1use std::path::Path;
2
3use cargo_util_terminal::report::AnnotationKind;
4use cargo_util_terminal::report::Group;
5use cargo_util_terminal::report::Level;
6use cargo_util_terminal::report::Patch;
7use cargo_util_terminal::report::Snippet;
8use toml_parser::Source;
9use toml_parser::Span;
10use toml_parser::decoder::Encoding;
11use toml_parser::parser::Event;
12use toml_parser::parser::EventKind;
13use toml_parser::parser::EventReceiver;
14use tracing::instrument;
15
16use super::CORRECTNESS;
17use crate::CargoResult;
18use crate::GlobalContext;
19use crate::diagnostics::Lint;
20use crate::diagnostics::LintLevelProduct;
21use crate::diagnostics::ManifestFor;
22use crate::diagnostics::ScopedDiagnosticStats;
23use crate::diagnostics::workspace_rel_path;
24use crate::workspace::MaybePackage;
25use crate::workspace::Workspace;
26
27pub static LINT: &Lint = &Lint {
28 name: "text_direction_codepoint_in_literal",
29 primary_group: &CORRECTNESS,
30 msrv: Some(super::CARGO_LINTS_MSRV),
31 feature_gate: None,
32 docs: Some(
33 r#"
34### What it does
35Detects Unicode codepoints in literals in manifests that change the visual representation of text on screen
36in a way that does not correspond to their on memory representation.
37
38### Why is this bad?
39Unicode allows changing the visual flow of text on screen
40in order to support scripts that are written right-to-left,
41but a specially crafted literal can make code that will be compiled appear to be part of a literal,
42depending on the software used to read the code.
43To avoid potential problems or confusion,
44such as in CVE-2021-42574,
45by default we deny their use.
46"#,
47 ),
48};
49
50#[instrument(skip_all)]
51pub(crate) fn lint_manifest(
52 ws: &Workspace<'_>,
53 manifest: ManifestFor<'_>,
54 manifest_path: &Path,
55 level: LintLevelProduct,
56 pkg_stats: &mut ScopedDiagnosticStats<'_>,
57 gctx: &GlobalContext,
58) -> CargoResult<()> {
59 let LintLevelProduct {
60 level: lint_level,
61 source,
62 } = level;
63
64 if matches!(
65 &manifest,
66 ManifestFor::Workspace {
67 maybe_pkg: MaybePackage::Package { .. },
68 ..
69 }
70 ) {
71 return Ok(());
73 }
74
75 let Some(contents) = manifest.contents() else {
76 return Ok(());
77 };
78
79 let bidi_spans = contents
80 .char_indices()
81 .filter(|(_i, c)| {
82 UNICODE_BIDI_CODEPOINTS
83 .iter()
84 .any(|(bidi, _, _name)| c == bidi)
85 })
86 .map(|(i, c)| (i, i + c.len_utf8()))
87 .collect::<Vec<_>>();
88 if bidi_spans.is_empty() {
89 return Ok(());
90 }
91
92 let toml_source = Source::new(contents);
93 let events = bidi_events(&toml_source, &bidi_spans);
94 let manifest_path = workspace_rel_path(ws, manifest_path);
95 let mut emitted_source = None;
96 for event in events {
97 let token_span = event.token.span();
98 let token_span = token_span.start()..token_span.end();
99 let mut snippet = Snippet::source(contents).path(&manifest_path).annotation(
100 AnnotationKind::Context
101 .span(token_span.clone())
102 .label("this literal contains an invisible unicode text flow control codepoint"),
103 );
104 for bidi_span in event.bidi_spans {
105 let bidi_span = bidi_span.0..bidi_span.1;
106 let escaped = format!("{:?}", &contents[bidi_span.clone()]);
107 snippet = snippet.annotation(AnnotationKind::Primary.span(bidi_span).label(escaped));
108 }
109 let mut help_snippet = Snippet::source(contents).path(&manifest_path);
110 if let Some(original_raw) = toml_source.get(&event.token) {
111 let mut decoded = String::new();
112 let replacement = match event.token.kind() {
113 toml_parser::parser::EventKind::SimpleKey => {
114 use toml_writer::ToTomlKey as _;
115 original_raw.decode_key(&mut decoded, &mut ());
116 let builder = toml_writer::TomlKeyBuilder::new(&decoded);
117 let replacement = builder.as_basic();
118 Some(replacement.to_toml_key())
119 }
120 toml_parser::parser::EventKind::Scalar => {
121 use toml_writer::ToTomlValue as _;
122 let kind = original_raw.decode_scalar(&mut decoded, &mut ());
123 if matches!(kind, toml_parser::decoder::ScalarKind::String) {
