{4,} RCURLY;
+  UNICODE_ESCAPE = '\\u' (FOUR_DIGIT_UNICODE | N_DIGIT_UNICODE);
+  # 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 | ^'"');
+  STRING_CHAR =   ((ESCAPED_QUOTE | ^QUOTE) - BACKSLASH) | UNICODE_ESCAPE | STRING_ESCAPE;
   VAR_SIGN =      '$';
   DIR_SIGN =      '@';
   ELLIPSIS =      '...';
   EQUALS =        '=';
   BANG =          '!';
+  PIPE =          '|';
+  AMP =           '&';
 
   QUOTED_STRING = QUOTE STRING_CHAR* QUOTE;
-
+  BLOCK_STRING = BLOCK_QUOTE BLOCK_STRING_CHAR* QUOTE{0,2} BLOCK_QUOTE;
   # 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_token.call(:INT) };
-    FLOAT         => { emit_token.call(:FLOAT) };
-    ON            => { emit_token.call(:ON) };
-    FRAGMENT      => { emit_token.call(:FRAGMENT) };
-    TRUE          => { emit_token.call(:TRUE) };
-    FALSE         => { emit_token.call(:FALSE) };
-    NULL          => { emit_token.call(:NULL) };
-    QUERY         => { emit_token.call(:QUERY) };
-    MUTATION      => { emit_token.call(:MUTATION) };
-    SUBSCRIPTION  => { emit_token.call(:SUBSCRIPTION) };
-    RCURLY        => { emit_token.call(:RCURLY) };
-    LCURLY        => { emit_token.call(:LCURLY) };
-    RPAREN        => { emit_token.call(:RPAREN) };
-    LPAREN        => { emit_token.call(:LPAREN) };
-    RBRACKET      => { emit_token.call(:RBRACKET) };
-    LBRACKET      => { emit_token.call(:LBRACKET) };
-    COLON         => { emit_token.call(:COLON) };
-    QUOTED_STRING => { emit_string(ts + 1, te - 1, meta) };
-    VAR_SIGN      => { emit_token.call(:VAR_SIGN) };
-    DIR_SIGN      => { emit_token.call(:DIR_SIGN) };
-    ELLIPSIS      => { emit_token.call(:ELLIPSIS) };
-    EQUALS        => { emit_token.call(:EQUALS) };
-    BANG          => { emit_token.call(:BANG) };
-    IDENTIFIER    => { emit_token.call(:IDENTIFIER) };
+    INT           => { emit(:INT, ts, te, meta) };
+    FLOAT         => { emit(:FLOAT, ts, te, meta) };
+    ON            => { emit(:ON, ts, te, meta, "on") };
+    FRAGMENT      => { emit(:FRAGMENT, ts, te, meta, "fragment") };
+    TRUE          => { emit(:TRUE, ts, te, meta, "true") };
+    FALSE         => { emit(:FALSE, ts, te, meta, "false") };
+    NULL          => { emit(:NULL, ts, te, meta, "null") };
+    QUERY         => { emit(:QUERY, ts, te, meta, "query") };
+    MUTATION      => { emit(:MUTATION, ts, te, meta, "mutation") };
+    SUBSCRIPTION  => { emit(:SUBSCRIPTION, ts, te, meta, "subscription") };
+    SCHEMA        => { emit(:SCHEMA, ts, te, meta) };
+    SCALAR        => { emit(:SCALAR, ts, te, meta) };
+    TYPE          => { emit(:TYPE, ts, te, meta) };
+    EXTEND        => { emit(:EXTEND, ts, te, meta) };
+    IMPLEMENTS    => { emit(:IMPLEMENTS, ts, te, meta) };
+    INTERFACE     => { emit(:INTERFACE, ts, te, meta) };
+    UNION         => { emit(:UNION, ts, te, meta) };
+    ENUM          => { emit(:ENUM, ts, te, meta) };
+    INPUT         => { emit(:INPUT, ts, te, meta) };
+    DIRECTIVE     => { emit(:DIRECTIVE, ts, te, meta) };
+    REPEATABLE    => { emit(:REPEATABLE, ts, te, meta, "repeatable") };
+    RCURLY        => { emit(:RCURLY, ts, te, meta, "}") };
+    LCURLY        => { emit(:LCURLY, ts, te, meta, "{") };
+    RPAREN        => { emit(:RPAREN, ts, te, meta, ")") };
+    LPAREN        => { emit(:LPAREN, ts, te, meta, "(")};
+    RBRACKET      => { emit(:RBRACKET, ts, te, meta, "]") };
+    LBRACKET      => { emit(:LBRACKET, ts, te, meta, "[") };
+    COLON         => { emit(:COLON, ts, te, meta, ":") };
+    QUOTED_STRING => { emit_string(ts, te, meta, block: false) };
+    BLOCK_STRING  => { emit_string(ts, te, meta, block: true) };
+    VAR_SIGN      => { emit(:VAR_SIGN, ts, te, meta, "$") };
+    DIR_SIGN      => { emit(:DIR_SIGN, ts, te, meta, "@") };
+    ELLIPSIS      => { emit(:ELLIPSIS, ts, te, meta, "...") };
+    EQUALS        => { emit(:EQUALS, ts, te, meta, "=") };
+    BANG          => { emit(:BANG, ts, te, meta, "!") };
+    PIPE          => { emit(:PIPE, ts, te, meta, "|") };
+    AMP           => { emit(:AMP, ts, te, meta, "&") };
+    IDENTIFIER    => { emit(:IDENTIFIER, ts, te, meta) };
+    COMMENT       => { record_comment(ts, te, meta) };
 
