Repository navigation
Expand file tree
/
Copy pathdocstring.rs
More file actions
234 lines (219 loc) · 8.15 KB
/
Copy pathdocstring.rs
File metadata and controls
234 lines (219 loc) · 8.15 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
//! getPrecedingDocstring / cleanCommentMarkers — faithful port of
//! src/extraction/tree-sitter-helpers.ts (#780 wrapper-climb semantics).
use regex::Regex;
use std::sync::OnceLock;
use tree_sitter::Node;
/// DOCSTRING_WRAPPER_TYPES (tree-sitter-helpers.ts).
fn is_wrapper(kind: &str) -> bool {
matches!(
kind,
"export_statement"
| "decorated_definition"
| "lexical_declaration"
| "variable_declaration"
| "variable_declarator"
| "ambient_declaration"
)
}
fn is_comment(kind: &str) -> bool {
matches!(
kind,
"comment" | "line_comment" | "block_comment" | "documentation_comment"
)
}
struct Cleaners {
block_open: Regex,
block_close: Regex,
lua_open: Regex,
lua_close: Regex,
paren_star_open: Regex,
paren_star_close: Regex,
brace_open: Regex,
brace_close: Regex,
slashes: Regex,
dashes: Regex,
hash: Regex,
percent: Regex,
star_cont: Regex,
}
fn cleaners() -> &'static Cleaners {
static C: OnceLock<Cleaners> = OnceLock::new();
C.get_or_init(|| Cleaners {
block_open: Regex::new(r"^/\*+!?").unwrap(),
block_close: Regex::new(r"\*+/$").unwrap(),
lua_open: Regex::new(r"^--\[=*\[").unwrap(),
lua_close: Regex::new(r"\]=*\]$").unwrap(),
paren_star_open: Regex::new(r"^\(\*").unwrap(),
paren_star_close: Regex::new(r"\*\)$").unwrap(),
brace_open: Regex::new(r"^\{").unwrap(),
brace_close: Regex::new(r"\}$").unwrap(),
slashes: Regex::new(r"\A//[/!]?\s?").unwrap(),
dashes: Regex::new(r"\A--\s?").unwrap(),
hash: Regex::new(r"\A#\s?").unwrap(),
percent: Regex::new(r"\A%+\s?").unwrap(),
star_cont: Regex::new(r"\A\s*\*\s?").unwrap(),
})
}
/// JS multiline `^` anchors after \n, \r, U+2028, U+2029; the regex crate's
/// `(?m)^` anchors after `\n` only. On CRLF content the JS engine finds a line
/// start after the `\r`, so a greedy leading `\s*` (the block-continuation
/// rule) consumes the `\n` and leaves the bare `\r` in the docstring —
/// byte-parity on CRLF checkouts (every Windows autocrlf clone) depends on
/// reproducing exactly that.
fn is_js_line_terminator(ch: char) -> bool {
matches!(ch, '\n' | '\r' | '\u{2028}' | '\u{2029}')
}
/// JS-semantics `str.replace(/^<pat>/gm, "")`: try the \A-anchored `pat` at
/// position 0 and after every JS line terminator, left to right, resuming
/// after each match's end — a faithful /g replace. (Remaining known
/// divergence: JS `\s` includes U+FEFF, Rust's does not; an embedded BOM
/// inside a comment is accepted as unreachable.)
fn js_multiline_strip(s: &str, pat: &Regex) -> String {
let mut out = String::with_capacity(s.len());
let mut last = 0usize;
let mut pos = 0usize;
while pos <= s.len() {
let at_line_start = pos == 0
|| s[..pos].chars().next_back().is_some_and(is_js_line_terminator);
if at_line_start {
if let Some(m) = pat.find(&s[pos..]) {
if !m.is_empty() {
out.push_str(&s[last..pos]);
last = pos + m.end();
pos = last;
continue;
}
}
}
match s[pos..].chars().next() {
Some(c) => pos += c.len_utf8(),
None => break,
}
}
out.push_str(&s[last..]);
out
}
/// cleanCommentMarkers — strip comment syntax, keep the prose.
