Skip to content

Commit 23a68dc

Browse files
haldunhaldun
committed
Faster lex_keyword
1 parent a535949 commit 23a68dc

File tree

1 file changed

+50
-47
lines changed

1 file changed

+50
-47
lines changed

‎src/prism.c

Lines changed: 50 additions & 47 deletions
Original file line numberDiff line numberDiff line change
@@ -6725,17 +6725,21 @@ lex_global_variable(pm_parser_t *parser) {
67256725

67266726
/**
67276727
* This function checks if the current token matches a keyword. If it does, it
6728-
* returns true. Otherwise, it returns false. The arguments are as follows:
6728+
* returns the token type. Otherwise, it returns PM_TOKEN_EOF. The arguments are as follows:
67296729
*
6730+
* * `parser` - the parser object
6731+
* * `current_start` - pointer to the start of the current token
67306732
* * `value` - the literal string that we're checking for
6731-
* * `width` - the length of the token
6733+
* * `vlen` - the length of the token
67326734
* * `state` - the state that we should transition to if the token matches
6735+
* * `type` - the expected token type
6736+
* * `modifier_type` - the expected modifier token type
67336737
*/
67346738
static inline pm_token_type_t
6735-
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) {
6736-
pm_lex_state_t last_state = parser->lex_state;
6739+
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) {
6740+
if (memcmp(current_start, value, vlen) == 0) {
6741+
pm_lex_state_t last_state = parser->lex_state;
67376742

6738-
if (parser->current.start + vlen <= parser->end && memcmp(parser->current.start, value, vlen) == 0) {
67396743
if (parser->lex_state & PM_LEX_STATE_FNAME) {
67406744
lex_state_set(parser, PM_LEX_STATE_ENDFN);
67416745
} else {
@@ -6791,7 +6795,7 @@ lex_identifier(pm_parser_t *parser, bool previous_command_start) {
67916795
}
67926796

67936797
if (parser->lex_state != PM_LEX_STATE_DOT) {
6794-
if (width == 8 && (lex_keyword(parser, "defined?", width, PM_LEX_STATE_ARG, PM_TOKEN_KEYWORD_DEFINED, PM_TOKEN_EOF) != PM_TOKEN_EOF)) {
6798+
if (width == 8 && (lex_keyword(parser, current_start, "defined?", width, PM_LEX_STATE_ARG, PM_TOKEN_KEYWORD_DEFINED, PM_TOKEN_EOF) != PM_TOKEN_EOF)) {
67956799
return PM_TOKEN_KEYWORD_DEFINED;
67966800
}
67976801
}
@@ -6819,67 +6823,66 @@ lex_identifier(pm_parser_t *parser, bool previous_command_start) {
68196823

68206824
if (parser->lex_state != PM_LEX_STATE_DOT) {
68216825
pm_token_type_t type;
6822-
68236826
switch (width) {
68246827
case 2:
6825-
if (lex_keyword(parser, "do", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_DO, PM_TOKEN_EOF) != PM_TOKEN_EOF) {
6828+
if (lex_keyword(parser, current_start, "do", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_DO, PM_TOKEN_EOF) != PM_TOKEN_EOF) {
68266829
if (pm_do_loop_stack_p(parser)) {
68276830
return PM_TOKEN_KEYWORD_DO_LOOP;
68286831
}
68296832
return PM_TOKEN_KEYWORD_DO;
68306833
}
68316834

