| 1 | // SPDX-License-Identifier: GPL-3.0-or-later |
| 2 | |
| 3 | package main |
| 4 | |
| 5 | import ( |
| 6 | "archive/tar" |
| 7 | "archive/zip" |
| 8 | "bytes" |
| 9 | "compress/gzip" |
| 10 | "errors" |
| 11 | "fmt" |
| 12 | "io" |
| 13 | "net/http" |
| 14 | "net/url" |
| 15 | "os" |
| 16 | "path" |
| 17 | "regexp" |
| 18 | "sort" |
| 19 | "strconv" |
| 20 | "strings" |
| 21 | "time" |
| 22 | ) |
| 23 | |
| 24 | var errTarMMDBNotFound = errors.New("tar payload has no mmdb member") |
| 25 | |
| 26 | type downloader struct { |
| 27 | client *http.Client |
| 28 | userAgent string |
| 29 | } |
| 30 | |
| 31 | func newDownloader(httpCfg httpConfig) *downloader { |
| 32 | return &downloader{ |
| 33 | client: &http.Client{Timeout: httpCfg.timeout}, |
| 34 | userAgent: strings.TrimSpace(httpCfg.userAgent), |
| 35 | } |
| 36 | } |
| 37 | |
| 38 | func (d *downloader) readDataset(source sourceEntry) (generationDatasetRef, []byte, error) { |
| 39 | resolved, err := d.resolveSource(source) |
| 40 | if err != nil { |
| 41 | return generationDatasetRef{}, nil, err |
| 42 | } |
| 43 | |
| 44 | raw, err := d.readRaw(resolved.fetchPath, resolved.fetchURL) |
| 45 | if err != nil { |
| 46 | return generationDatasetRef{}, nil, err |
| 47 | } |
| 48 | payload, err := decodePayloadForSource(source, raw) |
| 49 | if err != nil { |
| 50 | return generationDatasetRef{}, nil, err |
| 51 | } |
| 52 | return resolved.ref, payload, nil |
| 53 | } |
| 54 | |
| 55 | type resolvedSource struct { |
| 56 | ref generationDatasetRef |
| 57 | fetchURL string |
| 58 | fetchPath string |
| 59 | } |
| 60 | |
| 61 | func (d *downloader) resolveSource(source sourceEntry) (resolvedSource, error) { |
| 62 | spec, ok := builtInSource(source.provider, source.artifact) |
| 63 | if !ok { |
| 64 | return resolvedSource{}, fmt.Errorf( |
| 65 | "unsupported provider/artifact %q/%q", |
| 66 | source.provider, |
| 67 | source.artifact, |
| 68 | ) |
| 69 | } |
| 70 | |
| 71 | ref := generationDatasetRef{ |
| 72 | Name: source.name, |
| 73 | Family: source.family, |
| 74 | Provider: source.provider, |
| 75 | Artifact: source.artifact, |
| 76 | Format: source.format, |
| 77 | } |
| 78 | |
| 79 | switch { |
| 80 | case source.path != "": |
| 81 | path, err := expandEnvPlaceholders(source.path) |
| 82 | if err != nil { |
| 83 | return resolvedSource{}, err |
| 84 | } |
| 85 | ref.Source = "path" |
| 86 | ref.Path = path |
| 87 | return resolvedSource{ |
| 88 | ref: ref, |
| 89 | fetchPath: path, |
| 90 | }, nil |
| 91 | case source.url != "": |
| 92 | fetchURL, err := expandEnvPlaceholders(source.url) |
| 93 | if err != nil { |
| 94 | return resolvedSource{}, err |
| 95 | } |
| 96 | ref.Source = "url" |
| 97 | ref.URL = sanitizeURLForMetadata(source.url) |
| 98 | return resolvedSource{ |
| 99 | ref: ref, |
| 100 | fetchURL: fetchURL, |
| 101 | }, nil |
| 102 | case spec.directURL != "": |
| 103 | fetchURL, err := expandEnvPlaceholders(spec.directURL) |
| 104 | if err != nil { |
| 105 | return resolvedSource{}, err |
| 106 | } |
| 107 | ref.Source = "builtin" |
| 108 | ref.URL = sanitizeURLForMetadata(spec.directURL) |
| 109 | return resolvedSource{ |
| 110 | ref: ref, |
| 111 | fetchURL: fetchURL, |
| 112 | }, nil |
| 113 | case spec.pageURL != "": |
| 114 | var ( |
| 115 | resolvedURL string |
| 116 | err error |
| 117 | ) |
| 118 | if source.provider == providerCAIDA && source.artifact == artifactCAIDAPrefix2AS { |
| 119 | resolvedURL, err = d.resolveCAIDAPrefix2ASURL(spec.pageURL) |
| 120 | } else { |
