`nishi.nx` source

↩ module page · 532 lines · 21789 B
1// nishi.nx -- SOVEREIGN browse CLI + interactive navigator (the Nishi browser).
2// (B)-AUTHORIZED tutor build; flagged backfill debt (re-author from spec later).
3//
4//   nishi <https-url>        one-shot: fetch + render + link list, then exit
5//   nishi                    interactive REPL (prompt for first URL)
6//   nishi -i <https-url>     interactive, starting at the given URL
7//
8// Interactive commands:  <number> follow that link | <url> open | b back |
9//                        r reload | q quit  (EOF also quits)
10//
11//   live DNS -> sovereign TLS 1.3 -> HTTP/1.1 GET -> follow 3xx (abs+relative)
12//   -> de-chunk -> gunzip (Content-Encoding) -> charset (header + <meta>) ->
13//   UTF-8 -> HTML->text + absolute link list.  No curl/zlib/Playwright/gcc.
14//
15// Composes existing sovereign organs only.  Artifact = Linux x86-64 ELF (WSL).
16import "nx_syscalls.nx"
17import "nx_itoa_lib.nx"   // shared MSB-first emitter (zero-alloc)
18import "nx_x509_trust_store.nx"
19import "nx_trust_store_load_from_certdata.nx"
20import "nx_https_get.nx"
21import "nx_http_response_parse.nx"
22import "nx_html_to_text.nx"
23import "nx_redirect_resolve.nx"
24import "nx_gzip_wrap.nx"
25import "nx_charset.nx"
26import "nx_html_extract_links.nx"
27import "nx_js_eval.nx"   // R-NISH-JS1: run the page's inline <script> against the parsed DOM
28
29func _p(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 }
30// MIGRATED to the shared emitter (debt 1785563586). The old body mmapped a scratch buffer
31// per call and never freed it. At PAGE granularity that is 4096B leaked PER CALL -- the
32// defect that took 28.5GB of a 36GB host in nx_ts_lumadiff (2MB input, ~3.66M calls).
33// nxi_* is MSB-first, allocates NOTHING, and emits identical bytes including the sign.
34func _pn(fd: i64, v: i64) -> i64 { nxi_fd(fd, v); return 0 }
35
36// Read one line from stdin into out (NUL-terminated, CR/LF stripped).
37// Returns the line length (0 for an empty line), or -1 on EOF with no data.
38func _readline(out: *u8, cap: i64) -> i64 {
  var n: i64 = 0
  let ch: *u8 = sys_mmap(8)
  var go: i64 = 1
  var eof: i64 = 0
  while go == 1 {
      let r: i64 = sys_read(0, ch, 1)
      if r <= 0 { go = 0; if n == 0 { eof = 1 } }
      else {
          let c: i64 = ch[0] & 0xff
          if c == 0x0a { go = 0 }
          else {
              if c != 0x0d {
                  if n < (cap - 1) { out[n] = c as u8; n = n + 1 }
              }
          }
      }
  }
  out[n] = 0 as u8
  if eof == 1 { return 0 - 1 }
  return n
59}
60
61// Parse an all-digits string -> number, or -1 if not purely numeric.
62func _parse_num(s: *u8, n: i64) -> i64 {
  if n <= 0 { return 0 - 1 }
  var v: i64 = 0
  var i: i64 = 0
  while i < n {
      let c: i64 = s[i] & 0xff
      if c < 0x30 { return 0 - 1 }
      if c > 0x39 { return 0 - 1 }
      v = (v * 10) + (c - 0x30)
      i = i + 1
  }
  return v
74}
75
76// 1 iff s starts with "http" (covers http/https).
77func _is_http(s: *u8, n: i64) -> i64 {
  if n < 7 { return 0 }
  if _lc(s[0] & 0xff) != 104 { return 0 }   // h
  if _lc(s[1] & 0xff) != 116 { return 0 }   // t
  if _lc(s[2] & 0xff) != 116 { return 0 }   // t
  if _lc(s[3] & 0xff) != 112 { return 0 }   // p
  return 1
84}
85
86// If the parsed response is a followable redirect, copy the resolved https
87// target into `cur` and return 1; else 0.  (https-only gate in nx_redirect_resolve.)
