| 1 | // SPDX-License-Identifier: GPL-3.0-or-later |
| 2 | |
| 3 | package main |
| 4 | |
| 5 | import ( |
| 6 | "archive/tar" |
| 7 | "archive/zip" |
| 8 | "bufio" |
| 9 | "bytes" |
| 10 | "compress/gzip" |
| 11 | "encoding/csv" |
| 12 | "fmt" |
| 13 | "io" |
| 14 | "math/big" |
| 15 | "net" |
| 16 | "net/netip" |
| 17 | "path" |
| 18 | "strconv" |
| 19 | "strings" |
| 20 | |
| 21 | "github.com/oschwald/maxminddb-golang" |
| 22 | "go4.org/netipx" |
| 23 | ) |
| 24 | |
| 25 | type dbipAsnMMDBRecord struct { |
| 26 | AutonomousSystemNumber *uint32 `maxminddb:"autonomous_system_number"` |
| 27 | AutonomousSystemOrganization string `maxminddb:"autonomous_system_organization"` |
| 28 | } |
| 29 | |
| 30 | type dbipCountryMMDBValue struct { |
| 31 | ISOCode string `maxminddb:"iso_code"` |
| 32 | } |
| 33 | |
| 34 | type dbipCityMMDBValue struct { |
| 35 | Names map[string]string `maxminddb:"names"` |
| 36 | } |
| 37 | |
| 38 | type dbipSubdivisionMMDBValue struct { |
| 39 | ISOCode string `maxminddb:"iso_code"` |
| 40 | Names map[string]string `maxminddb:"names"` |
| 41 | } |
| 42 | |
| 43 | type dbipLocationMMDBValue struct { |
| 44 | Latitude float64 `maxminddb:"latitude"` |
| 45 | Longitude float64 `maxminddb:"longitude"` |
| 46 | } |
| 47 | |
| 48 | type dbipGeoMMDBRecord struct { |
| 49 | Country *dbipCountryMMDBValue `maxminddb:"country"` |
| 50 | City *dbipCityMMDBValue `maxminddb:"city"` |
| 51 | Subdivisions []dbipSubdivisionMMDBValue `maxminddb:"subdivisions"` |
| 52 | Region string `maxminddb:"region"` |
| 53 | Location *dbipLocationMMDBValue `maxminddb:"location"` |
| 54 | } |
| 55 | |
| 56 | func loadRanges( |
| 57 | cfg config, |
| 58 | dl *downloader, |
| 59 | ) ([]asnRange, []geoRange, []generationDatasetRef, error) { |
| 60 | asnSources := make([][]asnRange, 0) |
| 61 | geoSources := make([][]geoRange, 0) |
| 62 | sourceRefs := make([]generationDatasetRef, 0, len(cfg.sources)) |
| 63 | |
| 64 | for i, source := range cfg.sources { |
| 65 | ref, content, err := dl.readDataset(source) |
| 66 | if err != nil { |
| 67 | return nil, nil, nil, fmt.Errorf("source %d (%s): %w", i, sourceLabel(source, i), err) |
| 68 | } |
| 69 | sourceRefs = append(sourceRefs, ref) |
| 70 | |
| 71 | switch source.family { |
| 72 | case sourceFamilyASN: |
| 73 | asnRanges, err := parseASNSource(source, content) |
| 74 | if err != nil { |
| 75 | return nil, nil, nil, fmt.Errorf("source %d (%s): %w", i, sourceLabel(source, i), err) |
| 76 | } |
| 77 | asnSources = append(asnSources, asnRanges) |
| 78 | case sourceFamilyGeo: |
| 79 | geoRanges, err := parseGeoSource(source, content) |
| 80 | if err != nil { |
| 81 | return nil, nil, nil, fmt.Errorf("source %d (%s): %w", i, sourceLabel(source, i), err) |
| 82 | } |
| 83 | geoSources = append(geoSources, geoRanges) |
| 84 | default: |
| 85 | return nil, nil, nil, fmt.Errorf( |
| 86 | "source %d (%s): unsupported family %q", |
| 87 | i, |
| 88 | sourceLabel(source, i), |
| 89 | source.family, |
| 90 | ) |
| 91 | } |
| 92 | } |
| 93 | |
| 94 | mergedASN, err := mergeAsnSources(asnSources) |
| 95 | if err != nil { |
| 96 | return nil, nil, nil, err |
| 97 | } |
| 98 | mergedGeo, err := mergeGeoSources(geoSources) |
| 99 | if err != nil { |
| 100 | return nil, nil, nil, err |
| 101 | } |
| 102 | |
| 103 | return mergedASN, mergedGeo, sourceRefs, nil |
| 104 | } |
| 105 | |
| 106 | func sourceLabel(source sourceEntry, index int) string { |
| 107 | if source.name != "" { |
| 108 | return source.name |
| 109 | } |
| 110 | return fmt.Sprintf("%s-%d", source.family, index) |
| 111 | } |
| 112 | |
| 113 | func parseASNSource(source sourceEntry, payload []byte) ([]asnRange, error) { |
| 114 | switch { |
| 115 | case source.provider == providerIPToASN && source.artifact == artifactIPToASNCombined: |
| 116 | if source.format != formatTSV { |
| 117 | return nil, fmt.Errorf("iptoasn combined requires tsv format, got %q", source.format) |
| 118 | } |
| 119 | return parseIPToASNCombinedTSVAsn(payload) |
| 120 | case source.provider == providerDBIP && source.artifact == artifactDBIPASNLite: |
| 121 | switch source.format { |
| 122 | case formatCSV: |
| 123 | return parseDBIPAsnCSV(payload) |
| 124 | case formatMMDB: |
| 125 | return parseDBIPAsnMMDB(payload) |
| 126 | default: |
| 127 | return nil, fmt.Errorf("unsupported dbip ASN format %q", source.format) |
| 128 | } |
| 129 | case source.provider == providerCAIDA && source.artifact == artifactCAIDAPrefix2AS: |
| 130 | if source.format != formatTSV { |
| 131 | return nil, fmt.Errorf("caida prefix2as requires tsv format, got %q", source.format) |
| 132 | } |
| 133 | return parseCAIDAPrefix2AS(payload) |
| 134 | case source.provider == providerMaxMind && source.artifact == artifactMaxMindGeoLite2ASN: |