     NEWLINE => {
       meta[:line] += 1
@@ -70,13 +115,13 @@
     };
 
     BLANK   => { meta[:col] += te - ts };
-    COMMENT => { meta[:col] += te - ts };
 
-    UNKNOWN_CHAR => { emit_token.call(:UNKNOWN_CHAR) };
+    UNKNOWN_CHAR => { emit(:UNKNOWN_CHAR, ts, te, meta) };
 
   *|;
 }%%
 
+# frozen_string_literal: true
 
 module GraphQL
   module Language
@@ -89,7 +134,25 @@ module GraphQL
       # To avoid allocating more strings, this modifies the string passed into it
       def self.replace_escaped_characters_in_place(raw_string)
         raw_string.gsub!(ESCAPES, ESCAPES_REPLACE)
-        raw_string.gsub!(UTF_8, &UTF_8_REPLACE)
+        raw_string.gsub!(UTF_8) do |_matched_str|
+          codepoint_1 = ($1 || $2).to_i(16)
+          codepoint_2 = $3
+
+          if codepoint_2
+            codepoint_2 = codepoint_2.to_i(16)
+            if (codepoint_1 >= 0xD800 && codepoint_1 <= 0xDBFF) && # leading surrogate
+                (codepoint_2 >= 0xDC00 && codepoint_2 <= 0xDFFF) # trailing surrogate
+              # A surrogate pair
+              combined = ((codepoint_1 - 0xD800) * 0x400) + (codepoint_2 - 0xDC00) + 0x10000
+              [combined].pack('U'.freeze)
+            else
+              # Two separate code points
+              [codepoint_1].pack('U'.freeze) + [codepoint_2].pack('U'.freeze)
+            end
+          else
+            [codepoint_1].pack('U'.freeze)
+          end
+        end
         nil
       end
 
@@ -98,34 +161,53 @@ module GraphQL
       %% write data;
 
       def self.run_lexer(query_string)
-        data = query_string.unpack("c*")
+        data = query_string.unpack(PACK_DIRECTIVE)
         eof = data.length
 
+        # Since `Lexer` is a module, store all lexer state
+        # in this local variable:
         meta = {
           line: 1,
           col: 1,
           data: data,
-          tokens: []
+          tokens: [],
+          previous_token: nil,
         }
 
-        %% write init;
+        p ||= 0
+        pe ||= data.length
 
-        emit_token = -> (name) {
-          emit(name, ts, te, meta)
-        }
+        %% write init;
 
