nx_research_fetch_ext.nx
buildroot/runtime/nx_research_fetch_ext.nx
about
nx_research_fetch_ext.nx -- DYNAMIC / GROWABLE fetch extensions (operator 2026-07-07: "make the fetcher
dynamic and we can keep growing and keep going"). Additive lib over nx_research_engine (no main).
rf_fetch_bank_paged: fetch an offset/length-paginated API (HF datasets-server) across N pages, appending
each raw HTTP response to knowledge/fetched/<name>.raw. The record-scanning benchmark parsers
(db_find_key "question"/"context") skip the intervening HTTP headers and find records across ALL pages,
so concatenation "just works" -> the researcher's eval sets grow to any size with bounded memory.
license_tier: ORIGINAL
dependencies 1 imports · 4 importers
imports: nx_research_engine.nx
imported by: nx_musique_fetch.nxnx_paged_fetch_gate.nxnx_qabench_corpus_fetch.nxnx_qabench_fetch_big.nx
structs
| none |
consts
| 9 | const K_MAGIC_4096: i64 = 4096 |
| 10 | const K_MAGIC_8388608: i64 = 8388608 |
functions
| 13 | func rfx_cat(dst: *u8, off: i64, s: *u8) -> i64 called by 1: rf_fetch_bank_paged_from |
| 19 | func rfx_uitoa(dst: *u8, off: i64, v: i64) -> i64 called by 1: rf_fetch_bank_paged_from |
| 38 | func rf_fetch_bank_paged(urlbase: *u8, name: *u8, store: *TrustStore, pagesize: i64, npages: i64) -> i64 |
| 44 | func rf_fetch_bank_paged_from(urlbase: *u8, name: *u8, store: *TrustStore, pagesize: i64, npages: i64, startpage: i64) -> i64 |