1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
/// The column holding each row's source file, used to skip files already loaded by `append`.
pub const SOURCE_FILE_COLUMN: &str = "_sqlload_source_file";
/// The column holding the version of each row's source file (its ETag, or size and modification time), to notice files rewritten in place.
pub const SOURCE_VERSION_COLUMN: &str = "_sqlload_source_version";
/// Separates a file name from its version in the loaded-files query; versions never contain it.
pub const VERSION_SEPARATOR: char = '|';
/// The file name suffix of Parquet data files.
pub const PARQUET_SUFFIX: &str = ".parquet";
/// The file name suffixes of JSON data files, for both `ndjson` and `json`.
pub const JSON_SUFFIXES: & = &;
/// The file name suffix of CSV data files.
pub const CSV_SUFFIX: &str = ".csv";
/// The file name suffix of gzip-compressed files.
pub const GZIP_SUFFIX: &str = ".gz";
/// The file name suffix of zstd-compressed files.
pub const ZSTD_SUFFIX: &str = ".zst";
/// Records read from each sampled JSON or CSV file to work out its column types.
pub const SAMPLE_ROWS: usize = 10_000;
/// The longest JSON object or CSV record read, so a broken file (an unclosed quote or brace) fails instead of filling memory.
pub const MAX_RECORD_BYTES: usize = 64 * 1024 * 1024;
/// Bytes read from a JSON or CSV file (after decompression) at a time.
pub const READ_CHUNK_BYTES: usize = 256 * 1024;
/// Files sampled to work out the schema (their footers, or their first records): the newest by key, so recent schema changes are seen.
pub const SCHEMA_SAMPLE_FILES: usize = 20;
/// Footers read at once when a destination reads many.
pub const CONCURRENT_FOOTER_READS: usize = 16;