code wiki / (root) / nx_block_density_gate.nx
`nx_block_density_gate.nx` source

↩ module page · 159 lines · 10650 B
1// nx_block_density_gate.nx -- GATE for nx_block_density (the /compare/webscraping R6 contract bd_fit_text).
2// Network-free; fixtures are assembled at runtime. The accept rule pre-declared on the board: on the
3// boilerplate fixture every article sentence survives and the nav, link-farm, aside and footer blocks are
4// dropped; a page with no boilerplate comes back BYTE-IDENTICAL to the plain renderer; a page whose only
5// content block is boilerplate falls back to the full text instead of returning nothing.
6// license_tier: ORIGINAL
7import "nx_syscalls.nx"
8import "nx_block_density.nx"
9import "nx_gate_verdict.nx"
10
11const BG_CAP: i64 = 65536
12
13func bg_len(s: *u8) -> i64 { var n: i64 = 0; while s[n] != (0 as u8) { n = n + 1 } return n }
14// 1 when needle occurs in hay[0..n)
15func bg_has(hay: *u8, n: i64, needle: *u8) -> i64 {
  let nl: i64 = bg_len(needle)
  if nl == 0 { return 1 }
  var i: i64 = 0
  while i + nl <= n {
      var k: i64 = 0
      var m: i64 = 1
      while k < nl { if hay[i + k] != needle[k] { m = 0; k = nl } else { k = k + 1 } }
      if m == 1 { return 1 }
      i = i + 1
  }
  return 0
27}
28func bg_eq(a: *u8, an: i64, b: *u8, bn: i64) -> i64 {
  if an != bn { return 0 }
  var i: i64 = 0
  while i < an { if a[i] != b[i] { return 0 } i = i + 1 }
  return 1
33}
34func bg_cat(d: *u8, o: i64, s: *u8) -> i64 { var i: i64 = 0; var p: i64 = o; while s[i] != (0 as u8) { d[p] = s[i]; p = p + 1; i = i + 1 } d[p] = 0 as u8; return p }
35
36func main() -> i64 {
  gv_head("=== nx_block_density gate (Kohlschuetter block features, fit text, network-free) ===" as *u8)
  let ctr: *i64 = gv_ctr()
  bd_conf_reset()
  let src: i64 = bd_load_conf()
  gv_puts("  conf_src=" as *u8)
  if src == 1 { gv_puts("file" as *u8) } else { gv_puts("defaults" as *u8) }
  gv_puts(" ld_curr_max=" as *u8); gv_num(bd_ld_curr_max_g); gv_puts(" ld_prev_max=" as *u8); gv_num(bd_ld_prev_max_g)
  gv_puts(" words_curr_short=" as *u8); gv_num(bd_words_curr_short_g); gv_puts(" words_next_short=" as *u8); gv_num(bd_words_next_short_g)
  gv_puts(" words_prev_short=" as *u8); gv_num(bd_words_prev_short_g); gv_puts(" words_curr_long=" as *u8); gv_num(bd_words_curr_long_g)
  gv_puts(" words_next_long=" as *u8); gv_num(bd_words_next_long_g); gv_puts("\n" as *u8)
47
  // ---- T1 the published rule, explicit thresholds (conf-independent): every leaf of the tree ----
  var t1: i64 = 1
  if bd_classify_t(10, 0, 30, 500, 30, 333, 556, 16, 15, 4, 40, 17) != 0 { t1 = 0 }   // curr linky -> boilerplate, whatever else
  if bd_classify_t(3, 0, 10, 0, 10, 333, 556, 16, 15, 4, 40, 17) != 0 { t1 = 0 }      // short, short next, short prev -> boilerplate
  if bd_classify_t(10, 0, 10, 0, 10, 333, 556, 16, 15, 4, 40, 17) != 1 { t1 = 0 }     // short, short next, but prev has words -> content
  if bd_classify_t(0, 0, 10, 0, 20, 333, 556, 16, 15, 4, 40, 17) != 1 { t1 = 0 }      // short but next is long -> content
  if bd_classify_t(0, 0, 20, 0, 0, 333, 556, 16, 15, 4, 40, 17) != 1 { t1 = 0 }       // long -> content
  if bd_classify_t(5, 700, 30, 0, 10, 333, 556, 16, 15, 4, 40, 17) != 0 { t1 = 0 }    // prev linky, curr <= 40, next short -> boilerplate
  if bd_classify_t(5, 700, 30, 0, 20, 333, 556, 16, 15, 4, 40, 17) != 1 { t1 = 0 }    // prev linky, next long -> content
  if bd_classify_t(5, 700, 50, 0, 0, 333, 556, 16, 15, 4, 40, 17) != 1 { t1 = 0 }     // prev linky, curr > 40 -> content