         %% write exec;
 
         meta[:tokens]
       end
 
-      def self.emit(token_name, ts, te, meta)
-        meta[:tokens] << GraphQL::Language::Token.new(
-          name: token_name,
-          value: meta[:data][ts...te].pack("c*"),
-          line: meta[:line],
-          col: meta[:col],
+      def self.record_comment(ts, te, meta)
+        token = GraphQL::Language::Token.new(
+          :COMMENT,
+          meta[:data][ts, te - ts].pack(PACK_DIRECTIVE).force_encoding(UTF_8_ENCODING),
+          meta[:line],
+          meta[:col],
+          meta[:previous_token],
         )
+
+        meta[:previous_token] = token
+
+        meta[:col] += te - ts
+      end
+
+      def self.emit(token_name, ts, te, meta, token_value = nil)
+        token_value ||= meta[:data][ts, te - ts].pack(PACK_DIRECTIVE).force_encoding(UTF_8_ENCODING)
+        meta[:tokens] << token = GraphQL::Language::Token.new(
+          token_name,
+          token_value,
+          meta[:line],
+          meta[:col],
+          meta[:previous_token],
+        )
+        meta[:previous_token] = token
         # Bump the column counter for the next token
         meta[:col] += te - ts
       end
@@ -142,30 +224,56 @@ module GraphQL
         "\\t" => "\t",
       }
 
-      UTF_8 = /\\u[\dAa-f]{4}/i
-      UTF_8_REPLACE = -> (m) { [m[-4..-1].to_i(16)].pack('U'.freeze) }
-
-      def self.emit_string(ts, te, meta)
-        value = meta[:data][ts...te].pack("c*").force_encoding("UTF-8")
-        if value =~ /\\u|\\./ && value !~ ESCAPES
-          meta[:tokens] << GraphQL::Language::Token.new(
-            name: :BAD_UNICODE_ESCAPE,
-            value: value,
-            line: meta[:line],
-            col: meta[:col],
+      UTF_8 = /\\u(?:([\dAa-f]{4})|\{([\da-f]{4,})\})(?:\\u([\dAa-f]{4}))?/i
+
+
+      VALID_STRING = /\A(?:[^\\]|#{ESCAPES}|#{UTF_8})*\z/o
+
+      PACK_DIRECTIVE = "c*"
+      UTF_8_ENCODING = "UTF-8"
+
+      def self.emit_string(ts, te, meta, block:)
+        quotes_length = block ? 3 : 1
+        value = meta[:data][ts + quotes_length, te - ts - 2 * quotes_length].pack(PACK_DIRECTIVE).force_encoding(UTF_8_ENCODING) || ''
+        line_incr = 0
+        if block && !value.empty?
+          line_incr = value.count("\n")
+          value = GraphQL::Language::BlockString.trim_whitespace(value)
+        end
+
+        if !value.valid_encoding? || !value.match?(VALID_STRING)
+          meta[:tokens] << token = GraphQL::Language::Token.new(
+            :BAD_UNICODE_ESCAPE,
+            value,
+            meta[:line],
+            meta[:col],
+            meta[:previous_token],
           )
         else
           replace_escaped_characters_in_place(value)
 
-          meta[:tokens] << GraphQL::Language::Token.new(
-            name: :STRING,
-            value: value,
-            line: meta[:line],
-            col: meta[:col],
-          )
+          if !value.valid_encoding?
+            meta[:tokens] << token = GraphQL::Language::Token.new(
+              :BAD_UNICODE_ESCAPE,
+              value,
+              meta[:line],
+              meta[:col],
+              meta[:previous_token],
+            )
+          else
+            meta[:tokens] << token = GraphQL::Language::Token.new(
+              :STRING,
+              value,
+              meta[:line],
+              meta[:col],
+              meta[:previous_token],
+            )
+          end
         end
 
+        meta[:previous_token] = token
         meta[:col] += te - ts
+        meta[:line] += line_incr
       end
     end
   end