pub fn clean_comment_markers(comment: &str) -> String {
let c = cleaners();
let mut s = comment.trim().to_string();
if s.starts_with("/*") {
s = c.block_open.replace(&s, "").into_owned();
s = c.block_close.replace(&s, "").into_owned();
} else if s.starts_with("--[") {
s = c.lua_open.replace(&s, "").into_owned();
s = c.lua_close.replace(&s, "").into_owned();
} else if s.starts_with("(*") {
s = c.paren_star_open.replace(&s, "").into_owned();
s = c.paren_star_close.replace(&s, "").into_owned();
} else if s.starts_with('{') {
s = c.brace_open.replace(&s, "").into_owned();
s = c.brace_close.replace(&s, "").into_owned();
}
s = js_multiline_strip(&s, &c.slashes);
s = js_multiline_strip(&s, &c.dashes);
s = js_multiline_strip(&s, &c.hash);
s = js_multiline_strip(&s, &c.percent);
s = js_multiline_strip(&s, &c.star_cont);
s.trim().to_string()
}
/// getPrecedingDocstring — collect the comment run immediately preceding the
/// node (climbing out of declaration wrappers first), cleaned and joined.
/// Returns None when there is no preceding comment (a PRESENT-but-empty
/// docstring after cleaning still returns Some(""), matching the TS helper).
pub fn preceding_docstring(node: Node, src: &str) -> Option<String> {
preceding_docstring_stepping_over(node, src, &[])
}
/// getPrecedingDocstring's `stepOver` — sibling kinds that may stand between
/// the declaration and its comments without ending the run (Dart's
/// `annotation`: `/// Builds it.` `@override` `Widget build(…)`). They are not
/// part of the docstring; the comments on either side of one still join.
pub fn preceding_docstring_stepping_over(
node: Node,
src: &str,
step_over: &[&str],
) -> Option<String> {
let mut anchor = node;
while let Some(parent) = anchor.parent() {
if is_wrapper(parent.kind()) {
anchor = parent;
} else {
break;
}
}
let mut comments: Vec<&str> = Vec::new();
let mut sibling = anchor.prev_named_sibling();
while let Some(s) = sibling {
if is_comment(s.kind()) {
comments.push(&src[s.byte_range()]);
sibling = s.prev_named_sibling();
} else if step_over.contains(&s.kind()) {
sibling = s.prev_named_sibling();
} else {
break;
}
}
if comments.is_empty() {
return None;
}
comments.reverse(); // collected nearest-first; TS unshifts to keep source order
Some(
comments
.iter()
.map(|c| clean_comment_markers(c))
.collect::<Vec<_>>()
.join("\n")
.trim()
.to_string(),
)
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn strips_line_and_block_markers() {
assert_eq!(clean_comment_markers("// hello"), "hello");
assert_eq!(clean_comment_markers("/// doc line"), "doc line");
assert_eq!(
clean_comment_markers("/**\n * Adds things.\n * @param a first\n */"),
"Adds things.\n@param a first"
);
}
/// CRLF parity with the JS reference: multiline `^` matches after `\r`,
/// so the block-continuation `\s*` eats the `\n` and the bare `\r`
/// survives in the cleaned docstring (pinned against the wasm extractor
/// on a CRLF checkout — the Windows autocrlf shape).
#[test]
fn crlf_matches_js_reference() {
assert_eq!(
clean_comment_markers("/**\r\n * Class docs.\r\n * Multi-line.\r\n */"),
"Class docs.\rMulti-line."
);
assert_eq!(
clean_comment_markers("// a\r\n// b"),
"a\r\nb"
);
}
/// A `step_over` kind is passed without ending the run, and the comments
/// on either side of it join; any other node still ends it. With no
/// step-over kinds the walk stops at the first non-comment, as before.
#[test]
fn steps_over_only_the_listed_kinds() {
let grammar = crate::langs::grammar_for("dart").expect("dart grammar");
let mut parser = tree_sitter::Parser::new();
parser.set_language(&grammar).unwrap();
let src = "void g() {}\n/// a\n@x\n// b\nvoid f() {}\n";
let tree = parser.parse(src, None).unwrap();
let root = tree.root_node();
let f = (0..root.named_child_count())
.filter_map(|i| root.named_child(i))
.filter(|n| n.kind() == "function_signature")
.nth(1)
.expect("f's signature");
assert_eq!(preceding_docstring(f, src).as_deref(), Some("b"));
assert_eq!(
preceding_docstring_stepping_over(f, src, &["annotation"]).as_deref(),
Some("a\nb")
);
}
}