crowdsec/cmd/crowdsec-cli/clidecision/import.go

314 lines
8.1 KiB
Go

package clidecision
import (
"bufio"
"bytes"
"context"
"encoding/json"
"errors"
"fmt"
"io"
"iter"
"os"
"slices"
"strings"
"time"
"github.com/jszwec/csvutil"
log "github.com/sirupsen/logrus"
"github.com/spf13/cobra"
"github.com/crowdsecurity/go-cs-lib/ptr"
"github.com/crowdsecurity/crowdsec/cmd/crowdsec-cli/args"
"github.com/crowdsecurity/crowdsec/pkg/models"
"github.com/crowdsecurity/crowdsec/pkg/types"
)
// decisionRaw is only used to unmarshall json/csv decisions
type decisionRaw struct {
Duration string `csv:"duration,omitempty" json:"duration,omitempty"`
Scenario string `csv:"reason,omitempty" json:"reason,omitempty"`
Scope string `csv:"scope,omitempty" json:"scope,omitempty"`
Type string `csv:"type,omitempty" json:"type,omitempty"`
Value string `csv:"value" json:"value"`
}
func parseDecisionList(content []byte, format string) ([]decisionRaw, error) {
ret := []decisionRaw{}
switch format {
case "values":
fmt.Fprintln(os.Stdout, "Parsing values")
scanner := bufio.NewScanner(bytes.NewReader(content))
for scanner.Scan() {
value := strings.TrimSpace(scanner.Text())
ret = append(ret, decisionRaw{Value: value})
}
if err := scanner.Err(); err != nil {
return nil, fmt.Errorf("unable to parse values: '%w'", err)
}
case "json":
fmt.Fprintln(os.Stdout, "Parsing json")
if err := json.Unmarshal(content, &ret); err != nil {
return nil, err
}
case "csv":
fmt.Fprintln(os.Stdout, "Parsing csv")
if err := csvutil.Unmarshal(content, &ret); err != nil {
return nil, fmt.Errorf("unable to parse csv: '%w'", err)
}
default:
return nil, fmt.Errorf("invalid format '%s', expected one of 'json', 'csv', 'values'", format)
}
return ret, nil
}
func excludeAllowlistedDecisions(decisions []*models.Decision, allowlistedValues []string) iter.Seq[*models.Decision] {
return func(yield func(*models.Decision) bool) {
for _, d := range decisions {
if slices.Contains(allowlistedValues, *d.Value) {
continue
}
if !yield(d) {
return
}
}
}
}
func (cli *cliDecisions) import_(ctx context.Context, input string, duration string, scope string, reason string, type_ string, batch int, format string) error {
var (
content []byte
fin *os.File
err error
)
if duration == "" {
return errors.New("default duration cannot be empty")
}
if scope == "" {
return errors.New("default scope cannot be empty")
}
if reason == "" {
return errors.New("default reason cannot be empty")
}
if type_ == "" {
return errors.New("default type cannot be empty")
}
// set format if the file has a json or csv extension
if format == "" {
if strings.HasSuffix(input, ".json") {
format = "json"
} else if strings.HasSuffix(input, ".csv") {
format = "csv"
}
}
if format == "" {
return errors.New("unable to guess format from file extension, please provide a format with --format flag")
}
if input == "-" {
fin = os.Stdin
input = "stdin"
} else {
fin, err = os.Open(input)
if err != nil {
return fmt.Errorf("unable to open %s: %w", input, err)
}
}
content, err = io.ReadAll(fin)
if err != nil {
return fmt.Errorf("unable to read from %s: %w", input, err)
}
decisionsListRaw, err := parseDecisionList(content, format)
if err != nil {
return err
}
if len(decisionsListRaw) == 0 {
return errors.New("no decisions found")
}
decisions := make([]*models.Decision, len(decisionsListRaw))
for i, d := range decisionsListRaw {
if d.Value == "" {
return fmt.Errorf("item %d: missing 'value'", i)
}
if d.Duration == "" {
d.Duration = duration
log.Debugf("item %d: missing 'duration', using default '%s'", i, duration)
}
if d.Scenario == "" {
d.Scenario = reason
log.Debugf("item %d: missing 'reason', using default '%s'", i, reason)
}
if d.Type == "" {
d.Type = type_
log.Debugf("item %d: missing 'type', using default '%s'", i, type_)
}
if d.Scope == "" {
d.Scope = scope
log.Debugf("item %d: missing 'scope', using default '%s'", i, scope)
}
