Files
job-search/internal/findATender/dateRepair.go
T

44 lines
1.2 KiB
Go

package findATender
import (
"bufio"
"bytes"
"io"
"regexp"
)
// Global regex to identify and catch date-only fields
var invalidDateRegex = regexp.MustCompile(`:\s*"(\d{4}-\d{2}-\d{2})"\s*([,}\s])`)
var invalidPublishedDateRegex = regexp.MustCompile(`"publishedDate":\s*""`)
// The UK Government FindATender service doesn't strictly follow OCDS date
// format. Sometimes it provides dates only, sometimes it sends empty strings
// instead of nulls.
type FindATenderStreamCleaner struct {
reader *bufio.Reader
buffer bytes.Buffer
}
func NewOCDSStreamCleaner(r io.Reader) *FindATenderStreamCleaner {
return &FindATenderStreamCleaner{reader: bufio.NewReader(r)}
}
func (sc *FindATenderStreamCleaner) Read(p []byte) (int, error) {
if sc.buffer.Len() == 0 {
line, err := sc.reader.ReadBytes('\n')
if len(line) > 0 {
// Convert "2026-11-16" -> "2026-11-16T00:00:00Z"
line = invalidDateRegex.ReplaceAll(line, []byte(`:"$1T00:00:00Z"`))
// Convert {publishedDate:""} -> "{publishedDate:null}"
line = invalidPublishedDateRegex.ReplaceAll(line, []byte(`"publishedDate":null`))
sc.buffer.Write(line)
}
if err != nil {
if sc.buffer.Len() == 0 {
return 0, err
}
}
}
return sc.buffer.Read(p)
}