Function find_symbols

src/classmap_scanner.rs:830–1246

The **full-scan**: a single-pass byte-level scanner that extracts fully-qualified class, function, and constant names from PHP source bytes. This is the extended version of [`find_classes`] (the PSR-4 scanner) that also recognises `function` declarations, `define()` calls, and top-level `const` statements. It is used for both non-Composer projects (full workspace scan) and Composer autoload file

(content: &[u8])

/// projects (full workspace scan) and Composer autoload files
/// (`autoload_files.php` and their `require_once` chains).
pub fn find_symbols(content: &[u8]) -> ScanResult {
    // Quick rejection — if the file has none of the relevant keywords
    // we can bail immediately.
    if !has_any_keyword(content) {
        return ScanResult::default();
    }

    let mut result = ScanResult::default();
    let mut namespace = String::new();
    let len = content.len();
    let mut i = 0;

    // Brace depth tracking for top-level `const` detection.
    // Depth 0 = top-level, depth 1 = inside a class/namespace block.
    let mut brace_depth: u32 = 0;
    // Whether we are inside a braced namespace block.
    let mut in_braced_namespace = false;
    // The brace depth at which the current namespace was opened.
    // `const` declarations at this depth (or depth 0 outside braced
    // namespaces) are top-level.
    let mut namespace_brace_depth: u32 = 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; // rest of file is a comment
            }
            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; // unclosed block comment
            }
            continue;
        }