rustbrain_core/obsidian/
wikilink.rs

1//! Obsidian WikiLink (`[[...]]`) extraction.
2//!
3//! Links inside fenced code blocks and inline code spans are ignored.
4
5use serde::{Deserialize, Serialize};
6
7/// Parsed Obsidian WikiLink representation.
8#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
9pub struct WikiLink {
10    /// Target note name / id fragment.
11    pub target_node: String,
12    /// Optional `#section` fragment.
13    pub section: Option<String>,
14    /// Optional display alias after `|`.
15    pub display_alias: Option<String>,
16}
17
18impl WikiLink {
19    /// Format as standard Obsidian markdown.
20    pub fn to_markdown(&self) -> String {
21        match (&self.section, &self.display_alias) {
22            (Some(sec), Some(alias)) => format!("[[{}#{}|{}]]", self.target_node, sec, alias),
23            (Some(sec), None) => format!("[[{}#{}]]", self.target_node, sec),
24            (None, Some(alias)) => format!("[[{}|{}]]", self.target_node, alias),
25            (None, None) => format!("[[{}]]", self.target_node),
26        }
27    }
28}
29
30/// Extract all WikiLinks from markdown, skipping fenced and inline code.
31pub fn extract_wikilinks(markdown: &str) -> Vec<WikiLink> {
32    let mut links = Vec::new();
33    let bytes = markdown.as_bytes();
34    let len = bytes.len();
35    let mut i = 0;
36    let mut in_fence = false;
37
38    while i < len {
39        // Fence toggle on lines starting with ```
40        if is_line_start(bytes, i) && i + 2 < len && &bytes[i..i + 3] == b"```" {
41            in_fence = !in_fence;
42            i += 3;
43            continue;
44        }
45
46        if in_fence {
47            i += 1;
48            continue;
49        }
50
51        // Inline code span: skip until next unescaped `
52        if bytes[i] == b'`' {
53            i += 1;
54            while i < len && bytes[i] != b'`' {
55                i += 1;
56            }
57            if i < len {
58                i += 1;
59            }
60            continue;
61        }
62
63        if i + 1 < len && bytes[i] == b'[' && bytes[i + 1] == b'[' {
64            let start = i + 2;
65            if let Some(rel_end) = markdown[start..].find("]]") {
66                let inner = &markdown[start..start + rel_end];
67                if !inner.trim().is_empty() && !inner.contains('\n') {
68                    let mut alias_parts = inner.splitn(2, '|');
69                    let target_and_sec = alias_parts.next().unwrap_or("").trim();
70                    let display_alias = alias_parts.next().map(|s| s.trim().to_string());
71
72                    let mut sec_parts = target_and_sec.splitn(2, '#');
73                    let target_node = sec_parts.next().unwrap_or("").trim().to_string();
74                    let section = sec_parts.next().map(|s| s.trim().to_string());
75
76                    if !target_node.is_empty() {
77                        links.push(WikiLink {
78                            target_node,
79                            section,
80                            display_alias,
81                        });
82                    }
83                }
84                i = start + rel_end + 2;
85                continue;
86            }
87        }
88        i += 1;
89    }
90
91    links
92}
93
94fn is_line_start(bytes: &[u8], i: usize) -> bool {
95    i == 0 || bytes[i - 1] == b'\n'
96}
97
98#[cfg(test)]
99mod tests {
100    use super::*;
101
102    #[test]
103    fn extract_wikilinks_basic() {
104        let text =
105            "This relates to [[LogCompaction#Section|Log Compaction]] and [[RaftConsensus]].";
106        let links = extract_wikilinks(text);
107        assert_eq!(links.len(), 2);
108        assert_eq!(links[0].target_node, "LogCompaction");
109        assert_eq!(links[0].section.as_deref(), Some("Section"));
110        assert_eq!(links[0].display_alias.as_deref(), Some("Log Compaction"));
111        assert_eq!(links[1].target_node, "RaftConsensus");
112        assert_eq!(
113            links[0].to_markdown(),
114            "[[LogCompaction#Section|Log Compaction]]"
115        );
116    }
117
118    #[test]
119    fn ignores_code_fences_and_inline() {
120        let text = "See [[Real]] and `[[NotThis]]` and:\n```\n[[AlsoNot]]\n```\n[[Yes]]";
121        let links = extract_wikilinks(text);
122        let names: Vec<_> = links.iter().map(|l| l.target_node.as_str()).collect();
123        assert_eq!(names, vec!["Real", "Yes"]);
124    }
125}
rustbrain_core/obsidian/wikilink.rs

rustbrain_core/obsidian/
wikilink.rs