| 121 | resolvedURL, err = d.resolveDBIPArtifactURL(spec.pageURL, source.artifact, source.format) |
| 122 | } |
| 123 | if err != nil { |
| 124 | return resolvedSource{}, err |
| 125 | } |
| 126 | ref.Source = "builtin" |
| 127 | ref.DownloadPage = sanitizeURLForMetadata(spec.pageURL) |
| 128 | ref.ResolvedURL = sanitizeURLForMetadata(resolvedURL) |
| 129 | return resolvedSource{ |
| 130 | ref: ref, |
| 131 | fetchURL: resolvedURL, |
| 132 | }, nil |
| 133 | default: |
| 134 | return resolvedSource{}, fmt.Errorf( |
| 135 | "provider/artifact %q/%q has no usable locator", |
| 136 | source.provider, |
| 137 | source.artifact, |
| 138 | ) |
| 139 | } |
| 140 | } |
| 141 | |
| 142 | func (d *downloader) resolveDBIPArtifactURL(pageURL, artifact, format string) (string, error) { |
| 143 | page, err := d.readHTTP(pageURL) |
| 144 | if err != nil { |
| 145 | return "", fmt.Errorf("failed to fetch DB-IP landing page %s: %w", pageURL, err) |
| 146 | } |
| 147 | |
| 148 | ext := format |
| 149 | if format == formatTSV { |
| 150 | return "", fmt.Errorf("dbip artifact %q does not support format %q", artifact, format) |
| 151 | } |
| 152 | pattern := fmt.Sprintf( |
| 153 | `https://download\.db-ip\.com/free/dbip-%s-\d{4}-\d{2}\.%s\.gz`, |
| 154 | regexp.QuoteMeta(strings.ToLower(strings.TrimSpace(artifact))), |
| 155 | regexp.QuoteMeta(ext), |
| 156 | ) |
| 157 | re := regexp.MustCompile(pattern) |
| 158 | match := re.Find(page) |
| 159 | if len(match) == 0 { |
| 160 | return "", fmt.Errorf( |
| 161 | "failed to resolve DB-IP %s/%s download from %s", |
| 162 | artifact, |
| 163 | format, |
| 164 | pageURL, |
| 165 | ) |
| 166 | } |
| 167 | return string(match), nil |
| 168 | } |
| 169 | |
| 170 | func (d *downloader) resolveCAIDAPrefix2ASURL(logURL string) (string, error) { |
| 171 | page, err := d.readHTTP(logURL) |
| 172 | if err != nil { |
| 173 | return "", fmt.Errorf("failed to fetch CAIDA prefix2as creation log %s: %w", redactURLForDisplay(logURL), err) |
| 174 | } |
| 175 | |
| 176 | type caidaCandidate struct { |
| 177 | path string |
| 178 | timestamp int64 |
| 179 | hasTime bool |
| 180 | } |
| 181 | |
| 182 | candidates := make([]caidaCandidate, 0) |
| 183 | for line := range strings.SplitSeq(string(page), "\n") { |
| 184 | fields := strings.Fields(strings.TrimSpace(line)) |
| 185 | if len(fields) == 0 { |
| 186 | continue |
| 187 | } |
| 188 | candidate := fields[len(fields)-1] |
| 189 | if strings.HasSuffix(candidate, ".pfx2as.gz") { |
| 190 | entry := caidaCandidate{path: candidate} |
| 191 | if len(fields) >= 2 { |
| 192 | if timestamp, err := strconv.ParseInt(fields[len(fields)-2], 10, 64); err == nil { |
| 193 | entry.timestamp = timestamp |
| 194 | entry.hasTime = true |
| 195 | } |
| 196 | } |
| 197 | candidates = append(candidates, entry) |
| 198 | } |
| 199 | } |
| 200 | if len(candidates) == 0 { |
| 201 | return "", fmt.Errorf("failed to resolve latest CAIDA prefix2as download from %s", redactURLForDisplay(logURL)) |
| 202 | } |
| 203 | sort.Slice(candidates, func(i, j int) bool { |
| 204 | if candidates[i].hasTime != candidates[j].hasTime { |
| 205 | return !candidates[i].hasTime |
| 206 | } |
| 207 | if candidates[i].timestamp != candidates[j].timestamp { |
| 208 | return candidates[i].timestamp < candidates[j].timestamp |
| 209 | } |
| 210 | return candidates[i].path < candidates[j].path |
| 211 | }) |
| 212 | latest := candidates[len(candidates)-1].path |
| 213 | |
| 214 | base, err := url.Parse(logURL) |
| 215 | if err != nil { |