88func _try_redirect(buf: *u8, n: i64, r: *i64, cur: *u8, urlcap: i64) -> i64 {
  let st: i64 = r[1]
  var isr: i64 = 0
  if st == 301 { isr = 1 }
  if st == 302 { isr = 1 }
  if st == 303 { isr = 1 }
  if st == 307 { isr = 1 }
  if st == 308 { isr = 1 }
  if isr == 0 { return 0 }
  let vo: *i64 = sys_mmap(8) as *i64
  let vl: *i64 = sys_mmap(8) as *i64
  let loc: *u8 = sys_mmap(16)
  loc[0]=76 as u8; loc[1]=111 as u8; loc[2]=99 as u8; loc[3]=97 as u8
  loc[4]=116 as u8; loc[5]=105 as u8; loc[6]=111 as u8; loc[7]=110 as u8
  if nx_http_response_header(buf, n, r, loc, 8, vo, vl) == 0 { return 0 }
  let lo: i64 = vo[0]
  let ll: i64 = vl[0]
  if ll >= urlcap { return 0 }
  let lp: *u8 = ((buf as i64) + lo) as *u8
  var blen: i64 = 0
  while (cur[blen] & 0xff) != 0 { blen = blen + 1 }
  let tmp: *u8 = sys_mmap(urlcap)
  let olp: *i64 = sys_mmap(8) as *i64
  if nx_redirect_resolve(cur, blen, lp, ll, tmp, urlcap, olp) == 0 { return 0 }
  let tl: i64 = olp[0]
  var ci: i64 = 0
  while ci < tl { cur[ci] = tmp[ci]; ci = ci + 1 }
  cur[tl] = 0 as u8
  _p(2, "nishi: -> redirect " as *u8); _pn(2, st)
  _p(2, " -> " as *u8); sys_write(2, tmp, tl); _p(2, "\n" as *u8)
  return 1
119}
120
121// Fetch `cur` (URL, modified in place by redirects), decode + render the page
122// to stdout, and fill the link buffers (absolute URLs).  Returns 0 on success
123// or a negative verdict on failure.  Trust store + entropy are loaded once by
124// the caller and passed in.  Sets nlinks_box[0] to the number of links.
125func nish_fetch_render(cur: *u8, urlcap: i64, store: *TrustStore, cr: *u8, priv: *u8,
                     lbuf: *u8, lcap: i64, loff: *i64, llen: *i64, lmax: i64,
                     nlinks_box: *i64, emit_links: i64, out_html: i64, out_binary: i64) -> i64 {
  nlinks_box[0] = 0
  let t_start: i64 = sys_now_ms()
  let now: i64 = sys_now_realtime_sec()
  let cap: i64 = 524288
  let buf: *u8 = sys_mmap(cap)
  let r: *i64 = nx_http_resp_alloc()
134
  let MAXREDIR: i64 = 6
  var redirs: i64 = 0
  var n: i64 = 0
  var pv: i64 = 0
  var loopdone: i64 = 0
  while loopdone == 0 {
      n = nx_https_get(cur, cr, priv, store, now, buf, cap)
      if n < 0 {
          _p(2, "nishi: request failed. verdict=" as *u8); _pn(2, 0 - n)
          _p(2, " (2=bad-url 3=connect-fail 4=handshake-fail 5=fetch-fail)\n" as *u8)
          return 0 - 4
      }
      pv = nx_http_response_parse(buf, n, r)
      var follow: i64 = 0
      if pv == NX_HTTP_RESP_OK {
          if redirs < MAXREDIR {
              if _try_redirect(buf, n, r, cur, urlcap) == 1 { redirs = redirs + 1; follow = 1 }
          }
      }
      if follow == 0 { loopdone = 1 }
  }
  let t_net: i64 = sys_now_ms()
157
  let base: i64 = buf as i64
  var body: *u8 = buf
  var body_len: i64 = n
  if pv == NX_HTTP_RESP_OK {
      let bo: i64 = r[6]
      let kind: i64 = r[8]
      body = (base + bo) as *u8
      body_len = n - bo
      if kind == NX_HTTP_BODY_CONTENT_LENGTH {
          body_len = r[7]
          if body_len > n - bo { body_len = n - bo }
      }
      if kind == NX_HTTP_BODY_CHUNKED {
          let dcap: i64 = 1048576
          let dbuf: *u8 = sys_mmap(dcap)
          let dl: i64 = nx_http_dechunk(body, n - bo, dbuf, dcap)
          if dl > 0 {
              body = dbuf
              body_len = dl
          }
      }
179
      // gunzip if Content-Encoding: gzip (KAT-proven nx_gzip_inflate).
