| 1 | // SPDX-License-Identifier: GPL-3.0-or-later |
| 2 | |
| 3 | package main |
| 4 | |
| 5 | import ( |
| 6 | "container/heap" |
| 7 | "fmt" |
| 8 | "net/netip" |
| 9 | "sort" |
| 10 | ) |
| 11 | |
| 12 | type asnValue struct { |
| 13 | asn uint32 |
| 14 | org string |
| 15 | } |
| 16 | |
| 17 | type geoValue struct { |
| 18 | country string |
| 19 | state string |
| 20 | city string |
| 21 | lat float64 |
| 22 | lon float64 |
| 23 | hasLoc bool |
| 24 | } |
| 25 | |
| 26 | type intervalPriority struct { |
| 27 | sourceIndex int |
| 28 | rowIndex int |
| 29 | } |
| 30 | |
| 31 | type intervalRecord[T comparable] struct { |
| 32 | start netip.Addr |
| 33 | end netip.Addr |
| 34 | value T |
| 35 | priority intervalPriority |
| 36 | } |
| 37 | |
| 38 | type intervalEvent struct { |
| 39 | addr netip.Addr |
| 40 | kind intervalEventKind |
| 41 | id int |
| 42 | } |
| 43 | |
| 44 | type intervalEventKind uint8 |
| 45 | |
| 46 | const ( |
| 47 | intervalEventRemove intervalEventKind = iota |
| 48 | intervalEventAdd |
| 49 | ) |
| 50 | |
| 51 | type intervalHeap[T comparable] struct { |
| 52 | ids []int |
| 53 | records []intervalRecord[T] |
| 54 | } |
| 55 | |
| 56 | func (h intervalHeap[T]) Len() int { return len(h.ids) } |
| 57 | |
| 58 | func (h intervalHeap[T]) Less(i, j int) bool { |
| 59 | return betterPriority( |
| 60 | h.records[h.ids[i]].priority, |
| 61 | h.records[h.ids[j]].priority, |
| 62 | ) |
| 63 | } |
| 64 | |
| 65 | func (h intervalHeap[T]) Swap(i, j int) { |
| 66 | h.ids[i], h.ids[j] = h.ids[j], h.ids[i] |
| 67 | } |
| 68 | |
| 69 | func (h *intervalHeap[T]) Push(x any) { |
| 70 | h.ids = append(h.ids, x.(int)) |
| 71 | } |
| 72 | |
| 73 | func (h *intervalHeap[T]) Pop() any { |
| 74 | last := len(h.ids) - 1 |
| 75 | id := h.ids[last] |
| 76 | h.ids = h.ids[:last] |
| 77 | return id |
| 78 | } |
| 79 | |
| 80 | func betterPriority(a, b intervalPriority) bool { |
| 81 | if a.sourceIndex != b.sourceIndex { |
| 82 | return a.sourceIndex < b.sourceIndex |
| 83 | } |
| 84 | return a.rowIndex > b.rowIndex |
| 85 | } |
| 86 | |
| 87 | func mergeAsnSources(sources [][]asnRange) ([]asnRange, error) { |
| 88 | records := make([]intervalRecord[asnValue], 0) |
| 89 | for sourceIndex, ranges := range sources { |
| 90 | for rowIndex, rec := range ranges { |
| 91 | if err := rec.validate(); err != nil { |
| 92 | return nil, fmt.Errorf("asn source %d row %d: %w", sourceIndex, rowIndex, err) |
| 93 | } |
| 94 | records = append(records, intervalRecord[asnValue]{ |
| 95 | start: rec.start, |
| 96 | end: rec.end, |
| 97 | value: asnValue{ |
| 98 | asn: rec.asn, |
| 99 | org: rec.org, |
| 100 | }, |
| 101 | priority: intervalPriority{ |
| 102 | sourceIndex: sourceIndex, |
| 103 | rowIndex: rowIndex, |
| 104 | }, |
| 105 | }) |
| 106 | } |
| 107 | } |
| 108 | |
| 109 | merged, err := mergeIntervals(records) |
| 110 | if err != nil { |
| 111 | return nil, err |
| 112 | } |
| 113 | |
| 114 | out := make([]asnRange, 0, len(merged)) |
| 115 | for _, rec := range merged { |
| 116 | out = append(out, asnRange{ |
| 117 | start: rec.start, |
| 118 | end: rec.end, |
| 119 | asn: rec.value.asn, |
| 120 | org: rec.value.org, |
