nx_charset_test.nx source

1// nx_charset_test.nx -- charset transcode + detect + sniff(meta) + render KAT. 2// expect_exit: 0 3// license_tier: ORIGINAL 4 5import "nx_syscalls.nx" 6import "nx_charset.nx" 7import "nx_html_to_text.nx" 8 9func cs_eq(g: *u8, gn: i64, e: *u8, en: i64) -> i64 { 10 if gn != en { return 0 } 11 var i: i64 = 0 12 while i < gn { if (g[i]&0xff)!=(e[i]&0xff) { return 0 } i = i + 1 } 13 return 1 14} 15 16func main() -> i64 { 17 let i_lat1_eacute: *u8 = (sys_mmap(9)) as *u8 18 i_lat1_eacute[0] = 0xe9 as u8 19 let e_lat1_eacute: *u8 = (sys_mmap(10)) as *u8 20 e_lat1_eacute[0] = 0xc3 as u8 21 e_lat1_eacute[1] = 0xa9 as u8 22 let o_lat1_eacute: *u8 = (sys_mmap(64)) as *u8 23 let n_lat1_eacute: i64 = nx_charset_to_utf8(i_lat1_eacute, 1, 1, o_lat1_eacute, 64) 24 if cs_eq(o_lat1_eacute, n_lat1_eacute, e_lat1_eacute, 2) != 1 { return 100 } 25 26 let i_lat1_ascii: *u8 = (sys_mmap(9)) as *u8 27 i_lat1_ascii[0] = 0x41 as u8 28 let e_lat1_ascii: *u8 = (sys_mmap(9)) as *u8 29 e_lat1_ascii[0] = 0x41 as u8 30 let o_lat1_ascii: *u8 = (sys_mmap(64)) as *u8 31 let n_lat1_ascii: i64 = nx_charset_to_utf8(i_lat1_ascii, 1, 1, o_lat1_ascii, 64) 32 if cs_eq(o_lat1_ascii, n_lat1_ascii, e_lat1_ascii, 1) != 1 { return 101 } 33 34 let i_lat1_cafe: *u8 = (sys_mmap(12)) as *u8 35 i_lat1_cafe[0] = 0x43 as u8 36 i_lat1_cafe[1] = 0x61 as u8 37 i_lat1_cafe[2] = 0x66 as u8 38 i_lat1_cafe[3] = 0xe9 as u8 39 let e_lat1_cafe: *u8 = (sys_mmap(13)) as *u8 40 e_lat1_cafe[0] = 0x43 as u8 41 e_lat1_cafe[1] = 0x61 as u8 42 e_lat1_cafe[2] = 0x66 as u8 43 e_lat1_cafe[3] = 0xc3 as u8 44 e_lat1_cafe[4] = 0xa9 as u8 45 let o_lat1_cafe: *u8 = (sys_mmap(64)) as *u8 46 let n_lat1_cafe: i64 = nx_charset_to_utf8(i_lat1_cafe, 4, 1, o_lat1_cafe, 64) 47 if cs_eq(o_lat1_cafe, n_lat1_cafe, e_lat1_cafe, 5) != 1 { return 102 } 48 49 let i_w1252_lsquo: *u8 = (sys_mmap(9)) as *u8 50 i_w1252_lsquo[0] = 0x91 as u8 51 let e_w1252_lsquo: *u8 = (sys_mmap(11)) as *u8 52 e_w1252_lsquo[0] = 0xe2 as u8 53 e_w1252_lsquo[1] = 0x80 as u8 54 e_w1252_lsquo[2] = 0x98 as u8 55 let o_w1252_lsquo: *u8 = (sys_mmap(64)) as *u8 56 let n_w1252_lsquo: i64 = nx_charset_to_utf8(i_w1252_lsquo, 1, 2, o_w1252_lsquo, 64) 57 if cs_eq(o_w1252_lsquo, n_w1252_lsquo, e_w1252_lsquo, 3) != 1 { return 103 } 58 59 let i_w1252_euro: *u8 = (sys_mmap(9)) as *u8 60 i_w1252_euro[0] = 0x80 as u8 61 let e_w1252_euro: *u8 = (sys_mmap(11)) as *u8 62 e_w1252_euro[0] = 0xe2 as u8 63 e_w1252_euro[1] = 0x82 as u8 64 e_w1252_euro[2] = 0xac as u8 65 let o_w1252_euro: *u8 = (sys_mmap(64)) as *u8 66 let n_w1252_euro: i64 = nx_charset_to_utf8(i_w1252_euro, 1, 2, o_w1252_euro, 64) 67 if cs_eq(o_w1252_euro, n_w1252_euro, e_w1252_euro, 3) != 1 { return 104 } 68 69 let i_w1252_mdash: *u8 = (sys_mmap(9)) as *u8 70 i_w1252_mdash[0] = 0x97 as u8 71 let e_w1252_mdash: *u8 = (sys_mmap(11)) as *u8 72 e_w1252_mdash[0] = 0xe2 as u8 73 e_w1252_mdash[1] = 0x80 as u8 74 e_w1252_mdash[2] = 0x94 as u8 75 let o_w1252_mdash: *u8 = (sys_mmap(64)) as *u8 76 let n_w1252_mdash: i64 = nx_charset_to_utf8(i_w1252_mdash, 1, 2, o_w1252_mdash, 64) 77 if cs_eq(o_w1252_mdash, n_w1252_mdash, e_w1252_mdash, 3) != 1 { return 105 } 78 79 let i_w1252_latin: *u8 = (sys_mmap(9)) as *u8 80 i_w1252_latin[0] = 0xe9 as u8 81 let e_w1252_latin: *u8 = (sys_mmap(10)) as *u8 82 e_w1252_latin[0] = 0xc3 as u8 83 e_w1252_latin[1] = 0xa9 as u8 84 let o_w1252_latin: *u8 = (sys_mmap(64)) as *u8 85 let n_w1252_latin: i64 = nx_charset_to_utf8(i_w1252_latin, 1, 2, o_w1252_latin, 64) 86 if cs_eq(o_w1252_latin, n_w1252_latin, e_w1252_latin, 2) != 1 { return 106 } 87 88 let i_w1252_ascii: *u8 = (sys_mmap(9)) as *u8 89 i_w1252_ascii[0] = 0x5a as u8 90 let e_w1252_ascii: *u8 = (sys_mmap(9)) as *u8 91 e_w1252_ascii[0] = 0x5a as u8 92 let o_w1252_ascii: *u8 = (sys_mmap(64)) as *u8 93 let n_w1252_ascii: i64 = nx_charset_to_utf8(i_w1252_ascii, 1, 2, o_w1252_ascii, 64) 94 if cs_eq(o_w1252_ascii, n_w1252_ascii, e_w1252_ascii, 1) != 1 { return 107 } 95 96 let i_utf8_pass: *u8 = (sys_mmap(10)) as *u8 97 i_utf8_pass[0] = 0xc3 as u8 98 i_utf8_pass[1] = 0xa9 as u8 99 let e_utf8_pass: *u8 = (sys_mmap(10)) as *u8 100 e_utf8_pass[0] = 0xc3 as u8 101 e_utf8_pass[1] = 0xa9 as u8 102 let o_utf8_pass: *u8 = (sys_mmap(64)) as *u8 103 let n_utf8_pass: i64 = nx_charset_to_utf8(i_utf8_pass, 2, 0, o_utf8_pass, 64) 104 if cs_eq(o_utf8_pass, n_utf8_pass, e_utf8_pass, 2) != 1 { return 108 } 105 106 let c_det_utf8: *u8 = (sys_mmap(32)) as *u8 107 c_det_utf8[0] = 0x74 as u8 108 c_det_utf8[1] = 0x65 as u8 109 c_det_utf8[2] = 0x78 as u8 110 c_det_utf8[3] = 0x74 as u8 111 c_det_utf8[4] = 0x2f as u8 112 c_det_utf8[5] = 0x68 as u8 113 c_det_utf8[6] = 0x74 as u8 114 c_det_utf8[7] = 0x6d as u8 115 c_det_utf8[8] = 0x6c as u8 116 c_det_utf8[9] = 0x3b as u8 117 c_det_utf8[10] = 0x20 as u8 118 c_det_utf8[11] = 0x63 as u8 119 c_det_utf8[12] = 0x68 as u8 120 c_det_utf8[13] = 0x61 as u8 121 c_det_utf8[14] = 0x72 as u8 122 c_det_utf8[15] = 0x73 as u8 123 c_det_utf8[16] = 0x65 as u8 124 c_det_utf8[17] = 0x74 as u8 125 c_det_utf8[18] = 