nx_library_search_test.nx source

1// nx_library_search_test.nx -- the Researcher handles a LIBRARY OF INFORMATION, not one source, and 2// searches it with S-class ranked retrieval. End to end, sovereign: 3// 1. FETCH the index, RECOGNIZE it is a library (>=3 [doc] entries) 4// 2. HARVEST every listed doc -- follow each entry, fetch it, MIRROR it into the Nishi Library 5// 3. build the corpus FROM THE LOCAL MIRRORS (bot-block-proof) 6// 4. answer 3 different SEARCH PROMPTS by TF-IDF RANKING the whole corpus -- each returns the single 7// best-matching doc, where naive substring search would be ambiguous (e.g. "abandon" is in 2 docs) 8// Exit 0 on 8/8. license_tier: ORIGINAL 9// 10// NOTE 2026-06-20: this is a LIVE-INTEGRATION test -- it requires a fixture HTTP server on 127.0.0.1:8099 11// serving a library index + the 3 docs. With no server up it reads 0/8 ("ambiguous across 0 docs") because 12// the harvest fetches nothing -- that is an unmet fixture dependency, NOT a search-logic regression. The 13// REPRODUCIBLE-BY-OUR-SYSTEM proof of the ranked-retrieval claim (no socket, real on-disk corpus, both TF-IDF 14// + production BM25) is nx_library_search_gate (8/8 GREEN). Run THIS one only with the :8099 fixture live. 15 16import "nx_browse_text.nx" 17import "nx_library_search.nx" 18import "nx_library_cache.nx" 19import "nx_syscalls.nx" 20 21func st_puts(s: *u8) -> i64 { var n: i64 = 0; while s[n] != (0 as u8) { n = n + 1 } sys_write(1, s, n); return 0 } 22func st_num(v: i64) -> i64 { let bb: *u8 = sys_mmap(28); var m: i64=v; if m<0 {m=0-m; sys_write(1,"-" as *u8,1)}; let t: *u8 = sys_mmap(28); var k: i64=0; if m==0 {t[0]=48;k=1}; while m>0 {t[k]=48+(m%10); m=m/10; k=k+1}; var i: i64=0; while i<k {bb[i]=t[k-1-i]; i=i+1}; sys_write(1, bb, k); return 0 } 23 24// build "knowledge/library/harvK.txt" into out (null-terminated); return length. 25func st_harv_path(out: *u8, k: i64) -> i64 { 26 let pre: *u8 = "knowledge/library/harv" as *u8 27 var i: i64 = 0; while pre[i] != (0 as u8) { out[i] = pre[i]; i = i + 1 } 28 out[i] = 48 + k; i = i + 1 29 let suf: *u8 = ".txt" as *u8 30 var j: i64 = 0; while suf[j] != (0 as u8) { out[i] = suf[j]; i = i + 1; j = j + 1 } 31 out[i] = 0 32 return i 33} 34 35func main() -> i64 { 36 st_puts("=== RESEARCHER: harvest a LIBRARY + S-class ranked search over the corpus ===\n" as *u8) 37 let addr: *u8 = sys_mmap(32); br_addr_ipv4(addr, 127, 0, 0, 1, 8099) 38 let scratch: *u8 = sys_mmap(16384) 39 let status: *i64 = sys_mmap(8) as *i64; let verdict: *i64 = sys_mmap(8) as *i64 40 41 // 1) FETCH the index, recognize a LIBRARY 42 let idx: *u8 = sys_mmap(16384) 43 let inb: i64 = nx_browse_text(addr, "/" as *u8, 1, "127.0.0.1" as *u8, 9, scratch, 16384, idx, 16384, status, verdict) 44 let islib: i64 = ls_is_library(idx, inb) 45 let entries: i64 = ls_entry_count(idx, inb) 46 st_puts(" 1. INDEX : bytes=" as *u8); st_num(inb); st_puts(" is-library=" as *u8); st_num(islib); st_puts(" entries=" as *u8); st_num(entries); st_puts("\n" as *u8) 47 48 // 2) HARVEST every listed doc -> mirror into the library; 3) build corpus FROM THE MIRRORS 