query-orchestration/internal/serviceconfig/objectstore/bucketkey.go

package objectstore

import (
	"errors"
	"fmt"
	"regexp"
	"strconv"
	"strings"
	"time"

	"queryorchestration/internal/client"

	"github.com/google/uuid"
)

type BucketLocation string
type BucketKey struct {
	ClientID  string
	Location  BucketLocation
	CreatedAt time.Time
	Part      *uint16
	EntityID  uuid.UUID
	FileType  *string
}

const (
	Import       BucketLocation = "import"
	TextTextract BucketLocation = "text/textract"
	TextOut      BucketLocation = "text/out"
	Export       BucketLocation = "export"
)

var PartExclusions = []BucketLocation{
	Export,
}

const TimeFormat = "2006-01-02T150405Z"
const DateFormat = "20060102"

func (c *BucketKey) String() string {
	return fmt.Sprintf("%s/%s", c.Prefix(), c.Filename())
}

func (c *BucketKey) Prefix() string {
	dateStr := c.CreatedAt.Format(DateFormat)

	partStr := ""
	isExcluded := false
	for _, location := range PartExclusions {
		if location == c.Location {
			isExcluded = true
			break
		}
	}
	if !isExcluded {
		if c.Part == nil {
			part := uint16(0)
			c.Part = &part
		}
		partStr = fmt.Sprintf("/%d", *c.Part)
	}

	return fmt.Sprintf("%s/%s/%s%s", c.ClientID, c.Location, dateStr, partStr)
}

func (c *BucketKey) Filename() string {
	location := strings.ReplaceAll(string(c.Location), "/", "-")
	name := fmt.Sprintf("%s_%s_%s_%s", c.CreatedAt.Format(TimeFormat), c.ClientID, location, c.EntityID)
	if c.FileType != nil {
		name = fmt.Sprintf("%s.%s", name, *c.FileType)
	}

	return name
}

func (c *BucketKey) parseFilename(name string) error {
	parts := strings.Split(name, ".")
	if len(parts) != 1 && len(parts) != 2 {
		return errors.New("incorrect amount of dots")
	} else if len(parts) == 2 {
		c.parseFileType(parts[1])
	}

	metadata := strings.Split(parts[0], "_")
	if len(metadata) != 4 {
		return errors.New("incorrect amount of metadata elements")
	}

	err := c.parseTime(metadata[0])
	if err != nil {
		return err
	}

	err = c.parseClientID(metadata[1])
	if err != nil {
		return err
	}

	err = c.parseLocation(metadata[2])
	if err != nil {
		return err
	}

	err = c.parseEntityID(metadata[3])
	if err != nil {
		return err
	}

	return nil
}
func (c *BucketKey) parseFileType(name string) {
	c.FileType = &name
}
func (c *BucketKey) parseTime(name string) error {
	parsedTime, err := time.Parse(TimeFormat, name)
	if err != nil {
		return fmt.Errorf("failed to parse date: %w", err)
	}
	c.CreatedAt = parsedTime

	return nil
}
func (c *BucketKey) parseClientID(name string) error {
	c.ClientID = name
	pattern := fmt.Sprintf(`^(%s)$`, client.CLIENT_ID_REGEX)
	reg := regexp.MustCompile(pattern)
	matches := reg.FindStringSubmatch(c.ClientID)
	if len(matches) != 2 {
		return errors.New("invalid client id")
	}

	return nil
}
func (c *BucketKey) parseLocation(name string) error {
	c.Location = BucketLocation(strings.ReplaceAll(name, "-", "/"))
	pattern := fmt.Sprintf(`^(%s|%s|%s|%s)$`, Import, TextTextract, TextOut, Export)
	reg := regexp.MustCompile(pattern)
	matches := reg.FindStringSubmatch(string(c.Location))
	if len(matches) != 2 {
		return errors.New("invalid location")
	}

	return nil
}
func (c *BucketKey) parseEntityID(name string) error {
	entityId, err := uuid.Parse(name)
	if err != nil {
		return err
	}
	c.EntityID = entityId

	return nil
}

func ParseBucketKey(key string) (BucketKey, error) {
	locations := "[a-z]+/?[a-z]+?"
	pattern := fmt.Sprintf(`^(%s)/(%s)/([0-9]{8})/(.+)$`, client.CLIENT_ID_REGEX, locations)
	reg := regexp.MustCompile(pattern)
	matches := reg.FindStringSubmatch(key)
	if len(matches) != 5 {
		return BucketKey{}, errors.New("does not match expectation")
	}

	formattedKey := BucketKey{}

	filename := `.+`
	reg = regexp.MustCompile(fmt.Sprintf(`^([0-9]+)/(%s)$`, filename))
	subMatches := reg.FindStringSubmatch(matches[4])
	if len(subMatches) == 3 {
		part, err := strconv.ParseUint(subMatches[1], 10, 8)
		if err != nil {
			return BucketKey{}, err
		}
		safePart := uint16(part)
		formattedKey.Part = &safePart

		err = formattedKey.parseFilename(subMatches[2])
		if err != nil {
			return BucketKey{}, err
		}
	} else {
		reg = regexp.MustCompile(fmt.Sprintf(`^(%s)$`, filename))
		subMatches := reg.FindStringSubmatch(matches[4])
		if len(subMatches) != 2 {
			return BucketKey{}, errors.New("does not match expectation")
		}

		err := formattedKey.parseFilename(subMatches[1])
		if err != nil {
			return BucketKey{}, err
		}
	}

	return formattedKey, nil
}