nx_value_parse_csv.nx

nx_value_parse_csv.nx -- bytes -> NxValue (ARRAY of OBJECTs). module: nishi-core.data.value_parse_csv depends: nishi-core.io.syscalls, nishi-core.data.value disk_kb: 5 capability: CORE_IO wired_status: FULLY_WIRED license_tier: PUBLIC_NISHI_SUBSTRATE genealogy_id: rfc_4180_csv_canonical + postel_robustness_principle + nishi_build_the_system_cardinal_2026 Brick #2b of the bits-up ingestion stack. Parallel to nx_value_parse_json but for delimiter-separated values. Input: CSV/TSV bytes (RFC 4180-ish; tolerant to BOM, CRLF/LF, and unterminated final row). Output: NxValue of kind ARRAY whose items are OBJECTs. Each OBJECT's keys come from the header row; each OBJECT's values are STRING (typing happens later). Configurable via NxCsvOpts: delimiter: byte; default 0x2C (',') -- pass 0x09 ('\t') for TSV quote_char: byte; default 0x22 ('"'); 0 to disable quote handling has_header: 1 = first row is field names (default) 0 = synthesize "col_0","col_1",... keys max_rows: hard cap on data rows parsed (per JPL Rule 2) Composition: nx_ingest_batch reads bytes, hands to either nx_value_parse_json OR nx_value_parse_csv based on extension -- same downstream NxValue tree, same flat emit.