magi-code 0.97.5

//! Bounded, read-only excerpts from files too large for hashline snapshots.
use super::{args::FILE_READ_MAX_BYTES, read_selector::LineSelector};
use crate::cancellation::AgentCancellation;
use std::{
    fs::File,
    io::{BufRead, BufReader, Read},
};

pub(super) const SCAN_MAX_BYTES: u64 = 16 * FILE_READ_MAX_BYTES;
// Leave room for the caller's incomplete-read notice inside the existing output cap.
const EXCERPT_MAX_BYTES: usize = FILE_READ_MAX_BYTES as usize - 256;

pub(super) struct RawExcerpt {
    pub text: String,
    pub request_complete: bool,
    pub bytes_scanned: u64,
    pub output_limited: bool,
}

pub(super) fn read_raw_excerpt(
    file: File,
    selector: LineSelector,
    cancellation: &AgentCancellation,
) -> anyhow::Result<RawExcerpt> {
    let (start, end) = match selector {
        LineSelector::Single(line) => (line, line),
        LineSelector::Range { start, end } => (start, end),
        LineSelector::FromStart { end } => (1, end),
        LineSelector::FromLine { start } => (start, usize::MAX),
        LineSelector::Count { start, count } => (start, start.saturating_add(count - 1)),
    };
    let mut reader = BufReader::new(file.take(SCAN_MAX_BYTES));
    let mut bytes_scanned = 0u64;
    let mut line_number = 1usize;
    let mut line = Vec::new();
    let mut output = String::new();
    loop {
        cancellation.check()?;
        // Bound each allocation and allow cancellation even on a single huge line.
        let chunk = reader.fill_buf()?;
        if chunk.is_empty() {
            if bytes_scanned == SCAN_MAX_BYTES {
                return Ok(RawExcerpt {
                    text: output,
                    request_complete: false,
                    bytes_scanned,
                    output_limited: false,
                });
            }
            if !line.is_empty() && line_number >= start {
                append_line(&mut output, &line)?;
            }
            return Ok(RawExcerpt {
                text: output,
                request_complete: true,
                bytes_scanned,
                output_limited: false,
            });
        }
        let consumed = chunk
            .iter()
            .position(|byte| *byte == b'\n')
            .map_or(chunk.len(), |n| n + 1);
        let ends_line = chunk[consumed - 1] == b'\n';
        if line_number >= start {
            let body = &chunk[..consumed - usize::from(ends_line)];
            if output
                .len()
                .saturating_add(line.len())
                .saturating_add(body.len())
                .saturating_add(1)
                > EXCERPT_MAX_BYTES
            {
                return Ok(RawExcerpt {
                    text: output,
                    request_complete: false,
                    bytes_scanned,
                    output_limited: true,
                });
            }
            line.extend_from_slice(body);
        }
        reader.consume(consumed);
        bytes_scanned += consumed as u64;
        if ends_line {
            if line_number >= start {
                append_line(&mut output, &line)?;
                line.clear();
            }
            if line_number == end {
                return Ok(RawExcerpt {
                    text: output,
                    request_complete: true,
                    bytes_scanned,
                    output_limited: false,
                });
            }
            line_number = line_number.saturating_add(1);
        }
    }
}

fn append_line(output: &mut String, line: &[u8]) -> anyhow::Result<()> {
    let line = line.strip_suffix(b"\r").unwrap_or(line);
    // A trailing separator is removed by the caller; empty selected lines still count.
    output.push_str(std::str::from_utf8(line)?);
    output.push('\n');
    Ok(())
}