| 216 | return "", fmt.Errorf("failed to parse CAIDA prefix2as creation log URL %s: %w", redactURLForDisplay(logURL), err) |
| 217 | } |
| 218 | base.RawQuery = "" |
| 219 | base.Fragment = "" |
| 220 | if !strings.HasSuffix(base.Path, "/") { |
| 221 | base.Path = path.Dir(base.Path) + "/" |
| 222 | } |
| 223 | ref, err := url.Parse(latest) |
| 224 | if err != nil { |
| 225 | return "", fmt.Errorf("failed to parse CAIDA prefix2as candidate %q: %w", latest, err) |
| 226 | } |
| 227 | return base.ResolveReference(ref).String(), nil |
| 228 | } |
| 229 | |
| 230 | func expandEnvPlaceholders(raw string) (string, error) { |
| 231 | missing := map[string]struct{}{} |
| 232 | expanded := os.Expand(raw, func(name string) string { |
| 233 | value, ok := os.LookupEnv(name) |
| 234 | if !ok || strings.TrimSpace(value) == "" { |
| 235 | missing[name] = struct{}{} |
| 236 | return "" |
| 237 | } |
| 238 | return value |
| 239 | }) |
| 240 | if len(missing) == 0 { |
| 241 | return expanded, nil |
| 242 | } |
| 243 | names := make([]string, 0, len(missing)) |
| 244 | for name := range missing { |
| 245 | names = append(names, name) |
| 246 | } |
| 247 | sort.Strings(names) |
| 248 | return "", fmt.Errorf("missing environment variable(s): %s", strings.Join(names, ", ")) |
| 249 | } |
| 250 | |
| 251 | func sanitizeURLForMetadata(raw string) string { |
| 252 | parsed, err := url.Parse(raw) |
| 253 | if err != nil || parsed.Host == "" { |
| 254 | return "<redacted>" |
| 255 | } |
| 256 | parsed.User = nil |
| 257 | if parsed.RawQuery != "" { |
| 258 | parsed.RawQuery = "redacted" |
| 259 | } |
| 260 | parsed.Fragment = "" |
| 261 | return parsed.String() |
| 262 | } |
| 263 | |
| 264 | func (d *downloader) readRaw(path, rawURL string) ([]byte, error) { |
| 265 | if path != "" { |
| 266 | content, err := os.ReadFile(path) |
| 267 | if err != nil { |
| 268 | return nil, fmt.Errorf("failed to read %s: %w", path, err) |
| 269 | } |
| 270 | return content, nil |
| 271 | } |
| 272 | return d.readHTTP(rawURL) |
| 273 | } |
| 274 | |
| 275 | func (d *downloader) readHTTP(rawURL string) ([]byte, error) { |
| 276 | displayURL := redactURLForDisplay(rawURL) |
| 277 | req, err := http.NewRequest(http.MethodGet, rawURL, nil) |
| 278 | if err != nil { |
| 279 | return nil, fmt.Errorf("failed to build request %s: %w", displayURL, err) |
| 280 | } |
| 281 | req.Header.Set("User-Agent", d.userAgent) |
| 282 | |
| 283 | start := time.Now() |
| 284 | resp, err := d.client.Do(req) |
| 285 | if err != nil { |
| 286 | return nil, fmt.Errorf("failed to fetch %s: %w", displayURL, err) |
| 287 | } |
| 288 | defer resp.Body.Close() |
| 289 | |
| 290 | if resp.StatusCode != http.StatusOK { |
| 291 | return nil, fmt.Errorf("failed to fetch %s: unexpected status %d", displayURL, resp.StatusCode) |
| 292 | } |
| 293 | |
| 294 | content, err := io.ReadAll(resp.Body) |
| 295 | if err != nil { |
| 296 | return nil, fmt.Errorf("failed to read %s: %w", displayURL, err) |
| 297 | } |
| 298 | _ = start |
| 299 | return content, nil |
| 300 | } |
| 301 | |
| 302 | func decodePayloadForSource(source sourceEntry, raw []byte) ([]byte, error) { |
| 303 | switch { |
| 304 | case source.provider == providerMaxMind && source.artifact == artifactMaxMindGeoLite2ASN: |
| 305 | return decodeMaxMindASNPayload(raw) |
| 306 | case source.provider == providerMaxMind && source.artifact == artifactMaxMindGeoLite2Country: |
| 307 | return raw, nil |
| 308 | case source.provider == providerIP2Location && source.artifact == artifactIP2LocationCountryLite: |
