🦠 The Definitive Gemini Protocol Toolkit
gemini
gemini-protocol
gemtext
parser
zero-dependency
toolkit
ast
converter
html
markdown
cli
networking
fix(ast): Improve UTF-8 handling
+48
-5
+8
-4
src/ast/container.rs
+8
-4
src/ast/container.rs
···
224
224
225
225
nodes.push(Node::Heading {
226
226
level,
227
-
// Here, we are `get`ing the `&str` starting at the `level`-th
228
-
// index, then trimming the start. These operations
229
-
// effectively off the line identifier.
230
-
text: line.get(level..).unwrap_or("").trim_start().to_string(),
227
+
// Here, the text after the heading markers is safely extracted.
228
+
// `chars().skip()` is used to safely handle UTF-8 boundaries.
229
+
text: line
230
+
.chars()
231
+
.skip(level)
232
+
.collect::<String>()
233
+
.trim_start()
234
+
.to_string(),
231
235
});
232
236
233
237
break;
+40
-1
tests/ast.rs
+40
-1
tests/ast.rs
···
95
95
assert_eq!(to, "");
96
96
assert_eq!(text, &None);
97
97
} else {
98
-
panic!("Expected Link node");
98
+
panic!("Expected link node");
99
+
}
100
+
}
101
+
102
+
#[test]
103
+
fn build_heading_with_unicode_and_edge_cases() {
104
+
// Unicode characters
105
+
let ast = Ast::from_string("# Hello, 世界!");
106
+
107
+
assert_eq!(ast.inner().len(), 1);
108
+
109
+
if let Node::Heading { level, text } = ast.inner().first().unwrap() {
110
+
assert_eq!(level, &1);
111
+
assert_eq!(text, "Hello, 世界!");
112
+
} else {
113
+
panic!("Expected heading node");
114
+
}
115
+
116
+
// Only hashes
117
+
let ast = Ast::from_string("###");
118
+
119
+
assert_eq!(ast.inner().len(), 1);
120
+
121
+
if let Node::Heading { level, text } = ast.inner().first().unwrap() {
122
+
assert_eq!(level, &3);
123
+
assert_eq!(text, "");
124
+
} else {
125
+
panic!("Expected heading node");
126
+
}
127
+
128
+
// Many hashes
129
+
let ast = Ast::from_string("########## Very Deep Heading");
130
+
131
+
assert_eq!(ast.inner().len(), 1);
132
+
133
+
if let Node::Heading { level, text } = ast.inner().first().unwrap() {
134
+
assert_eq!(level, &10);
135
+
assert_eq!(text, "Very Deep Heading");
136
+
} else {
137
+
panic!("Expected heading node");
99
138
}
100
139
}
101
140
}