Struct ChunkMetadata 

pub struct ChunkMetadata {
    pub source_type: Arc<str>,
    pub path: Option<Arc<str>>,
    pub commit: Option<Arc<str>>,
    pub author: Option<Arc<str>>,
    pub date: Option<Arc<str>>,
    pub base_offset: usize,
    pub base_line: usize,
    pub mtime_ns: Option<u64>,
    pub size_bytes: Option<u64>,
    pub decoded_span: Option<(usize, usize)>,
}

Metadata that tracks the source location for a scanned chunk.

Examples

use keyhog_core::ChunkMetadata;

let metadata = ChunkMetadata {
    source_type: "git-diff".into(),
    path: Some("src/lib.rs".into()),
    commit: Some("abc123".into()),
    author: Some("Dev".into()),
    date: Some("2026-03-26T00:00:00Z".into()),
    ..Default::default()
};

assert_eq!(&*metadata.source_type, "git-diff");

Fields

source_type: Arc<str>

Arc<str> (not String) so cloning a chunk’s metadata, done per decode sub-chunk, where every sub-chunk of a file shares the same source_type and path: is a refcount bump, not a fresh heap allocation + copy of each string. Mirrors the Arc<str> convention already used by MatchLocation in finding.rs; serialized through the same serde_arc_str helpers so no serde rc feature is needed.

path: Option<Arc<str>>

commit: Option<Arc<str>>

author: Option<Arc<str>>

date: Option<Arc<str>>

base_offset: usize

base_line: usize

Number of lines that precede base_offset in the original file - the line-number analog of base_offset. Zero for whole-file chunks (single-pass mmap, stdin, http, git diffs). Non-zero only when a source slices one file into multiple chunks (the filesystem >window_size windowed path), where each window after the first starts partway through the file. The scanner computes a match’s line number within the chunk text and adds this base so the reported line is the absolute file line, not the per-window one - exactly mirroring how base_offset makes the byte offset absolute. Without it, a secret on line 584307 of a 70 MiB file was reported at the window-local line (e.g. line 2), making findings impossible to locate.

mtime_ns: Option<u64>

File mtime in nanoseconds since UNIX epoch, when the source can surface it cheaply (filesystem walks). Optional because non-fs sources (stdin, http, git diffs) don’t have a meaningful mtime. Populated to drive the merkle-index metadata fast-path.

size_bytes: Option<u64>

File size in bytes, when known cheaply at chunk-production time. Same shape and rationale as mtime_ns.

decoded_span: Option<(usize, usize)>

For DECODE sub-chunks only: the [start, end) byte range of the freshly decoded text within data. A decode sub-chunk is a small window of already-scanned parent context with the decoded blob spliced in at this span; everything OUTSIDE the span was scanned (and any finding deduped) when the parent chunk was scanned, so the self-contained passes only need to rescan a focus window around this span instead of the whole splice. None for all non-decode chunks (whole-file, windowed, git-diff, …).

Trait Implementations

impl Clone for ChunkMetadata

fn clone(&self) -> ChunkMetadata

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ChunkMetadata

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for ChunkMetadata

fn default() -> ChunkMetadata

Returns the “default value” for a type.

impl Serialize for ChunkMetadata

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.