p_parse_inner: do not consume the follow token

This commit is contained in:
josh committed 2026-07-11 22:46:21 -04:00
1 parent 946eb4eef7
commit 2f9a29f7c5
5 files changed
+59 -4

No files matched your search

+22 -1
View File
@@ -50,13 +50,26 @@ int main()
/* parse_inner_R1("abb", [b]) succeeds: even though `b` is the lookahead
* that parse_Start uses to select R2 over R1 in the ambiguous state, from
* R1's start state the reduce to R1 is unconditional, and the follow-
* token shift retry at the R1-accepting state completes the parse. */
* token shift retry at the R1-accepting state completes the parse.
*
* The follow token that completed the parse must not be consumed from
* the input: p_position() should point to the follow token, and a
* subsequent p_lex() should return it. */
{
input = "abb";
context = p_context_new((uint8_t const *)input, strlen(input));
p_token_t follow_tokens[] = { TOKEN_b };
assert(p_parse_inner_R1(context, follow_tokens, 1u) == P_SUCCESS);
assert_eq(11u, (size_t)p_result_R1(context));
/* Follow token `b` is at column 3 (1-based). */
p_position_t pos = p_position(context);
assert_eq(1u, (size_t)pos.row);
assert_eq(3u, (size_t)pos.col);
p_token_info_t token_info;
assert(p_lex(context, &token_info) == P_SUCCESS);
assert_eq((size_t)TOKEN_b, (size_t)token_info.token);
assert_eq(1u, (size_t)token_info.position.row);
assert_eq(3u, (size_t)token_info.position.col);
p_context_delete(context);
}
@@ -69,6 +82,14 @@ int main()
p_token_t follow_tokens[] = { TOKEN_a };
assert(p_parse_inner_R1(context, follow_tokens, 1u) == P_SUCCESS);
assert_eq(11u, (size_t)p_result_R1(context));
/* Follow token `a` is at column 3 (1-based) and remains in the
* input. */
p_position_t pos = p_position(context);
assert_eq(1u, (size_t)pos.row);
assert_eq(3u, (size_t)pos.col);
p_token_info_t token_info;
assert(p_lex(context, &token_info) == P_SUCCESS);
assert_eq((size_t)TOKEN_a, (size_t)token_info.token);
p_context_delete(context);
}
+17 -1
View File
@@ -34,12 +34,23 @@ unittest
/* parse_inner_R1("abb", [b]) succeeds: `b` is the lookahead that
* parse_Start would use to select R2 over R1, but from R1's own start
* state R1 reduces unconditionally, and the follow-token shift retry at
* the R1-accepting state completes the parse. */
* the R1-accepting state completes the parse.
*
* The follow token that completed the parse must not be consumed: a
* subsequent p_lex() should return it. */
input = "abb";
context = p_context_new(input);
p_token_t[] follow_tokens_b = [TOKEN_b];
assert(p_parse_inner_R1(context, follow_tokens_b) == P_SUCCESS);
assert(p_result_R1(context) == 11);
p_position_t pos = p_position(context);
assert(pos.row == 1);
assert(pos.col == 3);
p_token_info_t token_info;
assert(p_lex(context, &token_info) == P_SUCCESS);
assert(token_info.token == TOKEN_b);
assert(token_info.position.row == 1);
assert(token_info.position.col == 3);
/* parse_inner_R1("aba", [a]) also succeeds. */
input = "aba";
@@ -47,6 +58,11 @@ unittest
p_token_t[] follow_tokens_a = [TOKEN_a];
assert(p_parse_inner_R1(context, follow_tokens_a) == P_SUCCESS);
assert(p_result_R1(context) == 11);
pos = p_position(context);
assert(pos.row == 1);
assert(pos.col == 3);
assert(p_lex(context, &token_info) == P_SUCCESS);
assert(token_info.token == TOKEN_a);
/* parse_inner_R1("ab", null) behaves like p_parse_R1("ab"). */
input = "ab";