[ruby/prism] Faster lex_keyword

https://github.com/ruby/prism/commit/23a68dcda2

1 files changed, 50 insertions, 47 deletions

diff --git a/prism/prism.c b/prism/prism.c
index f1379f10ff..2cd00b00a2 100644
--- a/prism/prism.c
+++ b/prism/prism.c
@@ -6844,17 +6844,21 @@ lex_global_variable(pm_parser_t *parser) {
 
 /**
  * This function checks if the current token matches a keyword. If it does, it
- * returns true. Otherwise, it returns false. The arguments are as follows:
+ * returns the token type. Otherwise, it returns PM_TOKEN_EOF. The arguments are as follows:
  *
+ * * `parser` - the parser object
+ * * `current_start` - pointer to the start of the current token
  * * `value` - the literal string that we're checking for
- * * `width` - the length of the token
+ * * `vlen` - the length of the token
  * * `state` - the state that we should transition to if the token matches
+ * * `type` - the expected token type
+ * * `modifier_type` - the expected modifier token type
  */
 static inline pm_token_type_t
-lex_keyword(pm_parser_t *parser, const char *value, size_t vlen, pm_lex_state_t state, pm_token_type_t type, pm_token_type_t modifier_type) {
-    pm_lex_state_t last_state = parser->lex_state;
+lex_keyword(pm_parser_t *parser, const uint8_t *current_start, const char *value, size_t vlen, pm_lex_state_t state, pm_token_type_t type, pm_token_type_t modifier_type) {
+    if (memcmp(current_start, value, vlen) == 0) {
+        pm_lex_state_t last_state = parser->lex_state;
 
-    if (parser->current.start + vlen <= parser->end && memcmp(parser->current.start, value, vlen) == 0) {
         if (parser->lex_state & PM_LEX_STATE_FNAME) {
             lex_state_set(parser, PM_LEX_STATE_ENDFN);
         } else {
@@ -6910,7 +6914,7 @@ lex_identifier(pm_parser_t *parser, bool previous_command_start) {
             }
 
             if (parser->lex_state != PM_LEX_STATE_DOT) {
-                if (width == 8 && (lex_keyword(parser, "defined?", width, PM_LEX_STATE_ARG, PM_TOKEN_KEYWORD_DEFINED, PM_TOKEN_EOF) != PM_TOKEN_EOF)) {
+                if (width == 8 && (lex_keyword(parser, current_start, "defined?", width, PM_LEX_STATE_ARG, PM_TOKEN_KEYWORD_DEFINED, PM_TOKEN_EOF) != PM_TOKEN_EOF)) {
                     return PM_TOKEN_KEYWORD_DEFINED;
                 }
             }
@@ -6938,67 +6942,66 @@ lex_identifier(pm_parser_t *parser, bool previous_command_start) {
 
     if (parser->lex_state != PM_LEX_STATE_DOT) {
         pm_token_type_t type;
-
         switch (width) {
             case 2:
-                if (lex_keyword(parser, "do", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_DO, PM_TOKEN_EOF) != PM_TOKEN_EOF) {
+                if (lex_keyword(parser, current_start, "do", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_DO, PM_TOKEN_EOF) != PM_TOKEN_EOF) {
                     if (pm_do_loop_stack_p(parser)) {
                         return PM_TOKEN_KEYWORD_DO_LOOP;
                     }
                     return PM_TOKEN_KEYWORD_DO;
                 }
 
