summaryrefslogtreecommitdiff
path: root/ruby_parser.c
diff options
context:
space:
mode:
authorHASUMI Hitoshi <[email protected]>2024-02-16 17:45:22 +0900
committerYuichiro Kaneko <[email protected]>2024-03-12 17:17:52 +0900
commit9a19cfd4cd1a16528cc997e3a510c3046b83cdec (patch)
tree1d061cebf79d8aee39be26887b539f12b48a6f9f /ruby_parser.c
parentf42164e03700469a7000b4f00148a8ca01d75044 (diff)
[Universal Parser] Reduce dependence on RArray in parse.y
- Introduce `rb_parser_ary_t` structure to partly eliminate RArray from parse.y - In this patch, `parser_params->tokens` and `parser_params->ast->node_buffer->tokens` are now `rb_parser_ary_t *` - Instead, `ast_node_all_tokens()` internally creates a Ruby Array object from the `rb_parser_ary_t` - Also, delete `rb_ast_tokens()` and `rb_ast_set_tokens()` in node.c - Implement `rb_parser_str_escape()` - This is a port of the `rb_str_escape()` function in string.c - `rb_parser_str_escape()` does not depend on `VALUE` (RString) - Instead, it uses `rb_parser_stirng_t *` - This function works when --dump=y option passed - Because WIP of the universal parser, similar functions like `rb_parser_tokens_free()` exist in both node.c and parse.y. Refactoring them may be needed in some way in the future - Although we considered redesigning the structure: `ast->node_buffer->tokens` into `ast->tokens`, we leave it as it is because `rb_ast_t` is an imemo. (We will address it in the future)
Diffstat (limited to 'ruby_parser.c')
-rw-r--r--ruby_parser.c25
1 files changed, 25 insertions, 0 deletions
diff --git a/ruby_parser.c b/ruby_parser.c
index 8e2371fd1d..d37dc388cd 100644
--- a/ruby_parser.c
+++ b/ruby_parser.c
@@ -461,6 +461,27 @@ str_coderange_scan_restartable(const char *s, const char *e, void *enc, int *cr)
return rb_str_coderange_scan_restartable(s, e, (rb_encoding *)enc, cr);
}
+static int
+enc_mbminlen(void *enc)
+{
+ return rb_enc_mbminlen((rb_encoding *)enc);
+}
+
+static bool
+enc_isascii(OnigCodePoint c, void *enc)
+{
+ return rb_enc_isascii(c, (rb_encoding *)enc);
+}
+
+static OnigCodePoint
+enc_mbc_to_codepoint(const char *p, const char *e, void *enc)
+{
+ const OnigUChar *up = RBIMPL_CAST((const OnigUChar *)p);
+ const OnigUChar *ue = RBIMPL_CAST((const OnigUChar *)e);
+
+ return ONIGENC_MBC_TO_CODE((rb_encoding *)enc, up, ue);
+}
+
VALUE rb_io_gets_internal(VALUE io);
extern VALUE rb_eArgError;
extern VALUE rb_mRubyVMFrozenCore;
@@ -596,6 +617,10 @@ static const rb_parser_config_t rb_global_parser_config = {
.encoding_set = encoding_set,
.encoding_is_ascii8bit = encoding_is_ascii8bit,
.usascii_encoding = usascii_encoding,
+ .enc_coderange_broken = ENC_CODERANGE_BROKEN,
+ .enc_mbminlen = enc_mbminlen,
+ .enc_isascii = enc_isascii,
+ .enc_mbc_to_codepoint = enc_mbc_to_codepoint,
.ractor_make_shareable = rb_ractor_make_shareable,