Fixed releases with 1e9999 ("no limit") being skipped as unreadable
This commit is contained in:
1 parent
79efcb5e91
commit
4db268550b
2 files changed
+157
-49
No files matched your search
+80
-41
@@ -8,6 +8,7 @@ import (
|
||||
"fmt"
|
||||
"io"
|
||||
"log/slog"
|
||||
"math"
|
||||
"regexp"
|
||||
"strconv"
|
||||
"strings"
|
||||
@@ -65,81 +66,119 @@ type safeReleasePackage struct {
|
||||
// dateOnly matches a calendar date with no time of day: "2006-10-02".
|
||||
var dateOnly = regexp.MustCompile(`^\d{4}-\d{2}-\d{2}$`)
|
||||
|
||||
// unmarshalLenientDates decodes raw json into target like json.Unmarshal, after
|
||||
// repairing date fields that time.Time cannot parse.
|
||||
func unmarshalLenientDates(raw json.RawMessage, target any) error {
|
||||
fixed, err := repairDates(raw)
|
||||
// unmarshalLenient decodes raw json into target like json.Unmarshal, after
|
||||
// repairing values that it cannot decode (see repairJSON).
|
||||
func unmarshalLenient(raw json.RawMessage, target any) error {
|
||||
fixed, err := repairJSON(raw)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
return json.Unmarshal(fixed, target)
|
||||
}
|
||||
|
||||
// repairDates returns a copy of raw in which every date field holding a
|
||||
// date-only value ("2006-10-02") is rewritten as midnight UTC
|
||||
// ("2006-10-02T00:00:00Z"), and every date field holding an empty string is
|
||||
// set to null. Values that already have a time, values in other fields, and
|
||||
// dates that appear inside free text are left alone. If nothing needs
|
||||
// repairing, raw is returned unchanged.
|
||||
// repairJSON returns a copy of raw with the values json.Unmarshal cannot
|
||||
// decode repaired:
|
||||
//
|
||||
// A date field is one whose name is "date" or starts with "date", contains
|
||||
// "Date", or ends with "Deadline".
|
||||
func repairDates(raw json.RawMessage) (json.RawMessage, error) {
|
||||
// - every date field holding a date-only value ("2006-10-02") is rewritten
|
||||
// as midnight UTC ("2006-10-02T00:00:00Z"), and every date field holding
|
||||
// an empty string is set to null. Values that already have a time, values
|
||||
// in other fields, and dates that appear inside free text are left alone.
|
||||
// A date field is one whose name is "date" or starts with "date",
|
||||
// contains "Date", or ends with "Deadline".
|
||||
// - every number too large for a float64 is replaced by math.MaxInt64 (or
|
||||
// math.MinInt64 if negative). OCDS writes "no limit" as 1e9999, as JSON
|
||||
// has no infinity; math.MaxInt64 still fits the int and float64 fields
|
||||
// it is found in, and is marshalled again where +Inf would not be.
|
||||
//
|
||||
// If nothing needs repairing, raw is returned unchanged.
|
||||
func repairJSON(raw json.RawMessage) (json.RawMessage, error) {
|
||||
decoder := json.NewDecoder(bytes.NewReader(raw))
|
||||
decoder.UseNumber() // keep numbers exactly as written, not as float64
|
||||
|
||||
var doc any
|
||||
if err := decoder.Decode(&doc); err != nil {
|
||||
return nil, fmt.Errorf("repairing dates: %w", err)
|
||||
return nil, fmt.Errorf("repairing json: %w", err)
|
||||
}
|
||||
|
||||
if _, err := decoder.Token(); !errors.Is(err, io.EOF) {
|
||||
return nil, errors.New(
|
||||
"findatender: repairing dates: unexpected data after the JSON value",
|
||||
"findatender: repairing json: unexpected data after the JSON value",
|
||||
)
|
||||
}
|
||||
|
||||
if !recursiveRepairDates(doc) {
|
||||
repaired, changed := recursiveRepair("", doc)
|
||||
if !changed {
|
||||
return raw, nil
|
||||
}
|
||||
|
||||
return json.Marshal(doc)
|
||||
return json.Marshal(repaired)
|
||||
}
|
||||
|
||||
// recursiveRepairDates fixes date fields in value in place and reports whether
|
||||
// it changed anything.
|
||||
func recursiveRepairDates(value any) bool {
|
||||
changed := false
|
||||
// recursiveRepair repairs value, found in the field named key ("" for an
|
||||
// array element or the document itself), and its contents (see repairJSON).
