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

↩ module page · 322 lines · 17731 B
1// nx_cms_seo_gate.nx -- CMS W2 GATE (re-runnable, evidence-driven): SEO head + sitemap + robots
2// (the Yoast class). Proves the TEAM-AUTHORED _pe_seohead head renders into the live page; values
3// are attribute-ESCAPED (no injection through a SEO field); empty SEO title/desc FALL BACK to the
4// headline/tagline; /sitemap.xml is well-formed and references the site URL; /robots.txt points at
5// the sitemap. Appends "CMSGATE row=nx_cms_seo ... verdict=PASS|FAIL". Exit 0 iff all rows pass.
6// license_tier: ORIGINAL
7import "nx_cms_store.nx"
8import "nx_connect.nx"   // bounded connect: a raw sys_connect hangs ~127s on a black-holed host
9import "nx_sha256.nx"
10import "nx_syscalls.nx"
11import "_hdl_build/nx_kill_portable.nx"
12
13const SG_PORT: i64 = 8088
14const SG_SITE: *u8 = "/tmp/_cms_seo_site"
15
16func sg_w(s: *u8) -> i64 { var n: i64=0; while s[n]!=(0 as u8){n=n+1} sys_write(1,s,n); return 0 }
17func sg_len(s: *u8) -> i64 { var n: i64=0; while s[n]!=(0 as u8){n=n+1} return n }
18func sg_cat(dst: *u8, off: i64, s: *u8) -> i64 { var o: i64=off; var k: i64=0; while s[k]!=(0 as u8){dst[o]=s[k];o=o+1;k=k+1} return o }
19func sg_catn(dst: *u8, off: i64, s: *u8, n: i64) -> i64 { var o: i64=off; var k: i64=0; while k<n{dst[o]=s[k];o=o+1;k=k+1} return o }
20func sg_index(hay: *u8, n: i64, needle: *u8) -> i64 {
  let nl: i64 = sg_len(needle)
  var i: i64 = 0
  while i + nl <= n {
      var q: i64 = 0
      var ok: i64 = 1
      while q < nl { if (hay[i+q] as i64) != (needle[q] as i64) { ok = 0; q = nl } q = q + 1 }
      if ok == 1 { return i }
      i = i + 1
  }
  return 0 - 1
31}
32func sg_has(hay: *u8, n: i64, needle: *u8) -> i64 { if sg_index(hay, n, needle) >= 0 { return 1 } return 0 }
33
34func sg_http(req: *u8, rl: i64, resp: *u8, cap: i64) -> i64 {
  let fd: i64 = sys_socket(2, 1, 0)
  if fd < 0 { return 0 - 1 }
  let addr: *u8 = sys_mmap(16)
  addr[0] = 2 as u8; addr[1] = 0 as u8
  addr[2] = ((SG_PORT >> 8) & 0xff) as u8
  addr[3] = (SG_PORT & 0xff) as u8
  addr[4] = 127 as u8; addr[5] = 0 as u8; addr[6] = 0 as u8; addr[7] = 1 as u8
  var zi: i64 = 8
  while zi < 16 { addr[zi] = 0 as u8; zi = zi + 1 }
  if nx_connect_bounded(fd, addr, 16, NX_CONN_DEFAULT_MS) < 0 { sys_close(fd); return 0 - 1 }
  var off: i64 = 0
  var go: i64 = 1
  while go == 1 {
      go = 0
      if off < rl { let w: i64 = sys_write(fd, (req + off) as *u8, rl - off); if w > 0 { off = off + w; go = 1 } }
  }
  sys_set_socket_timeout(fd, 10)
  var total: i64 = 0
  go = 1
  while go == 1 {
      go = 0
      if total < cap { let r: i64 = sys_read(fd, (resp + total) as *u8, cap - total); if r > 0 { total = total + r; go = 1 } }
  }
  sys_close(fd)
