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

↩ module page · 151 lines · 8346 B
1// nx_mvault_fetch_gate.nx -- proves the album-ingest parsing rules OFFLINE.
2//
3// The parsing rules are where downloader bugs actually live, and every one of
4// them fails SILENTLY -- you get fewer files, not an error. So these teeth are
5// the ways a naive scraper quietly loses data:
6//   T5  a relative shape it does not understand must REFUSE, not guess a url
7//   T8  a too-small buffer must REFUSE, not return half an album as complete
8//   T9  <article> must not match tag "a" (the classic prefix collision)
9//   T11 a mirror domain must resolve to its family's adapter, not miss
10// license_tier: ORIGINAL
11import "nx_mvault_fetch.nx"
12import "nx_gate.nx"
13import "nx_gate_verdict.nx"   // D001: canonical verdict emission + actlog frame
14
15const FG_PFX: *u8 = "knowledge/mvfetch-test-\x00"
16
17func fg_eq(a: *u8, b: *u8) -> i64 {
  var i: i64 = 0
  while a[i] != (0 as u8) { if a[i] != b[i] { return 0 } i = i + 1 }
  if b[i] != (0 as u8) { return 0 }
  return 1
22}
23func fg_lines(b: *u8, n: i64) -> i64 { var c: i64=0; var i: i64=0; while i<n { if b[i]==(10 as u8) { c=c+1 } i=i+1 } return c }
24func fg_has(b: *u8, n: i64, s: *u8) -> i64 {
  let sl: i64 = mvf_strlen(s)
  if sl == 0 { return 1 }
  var i: i64 = 0
  while i + sl <= n {
      var m: i64 = 1
      var j: i64 = 0
      while j < sl { if b[i+j]!=s[j] { m=0; j=sl } else { j=j+1 } }
      if m == 1 { return 1 }
      i = i + 1
  }
  return 0
36}
37
38func main() -> i64 {
  gw("=== nx_mvault_fetch_gate: album ingest parsing, proven without a network ===\n" as *u8)
  var pass: i64=0; var tot: i64=0
41
  let buf: *u8 = sys_mmap(65536)
  let alb: *u8 = "https://bunkr.si/a/aXbYcZ" as *u8
44
  let h: *u8 = sys_mmap(512)
  mvf_host_of(alb, h)
  tot=tot+1; if fg_eq(h, "bunkr.si" as *u8)==1 { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) }
  gw("T1 host_of extracts the host\n" as *u8)
49
  let og: *u8 = sys_mmap(512)
  mvf_origin_of(alb, og)
  tot=tot+1; if fg_eq(og, "https://bunkr.si" as *u8)==1 { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) }
  gw("T2 origin_of keeps scheme+host and drops the path\n" as *u8)
54
  let r: *u8 = sys_mmap(MVF_URL_CAP)
  mvf_resolve(alb, "https://cdn.example/x.mp4" as *u8, r, MVF_URL_CAP)
  tot=tot+1; if fg_eq(r, "https://cdn.example/x.mp4" as *u8)==1 { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) }
  gw("T3 an absolute url passes through untouched\n" as *u8)
59
  mvf_resolve(alb, "/f/item1" as *u8, r, MVF_URL_CAP)
  tot=tot+1; if fg_eq(r, "https://bunkr.si/f/item1" as *u8)==1 { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) }
  gw("T4 a root-relative href resolves against the album origin\n" as *u8)
63
  let odd: i64 = mvf_resolve(alb, "../up/one" as *u8, r, MVF_URL_CAP)
  tot=tot+1; if odd==0 { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) }
  gw("T5 TOOTH an un-understood relative shape REFUSES rather than guessing a url\n" as *u8)
67
  // an album page shaped like the real thing: item links, plus noise that a
  // sloppy scanner would swallow (an <article>, an off-prefix link, a script url)
  let html: *u8 = sys_mmap(8192)
  var ho: i64 = 0
  ho = mvf_cat(html, ho, "<html><body><article class=wrap>\n" as *u8)
  ho = mvf_cat(html, ho, "<a href=\"/f/one.mp4\">one</a>\n" as *u8)
  ho = mvf_cat(html, ho, "<a class=x href='/f/two.jpg'>two</a>\n" as *u8)
  ho = mvf_cat(html, ho, "<a href=\"/about\">about us</a>\n" as *u8)
  ho = mvf_cat(html, ho, "<a href=\"/f/three.webm\">three</a>\n" as *u8)
  ho = mvf_cat(html, ho, "</article></body></html>\n" as *u8)
78
  let n1: i64 = mvf_extract(html, ho, "a" as *u8, "href" as *u8, "/f/" as *u8, buf, 65536)
  tot=tot+1; if fg_lines(buf, n1)==3 { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) }
  gw("T6 extract pulls exactly the three /f/ item links (got " as *u8); gn(fg_lines(buf,n1)); gw(")\n" as *u8)
82
  tot=tot+1; if fg_has(buf, n1, "/about" as *u8)==0 { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) }
  gw("T7 NEG-CONTROL the off-prefix /about link is excluded\n" as *u8)
85
  tot=tot+1; if fg_has(buf, n1, "/f/two.jpg" as *u8)==1 { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) }
  gw("T8 SINGLE-quoted attribute values parse too\n" as *u8)
88
  let tight: i64 = mvf_extract(html, ho, "a" as *u8, "href" as *u8, "/f/" as *u8, buf, 8)
  tot=tot+1; if tight==(0-1) { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) }
  gw("T9 TOOTH a too-small buffer REFUSES (-1) -- never a half album reported whole\n" as *u8)
92
  // <article> must NOT be matched by tag "a": the tag name has to end at a
  // separator. A scanner without this check silently harvests junk.