0x3d as u8 126 c_det_utf8[19] = 0x75 as u8 127 c_det_utf8[20] = 0x74 as u8 128 c_det_utf8[21] = 0x66 as u8 129 c_det_utf8[22] = 0x2d as u8 130 c_det_utf8[23] = 0x38 as u8 131 if nx_charset_from_content_type(c_det_utf8, 24) != 0 { return 200 } 132 133 let c_det_lat1: *u8 = (sys_mmap(36)) as *u8 134 c_det_lat1[0] = 0x74 as u8 135 c_det_lat1[1] = 0x65 as u8 136 c_det_lat1[2] = 0x78 as u8 137 c_det_lat1[3] = 0x74 as u8 138 c_det_lat1[4] = 0x2f as u8 139 c_det_lat1[5] = 0x68 as u8 140 c_det_lat1[6] = 0x74 as u8 141 c_det_lat1[7] = 0x6d as u8 142 c_det_lat1[8] = 0x6c as u8 143 c_det_lat1[9] = 0x3b as u8 144 c_det_lat1[10] = 0x63 as u8 145 c_det_lat1[11] = 0x68 as u8 146 c_det_lat1[12] = 0x61 as u8 147 c_det_lat1[13] = 0x72 as u8 148 c_det_lat1[14] = 0x73 as u8 149 c_det_lat1[15] = 0x65 as u8 150 c_det_lat1[16] = 0x74 as u8 151 c_det_lat1[17] = 0x3d as u8 152 c_det_lat1[18] = 0x69 as u8 153 c_det_lat1[19] = 0x73 as u8 154 c_det_lat1[20] = 0x6f as u8 155 c_det_lat1[21] = 0x2d as u8 156 c_det_lat1[22] = 0x38 as u8 157 c_det_lat1[23] = 0x38 as u8 158 c_det_lat1[24] = 0x35 as u8 159 c_det_lat1[25] = 0x39 as u8 160 c_det_lat1[26] = 0x2d as u8 161 c_det_lat1[27] = 0x31 as u8 162 if nx_charset_from_content_type(c_det_lat1, 28) != 1 { return 201 } 163 164 let c_det_w1252: *u8 = (sys_mmap(39)) as *u8 165 c_det_w1252[0] = 0x74 as u8 166 c_det_w1252[1] = 0x65 as u8 167 c_det_w1252[2] = 0x78 as u8 168 c_det_w1252[3] = 0x74 as u8 169 c_det_w1252[4] = 0x2f as u8 170 c_det_w1252[5] = 0x68 as u8 171 c_det_w1252[6] = 0x74 as u8 172 c_det_w1252[7] = 0x6d as u8 173 c_det_w1252[8] = 0x6c as u8 174 c_det_w1252[9] = 0x3b as u8 175 c_det_w1252[10] = 0x20 as u8 176 c_det_w1252[11] = 0x63 as u8 177 c_det_w1252[12] = 0x68 as u8 178 c_det_w1252[13] = 0x61 as u8 179 c_det_w1252[14] = 0x72 as u8 180 c_det_w1252[15] = 0x73 as u8 181 c_det_w1252[16] = 0x65 as u8 182 c_det_w1252[17] = 0x74 as u8 183 c_det_w1252[18] = 0x3d as u8 184 c_det_w1252[19] = 0x77 as u8 185 c_det_w1252[20] = 0x69 as u8 186 c_det_w1252[21] = 0x6e as u8 187 c_det_w1252[22] = 0x64 as u8 188 c_det_w1252[23] = 0x6f as u8 189 c_det_w1252[24] = 0x77 as u8 190 c_det_w1252[25] = 0x73 as u8 191 c_det_w1252[26] = 0x2d as u8 192 c_det_w1252[27] = 0x31 as u8 193 c_det_w1252[28] = 0x32 as u8 194 c_det_w1252[29] = 0x35 as u8 195 c_det_w1252[30] = 0x32 as u8 196 if nx_charset_from_content_type(c_det_w1252, 31) != 2 { return 202 } 197 198 let c_det_none: *u8 = (sys_mmap(17)) as *u8 199 c_det_none[0] = 0x74 as u8 200 c_det_none[1] = 0x65 as u8 201 