| 135 | if source.format != formatMMDB { |
| 136 | return nil, fmt.Errorf("maxmind geolite2 ASN requires mmdb format, got %q", source.format) |
| 137 | } |
| 138 | return parseDBIPAsnMMDB(payload) |
| 139 | default: |
| 140 | return nil, fmt.Errorf( |
| 141 | "unsupported ASN source %q/%q", |
| 142 | source.provider, |
| 143 | source.artifact, |
| 144 | ) |
| 145 | } |
| 146 | } |
| 147 | |
| 148 | func parseGeoSource(source sourceEntry, payload []byte) ([]geoRange, error) { |
| 149 | switch { |
| 150 | case source.provider == providerIPToASN && source.artifact == artifactIPToASNCombined: |
| 151 | if source.format != formatTSV { |
| 152 | return nil, fmt.Errorf("iptoasn combined requires tsv format, got %q", source.format) |
| 153 | } |
| 154 | return parseIPToASNCombinedTSVGeo(payload) |
| 155 | case source.provider == providerDBIP && source.artifact == artifactDBIPCountryLite: |
| 156 | switch source.format { |
| 157 | case formatCSV: |
| 158 | return parseDBIPCountryCSV(payload) |
| 159 | case formatMMDB: |
| 160 | return parseDBIPGeoMMDB(payload) |
| 161 | default: |
| 162 | return nil, fmt.Errorf("unsupported dbip GEO format %q", source.format) |
| 163 | } |
| 164 | case source.provider == providerDBIP && source.artifact == artifactDBIPCityLite: |
| 165 | switch source.format { |
| 166 | case formatCSV: |
| 167 | return parseDBIPCityCSV(payload) |
| 168 | case formatMMDB: |
| 169 | return parseDBIPGeoMMDB(payload) |
| 170 | default: |
| 171 | return nil, fmt.Errorf("unsupported dbip GEO format %q", source.format) |
| 172 | } |
| 173 | case source.provider == providerMaxMind && source.artifact == artifactMaxMindGeoLite2Country: |
| 174 | if source.format != formatCSV { |
| 175 | return nil, fmt.Errorf("maxmind geolite2 country requires csv format, got %q", source.format) |
| 176 | } |
| 177 | return parseMaxMindCountryCSVZip(payload) |
| 178 | case source.provider == providerIP2Location && source.artifact == artifactIP2LocationCountryLite: |
| 179 | if source.format != formatCSV { |
| 180 | return nil, fmt.Errorf("ip2location country-lite requires csv format, got %q", source.format) |
| 181 | } |
| 182 | return parseIP2LocationCountryZip(payload) |
| 183 | case source.provider == providerIPDeny && source.artifact == artifactIPDenyCountryZones: |
| 184 | if source.format != formatCIDR { |
| 185 | return nil, fmt.Errorf("ipdeny country-zones requires cidr format, got %q", source.format) |
| 186 | } |
| 187 | return parseIPDenyCountryTarGZ(payload) |
| 188 | case source.provider == providerIPIP && source.artifact == artifactIPIPCountry: |
| 189 | if source.format != formatTXT { |
| 190 | return nil, fmt.Errorf("ipip country requires txt format, got %q", source.format) |
| 191 | } |
| 192 | return parseIPIPCountryZip(payload) |
| 193 | default: |
| 194 | return nil, fmt.Errorf( |
| 195 | "unsupported GEO source %q/%q", |
| 196 | source.provider, |
| 197 | source.artifact, |
| 198 | ) |
| 199 | } |
| 200 | } |
| 201 | |
| 202 | func estimatedRangeCapacity(size uint64, averageLineBytes, maxCapacity int) int { |
| 203 | if size == 0 || averageLineBytes <= 0 || maxCapacity <= 0 { |
| 204 | return 0 |
| 205 | } |
| 206 | capacity := size / uint64(averageLineBytes) |
| 207 | if capacity > uint64(maxCapacity) { |
| 208 | return maxCapacity |
| 209 | } |
| 210 | return int(capacity) |
| 211 | } |
| 212 | |
| 213 | func parseIPToASNCombinedTSVAsn(payload []byte) ([]asnRange, error) { |
| 214 | asnRanges := make([]asnRange, 0, estimatedRangeCapacity(uint64(len(payload)), 64, 1<<20)) |
| 215 | |
| 216 | scanner := bufio.NewScanner(bytes.NewReader(payload)) |
| 217 | lineNo := 0 |
| 218 | for scanner.Scan() { |
| 219 | lineNo++ |
| 220 | line := strings.TrimSpace(scanner.Text()) |
| 221 | if line == "" || strings.HasPrefix(line, "#") { |
| 222 | continue |
| 223 | } |
| 224 | parts := strings.Split(line, "\t") |
| 225 | if len(parts) < 5 { |
| 226 | return nil, fmt.Errorf("iptoasn line %d: expected at least 5 columns", lineNo) |
| 227 | } |
| 228 | start, end, err := parseRangeEndpoints(parts[0], parts[1]) |
| 229 | if err != nil { |
| 230 | return nil, fmt.Errorf("iptoasn line %d: %w", lineNo, err) |
| 231 | } |
| 232 | asn, err := parseASN(parts[2]) |
| 233 | if err != nil { |
| 234 | return nil, fmt.Errorf("iptoasn line %d: %w", lineNo, err) |
| 235 | } |
| 236 | org := strings.TrimSpace(parts[4]) |
| 237 | |
| 238 | asnRange := asnRange{start: start, end: end, asn: asn, org: org} |
| 239 | if err := asnRange.validate(); err != nil { |
| 240 | return nil, fmt.Errorf("iptoasn line %d: %w", lineNo, err) |
| 241 | } |
| 242 | asnRanges = append(asnRanges, asnRange) |
| 243 | } |
| 244 | if err := scanner.Err(); err != nil { |
| 245 | return nil, fmt.Errorf("failed to scan iptoasn payload: %w", err) |
