1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426
|
package csvutil
import (
"reflect"
"sort"
)
const defaultBufSize = 4096
type encField struct {
field
encodeFunc
}
type encCache struct {
fields []encField
buf []byte
index []int
record []string
}
func newEncCache(k typeKey, funcMap map[reflect.Type]reflect.Value, funcs []reflect.Value, header []string) (_ *encCache, err error) {
fields := cachedFields(k)
encFields := make([]encField, 0, len(fields))
// if header is not empty, we are going to track columns in a set and we will
// track which columns are covered by type fields.
set := make(map[string]bool, len(header))
for _, s := range header {
set[s] = false
}
for _, f := range fields {
if _, ok := set[f.name]; len(header) > 0 && !ok {
continue
}
set[f.name] = true
fn, err := encodeFn(f.baseType, true, funcMap, funcs)
if err != nil {
return nil, err
}
encFields = append(encFields, encField{
field: f,
encodeFunc: fn,
})
}
if len(header) > 0 {
// look for columns that were defined in a header but are not present
// in the provided data type. In case we find any, we will set it to
// a no-op encoder that always produces an empty column.
for k, b := range set {
if b {
continue
}
encFields = append(encFields, encField{
field: field{
name: k,
},
encodeFunc: nopEncode,
})
}
sortEncFields(header, encFields)
}
return &encCache{
fields: encFields,
buf: make([]byte, 0, defaultBufSize),
index: make([]int, len(encFields)),
record: make([]string, len(encFields)),
}, nil
}
// sortEncFields sorts the provided fields according to the given header.
// at this stage header expects to contain matching fields, so both slices
// are expected to be of the same length.
func sortEncFields(header []string, fields []encField) {
set := make(map[string]int, len(header))
for i, s := range header {
set[s] = i
}
sort.Slice(fields, func(i, j int) bool {
return set[fields[i].name] < set[fields[j].name]
})
}
// Encoder writes structs CSV representations to the output stream.
type Encoder struct {
// Tag defines which key in the struct field's tag to scan for names and
// options (Default: 'csv').
Tag string
// If AutoHeader is true, a struct header is encoded during the first call
// to Encode automatically (Default: true).
AutoHeader bool
w Writer
c *encCache
header []string
noHeader bool
typeKey typeKey
funcMap map[reflect.Type]reflect.Value
ifaceFuncs []reflect.Value
}
// NewEncoder returns a new encoder that writes to w.
func NewEncoder(w Writer) *Encoder {
return &Encoder{
w: w,
noHeader: true,
AutoHeader: true,
}
}
// Register registers a custom encoding function for a concrete type or interface.
// The argument f must be of type:
// func(T) ([]byte, error)
//
// T must be a concrete type such as Foo or *Foo, or interface that has at
// least one method.
//
// During encoding, fields are matched by the concrete type first. If match is not
// found then Encoder looks if field implements any of the registered interfaces
// in order they were registered.
//
// Register panics if:
// - f does not match the right signature
// - f is an empty interface
// - f was already registered
//
// Register is based on the encoding/json proposal:
// https://github.com/golang/go/issues/5901.
func (e *Encoder) Register(f interface{}) {
v := reflect.ValueOf(f)
typ := v.Type()
if typ.Kind() != reflect.Func ||
typ.NumIn() != 1 || typ.NumOut() != 2 ||
typ.Out(0) != _bytes || typ.Out(1) != _error {
panic("csvutil: func must be of type func(T) ([]byte, error)")
}
argType := typ.In(0)
if argType.Kind() == reflect.Interface && argType.NumMethod() == 0 {
panic("csvutil: func argument type must not be an empty interface")
}
if e.funcMap == nil {
e.funcMap = make(map[reflect.Type]reflect.Value)
}
if _, ok := e.funcMap[argType]; ok {
panic("csvutil: func " + typ.String() + " already registered")
}
e.funcMap[argType] = v
if argType.Kind() == reflect.Interface {
e.ifaceFuncs = append(e.ifaceFuncs, v)
}
}
// SetHeader overrides the provided data type's default header. Fields are
// encoded in the order of the provided header. If a column specified in the
// header doesn't exist in the provided type, it will be encoded as an empty
// column. Fields that are not part of the provided header are ignored.
// Encoder can't guarantee the right order if the provided header contains
// duplicate column names.
//
// SetHeader must be called before EncodeHeader and/or Encode in order to take
// effect.
func (enc *Encoder) SetHeader(header []string) {
cp := make([]string, len(header))
copy(cp, header)
enc.header = cp
}
// Encode writes the CSV encoding of v to the output stream. The provided
// argument v must be a struct, struct slice or struct array.
//
// Only the exported fields will be encoded.
//
// First call to Encode will write a header unless EncodeHeader was called first
// or AutoHeader is false. Header names can be customized by using tags
// ('csv' by default), otherwise original Field names are used.
//
// If header was provided through SetHeader then it overrides the provided data
// type's default header. Fields are encoded in the order of the provided header.
// If a column specified in the header doesn't exist in the provided type, it will
// be encoded as an empty column. Fields that are not part of the provided header
// are ignored. Encoder can't guarantee the right order if the provided header
// contains duplicate column names.
//
// Header and fields are written in the same order as struct fields are defined.
// Embedded struct's fields are treated as if they were part of the outer struct.
// Fields that are embedded types and that are tagged are treated like any
// other field, but they have to implement Marshaler or encoding.TextMarshaler
// interfaces.
//
// Marshaler interface has the priority over encoding.TextMarshaler.
//
// Tagged fields have the priority over non tagged fields with the same name.
//
// Following the Go visibility rules if there are multiple fields with the same
// name (tagged or not tagged) on the same level and choice between them is
// ambiguous, then all these fields will be ignored.