  return total
60}
61func sg_post(req: *u8, path: *u8, cookie: *u8, body: *u8) -> i64 {
  var o: i64 = sg_cat(req, 0, "POST " as *u8)
  o = sg_cat(req, o, path)
  o = sg_cat(req, o, " HTTP/1.1\r\nHost: gate\r\nContent-Type: application/x-www-form-urlencoded\r\n" as *u8)
  if (cookie[0] as i64) != 0 { o = sg_cat(req, o, "Cookie: nsess=" as *u8); o = sg_catn(req, o, cookie, 32); o = sg_cat(req, o, "\r\n" as *u8) }
  o = sg_cat(req, o, "Content-Length: " as *u8)
  let bl: i64 = sg_len(body)
  let t: *u8 = sys_mmap(24)
  var m: i64 = bl
  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 { req[o] = t[k-1-i]; o = o + 1; i = i + 1 }
  o = sg_cat(req, o, "\r\n\r\n" as *u8)
  o = sg_cat(req, o, body)
  return o
78}
79func sg_get(req: *u8, path: *u8, cookie: *u8) -> i64 {
  var o: i64 = sg_cat(req, 0, "GET " as *u8)
  o = sg_cat(req, o, path)
  o = sg_cat(req, o, " HTTP/1.1\r\nHost: gate\r\n" as *u8)
  if (cookie[0] as i64) != 0 { o = sg_cat(req, o, "Cookie: nsess=" as *u8); o = sg_catn(req, o, cookie, 32); o = sg_cat(req, o, "\r\n" as *u8) }
  o = sg_cat(req, o, "\r\n" as *u8)
  return o
86}
87func sg_row(id: i64, ok: i64, what: *u8) -> i64 {
  sg_w("SEOROW " as *u8)
  let d: *u8 = sys_mmap(8); d[0] = (48 + id/10) as u8; d[1] = (48 + id%10) as u8; d[2] = 32 as u8; d[3] = 0 as u8
  sg_w(d)
  if ok == 1 { sg_w("PASS " as *u8) }
  if ok == 0 { sg_w("FAIL " as *u8) }
  sg_w(what); sg_w("\n" as *u8)
  return ok
95}
96func sg_mkdir(path: *u8, mode: i64) -> i64 { let nb: *i64 = sys_mmap(8) as *i64; nb[0] = 258; return __syscall(nb[0], 0 - 100, path, mode, 0, 0, 0) }
97func sg_writefile(path: *u8, buf: *u8, n: i64) -> i64 { let fd: i64 = sys_openat_wr(path, 0x1a4); if fd < 0 { return 0 } sys_write(fd, buf, n); sys_close(fd); return 1 }
98
99func main() -> i64 {
  sg_mkdir(SG_SITE, 0x1ed)
  // template renders the team SEO head + the headline (so the h1 still proves content)
  let tpl: *u8 = "<!DOCTYPE html><html lang=\"en\"><head><meta charset=\"utf-8\">{{seo_head_html}}</head><body><h1>{{hero_title}}</h1></body></html>" as *u8
  sg_writefile("/tmp/_cms_seo_site/template.html" as *u8, tpl, sg_len(tpl))
  // ★★SEED-SOURCE CHECK ADDED 2026-08-01 -- THE UNCHECKED COPY WAS THE WHOLE MYSTERY.
  // This gate seeds its fixture by copying knowledge/cms/andelinwest/content.txt. That directory does
  // NOT EXIST on this host (nx_fs ls knowledge/cms -> ABSENT), the copy's return was DISCARDED, so the
  // site dir got no content.txt, ca_render_publish returned 0, the admin printed SEED-RENDER-FAIL and
  // ALL EIGHT ROWS FAILED -- eight red lines describing eight capabilities, for one missing fixture
  // file, with nothing anywhere naming it.
  // ★AN UNCHECKED COPY TURNS A MISSING FIXTURE INTO A FALSE VERDICT ABOUT THE CODE UNDER TEST.