| 246 | } |
| 247 | return asnRanges, nil |
| 248 | } |
| 249 | |
| 250 | func parseIPToASNCombinedTSVGeo(payload []byte) ([]geoRange, error) { |
| 251 | geoRanges := make([]geoRange, 0, estimatedRangeCapacity(uint64(len(payload)), 64, 1<<20)) |
| 252 | |
| 253 | scanner := bufio.NewScanner(bytes.NewReader(payload)) |
| 254 | lineNo := 0 |
| 255 | for scanner.Scan() { |
| 256 | lineNo++ |
| 257 | line := strings.TrimSpace(scanner.Text()) |
| 258 | if line == "" || strings.HasPrefix(line, "#") { |
| 259 | continue |
| 260 | } |
| 261 | parts := strings.Split(line, "\t") |
| 262 | if len(parts) < 5 { |
| 263 | return nil, fmt.Errorf("iptoasn line %d: expected at least 5 columns", lineNo) |
| 264 | } |
| 265 | start, end, err := parseRangeEndpoints(parts[0], parts[1]) |
| 266 | if err != nil { |
| 267 | return nil, fmt.Errorf("iptoasn line %d: %w", lineNo, err) |
| 268 | } |
| 269 | country := normalizeCountry(parts[3]) |
| 270 | if country == "" { |
| 271 | continue |
| 272 | } |
| 273 | rec := geoRange{start: start, end: end, country: country} |
| 274 | if err := rec.validate(); err != nil { |
| 275 | return nil, fmt.Errorf("iptoasn line %d: %w", lineNo, err) |
| 276 | } |
| 277 | geoRanges = append(geoRanges, rec) |
| 278 | } |
| 279 | if err := scanner.Err(); err != nil { |
| 280 | return nil, fmt.Errorf("failed to scan iptoasn payload: %w", err) |
| 281 | } |
| 282 | return geoRanges, nil |
| 283 | } |
| 284 | |
| 285 | func parseDBIPAsnCSV(payload []byte) ([]asnRange, error) { |
| 286 | reader := csv.NewReader(strings.NewReader(string(payload))) |
| 287 | reader.FieldsPerRecord = -1 |
| 288 | reader.TrimLeadingSpace = true |
| 289 | |
| 290 | out := make([]asnRange, 0, estimatedRangeCapacity(uint64(len(payload)), 80, 1<<18)) |
| 291 | lineNo := 0 |
| 292 | for { |
| 293 | row, err := reader.Read() |
| 294 | if err != nil { |
| 295 | if err == io.EOF { |
| 296 | break |
| 297 | } |
| 298 | return nil, fmt.Errorf("dbip asn line %d: %w", lineNo+1, err) |
| 299 | } |
| 300 | lineNo++ |
| 301 | if len(row) == 0 { |
| 302 | continue |
| 303 | } |
| 304 | if strings.HasPrefix(strings.TrimSpace(row[0]), "#") { |
| 305 | continue |
| 306 | } |
| 307 | if len(row) < 3 { |
| 308 | return nil, fmt.Errorf("dbip asn line %d: expected >= 3 columns", lineNo) |
| 309 | } |
| 310 | start, end, err := parseRangeEndpoints(row[0], row[1]) |
| 311 | if err != nil { |
| 312 | return nil, fmt.Errorf("dbip asn line %d: %w", lineNo, err) |
| 313 | } |
| 314 | asn, err := parseASN(row[2]) |
| 315 | if err != nil { |
| 316 | return nil, fmt.Errorf("dbip asn line %d: %w", lineNo, err) |
| 317 | } |
| 318 | org := "" |
| 319 | if len(row) > 3 { |
| 320 | org = strings.TrimSpace(row[3]) |
| 321 | } |
| 322 | rec := asnRange{start: start, end: end, asn: asn, org: org} |
| 323 | if err := rec.validate(); err != nil { |
| 324 | return nil, fmt.Errorf("dbip asn line %d: %w", lineNo, err) |
| 325 | } |
| 326 | out = append(out, rec) |
| 327 | } |
| 328 | return out, nil |
| 329 | } |
| 330 | |
| 331 | func parseCAIDAPrefix2AS(payload []byte) ([]asnRange, error) { |
| 332 | out := make([]asnRange, 0, estimatedRangeCapacity(uint64(len(payload)), 32, 1<<20)) |
| 333 | scanner := bufio.NewScanner(bytes.NewReader(payload)) |
| 334 | scanner.Buffer(make([]byte, 0, 64*1024), 1024*1024) |
| 335 | |
| 336 | lineNo := 0 |
| 337 | for scanner.Scan() { |
| 338 | lineNo++ |
| 339 | line := strings.TrimSpace(scanner.Text()) |
| 340 | if line == "" || strings.HasPrefix(line, "#") { |
| 341 | continue |
| 342 | } |
| 343 | |
| 344 | fields := strings.Fields(line) |
| 345 | if len(fields) < 3 { |
| 346 | return nil, fmt.Errorf("caida prefix2as line %d: expected >= 3 columns", lineNo) |
| 347 | } |
| 348 | |
| 349 | prefixLength, err := strconv.Atoi(strings.TrimSpace(fields[1])) |
| 350 | if err != nil { |
| 351 | return nil, fmt.Errorf("caida prefix2as line %d: invalid prefix length %q: %w", lineNo, fields[1], err) |
| 352 | } |
| 353 | prefix, err := netip.ParsePrefix(fmt.Sprintf("%s/%d", strings.TrimSpace(fields[0]), prefixLength)) |
| 354 | if err != nil { |
| 355 | return nil, fmt.Errorf("caida prefix2as line %d: invalid prefix: %w", lineNo, err) |
| 356 | } |
| 357 | start, end := rangeFromPrefix(prefix.Masked()) |
| 358 | |
| 359 | asn, err := parsePrimaryASN(fields[2]) |
| 360 | if err != nil { |
| 361 | return nil, fmt.Errorf("caida prefix2as line %d: %w", lineNo, err) |
| 362 | } |
| 363 | if asn == 0 { |
| 364 | continue |
| 365 | } |
| 366 | |
| 367 | rec := asnRange{start: start, end: end, asn: asn} |
| 368 | if err := rec.validate(); err != nil { |
| 369 | return nil, fmt.Errorf("caida prefix2as line %d: %w", lineNo, err) |
| 370 | } |
| 371 | out = append(out, rec) |
