code_kb_core/
slicer.rs

1use std::fs;
2use std::path::Path;
3use thiserror::Error;
4
5use crate::models::Symbol;
6
7#[derive(Debug, Error)]
8pub enum SliceError {
9    #[error("Failed to read file '{0}': {1}")]
10    Io(String, #[source] std::io::Error),
11    #[error("Invalid slice range: start {0} > end {1}")]
12    InvalidRange(usize, usize),
13    #[error("File '{0}' is empty")]
14    EmptyFile(String),
15}
16
17/// Snap byte offset to the nearest UTF-8 character boundary <= index.
18fn snap_to_char_boundary_floor(bytes: &[u8], mut idx: usize) -> usize {
19    if idx >= bytes.len() {
20        return bytes.len();
21    }
22    while idx > 0 && (bytes[idx] & 0xC0) == 0x80 {
23        idx -= 1;
24    }
25    idx
26}
27
28/// Snap byte offset to the nearest UTF-8 character boundary >= index.
29fn snap_to_char_boundary_ceil(bytes: &[u8], mut idx: usize) -> usize {
30    if idx >= bytes.len() {
31        return bytes.len();
32    }
33    while idx < bytes.len() && (bytes[idx] & 0xC0) == 0x80 {
34        idx += 1;
35    }
36    idx
37}
38
39/// Slices exact byte content from an in-memory buffer with safe UTF-8 snapping.
40pub fn slice_bytes_safe(
41    content: &[u8],
42    start_byte: usize,
43    end_byte: usize,
44) -> Result<&str, SliceError> {
45    if start_byte > end_byte {
46        return Err(SliceError::InvalidRange(start_byte, end_byte));
47    }
48
49    let start = snap_to_char_boundary_floor(content, start_byte.min(content.len()));
50    let end = snap_to_char_boundary_ceil(content, end_byte.min(content.len()));
51
52    std::str::from_utf8(&content[start..end]).map_err(|_| SliceError::InvalidRange(start, end))
53}
54
55/// Slice the entire symbol declaration (signature + body) from file on disk.
56pub fn slice_symbol(file_path: &Path, symbol: &Symbol) -> Result<String, SliceError> {
57    let bytes =
58        fs::read(file_path).map_err(|e| SliceError::Io(file_path.display().to_string(), e))?;
59
60    if bytes.is_empty() {
61        return Err(SliceError::EmptyFile(file_path.display().to_string()));
62    }
63
64    // Try byte span first
65    if symbol.end_byte <= bytes.len()
66        && symbol.start_byte <= symbol.end_byte
67        && symbol.end_byte > 0
68        && let Ok(slice) = slice_bytes_safe(&bytes, symbol.start_byte, symbol.end_byte)
69    {
70        return Ok(slice.to_string());
71    }
72
73    // Fallback to line slicing
74    slice_lines(&bytes, symbol.start_line, symbol.end_line)
75}
76
77/// Slice only the symbol's implementation body from file on disk.
78pub fn slice_symbol_body(file_path: &Path, symbol: &Symbol) -> Result<String, SliceError> {
79    let bytes =
80        fs::read(file_path).map_err(|e| SliceError::Io(file_path.display().to_string(), e))?;
81
82    if bytes.is_empty() {
83        return Err(SliceError::EmptyFile(file_path.display().to_string()));
84    }
85
86    // Check if body byte spans are present
87    if let (Some(body_start), Some(body_end)) = (symbol.body_start_byte, symbol.body_end_byte)
88        && body_end <= bytes.len()
89        && body_start <= body_end
90        && let Ok(slice) = slice_bytes_safe(&bytes, body_start, body_end)
91    {
92        return Ok(slice.to_string());
93    }
94
95    // Fallback to body line span
96    if let (Some(b_start), Some(b_end)) = (symbol.body_start_line, symbol.body_end_line) {
97        return slice_lines(&bytes, b_start, b_end);
98    }
99
100    // If no body defined, return entire symbol
101    slice_symbol(file_path, symbol)
102}
103
104/// Fallback 1-based line slicer
105pub fn slice_lines(bytes: &[u8], start_line: usize, end_line: usize) -> Result<String, SliceError> {
106    if start_line > end_line {
107        return Err(SliceError::InvalidRange(start_line, end_line));
108    }
109
110    let text = String::from_utf8_lossy(bytes);
111    let mut selected = Vec::new();
112
113    for (idx, line) in text.lines().enumerate() {
114        let line_num = idx + 1;
115        if line_num >= start_line && line_num <= end_line {
116            selected.push(line);
117        }
118        if line_num > end_line {
119            break;
120        }
121    }
122
123    Ok(selected.join("\n"))
124}
125
126#[cfg(test)]
127mod tests {
128    use super::*;
129
130    #[test]
131    fn test_slice_bytes_safe() {
132        let sample = "fn hello() {\n    println!(\"hi\");\n}\n";
133        let sliced = slice_bytes_safe(sample.as_bytes(), 0, sample.len()).unwrap();
134        assert_eq!(sliced, sample);
135    }
136
137    #[test]
138    fn test_slice_lines() {
139        let sample = "line 1\nline 2\nline 3\nline 4\n";
140        let sliced = slice_lines(sample.as_bytes(), 2, 3).unwrap();
141        assert_eq!(sliced, "line 2\nline 3");
142    }
143}
code_kb_core/slicer.rs

code_kb_core/
slicer.rs
