code wiki / _hdl_build / _wiki_cite_gate.nx
`_wiki_cite_gate.nx` source

↩ module page · 355 lines · 18660 B
1// _wiki_cite_gate.nx -- THE WIKIPEDIA-EXCEED INLINE-CITATION GATE.
2//
3// Re-proves, from a REAL run (no fabricated GREEN), the anti-pushout
4// differentiator: a cited supporting source is rendered INLINE in the wiki
5// (read in place, no leaving for a paywalled/rotted/dead link) -- but ONLY when
6// licensing permits hosting it. Judged on SUBSTRING presence/absence in the
7// actually-rendered HTML (ground truth), never a bare rc==const.
8//
9//   inline_ok          A CC-BY source S1 ("ALPHA-SUPPORT-EVIDENCE-12345"),
10//                      archived + license-tagged, rendered via [[cite:<cid>]]
11//                      -> the HTML CONTAINS the evidence string INSIDE a
12//                      <details ... nx-cite ...> block (inline reading works).
13//   license_respected  THE LIAR-KILL: a PROPRIETARY source S2
14//                      ("PROPRIETARY-SECRET-67890"), archived but NOT hostable,
15//                      cited -> the HTML CONTAINS a link/license-note for it but
16//                      does NOT CONTAIN the secret bytes ANYWHERE. Unlicensed
17//                      content is never inlined -- respected, not merely labeled.
18//   pending_ok         A random NON-archived cid cited -> the HTML CONTAINS
19//                      "(source archiving pending)".
20//
21// EXCEED vs Wikipedia: Wikipedia pushes you OUT to a citation that may be dead.
22// Here the supporting source is read IN the page, pulled from the no-link-rot
23// content-addressed archive, gated by a sealed license layer so we never host
24// what we are not licensed to host.
25//
26// Determinism: a FRESH store prefix keyed by the wall-clock epoch => an empty
27// archive each run (idempotent re-run, Rule 10). Production archive prefix is
28// knowledge/store/wikiarchive- (in nx_wiki_archive).
29//
30// Verdict line (judged by this marker; stdout + knowledge/status):
31//   WIKICITE inline_ok=<0|1> license_respected=<0|1> pending_ok=<0|1>
32//     verdict=GREEN|RED
33//
34// Pure NishiLang, NO SQL, NO .sh/.py/.js, no new .tsv/.conf. nx_sites_daemon
35// UNTOUCHED. Imports the cite-render organ under test (which transitively pulls
36// the license layer + archive + seg_store + canon_cid, each spliced ONCE).
37// license_tier: ORIGINAL
38import "nx_syscalls.nx"
39import "nx_wiki_cite_render.nx"
40// Served-path handler under test: drives the REAL post-markdown render so the
41// inline-card check is judged on the SERVED bytes (closing the measurement
42// artifact -- the old gate only checked the PRE-markdown cite-render buffer,
43// which could pass while the served card was html-escaped). The resolver
44// path-dedups the shared cite/license/archive chain (no rc6 double-import).