c_det_none[2] = 0x78 as u8 202 c_det_none[3] = 0x74 as u8 203 c_det_none[4] = 0x2f as u8 204 c_det_none[5] = 0x68 as u8 205 c_det_none[6] = 0x74 as u8 206 c_det_none[7] = 0x6d as u8 207 c_det_none[8] = 0x6c as u8 208 if nx_charset_from_content_type(c_det_none, 9) != 0 { return 203 } 209 210 let c_det_latin1: *u8 = (sys_mmap(33)) as *u8 211 c_det_latin1[0] = 0x74 as u8 212 c_det_latin1[1] = 0x65 as u8 213 c_det_latin1[2] = 0x78 as u8 214 c_det_latin1[3] = 0x74 as u8 215 c_det_latin1[4] = 0x2f as u8 216 c_det_latin1[5] = 0x68 as u8 217 c_det_latin1[6] = 0x74 as u8 218 c_det_latin1[7] = 0x6d as u8 219 c_det_latin1[8] = 0x6c as u8 220 c_det_latin1[9] = 0x3b as u8 221 c_det_latin1[10] = 0x20 as u8 222 c_det_latin1[11] = 0x63 as u8 223 c_det_latin1[12] = 0x68 as u8 224 c_det_latin1[13] = 0x61 as u8 225 c_det_latin1[14] = 0x72 as u8 226 c_det_latin1[15] = 0x73 as u8 227 c_det_latin1[16] = 0x65 as u8 228 c_det_latin1[17] = 0x74 as u8 229 c_det_latin1[18] = 0x3d as u8 230 c_det_latin1[19] = 0x4c as u8 231 c_det_latin1[20] = 0x61 as u8 232 c_det_latin1[21] = 0x74 as u8 233 c_det_latin1[22] = 0x69 as u8 234 c_det_latin1[23] = 0x6e as u8 235 c_det_latin1[24] = 0x31 as u8 236 if nx_charset_from_content_type(c_det_latin1, 25) != 1 { return 204 } 237 238 let c_det_cp1252: *u8 = (sys_mmap(41)) as *u8 239 c_det_cp1252[0] = 0x61 as u8 240 c_det_cp1252[1] = 0x70 as u8 241 c_det_cp1252[2] = 0x70 as u8 242 c_det_cp1252[3] = 0x6c as u8 243 c_det_cp1252[4] = 0x69 as u8 244 c_det_cp1252[5] = 0x63 as u8 245 c_det_cp1252[6] = 0x61 as u8 246 c_det_cp1252[7] = 0x74 as u8 247 c_det_cp1252[8] = 0x69 as u8 248 c_det_cp1252[9] = 0x6f as u8 249 c_det_cp1252[10] = 0x6e as u8 250 c_det_cp1252[11] = 0x2f as u8 251 c_det_cp1252[12] = 0x78 as u8 252 c_det_cp1252[13] = 0x68 as u8 253 c_det_cp1252[14] = 0x74 as u8 254 c_det_cp1252[15] = 0x6d as u8 255 c_det_cp1252[16] = 0x6c as u8 256 c_det_cp1252[17] = 0x3b as u8 257 c_det_cp1252[18] = 0x20 as u8 258 c_det_cp1252[19] = 0x63 as u8 259 c_det_cp1252[20] = 0x68 as u8 260 c_det_cp1252[21] = 0x61 as u8 261 c_det_cp1252[22] = 0x72 as u8 262 c_det_cp1252[23] = 0x73 as u8 263 c_det_cp1252[24] = 0x65 as u8 264 c_det_cp1252[25] = 0x74 as u8 265 c_det_cp1252[26] = 0x3d as u8 266 c_det_cp1252[27] = 0x63 as u8 267 c_det_cp1252[28] = 0x70 as u8 268 c_det_cp1252[29] = 0x31 as u8 269 c_det_cp1252[30] = 0x32 as u8 270 c_det_cp1252[31] = 0x35 as u8 271 c_det_cp1252[32] = 0x32 as u8 272 if nx_charset_from_content_type(c_det_cp1252, 33) != 2 { return 205 } 273 274 // sniff