  if bd_link_density(20, 5) != 250 { t1 = 0 }
  if bd_link_density(0, 5) != 0 { t1 = 0 }
  if bd_link_density(4, 9) != 1000 { t1 = 0 }                                          // clamps, never exceeds 1000
  gv_check("T1 the Kohlschuetter rule: all 8 leaves and word link density (11 KATs)" as *u8, t1, ctr)
62
  // ---- T2 bite: fires (boilerplate) on a link menu, silent on a prose paragraph ----
  var bad: i64 = 0
  if bd_classify(0, 0, 4, 1000, 19) == 0 { bad = 1 }
  var good: i64 = 0
  if bd_classify(0, 0, 25, 80, 22) == 0 { good = 1 }
  gv_bite("T2 the rule fires on a 4-word all-link menu and stays silent on a 25-word paragraph" as *u8, bad, good, ctr)
69
  // ---- fixture A: the boilerplate page ----
  let a: *u8 = sys_mmap(BG_CAP)
  var o: i64 = 0
  o = bg_cat(a, o, "<html><head><title>T</title><style>.x{color:red}</style><script>var a=1; alertword('scriptonly');</script></head><body>\n" as *u8)
  o = bg_cat(a, o, "<nav><ul><li><a href=\"/\">Home</a></li><li><a href=\"/news\">News</a></li><li><a href=\"/about\">About</a></li></ul></nav>\n" as *u8)
  o = bg_cat(a, o, "<div class=\"menu\"><a href=\"/a\">Alpha</a> <a href=\"/b\">Beta</a> <a href=\"/c\">Gamma</a> <a href=\"/d\">Delta</a></div>\n" as *u8)
  o = bg_cat(a, o, "<article><h1>Churning temperature and butter texture</h1>\n" as *u8)
  o = bg_cat(a, o, "<p>Butter churned at a lower temperature forms smaller fat globules, and the resulting texture is firmer at room temperature.</p>\n" as *u8)
  o = bg_cat(a, o, "<p>Updated 2026</p>\n" as *u8)
  o = bg_cat(a, o, "<p>The rheology study found no difference as a function of churning temperature once the cream had been aged for twelve hours.</p>\n" as *u8)
  o = bg_cat(a, o, "</article>\n<aside><p>Related: <a href=\"/x\">xlink</a> <a href=\"/y\">ylink</a></p></aside>\n" as *u8)
  o = bg_cat(a, o, "<footer><p>Copyright 2026 Example Inc. <a href=\"/privacy\">Privacy</a> <a href=\"/terms\">Terms</a></p></footer>\n</body></html>\n" as *u8)
  let an: i64 = o
  let fit: *u8 = sys_mmap(BG_CAP)
  let fn: i64 = bd_fit_text(a, an, fit, BG_CAP)
  gv_puts("  fixtureA blocks=" as *u8); gv_num(bd_last_blocks_g); gv_puts(" kept=" as *u8); gv_num(bd_last_kept_g)
  gv_puts(" semantic=" as *u8); gv_num(bd_last_semantic_g)
  gv_puts(" structural=" as *u8); gv_num(bd_last_structural_g); gv_puts(" dropped=" as *u8); gv_num(bd_last_dropped_g)
  gv_puts(" fallback=" as *u8); gv_num(bd_last_fallback_g); gv_puts(" fit_bytes=" as *u8); gv_num(fn); gv_puts("\n" as *u8)
  gv_subjects("fixtureA-blocks" as *u8, bd_last_blocks_g, ctr)
  var t3: i64 = 1
  if bg_has(fit, fn, "smaller fat globules" as *u8) == 0 { t3 = 0 }
  if bg_has(fit, fn, "no difference as a function of churning temperature" as *u8) == 0 { t3 = 0 }
  if bg_has(fit, fn, "Churning temperature and butter texture" as *u8) == 0 { t3 = 0 }
  gv_check("T3 every article sentence and the title survive the fit" as *u8, t3, ctr)
  var t4: i64 = 1
  if bg_has(fit, fn, "Alpha" as *u8) == 1 { t4 = 0 }
  if bg_has(fit, fn, "Gamma" as *u8) == 1 { t4 = 0 }
  gv_check("T4 the link-farm menu div is dropped BY SCORE (no semantic tag to lean on)" as *u8, t4, ctr)
  var t5: i64 = 1
  if bg_has(fit, fn, "Home" as *u8) == 1 { t5 = 0 }
  if bg_has(fit, fn, "Privacy" as *u8) == 1 { t5 = 0 }
  if bg_has(fit, fn, "xlink" as *u8) == 1 { t5 = 0 }
  gv_check("T5 nav, footer and aside content is dropped by the semantic rows" as *u8, t5, ctr)
  var t6: i64 = 0
  if bg_has(fit, fn, "Updated 2026" as *u8) == 1 { t6 = 1 }
  gv_check("T6 a two-word block between two long paragraphs is CONTENT (the next-words context leaf)" as *u8, t6, ctr)