| 372 | } |
| 373 | if err := scanner.Err(); err != nil { |
| 374 | return nil, fmt.Errorf("failed to scan caida prefix2as payload: %w", err) |
| 375 | } |
| 376 | return out, nil |
| 377 | } |
| 378 | |
| 379 | func parseDBIPCountryCSV(payload []byte) ([]geoRange, error) { |
| 380 | reader := csv.NewReader(strings.NewReader(string(payload))) |
| 381 | reader.FieldsPerRecord = -1 |
| 382 | reader.TrimLeadingSpace = true |
| 383 | |
| 384 | out := make([]geoRange, 0, estimatedRangeCapacity(uint64(len(payload)), 64, 1<<18)) |
| 385 | lineNo := 0 |
| 386 | for { |
| 387 | row, err := reader.Read() |
| 388 | if err != nil { |
| 389 | if err == io.EOF { |
| 390 | break |
| 391 | } |
| 392 | return nil, fmt.Errorf("dbip country line %d: %w", lineNo+1, err) |
| 393 | } |
| 394 | lineNo++ |
| 395 | if len(row) == 0 { |
| 396 | continue |
| 397 | } |
| 398 | if strings.HasPrefix(strings.TrimSpace(row[0]), "#") { |
| 399 | continue |
| 400 | } |
| 401 | if len(row) < 3 { |
| 402 | return nil, fmt.Errorf("dbip country line %d: expected >= 3 columns", lineNo) |
| 403 | } |
| 404 | start, end, err := parseRangeEndpoints(row[0], row[1]) |
| 405 | if err != nil { |
| 406 | return nil, fmt.Errorf("dbip country line %d: %w", lineNo, err) |
| 407 | } |
| 408 | country := normalizeCountry(row[2]) |
| 409 | if country == "" { |
| 410 | continue |
| 411 | } |
| 412 | rec := geoRange{start: start, end: end, country: country} |
| 413 | if err := rec.validate(); err != nil { |
| 414 | return nil, fmt.Errorf("dbip country line %d: %w", lineNo, err) |
| 415 | } |
| 416 | out = append(out, rec) |
| 417 | } |
| 418 | return out, nil |
| 419 | } |
| 420 | |
| 421 | func parseDBIPCityCSV(payload []byte) ([]geoRange, error) { |
| 422 | reader := csv.NewReader(strings.NewReader(string(payload))) |
| 423 | reader.FieldsPerRecord = -1 |
| 424 | reader.TrimLeadingSpace = true |
| 425 | |
| 426 | out := make([]geoRange, 0, estimatedRangeCapacity(uint64(len(payload)), 128, 1<<18)) |
| 427 | lineNo := 0 |
| 428 | for { |
| 429 | row, err := reader.Read() |
| 430 | if err != nil { |
| 431 | if err == io.EOF { |
| 432 | break |
| 433 | } |
| 434 | return nil, fmt.Errorf("dbip city line %d: %w", lineNo+1, err) |
| 435 | } |
| 436 | lineNo++ |
| 437 | if len(row) == 0 { |
| 438 | continue |
| 439 | } |
| 440 | if strings.HasPrefix(strings.TrimSpace(row[0]), "#") { |
| 441 | continue |
| 442 | } |
| 443 | if len(row) < 8 { |
| 444 | return nil, fmt.Errorf("dbip city line %d: expected >= 8 columns", lineNo) |
| 445 | } |
| 446 | |
| 447 | start, end, err := parseRangeEndpoints(row[0], row[1]) |
| 448 | if err != nil { |
| 449 | return nil, fmt.Errorf("dbip city line %d: %w", lineNo, err) |
| 450 | } |
| 451 | |
| 452 | country := normalizeCountry(row[3]) |
| 453 | state := strings.TrimSpace(row[4]) |
| 454 | city := strings.TrimSpace(row[5]) |
| 455 | |
| 456 | rec := geoRange{ |
| 457 | start: start, |
| 458 | end: end, |
| 459 | country: country, |
| 460 | state: state, |
| 461 | city: city, |
| 462 | } |
| 463 | |
| 464 | if latRaw := strings.TrimSpace(row[6]); latRaw != "" { |
| 465 | lat, err := strconv.ParseFloat(latRaw, 64) |
| 466 | if err != nil { |
| 467 | return nil, fmt.Errorf("dbip city line %d: invalid latitude %q: %w", lineNo, latRaw, err) |
| 468 | } |
| 469 | rec.latitude = lat |
| 470 | rec.hasLocation = true |
| 471 | } |
| 472 | if lonRaw := strings.TrimSpace(row[7]); lonRaw != "" { |
| 473 | lon, err := strconv.ParseFloat(lonRaw, 64) |
| 474 | if err != nil { |
| 475 | return nil, fmt.Errorf("dbip city line %d: invalid longitude %q: %w", lineNo, lonRaw, err) |
| 476 | } |
| 477 | rec.longitude = lon |
| 478 | rec.hasLocation = true |
| 479 | } |
| 480 | |
| 481 | if err := rec.validate(); err != nil { |
| 482 | return nil, fmt.Errorf("dbip city line %d: %w", lineNo, err) |
| 483 | } |
| 484 | out = append(out, rec) |
| 485 | } |
| 486 | return out, nil |
| 487 | } |
| 488 | |
| 489 | func parseDBIPAsnMMDB(payload []byte) ([]asnRange, error) { |
| 490 | reader, err := maxminddb.FromBytes(payload) |
| 491 | if err != nil { |
| 492 | return nil, fmt.Errorf("failed to open dbip ASN mmdb: %w", err) |
| 493 | } |
| 494 | |
| 495 | out := make([]asnRange, 0) |
| 496 | networks := reader.Networks(maxminddb.SkipAliasedNetworks) |
| 497 | for networks.Next() { |
| 498 | var record dbipAsnMMDBRecord |
| 499 | network, err := networks.Network(&record) |
| 500 | if err != nil { |
| 501 | return nil, fmt.Errorf("failed to decode dbip ASN network: %w", err) |
| 502 | } |
| 503 | start, end, err := rangeFromIPNet(network) |
| 504 | if err != nil { |
| 505 | return nil, err |
| 506 | } |
| 507 | asn := uint32(0) |
| 508 | if record.AutonomousSystemNumber != nil { |
| 509 | asn = *record.AutonomousSystemNumber |