sn_hdr_lat1 -> 1 275 let b_sn_hdr_lat1: *u8 = (sys_mmap(37)) as *u8 276 b_sn_hdr_lat1[0] = 0x74 as u8 277 b_sn_hdr_lat1[1] = 0x65 as u8 278 b_sn_hdr_lat1[2] = 0x78 as u8 279 b_sn_hdr_lat1[3] = 0x74 as u8 280 b_sn_hdr_lat1[4] = 0x2f as u8 281 b_sn_hdr_lat1[5] = 0x68 as u8 282 b_sn_hdr_lat1[6] = 0x74 as u8 283 b_sn_hdr_lat1[7] = 0x6d as u8 284 b_sn_hdr_lat1[8] = 0x6c as u8 285 b_sn_hdr_lat1[9] = 0x3b as u8 286 b_sn_hdr_lat1[10] = 0x20 as u8 287 b_sn_hdr_lat1[11] = 0x63 as u8 288 b_sn_hdr_lat1[12] = 0x68 as u8 289 b_sn_hdr_lat1[13] = 0x61 as u8 290 b_sn_hdr_lat1[14] = 0x72 as u8 291 b_sn_hdr_lat1[15] = 0x73 as u8 292 b_sn_hdr_lat1[16] = 0x65 as u8 293 b_sn_hdr_lat1[17] = 0x74 as u8 294 b_sn_hdr_lat1[18] = 0x3d as u8 295 b_sn_hdr_lat1[19] = 0x69 as u8 296 b_sn_hdr_lat1[20] = 0x73 as u8 297 b_sn_hdr_lat1[21] = 0x6f as u8 298 b_sn_hdr_lat1[22] = 0x2d as u8 299 b_sn_hdr_lat1[23] = 0x38 as u8 300 b_sn_hdr_lat1[24] = 0x38 as u8 301 b_sn_hdr_lat1[25] = 0x35 as u8 302 b_sn_hdr_lat1[26] = 0x39 as u8 303 b_sn_hdr_lat1[27] = 0x2d as u8 304 b_sn_hdr_lat1[28] = 0x31 as u8 305 if nx_charset_sniff(b_sn_hdr_lat1, 29) != 1 { return 300 } 306 307 // sniff sn_hdr_none -> -1 308 let b_sn_hdr_none: *u8 = (sys_mmap(17)) as *u8 309 b_sn_hdr_none[0] = 0x74 as u8 310 b_sn_hdr_none[1] = 0x65 as u8 311 b_sn_hdr_none[2] = 0x78 as u8 312 b_sn_hdr_none[3] = 0x74 as u8 313 b_sn_hdr_none[4] = 0x2f as u8 314 b_sn_hdr_none[5] = 0x68 as u8 315 b_sn_hdr_none[6] = 0x74 as u8 316 b_sn_hdr_none[7] = 0x6d as u8 317 b_sn_hdr_none[8] = 0x6c as u8 318 if nx_charset_sniff(b_sn_hdr_none, 9) != -1 { return 301 } 319 320 // sniff sn_meta_short -> 1 321 let b_sn_meta_short: *u8 = (sys_mmap(35)) as *u8 322 b_sn_meta_short[0] = 0x3c as u8 323 b_sn_meta_short[1] = 0x6d as u8 324 b_sn_meta_short[2] = 0x65 as u8 325 b_sn_meta_short[3] = 0x74 as u8 326 b_sn_meta_short[4] = 0x61 as u8 327 b_sn_meta_short[5] = 0x20 as u8 328 b_sn_meta_short[6] = 0x63 as u8 329 b_sn_meta_short[7] = 0x68 as u8 330 b_sn_meta_short[8] = 0x61 as u8 331 b_sn_meta_short[9] = 0x72 as u8 332 b_sn_meta_short[10] = 0x73 as u8 333 b_sn_meta_short[11] = 0x65 as u8 334 b_sn_meta_short[12] = 0x74 as u8 335 b_sn_meta_short[13] = 0x3d as u8 336 b_sn_meta_short[14] = 0x22 as u8 337 b_sn_meta_short[15] = 0x69 as u8 338 b_sn_meta_short[16] = 0x73 as u8 339 b_sn_meta_short[17] = 0x6f as u8 340 b_sn_meta_short[18] = 0x2d as u8 341 b_sn_meta_short[19] = 0x38 as u8 342 b_sn_meta_short[20] = 0x38 as u8 343 b_sn_meta_short[21] = 0x35 as u8 344 b_sn_meta_short[22] = 0x39 as