      let cevo: *i64 = sys_mmap(8) as *i64
      let cevl: *i64 = sys_mmap(8) as *i64
      let cen: *u8 = sys_mmap(20)
      cen[0]=67 as u8;  cen[1]=111 as u8; cen[2]=110 as u8; cen[3]=116 as u8
      cen[4]=101 as u8; cen[5]=110 as u8; cen[6]=116 as u8; cen[7]=45 as u8
      cen[8]=69 as u8;  cen[9]=110 as u8; cen[10]=99 as u8; cen[11]=111 as u8
      cen[12]=100 as u8;cen[13]=105 as u8;cen[14]=110 as u8;cen[15]=103 as u8
      if nx_http_response_header(buf, n, r, cen, 16, cevo, cevl) == 1 {
          let cvl: i64 = cevl[0]
          var isgz: i64 = 0
          if cvl >= 4 {
              let cvp: *u8 = (base + cevo[0]) as *u8
              if _lc(cvp[0] as i64)==103 { if _lc(cvp[1] as i64)==122 {
                  if _lc(cvp[2] as i64)==105 { if _lc(cvp[3] as i64)==112 { isgz=1 } } } }
          }
          if isgz == 1 {
              let gr: *NxGzipResult = nx_gzip_inflate(body, body_len, 4194304)
              if gr != (0 as *NxGzipResult) {
                  if gr.error_code == NX_GZ_OK {
                      body = gr.output_data
                      body_len = gr.output_size
                  }
              }
          }
      }
  }
207
  // -B raw binary: emit the dechunked+gunzipped body VERBATIM (NO charset decode -- charset decode
  // corrupts binary, which is why -H returned a PNG with zeroed magic). For fetching images so the
  // decoder gets the exact bytes. MUST precede the charset step below.
  if out_binary == 1 {
      if pv == NX_HTTP_RESP_OK { if body_len > 0 { sys_write(1, body, body_len) } }
      nlinks_box[0] = 0
      return 0
  }
216
  // charset -> UTF-8 (Content-Type charset wins, else <meta charset>).
  if pv == NX_HTTP_RESP_OK {
      let ctvo: *i64 = sys_mmap(8) as *i64
      let ctvl: *i64 = sys_mmap(8) as *i64
      let ctn: *u8 = sys_mmap(20)
      ctn[0]=67 as u8;  ctn[1]=111 as u8; ctn[2]=110 as u8; ctn[3]=116 as u8
      ctn[4]=101 as u8; ctn[5]=110 as u8; ctn[6]=116 as u8; ctn[7]=45 as u8
      ctn[8]=84 as u8;  ctn[9]=121 as u8; ctn[10]=112 as u8; ctn[11]=101 as u8
      var cs: i64 = 0 - 1
      if nx_http_response_header(buf, n, r, ctn, 12, ctvo, ctvl) == 1 {
          let ctp: *u8 = (base + ctvo[0]) as *u8
          cs = nx_charset_sniff(ctp, ctvl[0])
      }
      if cs < 0 { cs = nx_charset_sniff(body, body_len) }
      if cs > 0 {
          let ccap: i64 = 2097152
          let cbuf: *u8 = sys_mmap(ccap)
          let cl: i64 = nx_charset_to_utf8(body, body_len, cs, cbuf, ccap)
          if cl > 0 {
              body = cbuf
              body_len = cl
          }
      }
  }
241
  // Save the fetched+cleaned HTML so the styled GRAPHICAL renderer (nx_render_wiki_styled) can
  // render THIS page next -- the fetch -> styled-render -> view browse loop, reusing nishi's proven
  // fetch (TLS/redirect/dechunk/charset) instead of replicating it. Additive; CLI text output below
  // is unchanged.