| 510 | } |
| 511 | rec := asnRange{ |
| 512 | start: start, |
| 513 | end: end, |
| 514 | asn: asn, |
| 515 | org: strings.TrimSpace(record.AutonomousSystemOrganization), |
| 516 | } |
| 517 | if rec.asn == 0 && rec.org == "" { |
| 518 | continue |
| 519 | } |
| 520 | if err := rec.validate(); err != nil { |
| 521 | return nil, err |
| 522 | } |
| 523 | out = append(out, rec) |
| 524 | } |
| 525 | if err := networks.Err(); err != nil { |
| 526 | return nil, fmt.Errorf("failed to iterate dbip ASN mmdb: %w", err) |
| 527 | } |
| 528 | return out, nil |
| 529 | } |
| 530 | |
| 531 | func parseDBIPGeoMMDB(payload []byte) ([]geoRange, error) { |
| 532 | reader, err := maxminddb.FromBytes(payload) |
| 533 | if err != nil { |
| 534 | return nil, fmt.Errorf("failed to open dbip GEO mmdb: %w", err) |
| 535 | } |
| 536 | |
| 537 | out := make([]geoRange, 0) |
| 538 | networks := reader.Networks(maxminddb.SkipAliasedNetworks) |
| 539 | for networks.Next() { |
| 540 | var record dbipGeoMMDBRecord |
| 541 | network, err := networks.Network(&record) |
| 542 | if err != nil { |
| 543 | return nil, fmt.Errorf("failed to decode dbip GEO network: %w", err) |
| 544 | } |
| 545 | start, end, err := rangeFromIPNet(network) |
| 546 | if err != nil { |
| 547 | return nil, err |
| 548 | } |
| 549 | |
| 550 | rec := geoRange{ |
| 551 | start: start, |
| 552 | end: end, |
| 553 | country: dbipCountryCode(record.Country), |
| 554 | state: dbipStateName(record), |
| 555 | city: dbipCityName(record.City), |
| 556 | } |
| 557 | if record.Location != nil { |
| 558 | rec.latitude = record.Location.Latitude |
| 559 | rec.longitude = record.Location.Longitude |
| 560 | rec.hasLocation = true |
| 561 | } |
| 562 | if rec.country == "" && rec.state == "" && rec.city == "" && !rec.hasLocation { |
| 563 | continue |
| 564 | } |
| 565 | if err := rec.validate(); err != nil { |
| 566 | return nil, err |
| 567 | } |
| 568 | out = append(out, rec) |
| 569 | } |
| 570 | if err := networks.Err(); err != nil { |
| 571 | return nil, fmt.Errorf("failed to iterate dbip GEO mmdb: %w", err) |
| 572 | } |
| 573 | return out, nil |
| 574 | } |
| 575 | |
| 576 | func parseMaxMindCountryCSVZip(payload []byte) ([]geoRange, error) { |
| 577 | archive, err := zip.NewReader(bytes.NewReader(payload), int64(len(payload))) |
| 578 | if err != nil { |
| 579 | return nil, fmt.Errorf("failed to open maxmind country csv zip: %w", err) |
| 580 | } |
| 581 | |
| 582 | countryByID, err := parseMaxMindCountryLocations(archive) |
| 583 | if err != nil { |
| 584 | return nil, err |
| 585 | } |
| 586 | |
| 587 | out := make([]geoRange, 0) |
| 588 | for _, suffix := range []string{ |
| 589 | "GeoLite2-Country-Blocks-IPv4.csv", |
| 590 | "GeoLite2-Country-Blocks-IPv6.csv", |
| 591 | } { |
| 592 | ranges, err := parseMaxMindCountryBlocks(archive, suffix, countryByID) |
| 593 | if err != nil { |
| 594 | return nil, err |
| 595 | } |
| 596 | out = append(out, ranges...) |
| 597 | } |
| 598 | return out, nil |
| 599 | } |
| 600 | |
| 601 | func parseMaxMindCountryLocations(archive *zip.Reader) (map[string]string, error) { |
| 602 | file, err := openZipEntrySuffix(archive, "GeoLite2-Country-Locations-en.csv") |
| 603 | if err != nil { |
| 604 | return nil, err |
| 605 | } |
| 606 | rc, err := file.Open() |
| 607 | if err != nil { |
| 608 | return nil, err |
| 609 | } |
| 610 | defer rc.Close() |
| 611 | |
| 612 | csvr := csv.NewReader(rc) |
| 613 | csvr.FieldsPerRecord = -1 |
| 614 | header, err := csvr.Read() |
| 615 | if err != nil { |
| 616 | return nil, fmt.Errorf("failed to read maxmind locations header: %w", err) |
| 617 | } |
| 618 | idx := csvHeaderIndex(header) |
| 619 | idIdx, ok := idx["geoname_id"] |
| 620 | if !ok { |
| 621 | return nil, fmt.Errorf("maxmind locations missing geoname_id column") |
| 622 | } |
| 623 | countryIdx, ok := idx["country_iso_code"] |
| 624 | if !ok { |
| 625 | return nil, fmt.Errorf("maxmind locations missing country_iso_code column") |
| 626 | } |
| 627 | |
| 628 | out := map[string]string{} |
| 629 | lineNo := 1 |
| 630 | for { |
| 631 | row, err := csvr.Read() |
| 632 | if err == io.EOF { |
| 633 | break |
| 634 | } |
| 635 | if err != nil { |
| 636 | return nil, fmt.Errorf("maxmind locations line %d: %w", lineNo+1, err) |
| 637 | } |
| 638 | lineNo++ |
| 639 | if len(row) <= idIdx || len(row) <= countryIdx { |
| 640 | continue |
| 641 | } |
| 642 | id := strings.TrimSpace(row[idIdx]) |
| 643 | country := normalizeCountry(row[countryIdx]) |
| 644 | if id != "" && country != "" { |
| 645 | out[id] = country |
| 646 | } |
| 647 | } |
| 648 | return out, nil |
| 649 | } |
| 650 | |
| 651 | func parseMaxMindCountryBlocks( |
| 652 | archive *zip.Reader, |
| 653 | suffix string, |