  let only: *u8 = sys_mmap(1024)
  let oo: i64 = mvf_cat(only, 0, "<article href=\"/f/nope\"></article>" as *u8)
  let n2: i64 = mvf_extract(only, oo, "a" as *u8, "href" as *u8, "/f/" as *u8, buf, 65536)
  tot=tot+1; if n2==0 { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) }
  gw("T10 TOOTH <article> is NOT matched by tag 'a' (prefix collision)\n" as *u8)
100
  // adapters: one row covers the mirror family
  mvf_adapter_put_pfx(FG_PFX, "bunkr.si" as *u8, "a" as *u8, "href" as *u8, "/f/" as *u8, "source" as *u8, "src" as *u8, "nishi" as *u8)
  let po: *i64 = sys_mmap(16) as *i64
  let lo: *i64 = sys_mmap(16) as *i64
  let hit: i64 = mvf_adapter_find_pfx(FG_PFX, "img.cdn.bunkr.si" as *u8, po, lo)
  tot=tot+1; if hit==1 { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) }
  gw("T11 TOOTH a mirror host img.cdn.bunkr.si resolves to the bunkr.si adapter\n" as *u8)
108
  let miss: i64 = mvf_adapter_find_pfx(FG_PFX, "unknownhost.example" as *u8, po, lo)
  tot=tot+1; if miss==0 { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) }
  gw("T12 an unknown host has NO adapter -- refused, never guessed\n" as *u8)
112
  // ---- the destination filename is chosen by a REMOTE SERVER ----
  // Untrusted input at a boundary (rule 12): a crafted name must never be able
  // to write outside the destination directory.
  let bn: *u8 = sys_mmap(1024)
  let b1: i64 = mvf_basename("https://cdn.example/a/b/clip.mp4" as *u8, bn, 1024)
  tot=tot+1; if b1 > 0 { if fg_eq(bn, "clip.mp4" as *u8)==1 { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) } } else { gw("  [FAIL] " as *u8) }
  gw("T13 basename takes the last path segment\n" as *u8)
120
  let b2: i64 = mvf_basename("https://cdn.example/a/pic.jpg?sig=abc123" as *u8, bn, 1024)
  tot=tot+1; if b2 > 0 { if fg_eq(bn, "pic.jpg" as *u8)==1 { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) } } else { gw("  [FAIL] " as *u8) }
  gw("T14 a signed-url ?query is stripped from the filename\n" as *u8)
124
  let b3: i64 = mvf_basename("https://cdn.example/a/.." as *u8, bn, 1024)
  tot=tot+1; if b3 == 0 { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) }
  gw("T15 TOOTH a '..' segment is REFUSED, never written\n" as *u8)
128
  let b4: i64 = mvf_basename("https://cdn.example/" as *u8, bn, 1024)
  tot=tot+1; if b4 == 0 { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) }
  gw("T16 TOOTH an EMPTY basename is refused rather than creating a nameless file\n" as *u8)
132
  // adapter fields must read back byte-exact, or the extraction rules silently
  // become the wrong rules
  let fv: *u8 = sys_mmap(256)
  // re-find rather than reuse po/lo -- T12's MISS leaves them stale, and a test
  // that reads stale state proves nothing about the thing it names
  let fpo: *i64 = sys_mmap(16) as *i64
  let flo: *i64 = sys_mmap(16) as *i64
  mvf_adapter_find_pfx(FG_PFX, "bunkr.si" as *u8, fpo, flo)
  mvf_field(fpo[0] as *u8, flo[0], "link_pfx" as *u8, fv, 256)
  tot=tot+1; if fg_eq(fv, "/f/" as *u8)==1 { pass=pass+1; gw("  [PASS] " as *u8) } else { gw("  [FAIL] " as *u8) }
  gw("T17 adapter fields read back byte-exact from the stored record\n" as *u8)
144
  let ctr: *i64 = gv_ctr()
  ctr[0] = pass
  ctr[1] = tot
  let rc: i64 = gv_verdict("ALBUM-PARSING" as *u8, ctr, "album parsing is data-driven, prefix-safe, and refuses partials" as *u8)
  sys_exit(rc)
  return rc
151}
nx_mvault_fetch_gate.nx source

`nx_mvault_fetch_gate.nx` source