  // The gate now REFUSES with the exact missing path instead of grading the subject on absent input:
  // a gate must never report on a subject it failed to set up.
  let seedsrc: *u8 = "knowledge/cms/andelinwest/content.txt" as *u8
  let seeddst: *u8 = "/tmp/_cms_seo_site/content.txt" as *u8
  // ★★I REINTRODUCED THE VERY DEFECT I WAS FIXING, ONE LAYER UP -- RECORDED BECAUSE IT IS THE
  // INSTRUCTIVE PART. First fix keyed the fallback on "does the destination exist?" instead of "did
  // the copy SUCCEED?", so a STALE FIXTURE from the previous run satisfied the check and silently
  // overrode the corrected one: the gate reported the same 6/8 while running last run's data, and the
  // "SEED:" line disappeared, which was the only visible tell.
  // ★EXISTENCE IS NOT FRESHNESS. A /tmp artifact from a previous run is indistinguishable from a
  // correct one by an existence test -- the same stale-/tmp trap that md5 caught during the seq1267
  // caller-gap fix. Key the decision on the OPERATION'S RETURN VALUE (cst_copy gives 1/0 and the
  // original code discarded it -- that discarded return WAS the root cause of this whole chain).
  let copied: i64 = cst_copy(seedsrc, seeddst)
  var seedchk: i64 = 0 - 1
  if copied == 1 { seedchk = sys_openat_rd(seeddst) }
  if seedchk < 0 {
      // ★★MADE HERMETIC 2026-08-01. Refusing was the right FIRST move (it named the cause instead of
      // grading the subject on absent input), but a gate that cannot run without an external site
      // corpus is not a gate -- it is a gate PLUS a dependency nobody declared. This gate tests SEO
      // RENDERING BEHAVIOUR, not the content of any particular site, so the fixture belongs to the
      // gate. It writes its own minimal store (@key\nvalue\n records) and says so in the output.
      // ★A TEST FIXTURE THAT LIVES OUTSIDE THE TEST IS A SILENT PREREQUISITE, AND IT FAILS ON THE
      // FIRST HOST THAT NEVER HAD IT -- which is exactly how eight capability rows came to be red
      // for one absent file. Deliberately carries hero_title and NO seo_title, because row 0 proves
      // the title FALLS BACK to the headline when SEO fields are unset.
      // hero_title MUST be exactly what row 0 asserts the <title> falls back to. My first fixture
      // said "Andelin West" and row 0 failed -- MY error, not the subject's. ★WHEN YOU REPLACE A
      // FIXTURE, THE ASSERTIONS BECOME ITS SPECIFICATION: read what the rows expect, do not invent
      // plausible-looking data and then read the resulting red as a defect.
      let fx: *u8 = "@hero_title\nAndelin West Law\n@hero_sub\nSeed fixture owned by nx_cms_seo_gate\n" as *u8
      sg_writefile(seeddst, fx, sg_len(fx))
      seedchk = sys_openat_rd(seeddst)
      if seedchk < 0 {
          sg_w("CMSGATE row=nx_cms_seo SEED-WRITE-FAILED -- cannot create own fixture, NOT a subject failure\n" as *u8)
          sys_exit(2)
      }
