mirror of
https://github.com/wader/fq.git
synced 2024-12-21 12:21:39 +03:00
b08ef00dd1
Replaces []Format with a Group type. A bit more type safe. Breaking change for RegisterFormat, now takes a first argument that is a "single" format group. Lots of naming cleanup. This is also preparation for decode group argument which will enable doing intresting probing, ex a format decoder could know it's decode as part of probe group (html could be probed possibly), or have "arg probe" group for decoder who inspect args to know if they should probe (-d /path/to/schema etc) to enable nice CLI-ergonomics.
167 lines
4.3 KiB
Go
167 lines
4.3 KiB
Go
package avro
|
|
|
|
import (
|
|
"bytes"
|
|
"compress/flate"
|
|
"embed"
|
|
"hash/crc32"
|
|
|
|
"github.com/golang/snappy"
|
|
"github.com/wader/fq/format"
|
|
"github.com/wader/fq/format/avro/decoders"
|
|
"github.com/wader/fq/format/avro/schema"
|
|
"github.com/wader/fq/pkg/bitio"
|
|
"github.com/wader/fq/pkg/decode"
|
|
"github.com/wader/fq/pkg/interp"
|
|
"github.com/wader/fq/pkg/scalar"
|
|
)
|
|
|
|
//go:embed avro_ocf.md
|
|
var avroOcfFS embed.FS
|
|
|
|
func init() {
|
|
interp.RegisterFormat(
|
|
format.AvroOcf,
|
|
&decode.Format{
|
|
Description: "Avro object container file",
|
|
Groups: []*decode.Group{format.Probe},
|
|
DecodeFn: decodeAvroOCF,
|
|
})
|
|
interp.RegisterFS(avroOcfFS)
|
|
}
|
|
|
|
type HeaderData struct {
|
|
Schema schema.SimplifiedSchema
|
|
Codec string
|
|
Sync []byte
|
|
}
|
|
|
|
const headerSchemaSpec = `
|
|
{
|
|
"type": "record",
|
|
"name": "org.apache.avro.file.Header",
|
|
"fields": [
|
|
{"name": "meta", "type": {"type": "map", "values": "string"}},
|
|
{"name": "sync", "type": {"type": "fixed", "name": "Sync", "size": 16}}
|
|
]
|
|
}`
|
|
|
|
func decodeHeader(d *decode.D) HeaderData {
|
|
d.FieldRawLen("magic", 4*8, d.AssertBitBuf([]byte{'O', 'b', 'j', 1}))
|
|
|
|
var headerData HeaderData
|
|
|
|
headerSchema, err := schema.FromSchemaString(headerSchemaSpec)
|
|
if err != nil {
|
|
d.Fatalf("Failed to parse header schema: %v", err)
|
|
}
|
|
decodeHeaderFn, err := decoders.DecodeFnForSchema(headerSchema)
|
|
if err != nil {
|
|
d.Fatalf("failed to parse header: %v", err)
|
|
}
|
|
|
|
header := decodeHeaderFn("header", d)
|
|
headerRecord, ok := header.(map[string]any)
|
|
if !ok {
|
|
d.Fatalf("header is not a map")
|
|
}
|
|
meta, ok := headerRecord["meta"].(map[string]any)
|
|
if !ok {
|
|
d.Fatalf("header.meta is not a map")
|
|
}
|
|
|
|
metaSchema, ok := meta["avro.schema"].(string)
|
|
if !ok {
|
|
d.Fatalf("missing meta avro.schema")
|
|
}
|
|
|
|
headerData.Schema, err = schema.FromSchemaString(metaSchema)
|
|
if err != nil {
|
|
d.Fatalf("failed to parse schema: %v", err)
|
|
}
|
|
if codec, ok := meta["avro.codec"]; ok {
|
|
headerData.Codec, ok = codec.(string)
|
|
if !ok {
|
|
d.Fatalf("avro.codec is not a string")
|
|
}
|
|
} else {
|
|
headerData.Codec = "null"
|
|
}
|
|
|
|
headerData.Sync, ok = headerRecord["sync"].([]byte)
|
|
if !ok {
|
|
d.Fatalf("header.sync is not a byte array")
|
|
}
|
|
return headerData
|
|
}
|
|
|
|
func decodeBlockCodec(d *decode.D, dataSize int64, codec string) *bytes.Buffer {
|
|
bb := &bytes.Buffer{}
|
|
if codec == "deflate" {
|
|
br := d.FieldRawLen("compressed", dataSize*8)
|
|
d.Copy(bb, flate.NewReader(bitio.NewIOReader(br)))
|
|
} else if codec == "snappy" {
|
|
// Everything but last 4 bytes which are the checksum
|
|
n := dataSize - 4
|
|
br := d.FieldRawLen("compressed", n*8)
|
|
|
|
// This could be simplified to be similar to deflate, however snappy's reader only works for streaming frames,
|
|
// not block data. See https://github.com/google/snappy/blob/main/framing_format.txt for details.
|
|
compressed := make([]byte, n)
|
|
if _, err := bitio.ReadFull(br, compressed, n*8); err != nil {
|
|
d.Fatalf("failed reading compressed data %v", err)
|
|
}
|
|
decompressed, err := snappy.Decode(nil, compressed)
|
|
if err != nil {
|
|
d.Fatalf("failed decompressing data: %v", err)
|
|
}
|
|
d.Copy(bb, bytes.NewReader(decompressed))
|
|
|
|
// Check the checksum
|
|
crc32W := crc32.NewIEEE()
|
|
d.Copy(crc32W, bytes.NewReader(bb.Bytes()))
|
|
d.FieldU32("crc", d.UintValidateBytes(crc32W.Sum(nil)), scalar.UintHex)
|
|
} else {
|
|
// Unknown codec, just dump the compressed data.
|
|
d.FieldRawLen("compressed", dataSize*8, scalar.BitBufDescription(codec+" encoded"))
|
|
return nil
|
|
}
|
|
return bb
|
|
}
|
|
|
|
func decodeAvroOCF(d *decode.D) any {
|
|
header := decodeHeader(d)
|
|
|
|
decodeFn, err := decoders.DecodeFnForSchema(header.Schema)
|
|
if err != nil {
|
|
d.Fatalf("unable to create codec: %v", err)
|
|
}
|
|
|
|
d.FieldStructArrayLoop("blocks", "block", func() bool { return d.NotEnd() }, func(d *decode.D) {
|
|
count := d.FieldSintFn("count", decoders.VarZigZag)
|
|
if count <= 0 {
|
|
return
|
|
}
|
|
size := d.FieldSintFn("size", decoders.VarZigZag)
|
|
i := int64(0)
|
|
|
|
if header.Codec != "null" {
|
|
if bb := decodeBlockCodec(d, size, header.Codec); bb != nil {
|
|
d.FieldArrayRootBitBufFn("data", bitio.NewBitReader(bb.Bytes(), -1), func(d *decode.D) {
|
|
for ; i < count; i++ {
|
|
decodeFn("data", d)
|
|
}
|
|
})
|
|
}
|
|
} else {
|
|
d.FieldArrayLoop("data", func() bool { return i < count }, func(d *decode.D) {
|
|
decodeFn("datum", d)
|
|
i++
|
|
})
|
|
}
|
|
d.FieldRawLen("sync", 16*8, d.AssertBitBuf(header.Sync))
|
|
})
|
|
|
|
return nil
|
|
}
|