| 654 | countryByID map[string]string, |
| 655 | ) ([]geoRange, error) { |
| 656 | file, err := openZipEntrySuffix(archive, suffix) |
| 657 | if err != nil { |
| 658 | return nil, err |
| 659 | } |
| 660 | rc, err := file.Open() |
| 661 | if err != nil { |
| 662 | return nil, err |
| 663 | } |
| 664 | defer rc.Close() |
| 665 | |
| 666 | csvr := csv.NewReader(rc) |
| 667 | csvr.FieldsPerRecord = -1 |
| 668 | header, err := csvr.Read() |
| 669 | if err != nil { |
| 670 | return nil, fmt.Errorf("failed to read maxmind blocks header %s: %w", suffix, err) |
| 671 | } |
| 672 | idx := csvHeaderIndex(header) |
| 673 | networkIdx, ok := idx["network"] |
| 674 | if !ok { |
| 675 | return nil, fmt.Errorf("maxmind blocks %s missing network column", suffix) |
| 676 | } |
| 677 | idColumns := []string{"geoname_id", "registered_country_geoname_id", "represented_country_geoname_id"} |
| 678 | |
| 679 | out := make([]geoRange, 0, estimatedRangeCapacity(file.UncompressedSize64, 96, 1<<18)) |
| 680 | lineNo := 1 |
| 681 | for { |
| 682 | row, err := csvr.Read() |
| 683 | if err == io.EOF { |
| 684 | break |
| 685 | } |
| 686 | if err != nil { |
| 687 | return nil, fmt.Errorf("maxmind blocks %s line %d: %w", suffix, lineNo+1, err) |
| 688 | } |
| 689 | lineNo++ |
| 690 | if len(row) <= networkIdx { |
| 691 | continue |
| 692 | } |
| 693 | |
| 694 | country := "" |
| 695 | for _, column := range idColumns { |
| 696 | columnIdx, ok := idx[column] |
| 697 | if !ok || len(row) <= columnIdx { |
| 698 | continue |
| 699 | } |
| 700 | if c := countryByID[strings.TrimSpace(row[columnIdx])]; c != "" { |
| 701 | country = c |
| 702 | break |
| 703 | } |
| 704 | } |
| 705 | if country == "" { |
| 706 | continue |
| 707 | } |
| 708 | |
| 709 | rec, err := geoRangeFromToken(row[networkIdx], country) |
| 710 | if err != nil { |
| 711 | return nil, fmt.Errorf("maxmind blocks %s line %d: %w", suffix, lineNo, err) |
| 712 | } |
| 713 | out = append(out, rec) |
| 714 | } |
| 715 | return out, nil |
| 716 | } |
| 717 | |
| 718 | func parseIP2LocationCountryZip(payload []byte) ([]geoRange, error) { |
| 719 | archive, err := zip.NewReader(bytes.NewReader(payload), int64(len(payload))) |
| 720 | if err != nil { |
| 721 | return nil, fmt.Errorf("failed to open ip2location country zip: %w", err) |
| 722 | } |
| 723 | file, err := openZipEntryBase(archive, "IP2LOCATION-LITE-DB1.CSV") |
| 724 | if err != nil { |
| 725 | return nil, err |
| 726 | } |
| 727 | rc, err := file.Open() |
| 728 | if err != nil { |
| 729 | return nil, err |
| 730 | } |
| 731 | defer rc.Close() |
| 732 | |
| 733 | csvr := csv.NewReader(rc) |
| 734 | csvr.FieldsPerRecord = -1 |
| 735 | out := make([]geoRange, 0, estimatedRangeCapacity(file.UncompressedSize64, 64, 1<<18)) |
| 736 | lineNo := 0 |
| 737 | for { |
| 738 | row, err := csvr.Read() |
| 739 | if err == io.EOF { |
| 740 | break |
| 741 | } |
| 742 | if err != nil { |
| 743 | return nil, fmt.Errorf("ip2location line %d: %w", lineNo+1, err) |
| 744 | } |
| 745 | lineNo++ |
| 746 | if len(row) < 3 { |
| 747 | continue |
| 748 | } |
| 749 | country := normalizeCountry(row[2]) |
| 750 | if country == "" { |
| 751 | continue |
| 752 | } |
| 753 | start, end, err := parseRangeEndpoints(row[0], row[1]) |
| 754 | if err != nil { |
| 755 | return nil, fmt.Errorf("ip2location line %d: %w", lineNo, err) |
| 756 | } |
| 757 | rec := geoRange{start: start, end: end, country: country} |
| 758 | if err := rec.validate(); err != nil { |
| 759 | return nil, fmt.Errorf("ip2location line %d: %w", lineNo, err) |
| 760 | } |
| 761 | out = append(out, rec) |
| 762 | } |
| 763 | return out, nil |
| 764 | } |
| 765 | |
| 766 | func parseIPDenyCountryTarGZ(payload []byte) ([]geoRange, error) { |
| 767 | gz, err := gzip.NewReader(bytes.NewReader(payload)) |
| 768 | if err != nil { |
| 769 | return nil, fmt.Errorf("failed to open ipdeny tar.gz: %w", err) |
| 770 | } |
| 771 | defer gz.Close() |
| 772 | |
| 773 | out := make([]geoRange, 0) |
| 774 | tr := tar.NewReader(gz) |
| 775 | for { |
| 776 | header, err := tr.Next() |
| 777 | if err == io.EOF { |
| 778 | break |
| 779 | } |
| 780 | if err != nil { |
| 781 | return nil, fmt.Errorf("failed to read ipdeny tar: %w", err) |
| 782 | } |
| 783 | name := path.Base(header.Name) |
| 784 | zoneName, ok := strings.CutSuffix(strings.ToLower(name), ".zone") |
| 785 | if header.Typeflag != tar.TypeReg || !ok { |
| 786 | continue |
| 787 | } |
| 788 | country := normalizeCountry(zoneName) |
| 789 | if country == "" { |
| 790 | continue |
| 791 | } |
| 792 | ranges, err := parseCountryTokenLines(tr, country, "ipdeny "+header.Name) |
| 793 | if err != nil { |
| 794 | return nil, err |
| 795 | } |
| 796 | out = append(out, ranges...) |
| 797 | } |
| 798 | return out, nil |
| 799 | } |