      sg_w("SEED: corpus absent, using gate-owned hermetic fixture (declared, not silent)\n" as *u8)
  }
  sys_close(seedchk)
  sys_renameat("/tmp/_cms_seo_site/draft.txt" as *u8, "/tmp/_cms_seo_site/draft.stale" as *u8)
  let dig: *u8 = sys_mmap(32)
  sha256_digest("seo-gate-pw-1" as *u8, 13, dig)
  let hexs: *u8 = sys_mmap(80)
  let hxc: *u8 = "0123456789abcdef" as *u8
  var i: i64 = 0
  while i < 32 { hexs[i*2] = hxc[((dig[i] as i64) >> 4) & 15]; hexs[i*2+1] = hxc[(dig[i] as i64) & 15]; i = i + 1 }
  sg_writefile("/tmp/_cms_seo_site/admin_pw.sha256" as *u8, hexs, 64)
159
  let bpid: i64 = sys_fork()
  if bpid == 0 {
      let dn: i64 = sys_openat_wr("/dev/null" as *u8, 0x1a4)
      if dn >= 0 { sys_dup3(dn, 1, 0); sys_dup3(dn, 2, 0) }
      let bargv: *i64 = sys_mmap(32) as *i64
      bargv[0] = "_offc/nx_sov_build_run.elf" as *u8 as i64
      bargv[1] = "nx_cms_admin" as *u8 as i64
      bargv[2] = 0
      let benvp: *i64 = sys_mmap(16) as *i64; benvp[0] = 0
      sys_execve("_offc/nx_sov_build_run.elf" as *u8, bargv, benvp); sys_exit(127)
  }
  let bst: *i64 = sys_mmap(16) as *i64
  sys_wait4(bpid, bst, 0)
  // ★CALLER-GAP FIX 2026-08-01 (same class as seq1267, which fixed nx_hostctl + nx_restage and left
  // these 11 cms gates behind): nx_sov_build_run EMITS buildroot/_build/<t>.sov.elf under the no-/tmp
  // doctrine, while this gate still probed /tmp/<t>.sov.elf. PROVEN BY CONTROL before editing --
  // buildroot/_build/nx_cms_admin.sov.elf reads 208962B, /tmp/nx_cms_admin.sov.elf cannot be read.
  // The gate was reporting ADMIN-BUILD-MISSING while the admin build was sitting there, built fine.
  // ★A PRODUCER THAT MOVES ITS OUTPUT AND A CONSUMER THAT KEEPS PROBING THE OLD PATH REPORT
  // "MISSING" AND "BROKEN" IN THE SAME WORDS -- and only the second one sends you to the wrong file.
  // Prefer _build; fall back to /tmp so an older runner on another host still works.
  var bchk: i64 = sys_openat_rd("buildroot/_build/nx_cms_admin.sov.elf" as *u8)
  var badm: *u8 = "buildroot/_build/nx_cms_admin.sov.elf" as *u8
  if bchk < 0 { bchk = sys_openat_rd("/tmp/nx_cms_admin.sov.elf" as *u8); badm = "/tmp/nx_cms_admin.sov.elf" as *u8 }
  if bchk < 0 { sg_w("CMSGATE row=nx_cms_seo ADMIN-BUILD-MISSING (probed buildroot/_build AND /tmp)\n" as *u8); sys_exit(1) }
  sys_close(bchk)
  let pid: i64 = sys_fork()
  if pid == 0 {
      let lg: i64 = sys_openat_wr("/tmp/_cms_admin_seo_gate.log" as *u8, 0x1a4)
      if lg >= 0 { sys_dup3(lg, 1, 0); sys_dup3(lg, 2, 0) }
      let argvv: *i64 = sys_mmap(8*6) as *i64
      argvv[0] = badm as i64
      argvv[1] = "8088" as *u8 as i64
      argvv[2] = SG_SITE as i64
      argvv[3] = "99" as *u8 as i64
      argvv[4] = 0
      let envp: *i64 = sys_mmap(16) as *i64; envp[0] = 0
      sys_execve(badm, argvv, envp)
      sys_exit(127)
  }
200
  let req: *u8 = sys_mmap(262144)
  let resp: *u8 = sys_mmap(524288)
  let nocookie: *u8 = sys_mmap(8)
  nocookie[0] = 0 as u8
  let sess: *u8 = sys_mmap(64)