45import "wiki/nx_wiki_index_builder.nx"
46import "wiki/nx_wiki_doc_handler.nx"
47import "wiki/nx_artifact_store.nx"
48
49const WCG_LOG: *u8 = "knowledge/status/wiki_cite_gate.log"
50const WCG_RESP_CAP: i64 = 2097152   // 2 MiB served-response buffer
51const WCG_CAP: i64 = 64           // manifest segment-scan cap (>> our few commits)
52const WCG_OUTCAP: i64 = 1048576   // rendered-HTML output buffer (1 MiB headroom)
53
54// ---- io helpers (mirror _wiki_archive_gate.nx) ----
55func wcg_w(fd: i64, s: *u8) -> i64 {
  var n: i64 = 0
  while s[n] != (0 as u8) { n = n + 1 }
  sys_write(fd, s, n)
  return 0
60}
61func wcg_n(fd: i64, v: i64) -> i64 {
  let bb: *u8 = sys_mmap(28)
  var m: i64 = v
  if m < 0 { m = 0 - m; sys_write(fd, "-" as *u8, 1) }
  let t: *u8 = sys_mmap(28)
  var k: i64 = 0
  if m == 0 { t[0] = 48 as u8; k = 1 }
  while m > 0 { t[k] = (48 + (m % 10)) as u8; m = m / 10; k = k + 1 }
  var i: i64 = 0
  while i < k { bb[i] = t[k - 1 - i]; i = i + 1 }
  sys_write(fd, bb, k)
  return 0
73}
74// write to BOTH stdout (gate evidence) and the durable log
75func wcg_w2(lfd: i64, s: *u8) -> i64 { wcg_w(1, s); if lfd >= 0 { wcg_w(lfd, s) } return 0 }
76func wcg_n2(lfd: i64, v: i64) -> i64 { wcg_n(1, v); if lfd >= 0 { wcg_n(lfd, v) } return 0 }
77
78func wcg_p(s: *u8) -> i64 { wcg_w(1, s); return 0 }
79func wcg_pn(v: i64) -> i64 { wcg_n(1, v); return 0 }
80func wcg_slen(s: *u8) -> i64 { var n: i64 = 0; while s[n] != (0 as u8) { n = n + 1 } return n }
81
82// concat NUL-terminated s into dst at off; returns new off
83func wcg_cat(dst: *u8, off: i64, s: *u8) -> i64 {
  var i: i64 = 0
  while s[i] != (0 as u8) { dst[off + i] = s[i]; i = i + 1 }
  return off + i
87}
88// append decimal of v into dst at off; returns new off
89func wcg_catn(dst: *u8, off: i64, v: i64) -> i64 {
  var m: i64 = v
  var o: i64 = off
  if m < 0 { dst[o] = 45 as u8; o = o + 1; m = 0 - m }
  let t: *u8 = sys_mmap(28)
  var k: i64 = 0
  if m == 0 { t[0] = 48 as u8; k = 1 }
  while m > 0 { t[k] = (48 + (m % 10)) as u8; m = m / 10; k = k + 1 }
  var i: i64 = 0
  while i < k { dst[o + i] = t[k - 1 - i]; i = i + 1 }
  return o + k
100}
101
102// ===== GROUND-TRUTH SUBSTRING SEARCH ==========================================
103// Does buffer hay[0..hn] contain the NUL-terminated needle as a contiguous
104// substring? Returns the start index (>=0) or -1 if absent. THIS is how the
105// gate judges -- on what is actually in the rendered HTML, not a return code.
106func wcg_find(hay: *u8, hn: i64, needle: *u8) -> i64 {
  let nn: i64 = wcg_slen(needle)
  if nn == 0 { return 0 - 1 }
  if nn > hn { return 0 - 1 }
  var i: i64 = 0
  let last: i64 = hn - nn
  while i <= last {
      var j: i64 = 0
      var eq: i64 = 1
      while j < nn {
          if eq == 1 { if hay[i + j] != needle[j] { eq = 0 } }
          j = j + 1
      }
      if eq == 1 { return i }
      i = i + 1
  }
  return 0 - 1
123}
124// boolean contains
125func wcg_has(hay: *u8, hn: i64, needle: *u8) -> i64 {
  if wcg_find(hay, hn, needle) >= 0 { return 1 }
  return 0
128}
129
130func main() -> i64 {
  wcg_p("WIKICITE-GATE: start (inline supporting source, license-aware anti-pushout)\n" as *u8)
132
  // ---- FRESH per-run store prefix => empty archive each run (deterministic) ----
  let prefix: *u8 = sys_mmap(256)
  var po: i64 = 0
  po = wcg_cat(prefix, po, "knowledge/store/wikiarchive-citegate-" as *u8)
  po = wcg_catn(prefix, po, sys_now_realtime_sec())
  po = wcg_cat(prefix, po, "-" as *u8)
  prefix[po] = 0 as u8
  wcg_p("  store prefix = " as *u8); wcg_p(prefix); wcg_p("\n" as *u8)
141
  // shared rendered-HTML output buffer + a "used length" cell
  let out: *u8 = sys_mmap(WCG_OUTCAP)
  let used: *i64 = sys_mmap(16) as *i64
145
  // distinctive content strings the gate searches for in the HTML
  let evidence: *u8 = "ALPHA-SUPPORT-EVIDENCE-12345" as *u8
  let secret:   *u8 = "PROPRIETARY-SECRET-67890" as *u8
149
  // ===== STEP 1: archive S1 (CC-BY) + render a page citing it -> INLINE ======
  // S1 is a hostable supporting source; its evidence MUST appear inline inside
  // a <details ... nx-cite> card.
