Example showing the API in action.

ExampleDecoder_PeekKind demonstrates how to use PeekKind for look-ahead parsing without consuming values.

ExampleKind_IsContainer demonstrates container type detection.

ExampleKind_IsScalar demonstrates scalar type detection.

ExampleKind_String demonstrates human-readable Kind names.

This example demonstrates how to decode to an any.

This example shows how to decode to a struct.

This example demonstrates how to iterate over all networks in the database.

This example demonstrates how to iterate over all networks in the database which are contained within an arbitrary network.

This example demonstrates how to validate a MaxMind DB file and access metadata.

This example demonstrates how to use SkipEmptyValues to iterate only over networks that have actual data, skipping those with empty maps or arrays.

This example demonstrates how to use the Unmarshaler interface for custom decoding. Types implementing Unmarshaler automatically use custom decoding l

FuzzDatabase tests MMDB file parsing and IP address lookups. This targets file format parsing, database initialization, and lookup operations.

FuzzDecode tests the ReflectionDecoder.Decode method with fuzzed data. This targets data section parsing and reflection-based decoding logic.

FuzzDecodePath tests path-based decoding with fuzzed path segments. This targets edge cases in path traversal logic.

FuzzLookup tests IP address lookups without decoding results. This isolates the tree traversal and lookup logic from data decoding.

FuzzNetworks tests the Networks() iterator with malformed databases. This focuses specifically on tree traversal and iteration logic.

NewDecoder creates a new Decoder with the given buffer, offset, and options. Error messages include contextual offset information.

TestBoundsChecking verifies that buffer access is properly bounds-checked to prevent panics on malformed databases.

Test custom unmarshaling through Reader.Lookup.

TestCustomUnmarshalerWithIterator tests that custom unmarshalers using iterators work correctly in struct fields. This reproduces the original "no nex

TestDecodeCtrlData verifies all four size-encoding paths (inline, +29, +285, +65821) plus the extended-kind path. Each path also has a truncated-buffe

TestDecodeCtrlDataTruncated exercises the bounds checks: each case has just enough bytes to start decoding but not enough to read the trailing size or

TestDecodeKeyFastPathBoundary covers the 28/29-character boundary between decodeKey's inline fast path (size < 29) and its slow path (size >= 29). An

TestDecodeKeyFollowsPointer verifies that a pointer control byte at the key position bypasses the fast path and is followed to the target string. A re

TestDecodePointerKeyFastExtendedSize verifies that a pointer targeting a string whose size requires the extended encoding (size byte >= 29) is handled

TestDecodePointerKeyFastNonStringTarget verifies that a pointer to a non-KindString target bails out of the fast path. A regression that accepted any

TestDecodePointerKeyFastPointerSizes exercises decodePointerKeyFast for pointer sizes 2 and 4. Size 1 is covered by TestDecodeKeyFollowsPointer; size

TestDecodeSliceClearsTailPointers is the pointer-element analog of TestDecodeSliceClearsTailReferences. For []*T the slice element IS the GC root, so

TestDecodeSliceClearsTailReferences exercises the second Clear in decodeSlice (the tail clear when the new slice is shorter than the previous one). Wh

TestDecodeSliceReusesBackingArray confirms the perf invariant of the reuse path: a second decode into the same destination must reuse the existing bac

TestEmbeddedPointerSupport tests support for embedded pointer types.

TestFallbackToReflection verifies that types without UnmarshalMaxMindDB still work.

TestFastPathPreservesUnmarshalerForNamedTypes guards every fast-path kind against a regression where decodeStruct's isFastType short-circuit runs befo

TestFieldCaching tests the field caching mechanism works with new precedence rules.

TestFieldPrecedenceRules tests json/v2 style field precedence behavior.

TestLookupIPv4OnlyDB exercises the IPv4 specialized walks against IPv4-only databases (ipv4StartBitDepth=96, ipv4Start=0). The mixed-DB parity test gu

TestLookupIPv4VsIPv6Mixed guards the specialized IPv4 tree walks in traverseTree24, traverseTree28, and traverseTree32 against drift from the generic

See GitHub #115.

TestNestedUnmarshaler tests that UnmarshalMaxMindDB is called for nested struct fields.

TestNestedUnmarshalerInMap tests UnmarshalMaxMindDB for map values.

TestNestedUnmarshalerInSlice tests UnmarshalMaxMindDB for slice elements.

TestNestedUnmarshalerPointer tests UnmarshalMaxMindDB with pointer fields.