Function find_classes

src/classmap_scanner.rs:1258–1541

The **PSR-4 scanner**: a single-pass byte-level scanner that extracts fully-qualified class, interface, trait, and enum names from PHP source bytes. This is the classes-only scanner used by the PSR-4 directory walker and vendor package scanner. For a scanner that also extracts functions and constants, see [`find_symbols`] (the full-scan). Skips comments, strings, heredocs, and nowdocs inline wi

(content: &[u8])

/// Skips comments, strings, heredocs, and nowdocs inline without
/// allocating a separate "cleaned" buffer.
pub fn find_classes(content: &[u8]) -> Vec<String> {
    // Quick rejection — use SIMD to check if any class-like keywords exist
    if !has_class_keyword(content) {
        return Vec::new();
    }

    let mut classes = Vec::with_capacity(4);
    let mut namespace = String::new();
    let len = content.len();
    let mut i = 0;

    // State flags
    let mut in_line_comment = false;
    let mut in_block_comment = false;
    let mut in_single_string = false;
    let mut in_double_string = false;
    let mut in_heredoc = false;
    let mut heredoc_id: &[u8] = &[];

    while i < len {
        // ── Skip: line comment (memchr to newline) ──────────────────
        if in_line_comment {
            if let Some(pos) = memchr(b'\n', &content[i..]) {
                i += pos + 1;
            } else {
                break;
            }
            in_line_comment = false;
            continue;
        }

        // ── Skip: block comment (memchr to '*') ─────────────────────
        if in_block_comment {
            if let Some(pos) = memchr(b'*', &content[i..]) {
                i += pos;
                if i + 1 < len && content[i + 1] == b'/' {
                    in_block_comment = false;
                    i += 2;
                } else {
                    i += 1;
                }
            } else {
                break;
            }
            continue;
        }

        // ── Skip: single-quoted string (memchr to '\'' or '\\') ────
        if in_single_string {
            match memchr2_single_string(&content[i..]) {
                Some((offset, b'\\')) => {
                    i += offset + 2;
                }
                Some((offset, _)) => {
                    i += offset + 1;
                    in_single_string = false;
                }
                None => break,