diff options
| author | JSDurand <mmemmew@gmail.com> | 2026-11-19 09:29:47 +0800 |
|---|---|---|
| committer | JSDurand <mmemmew@gmail.com> | 2026-11-19 09:29:47 +0800 |
| commit | 5b9a806af13d38163156302097e405a7b145b6cc (patch) | |
| tree | c079f3c4e027dc730892630f667b0e6401736043 /src/adapter/org.rs | |
Tantivy full text index, link graph, inline query filters, Ollama
embeddings with paragraph chunks, and an Elisp summary/article UI.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Diffstat (limited to 'src/adapter/org.rs')
| -rw-r--r-- | src/adapter/org.rs | 84 |
1 files changed, 84 insertions, 0 deletions
diff --git a/src/adapter/org.rs b/src/adapter/org.rs new file mode 100644 index 0000000..a999814 --- /dev/null +++ b/src/adapter/org.rs @@ -0,0 +1,84 @@ +//! Org adapter: each top-level heading is a note. Metadata lives in +//! the heading (title, tags) and in its property drawer. + +use super::{Fields, build}; +use crate::note::Note; + +fn is_tag_string(s: &str) -> bool { + s.len() > 2 + && s.starts_with(':') + && s.ends_with(':') + && s.chars() + .all(|c| c.is_alphanumeric() || "_@#%:.".contains(c)) +} + +/// Split a heading line into its title and its tags. +fn parse_heading(line: &str) -> (String, Vec<String>) { + let h = line.trim_start_matches('*').trim(); + if let Some(idx) = h.rfind(char::is_whitespace) { + let last = &h[idx + 1..]; + if is_tag_string(last) { + let tags = last + .split(':') + .filter(|t| !t.is_empty()) + .map(|t| t.to_string()) + .collect(); + return (h[..idx].trim().to_string(), tags); + } + } + (h.to_string(), Vec::new()) +} + +fn is_planning(line: &str) -> bool { + let t = line.trim_start(); + t.starts_with("SCHEDULED:") || t.starts_with("DEADLINE:") || t.starts_with("CLOSED:") +} + +pub fn parse(path: &str, text: &str) -> Vec<Note> { + // Find the top-level headings: a single star at column 0. + let mut heads: Vec<(usize, usize)> = Vec::new(); + let mut off = 0; + for (ln, line) in text.split_inclusive('\n').enumerate() { + let t = line.trim_end(); + if line.starts_with("* ") || t == "*" { + heads.push((off, ln + 1)); + } + off += line.len(); + } + let mut notes = Vec::new(); + for (i, &(start, line_no)) in heads.iter().enumerate() { + let end = heads.get(i + 1).map(|h| h.0).unwrap_or(text.len()); + let chunk = &text[start..end]; + let mut lines = chunk.lines(); + let heading = lines.next().unwrap_or(""); + let (title, tags) = parse_heading(heading); + let mut fields = Fields { + title, + tags, + ..Default::default() + }; + let mut rest: Vec<&str> = lines.collect(); + // Skip planning lines, then read the property drawer, if any. + let mut k = 0; + while k < rest.len() && is_planning(rest[k]) { + k += 1; + } + let mut body_from = 0; + if k < rest.len() && rest[k].trim() == ":PROPERTIES:" { + let mut j = k + 1; + while j < rest.len() && rest[j].trim() != ":END:" { + let l = rest[j].trim(); + if let Some(r) = l.strip_prefix(':') { + if let Some((key, val)) = r.split_once(':') { + fields.set(key, val); + } + } + j += 1; + } + body_from = (j + 1).min(rest.len()); + } + let body = rest.split_off(body_from).join("\n"); + notes.push(build(path, text, start, end, line_no, fields, body)); + } + notes +} |
