fix: fixed verifcation pattern logic for bulksms

[sahil9001]

Description:

Fixes #3477

Checklist:

• Tests passing (make test-community)?
• Lint passing (make lint this requires golangci-lint)?

[kashifkhan0771]

I generated over 10 tokens, and it appears the new regex is correct. So far, I’ve only received tokens containing the special characters !_*#, though I believe other special characters are possible.

Potential Improvements for the Detector:

• Instead of manually encoding the key and ID for basic authentication, we can leverage the SetBasicAuth function to simplify the process.
• Currently, we consider the result verified if the API returns a status code between 200 and 300. We could improve this by being more specific. Let's add the comment to reference the API docs as well.
• Based on my observations, both the ID and key are unique for each token generation. If a key is verified successfully with any ID, we can skip the remaining verification attempts for that key and proceed to the next one.

[sahil9001]

Thanks for the review @kashifkhan0771 , I have addressed the changes

[kashifkhan0771]

Please remove these commented lines. Also I guess you missed the third point from my last review.

[sahil9001]

Sure, I am unclear about the last point, are you talking about the tests? Can you please elaborate more?

[kashifkhan0771]

The secret (Basic Auth token) is a combination of two components here, an ID and a Key. Both the ID and Key are unique for each secret as I observed after generating multiple tokens. Currently, we loop through each ID and attempt to verify it against all the available keys. [O(nxm)]

If we find 3 IDs and 3 Keys, we start by verifying the first ID with all the keys. If, on the first attempt, we successfully verify that the first ID and the first Key form a valid Basic Auth token, there is no need to check the remaining keys for that ID since it has already been verified. We can then move on to the next ID.

To further optimize this process, it would be more better to avoid comparing the remaining IDs with any keys that have already been verified with previous IDs. This would reduce the number of unnecessary comparisons and improve efficiency.

[sahil9001]

Thanks, I have created a map of string -> boolean to mark the keys if a combination matches.

[kashifkhan0771]

Few suggestions:

• Can remove trim space funcs as there is no space in regex
• Instead of creating map[string]bool, we can define map[string]struct{} and filter unique values before processing the matches. This will also keep it same as other detectors.

Suggested Code: (This is a suggestion code, you can run it locally and verify and make changes if required)

	var uniqueIds = make(map[string]struct{})
	var uniqueKeys = make(map[string]struct{})

	for _, match := range idPat.FindAllStringSubmatch(dataStr, -1) {
		uniqueIds[match[1]] = struct{}{}
	}

	for _, match := range keyPat.FindAllStringSubmatch(dataStr, -1) {
		uniqueKeys[match[1]] = struct{}{}
	}

	for id := range uniqueIds {
		for key := range uniqueKeys {
			s1 := detectors.Result{
				DetectorType: detectorspb.DetectorType_Bulksms,
				Raw:          []byte(key),
				RawV2:        []byte(key + id),
			}

			if verify {
				req, err := http.NewRequestWithContext(ctx, "GET", "https://api.bulksms.com/v1/messages", nil)
				if err != nil {
					continue
				}
				req.SetBasicAuth(id, key)
				res, err := client.Do(req)
				if err == nil {
					defer func() {
						_, _ = io.Copy(io.Discard, res.Body)
						_ = res.Body.Close()
					}()

					if res.StatusCode == http.StatusOK {
						s1.Verified = true
						results = append(results, s1)
						// move to next id, by skipping remaining key's
						break
					}
				} else {
					s1.SetVerificationError(err, key)
				}
			}

			results = append(results, s1)
		}
	}

[sahil9001]

Thanks @kashifkhan0771 , this method is better, I agree. I have made the changes and tested it too, please check.

[kashifkhan0771]

LGTM!