u8 345 b_sn_meta_short[23] = 0x2d as u8 346 b_sn_meta_short[24] = 0x31 as u8 347 b_sn_meta_short[25] = 0x22 as u8 348 b_sn_meta_short[26] = 0x3e as u8 349 if nx_charset_sniff(b_sn_meta_short, 27) != 1 { return 302 } 350 351 // sniff sn_meta_utf8 -> 0 352 let b_sn_meta_utf8: *u8 = (sys_mmap(28)) as *u8 353 b_sn_meta_utf8[0] = 0x3c as u8 354 b_sn_meta_utf8[1] = 0x6d as u8 355 b_sn_meta_utf8[2] = 0x65 as u8 356 b_sn_meta_utf8[3] = 0x74 as u8 357 b_sn_meta_utf8[4] = 0x61 as u8 358 b_sn_meta_utf8[5] = 0x20 as u8 359 b_sn_meta_utf8[6] = 0x63 as u8 360 b_sn_meta_utf8[7] = 0x68 as u8 361 b_sn_meta_utf8[8] = 0x61 as u8 362 b_sn_meta_utf8[9] = 0x72 as u8 363 b_sn_meta_utf8[10] = 0x73 as u8 364 b_sn_meta_utf8[11] = 0x65 as u8 365 b_sn_meta_utf8[12] = 0x74 as u8 366 b_sn_meta_utf8[13] = 0x3d as u8 367 b_sn_meta_utf8[14] = 0x75 as u8 368 b_sn_meta_utf8[15] = 0x74 as u8 369 b_sn_meta_utf8[16] = 0x66 as u8 370 b_sn_meta_utf8[17] = 0x2d as u8 371 b_sn_meta_utf8[18] = 0x38 as u8 372 b_sn_meta_utf8[19] = 0x3e as u8 373 if nx_charset_sniff(b_sn_meta_utf8, 20) != 0 { return 303 } 374 375 // sniff sn_meta_heq -> 2 376 let b_sn_meta_heq: *u8 = (sys_mmap(95)) as *u8 377 b_sn_meta_heq[0] = 0x3c as u8 378 b_sn_meta_heq[1] = 0x68 as u8 379 b_sn_meta_heq[2] = 0x65 as u8 380 b_sn_meta_heq[3] = 0x61 as u8 381 b_sn_meta_heq[4] = 0x64 as u8 382 b_sn_meta_heq[5] = 0x3e as u8 383 b_sn_meta_heq[6] = 0x3c as u8 384 b_sn_meta_heq[7] = 0x6d as u8 385 b_sn_meta_heq[8] = 0x65 as u8 386 b_sn_meta_heq[9] = 0x74 as u8 387 b_sn_meta_heq[10] = 0x61 as u8 388 b_sn_meta_heq[11] = 0x20 as u8 389 b_sn_meta_heq[12] = 0x68 as u8 390 b_sn_meta_heq[13] = 0x74 as u8 391 b_sn_meta_heq[14] = 0x74 as u8 392 b_sn_meta_heq[15] = 0x70 as u8 393 b_sn_meta_heq[16] = 0x2d as u8 394 b_sn_meta_heq[17] = 0x65 as u8 395 b_sn_meta_heq[18] = 0x71 as u8 396 b_sn_meta_heq[19] = 0x75 as u8 397 b_sn_meta_heq[20] = 0x69 as u8 398 b_sn_meta_heq[21] = 0x76 as u8 399 b_sn_meta_heq[22] = 0x3d as u8 400 b_sn_meta_heq[23] = 0x22 as u8 401 b_sn_meta_heq[24] = 0x43 as u8 402 b_sn_meta_heq[25] = 0x6f as u8 403 b_sn_meta_heq[26] = 0x6e as u8 404 b_sn_meta_heq[27] = 0x74 as u8 405 b_sn_meta_heq[28] = 0x65 as u8 406 b_sn_meta_heq[29] = 0x6e as u8 407 b_sn_meta_heq[30] = 0x74 as u8 408 b_sn_meta_heq[31] = 0x2d as u8 409 b_sn_meta_heq[32] = 0x54 as u8 410 b_sn_meta_heq[33] = 0x79 as u8 411 b_sn_meta_heq[34] = 0x70 as u8 412 b_sn_meta_heq[35] = 0x65 as u8 413 b_sn_meta_heq[36] = 0x22 as u8 414 b_sn_meta_heq[37] = 0x20 as u8 415 b_sn_meta_heq[38] = 0x63 as