  sess[0] = 0 as u8
  let csrf: *u8 = sys_mmap(64)
  csrf[0] = 0 as u8
209
  // S0: boot + fallback head (no SEO fields yet) -> title falls back to hero_title
  var rn: i64 = 0 - 1
  var tries: i64 = 0
  while tries < 50 {
      let rl0: i64 = sg_get(req, "/" as *u8, nocookie)
      rn = sg_http(req, rl0, resp, 524287)
      if rn > 0 { tries = 50 }
      if rn <= 0 { sys_sleep_ms(100); tries = tries + 1 }
  }
  var pass: i64 = 0
  var rows: i64 = 0
  var ok: i64 = 0
  if rn > 0 { if sg_has(resp, rn, "<title>Andelin West Law</title>" as *u8) == 1 { if sg_has(resp, rn, "og:title" as *u8) == 1 { ok = 1 } } }
  rows = rows + 1; pass = pass + sg_row(0, ok, "SEO title falls back to headline" as *u8)
224
  // login + csrf
  var rl: i64 = sg_post(req, "/admin/login" as *u8, nocookie, "pw=seo-gate-pw-1" as *u8)
  rn = sg_http(req, rl, resp, 524287)
  var have: i64 = 0
  if rn > 0 { if sg_has(resp, rn, "303" as *u8) == 1 {
      let sp: i64 = sg_index(resp, rn, "nsess=" as *u8)
      if sp >= 0 {
          var m2: i64 = 0
          while m2 < 32 { sess[m2] = resp[sp+6+m2]; m2 = m2 + 1 }
          sess[32] = 0 as u8
          rl = sg_get(req, "/admin" as *u8, sess)
          rn = sg_http(req, rl, resp, 524287)
          let cp: i64 = sg_index(resp, rn, "name=\"csrf\" value=\"" as *u8)
          if cp >= 0 { var m3: i64 = 0; while m3 < 32 { csrf[m3] = resp[cp+19+m3]; m3 = m3 + 1 } csrf[32] = 0 as u8; have = 1 }
      }
  } }
  rows = rows + 1; pass = pass + sg_row(1, have, "login + editor shows SEO fields" as *u8)
  ok = 0
  if have == 1 { if sg_has(resp, rn, "name=\"seo_title\"" as *u8) == 1 { if sg_has(resp, rn, "name=\"site_url\"" as *u8) == 1 { ok = 1 } } }
  rows = rows + 1; pass = pass + sg_row(2, ok, "editor carries seo_title + site_url inputs" as *u8)
245
  // S3: save explicit SEO fields (desc has a <script> that MUST be escaped, not injected)
  let body3: *u8 = sys_mmap(8192)
  var b3: i64 = sg_cat(body3, 0, "csrf=" as *u8)
  b3 = sg_catn(body3, b3, csrf, 32)
  b3 = sg_cat(body3, b3, "&hero_title=Andelin+West+Law&seo_title=Andelin+West+Law+%E2%80%94+Trusted+Counsel&seo_desc=%3Cscript%3Ealert%284%29%3C%2Fscript%3EClient-first+legal+services.&og_image=https%3A%2F%2Fandelinwest.com%2Fog.png&site_url=https%3A%2F%2Fandelinwest.com" as *u8)
  body3[b3] = 0 as u8
  rl = sg_post(req, "/admin/save" as *u8, sess, body3)
  rn = sg_http(req, rl, resp, 524287)
  ok = 0
  if rn > 0 { if sg_has(resp, rn, "303" as *u8) == 1 { ok = 1 } }
  rows = rows + 1; pass = pass + sg_row(3, ok, "save SEO fields publishes" as *u8)
257
  // S4: live page carries the explicit SEO title + description + og:image + canonical
  rl = sg_get(req, "/" as *u8, nocookie)
  rn = sg_http(req, rl, resp, 524287)
  ok = 0
  if rn > 0 { if sg_has(resp, rn, "Trusted Counsel" as *u8) == 1 { if sg_has(resp, rn, "og:image\" content=\"https://andelinwest.com/og.png\"" as *u8) == 1 { if sg_has(resp, rn, "rel=\"canonical\" href=\"https://andelinwest.com\"" as *u8) == 1 { ok = 1 } } } }