-                if ((type = lex_keyword(parser, "if", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_IF, PM_TOKEN_KEYWORD_IF_MODIFIER)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "in", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_IN, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "or", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_OR, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "if", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_IF, PM_TOKEN_KEYWORD_IF_MODIFIER)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "in", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_IN, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "or", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_OR, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
                 break;
             case 3:
-                if ((type = lex_keyword(parser, "and", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_AND, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "def", width, PM_LEX_STATE_FNAME, PM_TOKEN_KEYWORD_DEF, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "end", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_END, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "END", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_END_UPCASE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "for", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_FOR, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "nil", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_NIL, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "not", width, PM_LEX_STATE_ARG, PM_TOKEN_KEYWORD_NOT, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "and", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_AND, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "def", width, PM_LEX_STATE_FNAME, PM_TOKEN_KEYWORD_DEF, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "end", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_END, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "END", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_END_UPCASE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "for", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_FOR, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "nil", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_NIL, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "not", width, PM_LEX_STATE_ARG, PM_TOKEN_KEYWORD_NOT, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
                 break;
             case 4:
-                if ((type = lex_keyword(parser, "case", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_CASE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "else", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_ELSE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "next", width, PM_LEX_STATE_MID, PM_TOKEN_KEYWORD_NEXT, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "redo", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_REDO, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "self", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_SELF, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "then", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_THEN, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "true", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_TRUE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "when", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_WHEN, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "case", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_CASE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "else", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_ELSE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "next", width, PM_LEX_STATE_MID, PM_TOKEN_KEYWORD_NEXT, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "redo", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_REDO, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "self", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_SELF, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "then", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_THEN, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "true", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_TRUE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "when", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_WHEN, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
                 break;
             case 5:
-                if ((type = lex_keyword(parser, "alias", width, PM_LEX_STATE_FNAME | PM_LEX_STATE_FITEM, PM_TOKEN_KEYWORD_ALIAS, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "begin", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_BEGIN, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "BEGIN", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_BEGIN_UPCASE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "break", width, PM_LEX_STATE_MID, PM_TOKEN_KEYWORD_BREAK, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "class", width, PM_LEX_STATE_CLASS, PM_TOKEN_KEYWORD_CLASS, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "elsif", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_ELSIF, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "false", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_FALSE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "retry", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_RETRY, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "super", width, PM_LEX_STATE_ARG, PM_TOKEN_KEYWORD_SUPER, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "undef", width, PM_LEX_STATE_FNAME | PM_LEX_STATE_FITEM, PM_TOKEN_KEYWORD_UNDEF, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "until", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_UNTIL, PM_TOKEN_KEYWORD_UNTIL_MODIFIER)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "while", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_WHILE, PM_TOKEN_KEYWORD_WHILE_MODIFIER)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "yield", width, PM_LEX_STATE_ARG, PM_TOKEN_KEYWORD_YIELD, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "alias", width, PM_LEX_STATE_FNAME | PM_LEX_STATE_FITEM, PM_TOKEN_KEYWORD_ALIAS, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "begin", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_BEGIN, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "BEGIN", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_BEGIN_UPCASE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "break", width, PM_LEX_STATE_MID, PM_TOKEN_KEYWORD_BREAK, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "class", width, PM_LEX_STATE_CLASS, PM_TOKEN_KEYWORD_CLASS, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "elsif", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_ELSIF, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "false", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_FALSE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "retry", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_RETRY, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "super", width, PM_LEX_STATE_ARG, PM_TOKEN_KEYWORD_SUPER, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "undef", width, PM_LEX_STATE_FNAME | PM_LEX_STATE_FITEM, PM_TOKEN_KEYWORD_UNDEF, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "until", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_UNTIL, PM_TOKEN_KEYWORD_UNTIL_MODIFIER)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "while", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_WHILE, PM_TOKEN_KEYWORD_WHILE_MODIFIER)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "yield", width, PM_LEX_STATE_ARG, PM_TOKEN_KEYWORD_YIELD, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
                 break;
             case 6:
-                if ((type = lex_keyword(parser, "ensure", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_ENSURE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "module", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_MODULE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "rescue", width, PM_LEX_STATE_MID, PM_TOKEN_KEYWORD_RESCUE, PM_TOKEN_KEYWORD_RESCUE_MODIFIER)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "return", width, PM_LEX_STATE_MID, PM_TOKEN_KEYWORD_RETURN, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "unless", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_UNLESS, PM_TOKEN_KEYWORD_UNLESS_MODIFIER)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "ensure", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_ENSURE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "module", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_MODULE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "rescue", width, PM_LEX_STATE_MID, PM_TOKEN_KEYWORD_RESCUE, PM_TOKEN_KEYWORD_RESCUE_MODIFIER)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "return", width, PM_LEX_STATE_MID, PM_TOKEN_KEYWORD_RETURN, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "unless", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_UNLESS, PM_TOKEN_KEYWORD_UNLESS_MODIFIER)) != PM_TOKEN_EOF) return type;
                 break;
             case 8:
-                if ((type = lex_keyword(parser, "__LINE__", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD___LINE__, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
-                if ((type = lex_keyword(parser, "__FILE__", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD___FILE__, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "__LINE__", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD___LINE__, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "__FILE__", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD___FILE__, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
                 break;
             case 12:
-                if ((type = lex_keyword(parser, "__ENCODING__", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD___ENCODING__, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
+                if ((type = lex_keyword(parser, current_start, "__ENCODING__", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD___ENCODING__, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
                 break;
         }
     }