  let s1_slug: *u8 = "src-s1" as *u8
  let s1_body: *u8 = "Supporting evidence: ALPHA-SUPPORT-EVIDENCE-12345 (full text of the cited source).\n" as *u8
  let cid_s1: *u8 = sys_mmap(80)
  let w1: *i64 = ss_begin()
  let r1: i64 = war_archive_page(w1, s1_slug, s1_body, wcg_slen(s1_body), cid_s1)
  let c1: i64 = ss_commit(prefix, w1, 0)
  wcg_p("  S1 archived rc=" as *u8); wcg_pn(r1); wcg_p(" commit=" as *u8); wcg_pn(c1); wcg_p(" cid=" as *u8); wcg_p(cid_s1); wcg_p("\n" as *u8)
  // tag S1 as CC-BY (hostable, attribution required)
  let lset1: i64 = nx_wiki_license_set(prefix, cid_s1, NXLIC_CC_BY, 1)
  wcg_p("  S1 license set CC_BY rc=" as *u8); wcg_pn(lset1); wcg_p(" (get=" as *u8); wcg_pn(nx_wiki_license_get(prefix, cid_s1)); wcg_p(")\n" as *u8)
163
  // build a page body:  "... [[cite:<cid_s1>]] ..."
  let pg1: *u8 = sys_mmap(512)
  var g1: i64 = 0
  g1 = wcg_cat(pg1, g1, "See the supporting source here: [[cite:" as *u8)
  g1 = wcg_cat(pg1, g1, cid_s1)
  g1 = wcg_cat(pg1, g1, "]] -- read it in place.\n" as *u8)
  pg1[g1] = 0 as u8
171
  let rr1: i64 = nx_wiki_cite_render(prefix, pg1, g1, out, WCG_OUTCAP, used)
  let hn1: i64 = used[0]
  wcg_p("  render1 rc=" as *u8); wcg_pn(rr1); wcg_p(" html_len=" as *u8); wcg_pn(hn1); wcg_p("\n" as *u8)
175
  // JUDGE inline_ok on GROUND TRUTH, on BOTH the organ output AND the SERVED
  // (post-markdown) bytes. (1) ORGAN: the cite-render organ emits a <details
  // nx-cite> card with the evidence inside. (2) SERVED (the artifact-closing
  // check): render the SAME [[cite:<cid>]] page through the REAL handler
  // (nx_wiki_doc_handle_cp, cite-prefix-injected) and require the served card be
  // LIVE -- "<details class=\"nx-cite\"" present, "&lt;details" ABSENT, evidence
  // inside -- so inline_ok can NO LONGER pass while the served card is escaped.
  var organ_inline: i64 = 0
  if rr1 == NXCITE_OK {
      let det_at: i64 = wcg_find(out, hn1, "<details class=\"nx-cite\">" as *u8)
      let ev_at: i64 = wcg_find(out, hn1, evidence)
      let body_at: i64 = wcg_find(out, hn1, "nx-cite-body" as *u8)
      if det_at >= 0 { if ev_at > det_at { if body_at >= 0 { if ev_at > body_at { organ_inline = 1 } } } }
      wcg_p("  inline(organ): details@" as *u8); wcg_pn(det_at); wcg_p(" body@" as *u8); wcg_pn(body_at); wcg_p(" evidence@" as *u8); wcg_pn(ev_at); wcg_p("\n" as *u8)
  }
  // SERVED: seed a doc store with the cited page, render via the real handler.