| 121 | }) |
| 122 | } |
| 123 | return out, nil |
| 124 | } |
| 125 | |
| 126 | func mergeGeoSources(sources [][]geoRange) ([]geoRange, error) { |
| 127 | records := make([]intervalRecord[geoValue], 0) |
| 128 | for sourceIndex, ranges := range sources { |
| 129 | for rowIndex, rec := range ranges { |
| 130 | if err := rec.validate(); err != nil { |
| 131 | return nil, fmt.Errorf("geo source %d row %d: %w", sourceIndex, rowIndex, err) |
| 132 | } |
| 133 | records = append(records, intervalRecord[geoValue]{ |
| 134 | start: rec.start, |
| 135 | end: rec.end, |
| 136 | value: geoValue{ |
| 137 | country: rec.country, |
| 138 | state: rec.state, |
| 139 | city: rec.city, |
| 140 | lat: rec.latitude, |
| 141 | lon: rec.longitude, |
| 142 | hasLoc: rec.hasLocation, |
| 143 | }, |
| 144 | priority: intervalPriority{ |
| 145 | sourceIndex: sourceIndex, |
| 146 | rowIndex: rowIndex, |
| 147 | }, |
| 148 | }) |
| 149 | } |
| 150 | } |
| 151 | |
| 152 | merged, err := mergeIntervals(records) |
| 153 | if err != nil { |
| 154 | return nil, err |
| 155 | } |
| 156 | |
| 157 | out := make([]geoRange, 0, len(merged)) |
| 158 | for _, rec := range merged { |
| 159 | out = append(out, geoRange{ |
| 160 | start: rec.start, |
| 161 | end: rec.end, |
| 162 | country: rec.value.country, |
| 163 | state: rec.value.state, |
| 164 | city: rec.value.city, |
| 165 | latitude: rec.value.lat, |
| 166 | longitude: rec.value.lon, |
| 167 | hasLocation: rec.value.hasLoc, |
| 168 | }) |
| 169 | } |
| 170 | return out, nil |
| 171 | } |
| 172 | |
| 173 | func mergeIntervals[T comparable](records []intervalRecord[T]) ([]intervalRecord[T], error) { |
| 174 | var v4 []intervalRecord[T] |
| 175 | var v6 []intervalRecord[T] |
| 176 | |
| 177 | for _, rec := range records { |
| 178 | if !rec.start.IsValid() || !rec.end.IsValid() { |
| 179 | return nil, fmt.Errorf("invalid interval %v-%v", rec.start, rec.end) |
| 180 | } |
| 181 | if rec.start.BitLen() != rec.end.BitLen() { |
| 182 | return nil, fmt.Errorf("mixed address family range: %s-%s", rec.start, rec.end) |
| 183 | } |
| 184 | switch rec.start.BitLen() { |
| 185 | case 32: |
| 186 | v4 = append(v4, rec) |
| 187 | case 128: |
| 188 | v6 = append(v6, rec) |
| 189 | default: |
| 190 | return nil, fmt.Errorf("unsupported address family bitlen %d", rec.start.BitLen()) |
| 191 | } |
| 192 | } |
| 193 | |
| 194 | out := make([]intervalRecord[T], 0, len(records)) |
| 195 | familyOut, err := mergeIntervalsForBitLen(v4, 32) |
| 196 | if err != nil { |
| 197 | return nil, err |
| 198 | } |
| 199 | out = append(out, familyOut...) |
| 200 | |
| 201 | familyOut, err = mergeIntervalsForBitLen(v6, 128) |
| 202 | if err != nil { |
| 203 | return nil, err |
| 204 | } |
| 205 | out = append(out, familyOut...) |
| 206 | |
| 207 | return out, nil |
| 208 | } |
| 209 | |
| 210 | func mergeIntervalsForBitLen[T comparable]( |
| 211 | records []intervalRecord[T], |
| 212 | bitLen int, |
| 213 | ) ([]intervalRecord[T], error) { |
| 214 | if len(records) == 0 { |
| 215 | return nil, nil |
| 216 | } |
| 217 | |
| 218 | events := make([]intervalEvent, 0, len(records)*2) |
| 219 | for id, rec := range records { |