  var t7: i64 = 0
  if bg_has(fit, fn, "scriptonly" as *u8) == 0 { if bg_has(fit, fn, "color:red" as *u8) == 0 { t7 = 1 } }
  gv_check("T7 script and style bodies are never text" as *u8, t7, ctr)
  var t8: i64 = 0
  if bd_last_blocks_g == bd_last_kept_g + bd_last_semantic_g + bd_last_structural_g + bd_last_dropped_g { t8 = 1 }
  gv_check("T8 partition sums: blocks == kept + semantic + structural + dropped" as *u8, t8, ctr)
  var t8b: i64 = 0
  if bd_last_fallback_g == 0 { t8b = 1 }
  gv_check("T8b the boilerplate page did NOT need the fallback" as *u8, t8b, ctr)
  var t8c: i64 = 0
  if bd_last_fit_html_g < an { t8c = 1 }
  gv_check("T8c the rebuilt HTML is smaller than the page (something was actually removed)" as *u8, t8c, ctr)
119
  // ---- fixture B: NEGATIVE CONTROL, no boilerplate -> byte-identical to the plain renderer ----
  let b: *u8 = sys_mmap(BG_CAP)
  var ob: i64 = 0
  ob = bg_cat(b, ob, "<html><body><h1>Two plain paragraphs</h1><p>Alpha prose sentence one is here and it runs on long enough to count as a real paragraph of content for the rule.</p>" as *u8)
  ob = bg_cat(b, ob, "<p>Beta prose sentence two is here and it also runs on long enough to count as a real paragraph of content for the rule.</p></body></html>" as *u8)
  let full: *u8 = sys_mmap(BG_CAP)
  let fulln: i64 = nx_html_to_text(b, ob, full, BG_CAP)
  let fitb: *u8 = sys_mmap(BG_CAP)
  let fitbn: i64 = bd_fit_text(b, ob, fitb, BG_CAP)
  gv_puts("  fixtureB full_bytes=" as *u8); gv_num(fulln); gv_puts(" fit_bytes=" as *u8); gv_num(fitbn); gv_puts(" fallback=" as *u8); gv_num(bd_last_fallback_g); gv_puts("\n" as *u8)
  gv_check("T9 NEG-CONTROL: a page with no boilerplate is BYTE-IDENTICAL after the fit" as *u8, bg_eq(full, fulln, fitb, fitbn), ctr)
  var t9b: i64 = 0
  if bd_last_fallback_g == 0 { t9b = 1 }
  gv_check("T9b ...and it got there by keeping every block, not by the fallback" as *u8, t9b, ctr)
134
  // ---- fixture C: the only content block is boilerplate -> FALLBACK to full text, never empty ----
  let c: *u8 = sys_mmap(BG_CAP)
  var oc: i64 = 0
  oc = bg_cat(c, oc, "<html><body><div class=\"m\"><a href=\"/a\">One</a> <a href=\"/b\">Two</a> <a href=\"/c\">Three</a></div></body></html>" as *u8)
  let fitc: *u8 = sys_mmap(BG_CAP)
  let fitcn: i64 = bd_fit_text(c, oc, fitc, BG_CAP)
  var t10: i64 = 0
  if fitcn > 0 { if bd_last_fallback_g == 1 { if bg_has(fitc, fitcn, "Three" as *u8) == 1 { t10 = 1 } } }
  gv_check("T10 a page whose every content block scores as boilerplate falls back to the FULL text and says so" as *u8, t10, ctr)
144
  // ---- T11 empty input ----
  let fite: *u8 = sys_mmap(64)
  var t11: i64 = 0
  if bd_fit_text(c, 0, fite, 64) == 0 { t11 = 1 }
  gv_check("T11 empty input -> 0 bytes, no crash" as *u8, t11, ctr)
150
  // ---- T12 map verb: one line per block, header carries the partition ----
  let map: *u8 = sys_mmap(BG_CAP)
  let mn: i64 = bd_map(a, an, map, BG_CAP)
  var t12: i64 = 0
  if bg_has(map, mn, "blk 0 " as *u8) == 1 { if bg_has(map, mn, " kept=" as *u8) == 1 { if bg_has(map, mn, "conf_src=" as *u8) == 1 { t12 = 1 } } }
  gv_check("T12 map prints the per-block table with the conf provenance and the partition" as *u8, t12, ctr)
157
  return gv_verdict("BLOCK-DENSITY-GATE" as *u8, ctr, "Kohlschuetter block features as permil with conf rows; article kept, nav and farms dropped, no-boilerplate page byte-identical, fallback never empty" as *u8)
159}
nx_block_density_gate.nx source

`nx_block_density_gate.nx` source