| 800 | |
| 801 | func parseIPIPCountryZip(payload []byte) ([]geoRange, error) { |
| 802 | archive, err := zip.NewReader(bytes.NewReader(payload), int64(len(payload))) |
| 803 | if err != nil { |
| 804 | return nil, fmt.Errorf("failed to open ipip country zip: %w", err) |
| 805 | } |
| 806 | file, err := openZipEntryBase(archive, "country.txt") |
| 807 | if err != nil { |
| 808 | return nil, err |
| 809 | } |
| 810 | rc, err := file.Open() |
| 811 | if err != nil { |
| 812 | return nil, err |
| 813 | } |
| 814 | defer rc.Close() |
| 815 | |
| 816 | out := make([]geoRange, 0, estimatedRangeCapacity(file.UncompressedSize64, 48, 1<<18)) |
| 817 | scanner := bufio.NewScanner(rc) |
| 818 | scanner.Buffer(make([]byte, 0, 64*1024), 1024*1024) |
| 819 | lineNo := 0 |
| 820 | for scanner.Scan() { |
| 821 | lineNo++ |
| 822 | fields := strings.Fields(strings.ReplaceAll(scanner.Text(), "\r", "")) |
| 823 | if len(fields) < 2 { |
| 824 | continue |
| 825 | } |
| 826 | country := normalizeCountry(fields[len(fields)-1]) |
| 827 | if country == "" { |
| 828 | continue |
| 829 | } |
| 830 | rec, err := geoRangeFromToken(fields[0], country) |
| 831 | if err != nil { |
| 832 | return nil, fmt.Errorf("ipip line %d: %w", lineNo, err) |
| 833 | } |
| 834 | out = append(out, rec) |
| 835 | } |
| 836 | if err := scanner.Err(); err != nil { |
| 837 | return nil, fmt.Errorf("failed to scan ipip country payload: %w", err) |
| 838 | } |
| 839 | return out, nil |
| 840 | } |
| 841 | |
| 842 | func dbipCountryCode(value *dbipCountryMMDBValue) string { |
| 843 | if value == nil { |
| 844 | return "" |
| 845 | } |
| 846 | return normalizeCountry(value.ISOCode) |
| 847 | } |
| 848 | |
| 849 | func dbipCityName(value *dbipCityMMDBValue) string { |
| 850 | if value == nil { |
| 851 | return "" |
| 852 | } |
| 853 | if name := strings.TrimSpace(value.Names["en"]); name != "" { |
| 854 | return name |
| 855 | } |
| 856 | for _, name := range value.Names { |
| 857 | name = strings.TrimSpace(name) |
| 858 | if name != "" { |
| 859 | return name |
| 860 | } |
| 861 | } |
| 862 | return "" |
| 863 | } |
| 864 | |
| 865 | func dbipStateName(record dbipGeoMMDBRecord) string { |
| 866 | if len(record.Subdivisions) > 0 { |
| 867 | if name := strings.TrimSpace(record.Subdivisions[0].Names["en"]); name != "" { |
| 868 | return name |
| 869 | } |
| 870 | for _, name := range record.Subdivisions[0].Names { |
| 871 | name = strings.TrimSpace(name) |
| 872 | if name != "" { |
| 873 | return name |
| 874 | } |
| 875 | } |
| 876 | if code := strings.TrimSpace(record.Subdivisions[0].ISOCode); code != "" { |
| 877 | return code |
| 878 | } |
| 879 | } |
| 880 | return strings.TrimSpace(record.Region) |
| 881 | } |
| 882 | |
| 883 | func rangeFromIPNet(network *net.IPNet) (netip.Addr, netip.Addr, error) { |
| 884 | prefix, err := netip.ParsePrefix(network.String()) |
| 885 | if err != nil { |
| 886 | return netip.Addr{}, netip.Addr{}, fmt.Errorf( |
| 887 | "failed to parse mmdb network %s: %w", |
| 888 | network.String(), |
| 889 | err, |
| 890 | ) |
| 891 | } |
| 892 | rng := netipx.RangeOfPrefix(prefix.Masked()) |
| 893 | return rng.From(), rng.To(), nil |
| 894 | } |
| 895 | |
| 896 | func parseRangeEndpoints(startRaw, endRaw string) (netip.Addr, netip.Addr, error) { |
| 897 | start, err := parseIP(startRaw) |
| 898 | if err != nil { |
| 899 | return netip.Addr{}, netip.Addr{}, fmt.Errorf("invalid start address %q: %w", startRaw, err) |
| 900 | } |
| 901 | end, err := parseIP(endRaw) |
| 902 | if err != nil { |
| 903 | return netip.Addr{}, netip.Addr{}, fmt.Errorf("invalid end address %q: %w", endRaw, err) |
| 904 | } |
| 905 | if start.BitLen() != end.BitLen() { |
| 906 | return netip.Addr{}, netip.Addr{}, fmt.Errorf("mixed address families %q and %q", startRaw, endRaw) |
| 907 | } |
| 908 | if compareAddrs(start, end) > 0 { |
| 909 | return netip.Addr{}, netip.Addr{}, fmt.Errorf("start %s after end %s", start, end) |
| 910 | } |
| 911 | return start, end, nil |
| 912 | } |
| 913 | |
| 914 | func parseIP(raw string) (netip.Addr, error) { |
| 915 | value := strings.TrimSpace(strings.Trim(raw, "\"")) |
| 916 | if value == "" { |
| 917 | return netip.Addr{}, fmt.Errorf("empty ip") |
| 918 | } |
| 919 | if addr, err := netip.ParseAddr(value); err == nil { |
| 920 | return addr.Unmap(), nil |
| 921 | } |
| 922 | |
| 923 | if num, ok := new(big.Int).SetString(value, 10); ok { |
| 924 | if num.Sign() < 0 { |
| 925 | return netip.Addr{}, fmt.Errorf("negative integer address") |
| 926 | } |
| 927 | if num.BitLen() <= 32 { |
| 928 | v := uint32(num.Uint64()) |
| 929 | return netip.AddrFrom4([4]byte{byte(v >> 24), byte(v >> 16), byte(v >> 8), byte(v)}), nil |
| 930 | } |
| 931 | if num.BitLen() <= 128 { |