  let cstore: *NxWikiDocStore = sys_mmap(2048) as *NxWikiDocStore
  let cs_rc: i64 = nx_wiki_doc_store_init(cstore, 16, 4096, 4096, 65536)
  let cpg: *u8 = sys_mmap(512)
  var cg: i64 = 0
  cg = wcg_cat(cpg, cg, "# Cite Gate Page\n\nSee the supporting source here: [[cite:" as *u8)
  cg = wcg_cat(cpg, cg, cid_s1)
  cg = wcg_cat(cpg, cg, "]] -- read it in place.\n" as *u8)
  cpg[cg] = 0 as u8
  let cpg_rid: i64 = nx_wiki_doc_store_add(cstore,
      "Cite Gate Page" as *u8, 14, "/wiki/cite-gate-pg" as *u8, 18, cpg, cg)
  let sresp: *u8 = sys_mmap(WCG_RESP_CAP)
  let sresp_n: *i64 = sys_mmap(8) as *i64
  sresp_n[0] = 0
  let sh_rc: i64 = nx_wiki_doc_handle_cp(cstore, 0 as *NxArtifactStore,
      "/wiki/cite-gate-pg" as *u8, 18, sresp, WCG_RESP_CAP, sresp_n, prefix)
  let sn: i64 = sresp_n[0]
  var served_inline: i64 = 0
  if sh_rc == NX_WDH_OK {
      let sdet_at: i64 = wcg_find(sresp, sn, "<details class=\"nx-cite\"" as *u8)
      let sesc_at: i64 = wcg_find(sresp, sn, "&lt;details" as *u8)   // MUST be absent
      var sev_at: i64 = 0 - 1
      if sdet_at >= 0 {
          let stail: *u8 = (sresp as i64 + sdet_at) as *u8
          let sre: i64 = wcg_find(stail, sn - sdet_at, evidence)
          if sre >= 0 { sev_at = sdet_at + sre }
      }
      if sdet_at >= 0 { if sesc_at < 0 { if sev_at > sdet_at { served_inline = 1 } } }
      wcg_p("  inline(SERVED): handle_rc=" as *u8); wcg_pn(sh_rc); wcg_p(" rid=" as *u8); wcg_pn(cpg_rid); wcg_p(" resp_len=" as *u8); wcg_pn(sn); wcg_p(" details@" as *u8); wcg_pn(sdet_at); wcg_p(" &lt;details@(should be -1)=" as *u8); wcg_pn(sesc_at); wcg_p(" evidence@" as *u8); wcg_pn(sev_at); wcg_p("\n" as *u8)
  }
  var inline_ok: i64 = 0
  if organ_inline == 1 { if served_inline == 1 { inline_ok = 1 } }
  wcg_p("  inline_ok = " as *u8); wcg_pn(inline_ok); wcg_p(" (organ=" as *u8); wcg_pn(organ_inline); wcg_p(" served=" as *u8); wcg_pn(served_inline); wcg_p(")\n" as *u8)
224
  // ===== STEP 2: archive S2 (PROPRIETARY) + cite it -> LINK ONLY, NO BYTES ====
  // THE LIAR-KILL. S2 is archived (so the bytes ARE in the store) but NOT
  // hostable; the render MUST emit a link/license-note and MUST NOT leak the
  // secret bytes anywhere in the HTML.
  let s2_slug: *u8 = "src-s2" as *u8
  let s2_body: *u8 = "Confidential: PROPRIETARY-SECRET-67890 must never be inlined without a license.\n" as *u8
  let cid_s2: *u8 = sys_mmap(80)
  let w2: *i64 = ss_begin()
  let r2: i64 = war_archive_page(w2, s2_slug, s2_body, wcg_slen(s2_body), cid_s2)
  let c2: i64 = ss_commit(prefix, w2, 2)
  wcg_p("  S2 archived rc=" as *u8); wcg_pn(r2); wcg_p(" commit=" as *u8); wcg_pn(c2); wcg_p(" cid=" as *u8); wcg_p(cid_s2); wcg_p("\n" as *u8)
  // tag S2 PROPRIETARY (NOT hostable)
  let lset2: i64 = nx_wiki_license_set(prefix, cid_s2, NXLIC_PROPRIETARY, 3)
  wcg_p("  S2 license set PROPRIETARY rc=" as *u8); wcg_pn(lset2); wcg_p(" (get=" as *u8); wcg_pn(nx_wiki_license_get(prefix, cid_s2)); wcg_p(" hostable=" as *u8); wcg_pn(nx_wiki_license_hostable(nx_wiki_license_get(prefix, cid_s2))); wcg_p(")\n" as *u8)
239
  // SANITY: prove the secret bytes ARE retrievable from the archive (so the
  // absence-from-HTML below is a real suppression, not an archiving miss).
