Documentation
¶
Overview ¶
Package tdtpjson converts TDTP packets to JSON arrays of objects.
One object per row, keys from the Schema field names (not positional pipe strings) — the consumer addresses columns by name, like an API. Values are honestly typed via schema.Converter: integers and floats as JSON numbers, booleans as bools, NULLs as null, dates as ISO strings, everything else as strings. Integers past the JS safe range (2^53-1) go out as strings so downstream jq/node does not silently round them; NaN/±Inf (no JSON representation) become null.
Rows stream through a single pass — no giant slice is ever built — so 100 MB pages convert in constant memory. Column order follows the schema (or --fields order); key order in JSON is therefore stable.
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func Write ¶
func Write(ctx context.Context, w io.Writer, pkt *packet.DataPacket, query *packet.Query, pretty bool) (int, error)
Write converts pkt to a JSON array on w, applying query (WHERE, ORDER BY, LIMIT, OFFSET, fields projection) exactly like the CSV converter does. It returns the number of data objects written. pretty indents one object per line for humans; compact is the default for pipelines.
The packet must hold plain row-major rows: callers decompress and expand compact/columnar layouts first (see WritePacket), or pass an already-plain packet. Encrypted content is refused.
func WritePacket ¶
func WritePacket(ctx context.Context, w io.Writer, pkt *packet.DataPacket, query *packet.Query, pretty bool) (int, error)
WritePacket is Write for a stored packet: decompresses, verifies local xxh3 integrity when stamped, and expands compact/columnar layouts first. Encrypted packets are refused (hashes would cover ciphertext).
Types ¶
This section is empty.
Source Files