fn build_prompt()

in crates/llm-ls/src/main.rs [141:233]

• 93 lines of code
• 17 McCabe index (conditional complexity)

fn build_prompt(
    pos: Position,
    text: &Rope,
    fim: &FimParams,
    tokenizer: Option<Arc<Tokenizer>>,
    context_window: usize,
) -> Result<String> {
    let t = Instant::now();
    if fim.enabled {
        let mut remaining_token_count = context_window - 3; // account for FIM tokens
        let mut before_iter = text.lines_at(pos.line as usize + 1).reversed();
        let mut after_iter = text.lines_at(pos.line as usize);
        let mut before_line = before_iter.next();
        if let Some(line) = before_line {
            let col = (pos.character as usize).clamp(0, line.len_chars());
            before_line = Some(line.slice(0..col));
        }
        let mut after_line = after_iter.next();
        if let Some(line) = after_line {
            let col = (pos.character as usize).clamp(0, line.len_chars());
            after_line = Some(line.slice(col..));
        }
        let mut before = vec![];
        let mut after = String::new();
        while before_line.is_some() || after_line.is_some() {
            if let Some(before_line) = before_line {
                let before_line = before_line.to_string();
                let tokens = if let Some(tokenizer) = tokenizer.clone() {
                    tokenizer.encode(before_line.clone(), false)?.len()
                } else {
                    before_line.len()
                };
                if tokens > remaining_token_count {
                    break;
                }
                remaining_token_count -= tokens;
                before.push(before_line);
            }
            if let Some(after_line) = after_line {
                let after_line = after_line.to_string();
                let tokens = if let Some(tokenizer) = tokenizer.clone() {
                    tokenizer.encode(after_line.clone(), false)?.len()
                } else {
                    after_line.len()
                };
                if tokens > remaining_token_count {
                    break;
                }
                remaining_token_count -= tokens;
                after.push_str(&after_line);
            }
            before_line = before_iter.next();
            after_line = after_iter.next();
        }
        let prompt = format!(
            "{}{}{}{}{}",
            fim.prefix,
            before.into_iter().rev().collect::<Vec<_>>().join(""),
            fim.suffix,
            after,
            fim.middle
        );
        let time = t.elapsed().as_millis();
        info!(prompt, build_prompt_ms = time, "built prompt in {time} ms");
        Ok(prompt)
    } else {
        let mut remaining_token_count = context_window;
        let mut before = vec![];
        let mut first = true;
        for mut line in text.lines_at(pos.line as usize + 1).reversed() {
            if first {
                let col = (pos.character as usize).clamp(0, line.len_chars());
                line = line.slice(0..col);
                first = false;
            }
            let line = line.to_string();
            let tokens = if let Some(tokenizer) = tokenizer.clone() {
                tokenizer.encode(line.clone(), false)?.len()
            } else {
                line.len()
            };
            if tokens > remaining_token_count {
                break;
            }
            remaining_token_count -= tokens;
            before.push(line);
        }
        let prompt = before.into_iter().rev().collect::<Vec<_>>().join("");
        let time = t.elapsed().as_millis();
        info!(prompt, build_prompt_ms = time, "built prompt in {time} ms");
        Ok(prompt)
    }
}