  let pps: *i64 = sys_mmap(16) as *i64
  let s2got: i64 = war_get_by_cid(prefix, cid_s2, pps, WCG_CAP)
  var s2_in_store: i64 = 0
  if s2got > 0 { if wcg_has(pps[0] as *u8, s2got, secret) == 1 { s2_in_store = 1 } }
  wcg_p("  S2 secret-in-store = " as *u8); wcg_pn(s2_in_store); wcg_p(" (archive readback len=" as *u8); wcg_pn(s2got); wcg_p(")\n" as *u8)
247
  let pg2: *u8 = sys_mmap(512)
  var g2: i64 = 0
  g2 = wcg_cat(pg2, g2, "Per the proprietary source [[cite:" as *u8)
  g2 = wcg_cat(pg2, g2, cid_s2)
  g2 = wcg_cat(pg2, g2, "]], details are restricted.\n" as *u8)
  pg2[g2] = 0 as u8
254
  let rr2: i64 = nx_wiki_cite_render(prefix, pg2, g2, out, WCG_OUTCAP, used)
  let hn2: i64 = used[0]
  wcg_p("  render2 rc=" as *u8); wcg_pn(rr2); wcg_p(" html_len=" as *u8); wcg_pn(hn2); wcg_p("\n" as *u8)
258
  // JUDGE license_respected (THE LIAR-KILL) on GROUND TRUTH, on BOTH the organ
  // output AND the SERVED bytes. (1) ORGAN: link/license-note present + secret
  // ABSENT in the cite-render output. (2) SERVED (artifact-closing): render an
  // S2-citing page through the REAL handler and require the secret to be ABSENT
  // from the served HTML while a link + note are present -- so a served leak can
  // no longer slip past while the organ-only check is green.
  var organ_license: i64 = 0
  if rr2 == NXCITE_OK {
      let link_at: i64 = wcg_find(out, hn2, "nx-cite-link" as *u8)
      let note_at: i64 = wcg_find(out, hn2, "licensing restricts inline hosting" as *u8)
      let leak_at: i64 = wcg_find(out, hn2, secret)
      if link_at >= 0 { if note_at >= 0 { if leak_at < 0 { organ_license = 1 } } }
      wcg_p("  license(organ): link@" as *u8); wcg_pn(link_at); wcg_p(" note@" as *u8); wcg_pn(note_at); wcg_p(" SECRET_LEAK@" as *u8); wcg_pn(leak_at); wcg_p(" (-1=absent=GOOD)\n" as *u8)
  }
  // SERVED liar-kill: add an S2-citing page to the doc store, render via handler.
  let cpg2: *u8 = sys_mmap(512)
  var cg2: i64 = 0
  cg2 = wcg_cat(cpg2, cg2, "# Proprietary Cite Page\n\nPer the proprietary source [[cite:" as *u8)
  cg2 = wcg_cat(cpg2, cg2, cid_s2)
  cg2 = wcg_cat(cpg2, cg2, "]], details are restricted.\n" as *u8)
  cpg2[cg2] = 0 as u8
  let cpg2_rid: i64 = nx_wiki_doc_store_add(cstore,
      "Proprietary Cite Page" as *u8, 21, "/wiki/cite-gate-prop" as *u8, 20, cpg2, cg2)
  let sresp2: *u8 = sys_mmap(WCG_RESP_CAP)
  let sresp2_n: *i64 = sys_mmap(8) as *i64
  sresp2_n[0] = 0
  let sh2_rc: i64 = nx_wiki_doc_handle_cp(cstore, 0 as *NxArtifactStore,
      "/wiki/cite-gate-prop" as *u8, 20, sresp2, WCG_RESP_CAP, sresp2_n, prefix)
  let sn2: i64 = sresp2_n[0]
  var served_license: i64 = 0
  if sh2_rc == NX_WDH_OK {
      let slink_at: i64 = wcg_find(sresp2, sn2, "nx-cite-link" as *u8)
      let snote_at: i64 = wcg_find(sresp2, sn2, "licensing restricts inline hosting" as *u8)
      let sleak_at: i64 = wcg_find(sresp2, sn2, secret)
      if slink_at >= 0 { if snote_at >= 0 { if sleak_at < 0 { served_license = 1 } } }