49 let ptrs: *i64 = sys_mmap(8 * 8) as *i64 50 let lens: *i64 = sys_mmap(8 * 8) as *i64 51 let pathbuf: *u8 = sys_mmap(128); let mpath: *u8 = sys_mmap(128) 52 var harvested: i64 = 0; var k: i64 = 0 53 while k < entries { 54 let plen: i64 = ls_entry_path(idx, inb, k, pathbuf, 120) 55 let tmp: *u8 = sys_mmap(16384) 56 let dn: i64 = nx_browse_text(addr, pathbuf, plen, "127.0.0.1" as *u8, 9, scratch, 16384, tmp, 16384, status, verdict) 57 st_harv_path(mpath, k) 58 let w: i64 = lc_mirror(mpath, tmp, dn) // download into the Nishi Library 59 let buf: *u8 = sys_mmap(16384) 60 let rn: i64 = lc_read_mirror(mpath, buf, 16384) // corpus comes from the LOCAL mirror 61 ptrs[k] = buf as i64; lens[k] = rn 62 if rn > 0 { if w == dn { harvested = harvested + 1 } } 63 st_puts(" harvested " as *u8); sys_write(1, pathbuf, plen); st_puts(" -> " as *u8); st_num(rn); st_puts(" bytes mirrored\n" as *u8) 64 k = k + 1 65 } 66 67 // 4) S-class SEARCH: 3 prompts, TF-IDF ranked over the corpus 68 let q1: *i64 = sys_mmap(4 * 8) as *i64; q1[0] = "abandon" as *u8 as i64; q1[1] = "load" as *u8 as i64 69 let q2: *i64 = sys_mmap(4 * 8) as *i64; q2[0] = "choices" as *u8 as i64; q2[1] = "decision" as *u8 as i64 70 let q3: *i64 = sys_mmap(4 * 8) as *i64; q3[0] = "credibility" as *u8 as i64; q3[1] = "conversion" as *u8 as i64 71 let b1: i64 = ls_best(ptrs, lens, entries, q1, 2) 72 let b2: i64 = ls_best(ptrs, lens, entries, q2, 2) 73 let b3: i64 = ls_best(ptrs, lens, entries, q3, 2) 74 let ambig: i64 = ls_df(ptrs, lens, entries, "abandon" as *u8) // naive substring: "abandon" in 2 docs 75 st_puts(" 4. SEARCH : 'abandon load'->doc" as *u8); st_num(b1); st_puts(" 'choices decision'->doc" as *u8); st_num(b2); st_puts(" 'credibility conversion'->doc" as *u8); st_num(b3); st_puts("\n" as *u8) 76 st_puts(" EXCEED: naive substring 'abandon' is ambiguous across " as *u8); st_num(ambig); st_puts(" docs; ranked search resolves to the single best (doc" as *u8); st_num(b1); st_puts(")\n" as *u8) 77 78 let r: *i64 = sys_mmap(8 * 8) as *i64 79 r[0] = 0; if islib == 1 { r[0] = 1 } // recognized a LIBRARY (not one source) 80 r[1] = 0; if entries == 3 { r[1] = 1 } // parsed the index 81 r[2] = 0; if harvested == 3 { r[2] = 1 } // harvested+mirrored every doc 82 r[3] = 0; if b1 == 0 { r[3] = 1 } // ranked search picks the abandonment doc 83 r[4] = 0; if b2 == 1 { r[4] = 1 } // ... the decision doc 84 r[5] = 0; if b3 == 2 { r[5] = 1 } // ... the trust doc 85 r[6] = 0; if ambig == 2 { if b1 == 0 { r[6] = 1 } } // naive ambiguous (2) but ranked resolves to 1 86 r[7] = 0; if lens[0] > 0 { if lens[1] > 0 { if lens[2] > 0 { r[7] = 1 } } } // corpus built from local mirrors 87 var pass: i64 = 0; var i: i64 = 0 88 while i < 8 { pass = pass + r[i]; i = i + 1 } 89 st_puts("----\n passed " as *u8); st_num(pass); st_puts("/8\n" as *u8) 90 if pass == 8 { st_puts(" S-CLASS SEARCH: a library recognized, every doc harvested into the Nishi Library, and 3 prompts each ranked to the right cited doc -- over LOCAL mirrors, no live web.\n" as *u8); sys_exit(0); return 0 } 91 st_puts(" FAIL\n" as *u8); sys_exit(1); return 1 92}