| 220 | events = append(events, intervalEvent{ |
| 221 | addr: rec.start, |
| 222 | kind: intervalEventAdd, |
| 223 | id: id, |
| 224 | }) |
| 225 | if next := rec.end.Next(); next.IsValid() { |
| 226 | events = append(events, intervalEvent{ |
| 227 | addr: next, |
| 228 | kind: intervalEventRemove, |
| 229 | id: id, |
| 230 | }) |
| 231 | } |
| 232 | } |
| 233 | |
| 234 | sort.Slice(events, func(i, j int) bool { |
| 235 | if records[events[i].id].start.BitLen() != records[events[j].id].start.BitLen() { |
| 236 | return records[events[i].id].start.BitLen() < records[events[j].id].start.BitLen() |
| 237 | } |
| 238 | if cmp := compareAddrs(events[i].addr, events[j].addr); cmp != 0 { |
| 239 | return cmp < 0 |
| 240 | } |
| 241 | return events[i].kind < events[j].kind |
| 242 | }) |
| 243 | |
| 244 | active := make(map[int]struct{}, len(records)) |
| 245 | candidates := &intervalHeap[T]{records: records} |
| 246 | heap.Init(candidates) |
| 247 | |
| 248 | maxAddr := maxAddrForBitLen(bitLen) |
| 249 | out := make([]intervalRecord[T], 0, len(records)) |
| 250 | |
| 251 | for i := 0; i < len(events); { |
| 252 | addr := events[i].addr |
| 253 | for i < len(events) && compareAddrs(events[i].addr, addr) == 0 { |
| 254 | switch events[i].kind { |
| 255 | case intervalEventRemove: |
| 256 | delete(active, events[i].id) |
| 257 | case intervalEventAdd: |
| 258 | active[events[i].id] = struct{}{} |
| 259 | heap.Push(candidates, events[i].id) |
| 260 | } |
| 261 | i++ |
| 262 | } |
| 263 | |
| 264 | bestID, ok := topActiveRecord(candidates, active) |
| 265 | if !ok { |
| 266 | continue |
| 267 | } |
| 268 | |
| 269 | end := maxAddr |
| 270 | if i < len(events) { |
| 271 | end = events[i].addr.Prev() |
| 272 | } |
| 273 | if compareAddrs(addr, end) > 0 { |
| 274 | continue |
| 275 | } |
| 276 | |
| 277 | best := records[bestID] |
| 278 | next := intervalRecord[T]{ |
| 279 | start: addr, |
| 280 | end: end, |
| 281 | value: best.value, |
| 282 | priority: best.priority, |
| 283 | } |
| 284 | if len(out) > 0 && |
| 285 | out[len(out)-1].value == next.value && |
| 286 | out[len(out)-1].end.Next() == next.start { |
| 287 | out[len(out)-1].end = next.end |
| 288 | continue |
| 289 | } |
| 290 | out = append(out, next) |
| 291 | } |
| 292 | |
| 293 | return out, nil |
| 294 | } |
| 295 | |
| 296 | func topActiveRecord[T comparable]( |
| 297 | candidates *intervalHeap[T], |
| 298 | active map[int]struct{}, |
| 299 | ) (int, bool) { |
| 300 | for candidates.Len() > 0 { |
| 301 | id := candidates.ids[0] |
| 302 | if _, ok := active[id]; ok { |
| 303 | return id, true |
| 304 | } |
| 305 | heap.Pop(candidates) |
| 306 | } |
| 307 | return 0, false |
| 308 | } |
| 309 | |
| 310 | func maxAddrForBitLen(bitLen int) netip.Addr { |
| 311 | switch bitLen { |
| 312 | case 32: |
| 313 | return netip.AddrFrom4([4]byte{0xff, 0xff, 0xff, 0xff}) |
| 314 | case 128: |
| 315 | return netip.AddrFrom16([16]byte{ |
| 316 | 0xff, 0xff, 0xff, 0xff, |
| 317 | 0xff, 0xff, 0xff, 0xff, |
| 318 | 0xff, 0xff, 0xff, 0xff, |
| 319 | 0xff, 0xff, 0xff, 0xff, |
| 320 | }) |
| 321 | default: |
| 322 | return netip.Addr{} |
| 323 | } |
| 324 | } |