Merge pull request #971 from tree-sitter/fix-query-analsyis-with-extras

Fix handling of extra nodes in query analysis
This commit is contained in:
Max Brunsfeld 2021-03-08 09:54:02 -08:00 committed by GitHub
commit 0b4794106d
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
2 changed files with 83 additions and 5 deletions

View file

@ -118,6 +118,18 @@ fn test_query_errors_on_invalid_syntax() {
]
.join("\n")
);
// tree-sitter/tree-sitter/issues/968
assert_eq!(
Query::new(get_language("c"), r#"(parameter_list [ ")" @foo)"#)
.unwrap_err()
.message,
[
r#"(parameter_list [ ")" @foo)"#,
r#" ^"#
]
.join("\n")
);
});
}
@ -642,6 +654,49 @@ fn test_query_matches_capturing_error_nodes() {
});
}
#[test]
fn test_query_matches_with_extra_children() {
allocations::record(|| {
let language = get_language("ruby");
let query = Query::new(
language,
"
(program(comment) @top_level_comment)
(argument_list (heredoc_body) @heredoc_in_args)
",
)
.unwrap();
assert_query_matches(
language,
&query,
"
# top-level
puts(
# not-top-level
<<-IN_ARGS, bar.baz
HELLO
IN_ARGS
)
puts <<-NOT_IN_ARGS
NO
NOT_IN_ARGS
",
&[
(0, vec![("top_level_comment", "# top-level")]),
(
1,
vec![(
"heredoc_in_args",
"\n HELLO\n IN_ARGS",
)],
),
],
);
});
}
#[test]
fn test_query_matches_with_named_wildcard() {
allocations::record(|| {
@ -3077,6 +3132,20 @@ fn test_query_step_is_definite() {
"#,
results_by_substring: &[("name:", true)],
},
Row {
description: "top-level non-terminal extra nodes",
language: get_language("ruby"),
pattern: r#"
(heredoc_body
(interpolation)
(heredoc_end) @end)
"#,
results_by_substring: &[
("(heredoc_body", false),
("(interpolation)", false),
("(heredoc_end)", true),
],
},
];
allocations::record(|| {

View file

@ -784,8 +784,10 @@ static bool ts_query__analyze_patterns(TSQuery *self, unsigned *error_offset) {
state_predecessor_map_add(&predecessor_map, next_state, state);
}
}
} else if (lookahead_iterator.next_state != 0 && lookahead_iterator.next_state != state) {
state_predecessor_map_add(&predecessor_map, lookahead_iterator.next_state, state);
} else if (lookahead_iterator.next_state != 0) {
if (lookahead_iterator.next_state != state) {
state_predecessor_map_add(&predecessor_map, lookahead_iterator.next_state, state);
}
const TSSymbol *aliases, *aliases_end;
ts_language_aliases_for_symbol(
self->language,
@ -959,6 +961,10 @@ static bool ts_query__analyze_patterns(TSQuery *self, unsigned *error_offset) {
}
#endif
// If no further progress can be made within the current recursion depth limit, then
// bump the depth limit by one, and continue to process the states the exceeded the
// limit. But only allow this if progress has been made since the last time the depth
// limit was increased.
if (states.size == 0) {
if (deeper_states.size > 0 && final_step_indices.size > prev_final_step_count) {
#ifdef DEBUG_ANALYZE_QUERY
@ -1019,12 +1025,12 @@ static bool ts_query__analyze_patterns(TSQuery *self, unsigned *error_offset) {
TSStateId next_parse_state;
if (lookahead_iterator.action_count) {
const TSParseAction *action = &lookahead_iterator.actions[lookahead_iterator.action_count - 1];
if (action->type == TSParseActionTypeShift && !action->shift.extra) {
next_parse_state = action->shift.state;
if (action->type == TSParseActionTypeShift) {
next_parse_state = action->shift.extra ? parse_state : action->shift.state;
} else {
continue;
}
} else if (lookahead_iterator.next_state != 0 && lookahead_iterator.next_state != parse_state) {
} else if (lookahead_iterator.next_state != 0) {
next_parse_state = lookahead_iterator.next_state;
} else {
continue;
@ -1127,6 +1133,8 @@ static bool ts_query__analyze_patterns(TSQuery *self, unsigned *error_offset) {
next_step->depth <= parent_depth + 1
) break;
}
} else if (next_parse_state == parse_state) {
continue;
}
for (;;) {
@ -1535,6 +1543,7 @@ static TSQueryError ts_query__parse_pattern(
stream_advance(stream);
break;
} else if (e) {
if (e == PARENT_DONE) e = TSQueryErrorSyntax;
array_delete(&branch_step_indices);
return e;
}