//
// Nil values will be encoded as empty strings. Same will happen if 'omitempty'
// tag is set, and the value is a default value like 0, false or nil interface.
//
// Bool types are encoded as 'true' or 'false'.
//
// Float types are encoded using strconv.FormatFloat with precision -1 and 'G'
// format. NaN values are encoded as 'NaN' string.
//
// Fields of type []byte are being encoded as base64-encoded strings.
//
// Fields can be excluded from encoding by using '-' tag option.
//
// Examples of struct tags:
//
// // Field appears as 'myName' header in CSV encoding.
// Field int `csv:"myName"`
//
// // Field appears as 'Field' header in CSV encoding.
// Field int
//
// // Field appears as 'myName' header in CSV encoding and is an empty string
// // if Field is 0.
// Field int `csv:"myName,omitempty"`
//
// // Field appears as 'Field' header in CSV encoding and is an empty string
// // if Field is 0.
// Field int `csv:",omitempty"`
//
// // Encode ignores this field.
// Field int `csv:"-"`
//
// // Encode treats this field exactly as if it was an embedded field and adds
// // "my_prefix_" to each field's name.
// Field Struct `csv:"my_prefix_,inline"`
//
// // Encode treats this field exactly as if it was an embedded field.
// Field Struct `csv:",inline"`
//
// Fields with inline tags that have a non-empty prefix must not be cyclic
// structures. Passing such values to Encode will result in an infinite loop.
//
// Encode doesn't flush data. The caller is responsible for calling Flush() if
// the used Writer supports it.
func (e *Encoder) Encode(v interface{}) error {
return e.encode(reflect.ValueOf(v))
}
// EncodeHeader writes the CSV header of the provided struct value to the output
// stream. The provided argument v must be a struct value.
//
// The first Encode method call will not write header if EncodeHeader was called
// before it. This method can be called in cases when a data set could be
// empty, but header is desired.
//
// EncodeHeader is like Header function, but it works with the Encoder and writes
// directly to the output stream. Look at Header documentation for the exact
// header encoding rules.
func (e *Encoder) EncodeHeader(v interface{}) error {
typ, err := valueType(v)
if err != nil {
return err
}
return e.encodeHeader(typ)
}
func (e *Encoder) encode(v reflect.Value) error {
val := walkValue(v)
if !val.IsValid() {
return &InvalidEncodeError{}
}
switch val.Kind() {
case reflect.Struct:
return e.encodeStruct(val)
case reflect.Array, reflect.Slice:
if walkType(val.Type().Elem()).Kind() != reflect.Struct {
return &InvalidEncodeError{v.Type()}
}
return e.encodeArray(val)
default:
return &InvalidEncodeError{v.Type()}
}
}
func (e *Encoder) encodeStruct(v reflect.Value) error {
if e.AutoHeader && e.noHeader {
if err := e.encodeHeader(v.Type()); err != nil {
return err
}
}
return e.marshal(v)
}
func (e *Encoder) encodeArray(v reflect.Value) error {
l := v.Len()
for i := 0; i < l; i++ {
if err := e.encodeStruct(walkValue(v.Index(i))); err != nil {
return err
}
}
return nil
}
func (e *Encoder) encodeHeader(typ reflect.Type) error {
fields, _, _, record, err := e.cache(typ)
if err != nil {
return err
}
for i, f := range fields {
record[i] = f.name
}
if err := e.w.Write(record); err != nil {
return err
}
e.noHeader = false
return nil
}
func (e *Encoder) marshal(v reflect.Value) error {
fields, buf, index, record, err := e.cache(v.Type())
if err != nil {
return err
}
for i, f := range fields {
v := walkIndex(v, f.index)
omitempty := f.tag.omitEmpty
if v.Kind() == reflect.Ptr || v.Kind() == reflect.Interface {
// We should disable omitempty for pointer and interface values,
// because if it's nil we will automatically encode it as an empty
// string. However, the initialized pointer should not be affected,
// even if it's a default value.
omitempty = false
}
if !v.IsValid() {
index[i] = 0
continue
}
b, err := f.encodeFunc(buf, v, omitempty)
if err != nil {
return err
}
index[i], buf = len(b)-len(buf), b
}
out := string(buf)
for i, n := range index {
record[i], out = out[:n], out[n:]
}
e.c.buf = buf[:0]
return e.w.Write(record)
}
func (e *Encoder) tag() string {
if e.Tag == "" {
return defaultTag
}
return e.Tag
}
func (e *Encoder) cache(typ reflect.Type) ([]encField, []byte, []int, []string, error) {
if k := (typeKey{e.tag(), typ}); k != e.typeKey {
c, err := newEncCache(k, e.funcMap, e.ifaceFuncs, e.header)
if err != nil {
return nil, nil, nil, nil, err
}
e.c, e.typeKey = c, k
}
return e.c.fields, e.c.buf[:0], e.c.index, e.c.record, nil
}
func walkIndex(v reflect.Value, index []int) reflect.Value {
for _, i := range index {
v = walkPtr(v)
if !v.IsValid() {
return reflect.Value{}
}
v = v.Field(i)
}
return v
}
func walkPtr(v reflect.Value) reflect.Value {
for v.Kind() == reflect.Ptr {
v = v.Elem()
}
return v
}
func walkValue(v reflect.Value) reflect.Value {
for {
switch v.Kind() {
case reflect.Ptr, reflect.Interface:
v = v.Elem()
default:
return v
}
}
}
func walkType(typ reflect.Type) reflect.Type {
for typ.Kind() == reflect.Ptr {
typ = typ.Elem()
}
return typ
}
|