  if pv == NX_HTTP_RESP_OK { if body_len > 0 {
      let _sfd: i64 = sys_openat_wr("knowledge/fetched/last_fetch.html\x00" as *u8, 420)
      if _sfd >= 0 { sys_write(_sfd, body, body_len); sys_close(_sfd) }
  } }
250
  // R1 (CSS-on-PE arc): raw-HTML output mode -- emit the decoded HTML body verbatim so the GUI's
  // CSS layout+paint pipeline (nx_browser br_layout/br_draw_fb) renders the REAL page, not flattened
  // text. Placed AFTER fetch+redirect+dechunk+gunzip+charset so `body` is final UTF-8 HTML.
  if out_html == 1 {
      nlinks_box[0] = 0
      sys_write(1, body, body_len)
      return 0
  }
  // render HTML -> readable text.  emit_links=1 (GUI) interleaves inline link
  // markers (STX url SOH anchor ETX) in the body; emit_links=0 (CLI) is plain.
  let tcap: i64 = 524288
  let tbuf: *u8 = sys_mmap(tcap)
  let tn: i64 = nx_html_to_text_x(body, body_len, tbuf, tcap, emit_links)
264
  // links (absolute, resolved against the final URL) -> caller's buffers
  // (needed for nlinks_box + the CLI's follow-link-by-number).
  var nlinks: i64 = 0
  if pv == NX_HTTP_RESP_OK {
      var curlen: i64 = 0
      while (cur[curlen] & 0xff) != 0 { curlen = curlen + 1 }
      nlinks = nx_html_extract_links(body, body_len, cur, curlen,
                                     lbuf, lcap, loff, llen, lmax)
  }
  nlinks_box[0] = nlinks
275
  // Output: title (first rendered line), then -- only in CLI mode -- the numbered
  // link list, then the article body.  In GUI inline mode the body already carries
  // clickable link markers, so the separate list is suppressed (cleaner article).
  if tn > 0 {
      var t1: i64 = 0
      var tf: i64 = 0
      while tf == 0 {
          if t1 >= tn { tf = 1 } else { if (tbuf[t1] & 0xff) == 0x0a { tf = 1 } else { t1 = t1 + 1 } }
      }
      sys_write(1, tbuf, t1)
      _p(1, "\n" as *u8)
      if emit_links == 0 {
          if nlinks > 0 {
              _p(1, "--- Links (" as *u8); _pn(1, nlinks); _p(1, ") ---\n" as *u8)
              var li: i64 = 0
              while li < nlinks {
                  _pn(1, li + 1); _p(1, ". " as *u8)
                  sys_write(1, ((lbuf as i64) + loff[li]) as *u8, llen[li])
                  _p(1, "\n" as *u8)
                  li = li + 1
              }
              _p(1, "\n" as *u8)
          }
      }
      if t1 < tn { sys_write(1, ((tbuf as i64) + t1 + 1) as *u8, tn - t1 - 1) }
      _p(1, "\n" as *u8)
  }
  if tn <= 0 {
      sys_write(1, body, body_len)
  }
  // R-NISH-JS1: execute the page's inline <script> blocks against the parsed DOM -- the sovereign
  // browser RUNS JavaScript. Scripts' console.log output appears on stdout (above); external src=
  // scripts are skipped; per-script errors + oversized bundles are swallowed (best-effort, never
  // crashes the browse). The i64-only engine errors out gracefully on bundles needing f64/regex.
  if pv == NX_HTTP_RESP_OK {
      if body_len > 0 {
          let nsc: i64 = js_run_page_scripts(body, body_len)
          _p(2, "nishi-js inline_scripts_executed=" as *u8); _pn(2, nsc); _p(2, "\n" as *u8)
      }
  }
  // native sovereign perf line (system performance, emitted to stderr via sys_now_ms)
  let t_end: i64 = sys_now_ms()
  _p(2, "nishi-perf net=" as *u8); _pn(2, t_net - t_start)
  _p(2, "ms proc=" as *u8); _pn(2, t_end - t_net)
  _p(2, "ms total=" as *u8); _pn(2, t_end - t_start)
  _p(2, "ms bytes=" as *u8); _pn(2, n); _p(2, "\n" as *u8)
  return 0
323}
324
325func main(argc: i64, argv: *i64) -> i64 {
  // trust store (Mozilla NSS bundle) -- loaded ONCE.
