nx_yt_probe.nx source

1// nx_yt_probe.nx -- capability test: fetch a YouTube watch page, extract ytInitialPlayerResponse (embedded 2// state), and report HOW FAR our extractor gets -- direct stream URLs vs signatureCipher (which needs the 3// base.js sig/nsig extractor). Honest assessment, no download. Run from nxc2 root. license_tier: ORIGINAL 4import "nx_syscalls.nx" 5import "nx_x509_trust_store.nx" 6import "nx_trust_store_load_from_certdata.nx" 7import "nx_https_fetch_follow.nx" 8import "nx_media_state.nx" 9import "nx_media_signal.nx" 10import "nx_media_extract.nx" 11const K_MAGIC_4194304: i64 = 4194304 12const K_MAGIC_8388608: i64 = 8388608 13const K_MAGIC_262144: i64 = 262144 14const K_MAGIC_4096: i64 = 4096 15const K_MAGIC_1024: i64 = 1024 16 17func pw(s: *u8) -> i64 { var n: i64=0; while s[n]!=(0 as u8){n=n+1} sys_write(1,s,n); return 0 } 18func pn(v: i64) -> i64 { let b: *u8=sys_mmap(24); var m: i64=v; if m<0{pw("-" as *u8);m=0-m} let t: *u8=sys_mmap(24); 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{b[i]=t[k-1-i];i=i+1} sys_write(1,b,k); return 0 } 19func pslen(s: *u8) -> i64 { var n: i64=0; while s[n]!=(0 as u8){n=n+1} return n } 20func count_occ(hay: *u8, hl: i64, needle: *u8) -> i64 { 21 let nl: i64 = pslen(needle) 22 var c: i64 = 0 23 var i: i64 = ms_find(hay, hl, needle, nl, 0) 24 while i >= 0 { c = c + 1; i = ms_find(hay, hl, needle, nl, i + nl) } 25 return c 26} 27 28func main(argc: i64, argv: *i64) -> i64 { 29 var url: *u8 = "https://www.youtube.com/watch?v=DtBUSSrSw34" as *u8 30 if argc >= 2 { url = argv[1] as *u8 } 31 let r: i64 = nx_trust_store_load_from_certdata("data/mozilla_certdata.txt" as *u8, 512, K_MAGIC_4194304) 32 if r <= 0 { pw("YT-PROBE: certdata load failed (run from nxc2 root)\n" as *u8); sys_exit(1); return 1 } 33 let store: *TrustStore = r as *TrustStore 34 let status: *i64 = sys_mmap(8) as *i64 35 36 pw("== nx_yt_probe ==\nurl: " as *u8); pw(url); pw("\n" as *u8) 37 let cap: i64 = K_MAGIC_8388608 38 let html: *u8 = sys_mmap(cap) 39 let hn: i64 = nx_https_fetch_follow(url, store, html, cap, 6, status) 40 pw("fetch status=" as *u8); pn(status[0]); pw(" bytes=" as *u8); pn(hn); pw("\n" as *u8) 41 if hn <= 0 { pw("YT-PROBE: fetch FAILED\n" as *u8); sys_exit(1); return 1 } 42 43 // ground-truth signals across the FULL HTML (independent of my extraction) 44 pw("-- FULL HTML ground-truth signals --\n" as *u8) 45 pw(" ytInitialPlayerResponse: " as *u8); pn(count_occ(html, hn, "ytInitialPlayerResponse" as *u8)); pw("\n" as *u8) 46 pw(" consent.youtube (block): " as *u8); pn(count_occ(html, hn, "consent.youtube" as *u8)); pw("\n" as *u8) 47 pw(" streamingData: " as *u8); pn(count_occ(html, hn, "streamingData" as *u8)); pw("\n" as *u8) 48 pw(" adaptiveFormats: " as *u8); pn(count_occ(html, hn, "adaptiveFormats" as *u8)); pw("\n" as *u8) 49 pw(" signatureCipher (ciphered URLs): " as *u8); pn(count_occ(html, hn, "signatureCipher" as *u8)); pw("\n" as *u8) 50 pw(" videoplayback (stream path): " as *u8); pn(count_occ(html, hn, "videoplayback" as *u8)); pw("\n" as *u8) 51 pw(" googlevideo (CDN host): " as *u8); pn(count_occ(html, hn, "googlevideo" as *u8)); pw("\n" as *u8) 52 pw(" base.js (player script): " as *u8); pn(count_occ(html, hn, "base.js" as *u8)); pw("\n" as *u8) 53 pw(" &n= (throttle param): " as *u8); pn(count_occ(html, hn, "n=" as *u8)); pw("\n" as *u8) 54 55 // extract the player response blob 56 let blob: *u8 = sys_mmap(cap) 57 let bl: i64 = ms_extract_state(html, hn, "ytInitialPlayerResponse" as *u8, blob, cap) 58 pw("-- ytInitialPlayerResponse blob --\n" as *u8) 59 pw(" extracted bytes: " as *u8); pn(bl); pw("\n" as *u8) 60 if bl > 0 { 61 pw(" \"signatureCipher\" (ciphered URLs): " as *u8); pn(count_occ(blob, bl, "signatureCipher" as *u8)); pw("\n" as *u8) 62 pw(" \"googlevideo.com\" (CDN host refs): " as *u8); pn(count_occ(blob, bl, "googlevideo.com" as *u8)); pw("\n" as *u8) 63 pw(" \"itag\" (format entries): " as *u8); pn(count_occ(blob, bl, "itag" as *u8)); pw("\n" as *u8) 64 pw(" \"mimeType\": " as *u8); pn(count_occ(blob, bl, "mimeType" as *u8)); pw("\n" as *u8) 65 pw(" \"\\\"n\\\":\" n-param present: " as *u8); pn(count_occ(blob, bl, "&n=" as *u8)); pw("\n" as *u8) 66 let streams: *u8 = sys_mmap(K_MAGIC_262144) 67 let ns: i64 = ms_find_streams(blob, bl, streams, K_MAGIC_262144) 68 pw(" DIRECT (un-ciphered) media URLs found: " as *u8); pn(ns); pw("\n" as *u8) 69 if ns > 0 { pw(" first direct URLs:\n" as *u8); pw(streams); pw("\n" as *u8) } 70 } 71 // intelligent orchestrator on the REAL page: layered + resilient classify+rank 72 pw("-- intelligent orchestrator (mx_extract) --\n" as *u8) 73 let best: *u8 = sys_mmap(K_MAGIC_4096) 74 let kb: *i64 = sys_mmap(8) as *i64 75 let conf: i64 = mx_extract(html, hn, best, K_MAGIC_4096, kb) 76 pw(" best media kind: " as *u8); pn(kb[0]); pw(" confidence: " as *u8); pn(conf); pw("\n" as *u8) 77 if conf > 0 { 78 pw(" best URL (first 64 chars): " as *u8) 79 var q: i64 = 0 80 while q < 64 { if (best[q]&0xff) == 0 { q = 64 } else { sys_write(1, ((best as i64)+q) as *u8, 1); q = q + 1 } } 81 pw("\n" as *u8) 82 } 83 let title: *u8 = sys_mmap(K_MAGIC_1024) 84 let tl: i64 = xt_extract_title(html, hn, title, K_MAGIC_1024) 85 pw(" smart-name title (" as *u8); pn(tl); pw(" chars): " as *u8); if tl > 0 { sys_write(1, title, tl) } pw("\n" as *u8) 86 pw("== assessment: extraction works if blob>0; conf>0 = orchestrator found a candidate; signatureCipher>0 = needs base.js sig ==\n" as *u8) 87 sys_exit(0) 88 return 0 89}