124 let builder = toml_writer::TomlStringBuilder::new(&decoded);
125 let replacement = match event.token.encoding() {
126 Some(toml_parser::decoder::Encoding::BasicString)
127 | Some(toml_parser::decoder::Encoding::LiteralString)
128 | None => builder.as_basic(),
129 Some(toml_parser::decoder::Encoding::MlBasicString)
130 | Some(toml_parser::decoder::Encoding::MlLiteralString) => {
131 builder.as_ml_basic()
132 }
133 };
134 Some(replacement.to_toml_value())
135 } else {
136 None
137 }
138 }
139 _ => None,
140 };
141 if let Some(mut replacement) = replacement {
142 for (bidi, escaped, _) in UNICODE_BIDI_CODEPOINTS {
143 replacement = replacement.replace(*bidi, escaped);
144 }
145 help_snippet = help_snippet.patch(Patch::new(token_span.clone(), replacement));
146 }
147 }
148
149 let level = lint_level.to_diagnostic_level();
150 let mut primary = Group::with_title(level.primary_title(
151 "unicode codepoint changing visible direction of text present in literal",
152 ))
153 .element(snippet);
154 if emitted_source.is_none() {
155 emitted_source = Some(LINT.emitted_source(lint_level, source));
156 primary = primary.element(Level::NOTE.message(emitted_source.as_ref().unwrap()));
157 }
158
159 let help = Group::with_title(Level::HELP.secondary_title("if you want to keep them but make them visible in your source code, you can escape them")).element(help_snippet);
160
161 let report = [primary, help];
162
163 pkg_stats.record_lint(lint_level);
164 gctx.shell().print_report(&report, lint_level.force())?;
165 }
166
167 Ok(())
168}
169
170const UNICODE_BIDI_CODEPOINTS: &[(char, &str, &str)] = &[
171 ('\u{202A}', r"\u{202A}", "LEFT-TO-RIGHT EMBEDDING"),
172 ('\u{202B}', r"\u{202B}", "RIGHT-TO-LEFT EMBEDDING"),
173 ('\u{202C}', r"\u{202C}", "POP DIRECTIONAL FORMATTING"),
174 ('\u{202D}', r"\u{202D}", "LEFT-TO-RIGHT OVERRIDE"),
175 ('\u{202E}', r"\u{202E}", "RIGHT-TO-LEFT OVERRIDE"),
176 ('\u{2066}', r"\u{2066}", "LEFT-TO-RIGHT ISOLATE"),
177 ('\u{2067}', r"\u{2067}", "RIGHT-TO-LEFT ISOLATE"),
178 ('\u{2068}', r"\u{2068}", "FIRST STRONG ISOLATE"),
179 ('\u{2069}', r"\u{2069}", "POP DIRECTIONAL ISOLATE"),
180];
181
182struct BiDiEvent {
183 token: Event,
184 bidi_spans: Vec<(usize, usize)>,
185}
186
187fn bidi_events(source: &Source<'_>, bidi_spans: &[(usize, usize)]) -> Vec<BiDiEvent> {
188 let mut bidi_spans = bidi_spans.iter();
189 let bidi_span = bidi_spans.next().copied();
190
191 let tokens = source.lex().into_vec();
192 let mut collector = BiDiCollector {
193 bidi_span,
194 bidi_spans,
195 events: Vec::new(),
196 };
197 let mut errors = ();
198 toml_parser::parser::parse_document(&tokens, &mut collector, &mut errors);
199
200 collector.events
201}
202
203struct BiDiCollector<'b> {
204 bidi_span: Option<(usize, usize)>,
205 bidi_spans: std::slice::Iter<'b, (usize, usize)>,
206 events: Vec<BiDiEvent>,
207}
208
209impl BiDiCollector<'_> {
210 fn process(&mut self, kind: EventKind, encoding: Option<Encoding>, span: Span) {
211 let mut event_bidi_spans = Vec::new();
212 while let Some(bidi_span) = self.bidi_span {
213 if bidi_span.0 < span.start() {
214 self.bidi_span = self.bidi_spans.next().copied();
215 continue;
216 } else if span.end() <= bidi_span.0 {
217 break;
218 }
219
220 event_bidi_spans.push(bidi_span);
221 self.bidi_span = self.bidi_spans.next().copied();
222 }
223
224 if !event_bidi_spans.is_empty() {
225 let token = Event::new_unchecked(kind, encoding, span);
226 self.events.push(BiDiEvent {
227 token,
228 bidi_spans: event_bidi_spans,
229 });
230 }
231 }
232}
233
234impl EventReceiver for BiDiCollector<'_> {
235 fn simple_key(
236 &mut self,
237 span: Span,
238 encoding: Option<Encoding>,
239 _error: &mut dyn toml_parser::ErrorSink,
240 ) {
241 self.process(EventKind::SimpleKey, encoding, span)
242 }
243 fn scalar(
244 &mut self,
245 span: Span,
246 encoding: Option<Encoding>,
247 _error: &mut dyn toml_parser::ErrorSink,
248 ) {
249 self.process(EventKind::Scalar, encoding, span)
250 }
251}