  var tr: i64 = nx_trust_store_load_from_certdata("data/mozilla_certdata.txt" as *u8, 300, 4194304)
  if tr <= 0 { tr = nx_trust_store_load_from_certdata("/tmp/mozilla_certdata.txt" as *u8, 300, 4194304) }
  if tr <= 0 {
      _p(2, "nishi: CA bundle load failed (need /tmp/mozilla_certdata.txt). verdict=" as *u8)
      _pn(2, 0 - tr); _p(2, "\n" as *u8)
      sys_exit(3); return 3
  }
  let store: *TrustStore = tr as *TrustStore
335
  // entropy -- loaded ONCE (32B client_random + 32B x25519 ephemeral priv).
  let cr: *u8 = sys_mmap(32)
  let priv: *u8 = sys_mmap(32)
  let ufd: i64 = sys_openat_rd("/dev/urandom" as *u8)
  if ufd >= 0 {
      sys_read(ufd, cr, 32)
      sys_read(ufd, priv, 32)
      sys_close(ufd)
  }
  if ufd < 0 {
      var z: i64 = 0
      while z < 32 { cr[z] = (0xC0 + z) as u8; priv[z] = (0xA0 + z) as u8; z = z + 1 }
  }
349
  // working buffers (persist across pages).
  let URLCAP: i64 = 4096
  let cur: *u8 = sys_mmap(URLCAP)
  let cmd: *u8 = sys_mmap(URLCAP)
  let LMAX: i64 = 150
  let lcap: i64 = 262144
  let lbuf: *u8 = sys_mmap(lcap)
  let loff: *i64 = sys_mmap(8 * LMAX) as *i64
  let llen: *i64 = sys_mmap(8 * LMAX) as *i64
  let nbox: *i64 = sys_mmap(8) as *i64
360
  // history stack (packed URLs).
  let HMAX: i64 = 64
  let hcap: i64 = 131072
  let hbuf: *u8 = sys_mmap(hcap)
  let hoff: *i64 = sys_mmap(8 * HMAX) as *i64
  let hlen: *i64 = sys_mmap(8 * HMAX) as *i64
  var hcount: i64 = 0
  var hw: i64 = 0
369
  // arg parse: interactive (-i) vs one-shot; -L = one-shot with inline link
  // markers (used by the graphical GUI front-end); optional initial URL.
  var interactive: i64 = 0
  var initial: *u8 = 0 as *u8
  var emit_links: i64 = 0
  var out_html: i64 = 0                            // -H: emit the raw decoded HTML (for the GUI CSS renderer)
  var out_binary: i64 = 0                          // -B: emit the raw body bytes verbatim (images/binary)
  if argc < 2 { interactive = 1 }
  else {
      let a1: *u8 = argv[1] as *u8
      var isi: i64 = 0
      var isL: i64 = 0
      var isH: i64 = 0
      var isB: i64 = 0
      if (a1[0] & 0xff) == 0x2d {
          if (a1[1] & 0xff) == 0x69 { isi = 1 }   // -i
          if (a1[1] & 0xff) == 0x4c { isL = 1 }   // -L
          if (a1[1] & 0xff) == 0x48 { isH = 1 }   // -H (raw HTML out)
          if (a1[1] & 0xff) == 0x42 { isB = 1 }   // -B (raw binary out)
      }
      if isi == 1 {
          interactive = 1
          if argc >= 3 { initial = argv[2] as *u8 }
      } else {
          if isL == 1 {
              emit_links = 1
              initial = ((a1 as i64) + 2) as *u8   // URL is the rest of argv[1] after "-L" (build_argv gives one token)
          } else {
              if isH == 1 {
                  out_html = 1
                  initial = ((a1 as i64) + 2) as *u8   // URL is the rest of argv[1] after "-H"
              } else {
                  if isB == 1 {
                      out_binary = 1
                      initial = ((a1 as i64) + 2) as *u8   // URL is the rest of argv[1] after "-B"
                  } else {
                      initial = a1
                  }
              }
          }
      }
  }
412
  var have_page: i64 = 0
  var cur_nlinks: i64 = 0
415
  // open the initial URL (if any).
  if initial != (0 as *u8) {
      var ul: i64 = 0
      while initial[ul] != (0 as u8) { cur[ul] = initial[ul]; ul = ul + 1 }
      cur[ul] = 0 as u8
      if nish_fetch_render(cur, URLCAP, store, cr, priv, lbuf, lcap, loff, llen, LMAX, nbox, emit_links, out_html, out_binary) == 0 {
          have_page = 1
          cur_nlinks = nbox[0]
          // push final cur onto history.