u8 416 b_sn_meta_heq[39] = 0x6f as u8 417 b_sn_meta_heq[40] = 0x6e as u8 418 b_sn_meta_heq[41] = 0x74 as u8 419 b_sn_meta_heq[42] = 0x65 as u8 420 b_sn_meta_heq[43] = 0x6e as u8 421 b_sn_meta_heq[44] = 0x74 as u8 422 b_sn_meta_heq[45] = 0x3d as u8 423 b_sn_meta_heq[46] = 0x22 as u8 424 b_sn_meta_heq[47] = 0x74 as u8 425 b_sn_meta_heq[48] = 0x65 as u8 426 b_sn_meta_heq[49] = 0x78 as u8 427 b_sn_meta_heq[50] = 0x74 as u8 428 b_sn_meta_heq[51] = 0x2f as u8 429 b_sn_meta_heq[52] = 0x68 as u8 430 b_sn_meta_heq[53] = 0x74 as u8 431 b_sn_meta_heq[54] = 0x6d as u8 432 b_sn_meta_heq[55] = 0x6c as u8 433 b_sn_meta_heq[56] = 0x3b as u8 434 b_sn_meta_heq[57] = 0x20 as u8 435 b_sn_meta_heq[58] = 0x63 as u8 436 b_sn_meta_heq[59] = 0x68 as u8 437 b_sn_meta_heq[60] = 0x61 as u8 438 b_sn_meta_heq[61] = 0x72 as u8 439 b_sn_meta_heq[62] = 0x73 as u8 440 b_sn_meta_heq[63] = 0x65 as u8 441 b_sn_meta_heq[64] = 0x74 as u8 442 b_sn_meta_heq[65] = 0x3d as u8 443 b_sn_meta_heq[66] = 0x77 as u8 444 b_sn_meta_heq[67] = 0x69 as u8 445 b_sn_meta_heq[68] = 0x6e as u8 446 b_sn_meta_heq[69] = 0x64 as u8 447 b_sn_meta_heq[70] = 0x6f as u8 448 b_sn_meta_heq[71] = 0x77 as u8 449 b_sn_meta_heq[72] = 0x73 as u8 450 b_sn_meta_heq[73] = 0x2d as u8 451 b_sn_meta_heq[74] = 0x31 as u8 452 b_sn_meta_heq[75] = 0x32 as u8 453 b_sn_meta_heq[76] = 0x35 as u8 454 b_sn_meta_heq[77] = 0x32 as u8 455 b_sn_meta_heq[78] = 0x22 as u8 456 b_sn_meta_heq[79] = 0x3e as u8 457 b_sn_meta_heq[80] = 0x3c as u8 458 b_sn_meta_heq[81] = 0x2f as u8 459 b_sn_meta_heq[82] = 0x68 as u8 460 b_sn_meta_heq[83] = 0x65 as u8 461 b_sn_meta_heq[84] = 0x61 as u8 462 b_sn_meta_heq[85] = 0x64 as u8 463 b_sn_meta_heq[86] = 0x3e as u8 464 if nx_charset_sniff(b_sn_meta_heq, 87) != 2 { return 304 } 465 466 // sniff sn_nocs -> -1 467 let b_sn_nocs: *u8 = (sys_mmap(44)) as *u8 468 b_sn_nocs[0] = 0x3c as u8 469 b_sn_nocs[1] = 0x68 as u8 470 b_sn_nocs[2] = 0x74 as u8 471 b_sn_nocs[3] = 0x6d as u8 472 b_sn_nocs[4] = 0x6c as u8 473 b_sn_nocs[5] = 0x3e as u8 474 b_sn_nocs[6] = 0x3c as u8 475 b_sn_nocs[7] = 0x62 as u8 476 b_sn_nocs[8] = 0x6f as u8 477 b_sn_nocs[9] = 0x64 as u8 478 b_sn_nocs[10] = 0x79 as u8 479 b_sn_nocs[11] = 0x3e as u8 480 b_sn_nocs[12] = 0x70 as u8 481 b_sn_nocs[13] = 0x6c as u8 482 b_sn_nocs[14] = 0x61 as u8 483 b_sn_nocs[15] = 0x69 as u8 484 b_sn_nocs[16] = 0x6e as u8 485 b_sn_nocs[17] = 0x20 as u8 486 b_sn_nocs[18] = 0x74 as u8 487 b_sn_nocs[19] = 0x65 as u8 488 b_sn_nocs[20] = 0x78 as u8 489 b_sn_nocs[21] = 0x74 as u8 490 b_sn_nocs[22] = 