a4e667de10
Single-commit clean baseline after security scrub of niche-tells, project codenames, internal jargon, and contributor-email leaks. Contents: - 100 Rust crates (_primitives/_rust/) - 37 agent manifests (_manifests/) + generated specs (_generated/) - 67 user-invocable skills (skills/) - 33 hooks (hooks/) - Composition blocks (_blocks/) - Documentation (docs/, README.md) - TS adapter packages (_ts_packages/) - Assembler (_assembler/) - Roles (_roles/) - Templates (_templates/) - Forgejo CI (.forgejo/) Author: Denis Parfionovich <info@greendragon.info> License: see LICENSE.
112 lines
3.3 KiB
Rust
112 lines
3.3 KiB
Rust
//! Markdown H2-heading splitter and description extractor.
|
|
//!
|
|
//! Provides pure text operations used by skill_extractor:
|
|
//! - split_by_h2: parse `## ` sections from markdown text
|
|
//! - first_sentences: extract up to 3 sentences for a skill description
|
|
//! - strip_markdown: remove markdown syntax for plain-text extraction
|
|
|
|
use crate::fragment_writer::sanitize;
|
|
|
|
/// Split markdown text into `(slug, heading, body)` tuples at `## ` boundaries.
|
|
pub fn split_by_h2(text: &str) -> Vec<(String, String, String)> {
|
|
let mut out = Vec::new();
|
|
let mut current_heading: Option<String> = None;
|
|
let mut buf = String::new();
|
|
|
|
for line in text.lines() {
|
|
if let Some(rest) = line.strip_prefix("## ") {
|
|
flush_section(&mut out, &mut current_heading, &mut buf);
|
|
current_heading = Some(rest.trim().to_string());
|
|
} else {
|
|
buf.push_str(line);
|
|
buf.push('\n');
|
|
}
|
|
}
|
|
flush_section(&mut out, &mut current_heading, &mut buf);
|
|
out
|
|
}
|
|
|
|
fn flush_section(
|
|
out: &mut Vec<(String, String, String)>,
|
|
heading: &mut Option<String>,
|
|
buf: &mut String,
|
|
) {
|
|
let body = std::mem::take(buf);
|
|
if let Some(h) = heading.take() {
|
|
let slug = sanitize(&h);
|
|
out.push((slug, h, body));
|
|
}
|
|
}
|
|
|
|
/// Extract first 1-3 sentences up to `max_chars` from body text.
|
|
pub fn first_sentences(text: &str, max_chars: usize) -> String {
|
|
let plain = strip_markdown(text);
|
|
let mut out = String::new();
|
|
let mut count = 0usize;
|
|
for part in plain.split_inclusive(&['.', '!', '?'][..]) {
|
|
let trimmed = part.trim();
|
|
if trimmed.is_empty() {
|
|
continue;
|
|
}
|
|
if out.len() + trimmed.len() + 1 > max_chars || count >= 3 {
|
|
break;
|
|
}
|
|
if !out.is_empty() {
|
|
out.push(' ');
|
|
}
|
|
out.push_str(trimmed);
|
|
count += 1;
|
|
}
|
|
if out.is_empty() {
|
|
plain.chars().take(max_chars).collect()
|
|
} else {
|
|
out
|
|
}
|
|
}
|
|
|
|
/// Strip markdown syntax (headings, code fences, links) for description use.
|
|
pub fn strip_markdown(text: &str) -> String {
|
|
let mut out = String::new();
|
|
for line in text.lines() {
|
|
let l = line.trim();
|
|
if l.starts_with('#') || l.starts_with("```") || l.starts_with("---") {
|
|
continue;
|
|
}
|
|
let clean = l.trim_start_matches(|c: char| c == '*' || c == '-' || c == '>');
|
|
let clean = clean.trim();
|
|
if !clean.is_empty() {
|
|
if !out.is_empty() {
|
|
out.push(' ');
|
|
}
|
|
out.push_str(clean);
|
|
}
|
|
}
|
|
out
|
|
}
|
|
|
|
#[cfg(test)]
|
|
mod tests {
|
|
use super::*;
|
|
|
|
#[test]
|
|
fn split_h2_basic() {
|
|
let md = "preamble\n## Alpha\nbody a\n## Beta\nbody b\n";
|
|
let secs = split_by_h2(md);
|
|
assert_eq!(secs.len(), 2);
|
|
assert_eq!(secs[0].1, "Alpha");
|
|
assert!(secs[0].2.contains("body a"));
|
|
}
|
|
|
|
#[test]
|
|
fn first_sentences_truncates() {
|
|
let s = first_sentences("Hello world. Second sentence. Third one. Fourth one.", 200);
|
|
assert!(s.contains("Hello world."));
|
|
}
|
|
|
|
#[test]
|
|
fn strip_removes_headings() {
|
|
let stripped = strip_markdown("## Heading\nNormal text.");
|
|
assert!(!stripped.contains("##"));
|
|
assert!(stripped.contains("Normal text."));
|
|
}
|
|
}
|