diff --git a/parse.y b/parse.y index e23439294b..e84468d2a5 100644 --- a/parse.y +++ b/parse.y @@ -7391,6 +7391,19 @@ whole_match_p(struct parser_params *p, const char *eos, long len, int indent) return strncmp(eos, ptr, len) == 0; } +static int +word_match_p(struct parser_params *p, const char *word, long len) +{ + if (strncmp(p->lex.pcur, word, len)) return 0; + if (p->lex.pcur + len == p->lex.pend) return 1; + int c = (unsigned char)p->lex.pcur[len]; + if (ISSPACE(c)) return 1; + switch (c) { + case '\0': case '\004': case '\032': return 1; + } + return 0; +} + #define NUM_SUFFIX_R (1<<0) #define NUM_SUFFIX_I (1<<1) #define NUM_SUFFIX_ALL 3 @@ -8976,7 +8989,7 @@ parser_yylex(struct parser_params *p) case '=': if (was_bol(p)) { /* skip embedded rd document */ - if (strncmp(p->lex.pcur, "begin", 5) == 0 && ISSPACE(p->lex.pcur[5])) { + if (word_match_p(p, "begin", 5)) { int first_p = TRUE; lex_goto_eol(p); @@ -8992,11 +9005,10 @@ parser_yylex(struct parser_params *p) compile_error(p, "embedded document meets end of file"); return 0; } - if (c != '=') continue; - if (c == '=' && strncmp(p->lex.pcur, "end", 3) == 0 && - (p->lex.pcur + 3 == p->lex.pend || ISSPACE(p->lex.pcur[3]))) { + if (c == '=' && word_match_p(p, "end", 3)) { break; } + pushback(p, c); } lex_goto_eol(p); dispatch_scan_event(p, tEMBDOC_END); diff --git a/test/ruby/test_parse.rb b/test/ruby/test_parse.rb index ee5f815fd6..2622ec529c 100644 --- a/test/ruby/test_parse.rb +++ b/test/ruby/test_parse.rb @@ -709,6 +709,15 @@ x = __ENCODING__ def test_embedded_rd assert_valid_syntax("=begin\n""=end") + assert_valid_syntax("=begin\n""=end\0") + assert_valid_syntax("=begin\n""=end\C-d") + assert_valid_syntax("=begin\n""=end\C-z") + end + + def test_embedded_rd_error + error = 'embedded document meets end of file' + assert_syntax_error("=begin\n", error) + assert_syntax_error("=begin", error) end def test_float