0x3c as u8 491 b_sn_nocs[23] = 0x2f as u8 492 b_sn_nocs[24] = 0x62 as u8 493 b_sn_nocs[25] = 0x6f as u8 494 b_sn_nocs[26] = 0x64 as u8 495 b_sn_nocs[27] = 0x79 as u8 496 b_sn_nocs[28] = 0x3e as u8 497 b_sn_nocs[29] = 0x3c as u8 498 b_sn_nocs[30] = 0x2f as u8 499 b_sn_nocs[31] = 0x68 as u8 500 b_sn_nocs[32] = 0x74 as u8 501 b_sn_nocs[33] = 0x6d as u8 502 b_sn_nocs[34] = 0x6c as u8 503 b_sn_nocs[35] = 0x3e as u8 504 if nx_charset_sniff(b_sn_nocs, 36) != -1 { return 305 } 505 506 // INTEGRATION: <meta charset=windows-1252> page (no HTTP header): 507 // sniff(body) -> WIN1252 -> transcode -> render -> clean UTF-8 508 let mh: *u8 = (sys_mmap(54)) as *u8 509 mh[0] = 0x3c as u8 510 mh[1] = 0x6d as u8 511 mh[2] = 0x65 as u8 512 mh[3] = 0x74 as u8 513 mh[4] = 0x61 as u8 514 mh[5] = 0x20 as u8 515 mh[6] = 0x63 as u8 516 mh[7] = 0x68 as u8 517 mh[8] = 0x61 as u8 518 mh[9] = 0x72 as u8 519 mh[10] = 0x73 as u8 520 mh[11] = 0x65 as u8 521 mh[12] = 0x74 as u8 522 mh[13] = 0x3d as u8 523 mh[14] = 0x22 as u8 524 mh[15] = 0x77 as u8 525 mh[16] = 0x69 as u8 526 mh[17] = 0x6e as u8 527 mh[18] = 0x64 as u8 528 mh[19] = 0x6f as u8 529 mh[20] = 0x77 as u8 530 mh[21] = 0x73 as u8 531 mh[22] = 0x2d as u8 532 mh[23] = 0x31 as u8 533 mh[24] = 0x32 as u8 534 mh[25] = 0x35 as u8 535 mh[26] = 0x32 as u8 536 mh[27] = 0x22 as u8 537 mh[28] = 0x3e as u8 538 mh[29] = 0x3c as u8 539 mh[30] = 0x70 as u8 540 mh[31] = 0x3e as u8 541 mh[32] = 0x43 as u8 542 mh[33] = 0x61 as u8 543 mh[34] = 0x66 as u8 544 mh[35] = 0xe9 as u8 545 mh[36] = 0x20 as u8 546 mh[37] = 0x97 as u8 547 mh[38] = 0x20 as u8 548 mh[39] = 0x66 as u8 549 mh[40] = 0x69 as u8 550 mh[41] = 0x6e as u8 551 mh[42] = 0x3c as u8 552 mh[43] = 0x2f as u8 553 mh[44] = 0x70 as u8 554 mh[45] = 0x3e as u8 555 let mcs: i64 = nx_charset_sniff(mh, 46) 556 if mcs != 2 { return 400 } 557 let mtb: *u8 = (sys_mmap(256)) as *u8 558 let mtl: i64 = nx_charset_to_utf8(mh, 46, mcs, mtb, 256) 559 if mtl <= 0 { return 401 } 560 let mrb: *u8 = (sys_mmap(256)) as *u8 561 let mrl: i64 = nx_html_to_text(mtb, mtl, mrb, 256) 562 let mex: *u8 = (sys_mmap(22)) as *u8 563 mex[0] = 0x43 as u8 564 mex[1] = 0x61 as u8 565 mex[2] = 0x66 as u8 566 mex[3] = 0xc3 as u8 567 mex[4] = 0xa9 as u8 568 mex[5] = 0x20 as u8 569 mex[6] = 0xe2 as u8 570 mex[7] = 0x80 as u8 571 mex[8] = 0x94 as u8 572 mex[9] = 0x20 as u8 573 mex[10] = 0x66 as u8 574 mex[11] = 0x69 as u8 575 mex[12] = 0x6e as u8 576 mex[13] = 0x0a as u8 577 if cs_eq(mrb, mrl, mex, 14) != 1 { return 402 } 578 579 return 0 580}