| 932 | b := num.FillBytes(make([]byte, 16)) |
| 933 | var arr [16]byte |
| 934 | copy(arr[:], b) |
| 935 | return netip.AddrFrom16(arr), nil |
| 936 | } |
| 937 | } |
| 938 | |
| 939 | return netip.Addr{}, fmt.Errorf("unsupported ip value") |
| 940 | } |
| 941 | |
| 942 | func parseASN(raw string) (uint32, error) { |
| 943 | value := strings.TrimSpace(strings.Trim(raw, "\"")) |
| 944 | if value == "" { |
| 945 | return 0, nil |
| 946 | } |
| 947 | if strings.HasPrefix(value, "AS") || strings.HasPrefix(value, "as") { |
| 948 | value = value[2:] |
| 949 | } |
| 950 | n, err := strconv.ParseUint(value, 10, 32) |
| 951 | if err != nil { |
| 952 | return 0, fmt.Errorf("invalid ASN %q: %w", raw, err) |
| 953 | } |
| 954 | return uint32(n), nil |
| 955 | } |
| 956 | |
| 957 | func parsePrimaryASN(raw string) (uint32, error) { |
| 958 | value := strings.TrimSpace(strings.Trim(raw, "\"{}")) |
| 959 | if value == "" { |
| 960 | return 0, nil |
| 961 | } |
| 962 | for _, sep := range []string{"_", ","} { |
| 963 | if idx := strings.Index(value, sep); idx >= 0 { |
| 964 | value = strings.TrimSpace(value[:idx]) |
| 965 | } |
| 966 | } |
| 967 | return parseASN(value) |
| 968 | } |
| 969 | |
| 970 | func csvHeaderIndex(header []string) map[string]int { |
| 971 | out := make(map[string]int, len(header)) |
| 972 | for i, name := range header { |
| 973 | out[strings.TrimSpace(name)] = i |
| 974 | } |
| 975 | return out |
| 976 | } |
| 977 | |
| 978 | func openZipEntryBase(archive *zip.Reader, name string) (*zip.File, error) { |
| 979 | for _, file := range archive.File { |
| 980 | if path.Base(file.Name) == name { |
| 981 | return file, nil |
| 982 | } |
| 983 | } |
| 984 | return nil, fmt.Errorf("zip entry %q not found", name) |
| 985 | } |
| 986 | |
| 987 | func openZipEntrySuffix(archive *zip.Reader, suffix string) (*zip.File, error) { |
| 988 | for _, file := range archive.File { |
| 989 | if strings.HasSuffix(file.Name, suffix) { |
| 990 | return file, nil |
| 991 | } |
| 992 | } |
| 993 | return nil, fmt.Errorf("zip entry with suffix %q not found", suffix) |
| 994 | } |
| 995 | |
| 996 | func parseCountryTokenLines(r io.Reader, country, label string) ([]geoRange, error) { |
| 997 | out := make([]geoRange, 0) |
| 998 | scanner := bufio.NewScanner(r) |
| 999 | scanner.Buffer(make([]byte, 0, 64*1024), 1024*1024) |
| 1000 | lineNo := 0 |
| 1001 | for scanner.Scan() { |
| 1002 | lineNo++ |
| 1003 | token := strings.TrimSpace(scanner.Text()) |
| 1004 | if token == "" || strings.HasPrefix(token, "#") { |
| 1005 | continue |
| 1006 | } |
| 1007 | fields := strings.Fields(token) |
| 1008 | if len(fields) == 0 { |
| 1009 | continue |
| 1010 | } |
| 1011 | rec, err := geoRangeFromToken(fields[0], country) |
| 1012 | if err != nil { |
| 1013 | return nil, fmt.Errorf("%s line %d: %w", label, lineNo, err) |
| 1014 | } |
| 1015 | out = append(out, rec) |
| 1016 | } |
| 1017 | if err := scanner.Err(); err != nil { |
| 1018 | return nil, fmt.Errorf("failed to scan %s: %w", label, err) |
| 1019 | } |
| 1020 | return out, nil |
| 1021 | } |
| 1022 | |
| 1023 | func geoRangeFromToken(raw, country string) (geoRange, error) { |
| 1024 | start, end, err := rangeFromToken(raw) |
| 1025 | if err != nil { |
| 1026 | return geoRange{}, err |
| 1027 | } |
| 1028 | rec := geoRange{start: start, end: end, country: country} |
| 1029 | if err := rec.validate(); err != nil { |
| 1030 | return geoRange{}, err |
| 1031 | } |
| 1032 | return rec, nil |
| 1033 | } |
| 1034 | |
| 1035 | func rangeFromToken(raw string) (netip.Addr, netip.Addr, error) { |
| 1036 | token := strings.TrimSpace(strings.Trim(raw, "\"")) |
| 1037 | if token == "" { |
| 1038 | return netip.Addr{}, netip.Addr{}, fmt.Errorf("empty range token") |
| 1039 | } |
| 1040 | |
| 1041 | if strings.Contains(token, "/") { |
| 1042 | prefix, err := netip.ParsePrefix(token) |
| 1043 | if err != nil { |
| 1044 | return netip.Addr{}, netip.Addr{}, err |
| 1045 | } |
| 1046 | start, end := rangeFromPrefix(prefix.Masked()) |
| 1047 | return start, end, nil |
| 1048 | } |
| 1049 | |
| 1050 | if strings.Contains(token, "-") { |
| 1051 | left, right, ok := strings.Cut(strings.ReplaceAll(token, " ", ""), "-") |
| 1052 | if !ok { |
| 1053 | return netip.Addr{}, netip.Addr{}, fmt.Errorf("invalid range token %q", raw) |
| 1054 | } |
| 1055 | return parseRangeEndpoints(left, right) |
| 1056 | } |
| 1057 | |
| 1058 | addr, err := parseIP(token) |
| 1059 | if err != nil { |
| 1060 | return netip.Addr{}, netip.Addr{}, err |
| 1061 | } |
| 1062 | return addr, addr, nil |
| 1063 | } |
| 1064 | |
| 1065 | func rangeFromPrefix(prefix netip.Prefix) (netip.Addr, netip.Addr) { |
| 1066 | rng := netipx.RangeOfPrefix(prefix.Masked()) |
| 1067 | return rng.From(), rng.To() |
| 1068 | } |