      wcg_p("  license(SERVED): handle_rc=" as *u8); wcg_pn(sh2_rc); wcg_p(" rid=" as *u8); wcg_pn(cpg2_rid); wcg_p(" resp_len=" as *u8); wcg_pn(sn2); wcg_p(" link@" as *u8); wcg_pn(slink_at); wcg_p(" note@" as *u8); wcg_pn(snote_at); wcg_p(" SECRET_LEAK@" as *u8); wcg_pn(sleak_at); wcg_p(" (-1=absent=GOOD)\n" as *u8)
  }
  var license_respected: i64 = 0
  if organ_license == 1 { if served_license == 1 { license_respected = 1 } }
  wcg_p("  license_respected = " as *u8); wcg_pn(license_respected); wcg_p(" (organ=" as *u8); wcg_pn(organ_license); wcg_p(" served=" as *u8); wcg_pn(served_license); wcg_p(")\n" as *u8)
299
  // ===== STEP 3: cite a NON-archived cid -> pending placeholder ==============
  let cid_none: *u8 = "nxc1-deadbeefdeadbeefdeadbeefdeadbeefdeadbeefdeadbeefdeadbeefdeadbeef" as *u8
  let pg3: *u8 = sys_mmap(512)
  var g3: i64 = 0
  g3 = wcg_cat(pg3, g3, "An as-yet-unarchived claim [[cite:" as *u8)
  g3 = wcg_cat(pg3, g3, cid_none)
  g3 = wcg_cat(pg3, g3, "]] is noted.\n" as *u8)
  pg3[g3] = 0 as u8
308
  let rr3: i64 = nx_wiki_cite_render(prefix, pg3, g3, out, WCG_OUTCAP, used)
  let hn3: i64 = used[0]
  var pending_ok: i64 = 0
  if rr3 == NXCITE_OK {
      if wcg_has(out, hn3, "(source archiving pending)" as *u8) == 1 { pending_ok = 1 }
  }
  wcg_p("  render3 rc=" as *u8); wcg_pn(rr3); wcg_p(" html_len=" as *u8); wcg_pn(hn3); wcg_p(" pending_ok=" as *u8); wcg_pn(pending_ok); wcg_p("\n" as *u8)
316
  // ===== EVIDENCE: print the inline card HTML + a CSS sample =================
  // (re-render S1 so `out` holds the inline card, then echo it for judging.)
  let rr1b: i64 = nx_wiki_cite_render(prefix, pg1, g1, out, WCG_OUTCAP, used)
  if rr1b == NXCITE_OK {
      wcg_p("  ---- rendered inline card (S1, CC-BY) ----\n" as *u8)
      sys_write(1, out, used[0])
      wcg_p("\n  ---- /card ----\n" as *u8)
  }
  let cssbuf: *u8 = sys_mmap(4096)
  let cssn: i64 = nx_wiki_cite_css(cssbuf, 4096)
  if cssn > 0 {
      wcg_p("  ---- sovereign-emitted card CSS ----\n" as *u8)
      sys_write(1, cssbuf, cssn)
      wcg_p("  ---- /CSS ----\n" as *u8)
  }
332
  // ===== VERDICT =====
  var green: i64 = 1
  if inline_ok != 1 { green = 0 }
  if license_respected != 1 { green = 0 }
  if pending_ok != 1 { green = 0 }
  // belt-and-suspenders: the secret must truly have been in the store, so the
  // suppression we measured was real (not an archiving miss masquerading).
  if s2_in_store != 1 { green = 0 }
341
  let lfd: i64 = sys_openat_append(WCG_LOG, 420)
  wcg_w2(lfd, "WIKICITE inline_ok=" as *u8);          wcg_n2(lfd, inline_ok)
  wcg_w2(lfd, " license_respected=" as *u8);          wcg_n2(lfd, license_respected)
  wcg_w2(lfd, " pending_ok=" as *u8);                 wcg_n2(lfd, pending_ok)
  if green == 1 { wcg_w2(lfd, " verdict=GREEN\n" as *u8) } else { wcg_w2(lfd, " verdict=RED\n" as *u8) }
  if lfd >= 0 { sys_close(lfd) }
348
  if green == 1 {
      sys_exit(0)
      return 0
  }
  sys_exit(1)
  return 1
355}
_wiki_cite_gate.nx source

`_wiki_cite_gate.nx` source