| 309 | return raw, nil |
| 310 | case source.provider == providerIPDeny && source.artifact == artifactIPDenyCountryZones: |
| 311 | return raw, nil |
| 312 | case source.provider == providerIPIP && source.artifact == artifactIPIPCountry: |
| 313 | return raw, nil |
| 314 | default: |
| 315 | return decodePayload(raw) |
| 316 | } |
| 317 | } |
| 318 | |
| 319 | func decodeMaxMindASNPayload(raw []byte) ([]byte, error) { |
| 320 | if len(raw) < 2 || raw[0] != 0x1f || raw[1] != 0x8b { |
| 321 | return raw, nil |
| 322 | } |
| 323 | content, err := decodeGzip(raw) |
| 324 | if err != nil { |
| 325 | return nil, err |
| 326 | } |
| 327 | mmdb, err := extractMMDBFromTar(content) |
| 328 | if err == nil { |
| 329 | return mmdb, nil |
| 330 | } |
| 331 | if !errors.Is(err, errTarMMDBNotFound) { |
| 332 | return nil, fmt.Errorf("failed to extract MaxMind ASN MMDB from tar payload: %w", err) |
| 333 | } |
| 334 | return content, nil |
| 335 | } |
| 336 | |
| 337 | func decodePayload(raw []byte) ([]byte, error) { |
| 338 | if len(raw) >= 2 && raw[0] == 0x1f && raw[1] == 0x8b { |
| 339 | return decodeGzip(raw) |
| 340 | } |
| 341 | if len(raw) >= 4 && bytes.Equal(raw[:4], []byte{'P', 'K', 0x03, 0x04}) { |
| 342 | return decodeZip(raw) |
| 343 | } |
| 344 | return raw, nil |
| 345 | } |
| 346 | |
| 347 | func decodeGzip(raw []byte) ([]byte, error) { |
| 348 | reader, err := gzip.NewReader(bytes.NewReader(raw)) |
| 349 | if err != nil { |
| 350 | return nil, fmt.Errorf("failed to open gzip payload: %w", err) |
| 351 | } |
| 352 | defer reader.Close() |
| 353 | |
| 354 | content, err := io.ReadAll(reader) |
| 355 | if err != nil { |
| 356 | return nil, fmt.Errorf("failed to read gzip payload: %w", err) |
| 357 | } |
| 358 | return content, nil |
| 359 | } |
| 360 | |
| 361 | func decodeZip(raw []byte) ([]byte, error) { |
| 362 | archive, err := zip.NewReader(bytes.NewReader(raw), int64(len(raw))) |
| 363 | if err != nil { |
| 364 | return nil, fmt.Errorf("failed to open zip payload: %w", err) |
| 365 | } |
| 366 | for _, f := range archive.File { |
| 367 | if f.FileInfo().IsDir() { |
| 368 | continue |
| 369 | } |
| 370 | rc, err := f.Open() |
| 371 | if err != nil { |
| 372 | return nil, fmt.Errorf("failed to open zip member %s: %w", f.Name, err) |
| 373 | } |
| 374 | defer rc.Close() |
| 375 | content, err := io.ReadAll(rc) |
| 376 | if err != nil { |
| 377 | return nil, fmt.Errorf("failed to read zip member %s: %w", f.Name, err) |
| 378 | } |
| 379 | return content, nil |
| 380 | } |
| 381 | return nil, fmt.Errorf("zip payload has no regular files") |
| 382 | } |
| 383 | |
| 384 | func extractMMDBFromTar(raw []byte) ([]byte, error) { |
| 385 | tr := tar.NewReader(bytes.NewReader(raw)) |
| 386 | for { |
| 387 | header, err := tr.Next() |
| 388 | if err == io.EOF { |
| 389 | return nil, errTarMMDBNotFound |
| 390 | } |
| 391 | if err != nil { |
| 392 | if !looksLikeTarPayload(raw) { |
| 393 | return nil, errTarMMDBNotFound |
| 394 | } |
| 395 | return nil, err |
| 396 | } |
| 397 | if header.Typeflag != tar.TypeReg || !strings.HasSuffix(strings.ToLower(header.Name), ".mmdb") { |
| 398 | continue |
| 399 | } |
| 400 | content, err := io.ReadAll(tr) |
| 401 | if err != nil { |
| 402 | return nil, fmt.Errorf("failed to read tar member %s: %w", header.Name, err) |
| 403 | } |
| 404 | return content, nil |
| 405 | } |
| 406 | } |
| 407 | |
| 408 | func looksLikeTarPayload(raw []byte) bool { |
| 409 | return len(raw) >= 512 && bytes.Equal(raw[257:262], []byte("ustar")) |
| 410 | } |