decisions[i] = &models.Decision{
Value: ptr.Of(d.Value),
Duration: ptr.Of(d.Duration),
Origin: ptr.Of(types.CscliImportOrigin),
Scenario: ptr.Of(d.Scenario),
Type: ptr.Of(d.Type),
Scope: ptr.Of(d.Scope),
Simulated: ptr.Of(false),
}
}
if len(decisions) > 1000 {
fmt.Fprintf(os.Stdout, "You are about to add %d decisions, this may take a while\n", len(decisions))
}
if batch == 0 {
batch = len(decisions)
} else {
fmt.Fprintf(os.Stdout, "batch size: %d\n", batch)
}
allowlistedValues := make([]string, 0)
for chunk := range slices.Chunk(decisions, batch) {
log.Debugf("Processing chunk of %d decisions", len(chunk))
decisionsStr := make([]string, 0, len(chunk))
for _, d := range chunk {
if *d.Scope != types.Ip && *d.Scope != types.Range {
continue
}
decisionsStr = append(decisionsStr, *d.Value)
}
// Skip if no IPs or ranges
if len(decisionsStr) == 0 {
continue
}
allowlistResp, _, err := cli.client.Allowlists.CheckIfAllowlistedBulk(ctx, decisionsStr)
if err != nil {
return err
}
for _, r := range allowlistResp.Results {
fmt.Fprintf(os.Stdout, "Value %s is allowlisted by %s\n", *r.Target, r.Allowlists)
allowlistedValues = append(allowlistedValues, *r.Target)
}
}
actualDecisions := slices.Collect(excludeAllowlistedDecisions(decisions, allowlistedValues))
for chunk := range slices.Chunk(actualDecisions, batch) {
importAlert := models.Alert{
CreatedAt: time.Now().UTC().Format(time.RFC3339),
Scenario: ptr.Of(fmt.Sprintf("import %s: %d IPs", input, len(chunk))),
Message: ptr.Of(""),
Events: []*models.Event{},
Source: &models.Source{
Scope: ptr.Of(""),
Value: ptr.Of(""),
},
StartAt: ptr.Of(time.Now().UTC().Format(time.RFC3339)),
StopAt: ptr.Of(time.Now().UTC().Format(time.RFC3339)),
Capacity: ptr.Of(int32(0)),
Simulated: ptr.Of(false),
EventsCount: ptr.Of(int32(len(chunk))),
Leakspeed: ptr.Of(""),
ScenarioHash: ptr.Of(""),
ScenarioVersion: ptr.Of(""),
Decisions: chunk,
}
_, _, err = cli.client.Alerts.Add(ctx, models.AddAlertsRequest{&importAlert})
if err != nil {
return err
}
}
fmt.Fprintf(os.Stdout, "Imported %d decisions", len(actualDecisions))
return nil
}
func (cli *cliDecisions) newImportCmd() *cobra.Command {
var (
input string
duration string
scope string
reason string
decisionType string
batch int
format string
)
cmd := &cobra.Command{
Use: "import [options]",
Short: "Import decisions from a file or pipe",
Long: "expected format:\n" +
"csv : any of duration,reason,scope,type,value, with a header line\n" +
"json :" + "`{" + `"duration": "24h", "reason": "my_scenario", "scope": "ip", "type": "ban", "value": "x.y.z.z"` + "}`",
Args: args.NoArgs,
DisableAutoGenTag: true,
Example: `decisions.csv:
duration,scope,value
24h,ip,1.2.3.4
$ cscli decisions import -i decisions.csv
decisions.json:
[{"duration": "4h", "scope": "ip", "type": "ban", "value": "1.2.3.4"}]
The file format is detected from the extension, but can be forced with the --format option
which is required when reading from standard input.
Raw values, standard input:
$ echo "1.2.3.4" | cscli decisions import -i - --format values
`,
RunE: func(cmd *cobra.Command, args []string) error {
return cli.import_(cmd.Context(), input, duration, scope, reason, decisionType, batch, format)
},
}
flags := cmd.Flags()
flags.SortFlags = false
flags.StringVarP(&input, "input", "i", "", "Input file")
flags.StringVarP(&duration, "duration", "d", "4h", "Decision duration: 1h,4h,30m")
flags.StringVar(&scope, "scope", types.Ip, "Decision scope: ip,range,username")
flags.StringVarP(&reason, "reason", "R", "manual", "Decision reason: <scenario-name>")
flags.StringVarP(&decisionType, "type", "t", "ban", "Decision type: ban,captcha,throttle")
flags.IntVar(&batch, "batch", 0, "Split import in batches of N decisions")
flags.StringVar(&format, "format", "", "Input format: 'json', 'csv' or 'values' (each line is a value, no headers)")
_ = cmd.MarkFlagRequired("input")
return cmd
}