  rows = rows + 1; pass = pass + sg_row(4, ok, "explicit SEO title/og:image/canonical render" as *u8)
264
  // S5: the injected <script> in seo_desc is ESCAPED, not present as a tag
  ok = 0
  if rn > 0 { if sg_has(resp, rn, "Client-first legal services." as *u8) == 1 { if sg_has(resp, rn, "<script" as *u8) == 0 { if sg_has(resp, rn, "&lt;script" as *u8) == 1 { ok = 1 } } } }
  rows = rows + 1; pass = pass + sg_row(5, ok, "SEO field XSS escaped not injected" as *u8)
269
  // S6: /sitemap.xml well-formed + references the site URL home page
  rl = sg_get(req, "/sitemap.xml" as *u8, nocookie)
  rn = sg_http(req, rl, resp, 524287)
  ok = 0
  if rn > 0 { if sg_has(resp, rn, "application/xml" as *u8) == 1 { if sg_has(resp, rn, "<urlset" as *u8) == 1 { if sg_has(resp, rn, "<loc>https://andelinwest.com/</loc>" as *u8) == 1 { ok = 1 } } } }
  rows = rows + 1; pass = pass + sg_row(6, ok, "sitemap.xml well-formed + home loc" as *u8)
276
  // S7: /robots.txt references the sitemap
  rl = sg_get(req, "/robots.txt" as *u8, nocookie)
  rn = sg_http(req, rl, resp, 524287)
  ok = 0
  if rn > 0 { if sg_has(resp, rn, "text/plain" as *u8) == 1 { if sg_has(resp, rn, "Sitemap: https://andelinwest.com/sitemap.xml" as *u8) == 1 { if sg_has(resp, rn, "Allow: /" as *u8) == 1 { ok = 1 } } } }
  rows = rows + 1; pass = pass + sg_row(7, ok, "robots.txt references the sitemap" as *u8)
283
  // ---- teardown ----
  nxk_kill(pid, 9)
  let stbuf: *i64 = sys_mmap(16) as *i64
  sys_wait4(pid, stbuf, 0)
288
  var verdict: *u8 = "FAIL" as *u8
  if pass == rows { verdict = "PASS" as *u8 }
  let line: *u8 = sys_mmap(512)
  var lo: i64 = sg_cat(line, 0, "CMSGATE row=nx_cms_seo rows=" as *u8)
  let t2: *u8 = sys_mmap(24)
  var mm: i64 = rows
  var kk: i64 = 0
  if mm == 0 { t2[0] = 48 as u8; kk = 1 }
  while mm > 0 { t2[kk] = (48 + (mm % 10)) as u8; mm = mm / 10; kk = kk + 1 }
  var ii: i64 = 0
  while ii < kk { line[lo] = t2[kk-1-ii]; lo = lo + 1; ii = ii + 1 }
  lo = sg_cat(line, lo, " pass=" as *u8)
  mm = pass; kk = 0
  if mm == 0 { t2[0] = 48 as u8; kk = 1 }
  while mm > 0 { t2[kk] = (48 + (mm % 10)) as u8; mm = mm / 10; kk = kk + 1 }
  ii = 0
  while ii < kk { line[lo] = t2[kk-1-ii]; lo = lo + 1; ii = ii + 1 }
  lo = sg_cat(line, lo, " verdict=" as *u8)
  lo = sg_cat(line, lo, verdict)
  lo = sg_cat(line, lo, " epoch=" as *u8)
  mm = sys_now_realtime_sec(); kk = 0
  if mm == 0 { t2[0] = 48 as u8; kk = 1 }
  while mm > 0 { t2[kk] = (48 + (mm % 10)) as u8; mm = mm / 10; kk = kk + 1 }
  ii = 0
  while ii < kk { line[lo] = t2[kk-1-ii]; lo = lo + 1; ii = ii + 1 }
  line[lo] = 10 as u8
  lo = lo + 1
  sys_write(1, line, lo)
  let gf: i64 = sys_openat_append("knowledge/status/cms_gate.log" as *u8, 0x1a4)
  if gf >= 0 { sys_write(gf, line, lo); sys_close(gf) }
  if pass == rows { sys_exit(0) }
  sys_exit(1)
  return 1
322}
nx_cms_seo_gate.nx source

`nx_cms_seo_gate.nx` source