|
||||
// Maps and slices are repaired in place; the repaired value is returned, and
|
||||
// whether anything changed.
|
||||
func recursiveRepair(key string, value any) (any, bool) {
|
||||
switch typed := value.(type) {
|
||||
case map[string]any:
|
||||
for key, fieldValue := range typed {
|
||||
if text, ok := fieldValue.(string); ok && isDateField(key) {
|
||||
switch {
|
||||
case dateOnly.MatchString(text):
|
||||
typed[key] = text + "T00:00:00Z"
|
||||
changed = true
|
||||
case text == "":
|
||||
typed[key] = nil
|
||||
changed = true
|
||||
}
|
||||
continue
|
||||
}
|
||||
if recursiveRepairDates(fieldValue) {
|
||||
changed := false
|
||||
for fieldKey, fieldValue := range typed {
|
||||
repaired, fieldChanged := recursiveRepair(fieldKey, fieldValue)
|
||||
if fieldChanged {
|
||||
typed[fieldKey] = repaired
|
||||
changed = true
|
||||
}
|
||||
}
|
||||
return typed, changed
|
||||
case []any:
|
||||
for _, element := range typed {
|
||||
if recursiveRepairDates(element) {
|
||||
changed := false
|
||||
for i, element := range typed {
|
||||
repaired, elementChanged := recursiveRepair("", element)
|
||||
if elementChanged {
|
||||
typed[i] = repaired
|
||||
changed = true
|
||||
}
|
||||
}
|
||||
return typed, changed
|
||||
case string:
|
||||
return repairDate(key, typed)
|
||||
case json.Number:
|
||||
return repairNumber(typed)
|
||||
default:
|
||||
return value, false
|
||||
}
|
||||
}
|
||||
|
||||
// repairDate repairs text if key is a date field (see repairJSON).
|
||||
func repairDate(key, text string) (any, bool) {
|
||||
if !isDateField(key) {
|
||||
return text, false
|
||||
}
|
||||
switch {
|
||||
case dateOnly.MatchString(text):
|
||||
return text + "T00:00:00Z", true
|
||||
case text == "":
|
||||
return nil, true
|
||||
default:
|
||||
return text, false
|
||||
}
|
||||
}
|
||||
|
||||
// repairNumber replaces a number too large for a float64 (see repairJSON).
|
||||
func repairNumber(number json.Number) (any, bool) {
|
||||
parsed, err := strconv.ParseFloat(number.String(), 64)
|
||||
switch {
|
||||
case !errors.Is(err, strconv.ErrRange) || !math.IsInf(parsed, 0):
|
||||
return number, false
|
||||
case parsed > 0:
|
||||
return json.Number(strconv.FormatInt(math.MaxInt64, 10)), true
|
||||
default:
|
||||
return json.Number(strconv.FormatInt(math.MinInt64, 10)), true
|
||||
}
|
||||
return changed
|
||||
}
|
||||
|
||||
// isDateField reports whether a field holds a date, by its name (see
|
||||
// repairDates).
|
||||
// repairJSON).
|
||||
func isDateField(name string) bool {
|
||||
return strings.HasPrefix(name, "date") ||
|
||||
strings.Contains(name, "Date") ||
|
||||
@@ -224,8 +263,8 @@ func missingIDsError(haveOCID, haveID bool) error {
|
||||
}
|
||||
|
||||
// safeUnpackReleases unpacks an API response into an OCDS ReleasePackage,
|
||||
// handling API irregularities (malformed dates, missing fields, invalid
|
||||
// releases, etc).
|
||||
// handling API irregularities (malformed dates, 1e9999 for no limit, missing
|
||||
// fields, invalid releases, etc).
|
||||
//
|
||||
// This method takes ownership of the response body and closes it when done.
|
||||
func safeUnpackReleases(
|
||||
@@ -265,7 +304,7 @@ func safeUnpackReleases(
|
||||
|
||||
for _, rawRelease := range releasePackage.Releases {
|
||||
var release ocds.Release
|
||||
err := unmarshalLenientDates(rawRelease, &release)
|
||||
err := unmarshalLenient(rawRelease, &release)
|
||||
if err != nil {
|
||||
ocid, releaseID, _ := releaseIDs(rawRelease)
|
||||
logger.Warn(
|
||||
|
||||
Reference in new issue
Block a user