6832-
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;
6833-
if ((type = lex_keyword(parser, "in", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_IN, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6834-
if ((type = lex_keyword(parser, "or", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_OR, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6835+
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;
6836+
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;
6837+
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;
68356838
break;
68366839
case 3:
6837-
if ((type = lex_keyword(parser, "and", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_AND, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6838-
if ((type = lex_keyword(parser, "def", width, PM_LEX_STATE_FNAME, PM_TOKEN_KEYWORD_DEF, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6839-
if ((type = lex_keyword(parser, "end", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_END, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6840-
if ((type = lex_keyword(parser, "END", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_END_UPCASE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6841-
if ((type = lex_keyword(parser, "for", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_FOR, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6842-
if ((type = lex_keyword(parser, "nil", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_NIL, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6843-
if ((type = lex_keyword(parser, "not", width, PM_LEX_STATE_ARG, PM_TOKEN_KEYWORD_NOT, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6840+
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;
6841+
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;
6842+
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;
6843+
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;
6844+
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;
6845+
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;
6846+
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;
68446847
break;
68456848
case 4:
6846-
if ((type = lex_keyword(parser, "case", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_CASE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6847-
if ((type = lex_keyword(parser, "else", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_ELSE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6848-
if ((type = lex_keyword(parser, "next", width, PM_LEX_STATE_MID, PM_TOKEN_KEYWORD_NEXT, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6849-
if ((type = lex_keyword(parser, "redo", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_REDO, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6850-
if ((type = lex_keyword(parser, "self", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_SELF, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6851-
if ((type = lex_keyword(parser, "then", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_THEN, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6852-
if ((type = lex_keyword(parser, "true", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_TRUE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6853-
if ((type = lex_keyword(parser, "when", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_WHEN, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6849+
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;
6850+
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;
6851+
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;
6852+
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;
6853+
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;
6854+
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;
6855+
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;
6856+
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;
68546857
break;
68556858
case 5:
6856-
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;
6857-
if ((type = lex_keyword(parser, "begin", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_BEGIN, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6858-
if ((type = lex_keyword(parser, "BEGIN", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_BEGIN_UPCASE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6859-
if ((type = lex_keyword(parser, "break", width, PM_LEX_STATE_MID, PM_TOKEN_KEYWORD_BREAK, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6860-
if ((type = lex_keyword(parser, "class", width, PM_LEX_STATE_CLASS, PM_TOKEN_KEYWORD_CLASS, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6861-
if ((type = lex_keyword(parser, "elsif", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_ELSIF, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6862-
if ((type = lex_keyword(parser, "false", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_FALSE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6863-
if ((type = lex_keyword(parser, "retry", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD_RETRY, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6864-
if ((type = lex_keyword(parser, "super", width, PM_LEX_STATE_ARG, PM_TOKEN_KEYWORD_SUPER, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6865-
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;
6866-
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;
6867-
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;
6868-
if ((type = lex_keyword(parser, "yield", width, PM_LEX_STATE_ARG, PM_TOKEN_KEYWORD_YIELD, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6859+
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;
6860+
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;
6861+
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;
6862+
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;
6863+
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;
6864+
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;
6865+
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;
6866+
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;
6867+
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;
6868+
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;
6869+
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;
6870+
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;
6871+
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;
68696872
break;
68706873
case 6:
6871-
if ((type = lex_keyword(parser, "ensure", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_ENSURE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6872-
if ((type = lex_keyword(parser, "module", width, PM_LEX_STATE_BEG, PM_TOKEN_KEYWORD_MODULE, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6873-
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;
6874-
if ((type = lex_keyword(parser, "return", width, PM_LEX_STATE_MID, PM_TOKEN_KEYWORD_RETURN, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6875-
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;
6874+
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;
6875+
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;
6876+
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;
6877+
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;
6878+
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;
68766879
break;
68776880
case 8:
6878-
if ((type = lex_keyword(parser, "__LINE__", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD___LINE__, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6879-
if ((type = lex_keyword(parser, "__FILE__", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD___FILE__, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6881+
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;
6882+
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;
68806883
break;
68816884
case 12:
6882-
if ((type = lex_keyword(parser, "__ENCODING__", width, PM_LEX_STATE_END, PM_TOKEN_KEYWORD___ENCODING__, PM_TOKEN_EOF)) != PM_TOKEN_EOF) return type;
6885+
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;
68836886
break;
68846887
}
68856888
}

0 commit comments

Comments
 (0)