          var cl0: i64 = 0
          while (cur[cl0] & 0xff) != 0 { cl0 = cl0 + 1 }
          if hcount < HMAX { if (hw + cl0) < hcap {
              hoff[hcount] = hw
              var q: i64 = 0
              while q < cl0 { hbuf[hw + q] = cur[q]; q = q + 1 }
              hlen[hcount] = cl0
              hw = hw + cl0
              hcount = hcount + 1
          } }
      }
  }
437
  if interactive == 0 { sys_exit(0); return 0 }
439
  _p(2, "nishi interactive -- <number> follow link | <url> open | b back | r reload | q quit\n" as *u8)
  var go: i64 = 1
  while go == 1 {
      _p(2, "\nnish> " as *u8)
      let clen: i64 = _readline(cmd, URLCAP)
      if clen < 0 { go = 0 }
      else {
          if clen > 0 {
              let c0: i64 = cmd[0] & 0xff
              var handled: i64 = 0
              var target_ready: i64 = 0
              var push: i64 = 1
452
              // q quit
              if handled == 0 { if clen == 1 { if c0 == 0x71 { go = 0; handled = 1 } } }
              // b back
              if handled == 0 { if clen == 1 { if c0 == 0x62 {
                  handled = 1
                  if hcount >= 2 {
                      hcount = hcount - 1
                      let idx: i64 = hcount - 1
                      let o: i64 = hoff[idx]
                      let l: i64 = hlen[idx]
                      var q: i64 = 0
                      while q < l { cur[q] = hbuf[o + q]; q = q + 1 }
                      cur[l] = 0 as u8
                      target_ready = 1
                      push = 0
                  } else { _p(2, "(no history)\n" as *u8) }
              } } }
              // r reload
              if handled == 0 { if clen == 1 { if c0 == 0x72 {
                  handled = 1
                  if have_page == 1 { target_ready = 1; push = 0 }
                  else { _p(2, "(no page yet)\n" as *u8) }
              } } }
              // url
              if handled == 0 { if _is_http(cmd, clen) == 1 {
                  handled = 1
                  var q: i64 = 0
                  while q < clen { cur[q] = cmd[q]; q = q + 1 }
                  cur[clen] = 0 as u8
                  target_ready = 1
                  push = 1
              } }
              // number -> follow link
              if handled == 0 {
                  let num: i64 = _parse_num(cmd, clen)
                  if num >= 1 {
                      handled = 1
                      if have_page == 1 { if num <= cur_nlinks {
                          let li: i64 = num - 1
                          let o: i64 = loff[li]
                          let l: i64 = llen[li]
                          var q: i64 = 0
                          while q < l { cur[q] = lbuf[o + q]; q = q + 1 }
                          cur[l] = 0 as u8
                          target_ready = 1
                          push = 1
                      } }
                      if target_ready == 0 { _p(2, "(no such link)\n" as *u8) }
                  }
              }
              // unknown
              if handled == 0 {
                  _p(2, "commands: <number> | <url> | b back | r reload | q quit\n" as *u8)
              }
507
              // navigate
              if target_ready == 1 {
                  if nish_fetch_render(cur, URLCAP, store, cr, priv, lbuf, lcap, loff, llen, LMAX, nbox, emit_links, out_html, out_binary) == 0 {
                      have_page = 1
                      cur_nlinks = nbox[0]
                      if push == 1 {
                          var cl1: i64 = 0
                          while (cur[cl1] & 0xff) != 0 { cl1 = cl1 + 1 }
                          if hcount < HMAX { if (hw + cl1) < hcap {
                              hoff[hcount] = hw
                              var q2: i64 = 0
                              while q2 < cl1 { hbuf[hw + q2] = cur[q2]; q2 = q2 + 1 }
                              hlen[hcount] = cl1
                              hw = hw + cl1
                              hcount = hcount + 1
                          } }
                      }
                  }
              }
          }
      }
  }
  _p(2, "bye\n" as *u8)
  sys_exit(0); return 0
532}
nishi.nx source

`nishi.nx` source
