Skip to content

Commit 8465469

Browse files
isasmendiagusisasmendiagus
committed
refactor(scanoss): Remove path anonymization from SCANOSS implementation
Removes the path anonymization functionality from the existing SCANOSS implementation as preparation for migrating to the Java SCANOSS SDK. Signed-off-by: Agustin Isasmendi <[email protected]>
1 parent 9d6a7e8 commit 8465469

File tree

4 files changed

+3
-49
lines changed

4 files changed

+3
-49
lines changed

‎plugins/scanners/scanoss/src/main/kotlin/ScanOss.kt‎

Lines changed: 2 additions & 28 deletions
Original file line numberDiff line numberDiff line change
@@ -20,14 +20,12 @@
2020
package org.ossreviewtoolkit.plugins.scanners.scanoss
2121

2222
import com.scanoss.Winnowing
23-
import com.scanoss.dto.ScanFileResult
2423
import com.scanoss.rest.ScanApi
2524
import com.scanoss.utils.JsonUtils
2625
import com.scanoss.utils.PackageDetails
2726

2827
import java.io.File
2928
import java.time.Instant
30-
import java.util.UUID
3129

3230
import org.apache.logging.log4j.kotlin.logger
3331

@@ -80,16 +78,6 @@ class ScanOss(
8078

8179
override val writeToStorage = config.writeToStorage
8280

83-
/**
84-
* The name of the file corresponding to the fingerprints can be sent to SCANOSS for more precise matches.
85-
* However, for anonymity, a unique identifier should be generated and used instead. This property holds the
86-
* mapping between the file paths and the unique identifiers. When receiving the response, the UUID will be
87-
* replaced by the actual file path.
88-
*
89-
* TODO: This behavior should be driven by a configuration parameter enabled by default.
90-
*/
91-
private val fileNamesAnonymizationMapping = mutableMapOf<UUID, String>()
92-
9381
override fun scanPath(path: File, context: ScanContext): ScanSummary {
9482
val startTime = Instant.now()
9583

@@ -110,27 +98,13 @@ class ScanOss(
11098
)
11199

112100
// Replace the anonymized UUIDs by their file paths.
113-
val results = JsonUtils.toScanFileResultsFromObject(JsonUtils.toJsonObject(result)).map {
114-
val uuid = UUID.fromString(it.filePath)
115-
116-
val fileName = fileNamesAnonymizationMapping[uuid] ?: throw IllegalArgumentException(
117-
"The ${descriptor.id} server returned UUID '$uuid' which is not present in the mapping."
118-
)
119-
120-
ScanFileResult(fileName, it.fileDetails)
121-
}
101+
val results = JsonUtils.toScanFileResultsFromObject(JsonUtils.toJsonObject(result))
122102

123103
val endTime = Instant.now()
124104
return generateSummary(startTime, endTime, results)
125105
}
126106

127-
internal fun generateRandomUUID() = UUID.randomUUID()
128-
129107
internal fun createWfpForFile(file: File): String {
130-
generateRandomUUID().let { uuid ->
131-
// TODO: Let's keep the original file extension to give SCANOSS some hint about the mime type.
132-
fileNamesAnonymizationMapping[uuid] = file.path
133-
return Winnowing.builder().build().wfpForFile(file.path, uuid.toString())
134-
}
108+
return Winnowing.builder().build().wfpForFile(file.path, file.path)
135109
}
136110
}

‎plugins/scanners/scanoss/src/test/assets/scanMulti/mappings/scanoss-multi-response.json‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -10,7 +10,7 @@
1010
},
1111
"response" : {
1212
"status" : 200,
13-
"body" : "{ \"c198b884-f6cf-496f-95eb-0e7968dd2ec6\": [ { \"id\": \"snippet\", \"status\": \"pending\", \"lines\": \"1-240\", \"oss_lines\": \"128-367\", \"matched\": \"99%\", \"purl\": [ \"pkg:github/scanoss/ort\" ], \"vendor\": \"scanoss\", \"component\": \"ort\", \"version\": \"e654028\", \"latest\": \"b12f8ee\", \"url\": \"https://github.com/scanoss/ort\", \"release_date\": \"2021-03-18\", \"file\": \"utils/src/main/kotlin/ArchiveUtils.kt\", \"url_hash\": \"37faa38a820322fa93bf7a8fa8290bb8\", \"file_hash\": \"871fb0c5188c2f620d9b997e225b0095\", \"source_hash\": \"2e91edbe430c4eb195a977d326d6d6c0\", \"file_url\": \"https://osskb.org/api/file_contents/871fb0c5188c2f620d9b997e225b0095\", \"licenses\": [ { \"name\": \"Apache-2.0\", \"patent_hints\": \"yes\", \"copyleft\": \"no\", \"checklist_url\": \"https://www.osadl.org/fileadmin/checklists/unreflicenses/Apache-2.0.txt\", \"osadl_updated\": \"2022-03-17 13:38\", \"source\": \"file_spdx_tag\" }, { \"name\": \"Apache-2.0\", \"patent_hints\": \"yes\", \"copyleft\": \"no\", \"checklist_url\": \"https://www.osadl.org/fileadmin/checklists/unreflicenses/Apache-2.0.txt\", \"osadl_updated\": \"2022-03-17 13:38\", \"source\": \"scancode\" } ], \"server\": { \"version\": \"4.4.2\", \"kb_version\": { \"monthly\": \"22.02\", \"daily\": \"22.03.25\" } } } ], \"5530105e-0752-4750-9c07-4e4604b879a5\": [ { \"id\": \"file\", \"status\": \"pending\", \"lines\": \"all\", \"oss_lines\": \"all\", \"matched\": \"100%\", \"purl\": [ \"pkg:github/scanoss/ort\" ], \"vendor\": \"scanoss\", \"component\": \"ort\", \"version\": \"e654028\", \"latest\": \"b12f8ee\", \"url\": \"https://github.com/scanoss/ort\", \"release_date\": \"2021-03-18\", \"file\": \"scanner/src/main/kotlin/ScannerFactory.kt\", \"url_hash\": \"37faa38a820322fa93bf7a8fa8290bb8\", \"file_hash\": \"5c8ab9be40df937e46c53509481107cd\", \"source_hash\": \"5c8ab9be40df937e46c53509481107cd\", \"file_url\": \"https://osskb.org/api/file_contents/5c8ab9be40df937e46c53509481107cd\", \"licenses\": [ { \"name\": \"Apache-2.0\", \"patent_hints\": \"yes\", \"copyleft\": \"no\", \"checklist_url\": \"https://www.osadl.org/fileadmin/checklists/unreflicenses/Apache-2.0.txt\", \"osadl_updated\": \"2022-03-17 13:38\", \"source\": \"file_spdx_tag\" }, { \"name\": \"Apache-2.0\", \"patent_hints\": \"yes\", \"copyleft\": \"no\", \"checklist_url\": \"https://www.osadl.org/fileadmin/checklists/unreflicenses/Apache-2.0.txt\", \"osadl_updated\": \"2022-03-17 13:38\", \"source\": \"scancode\" } ], \"server\": { \"version\": \"4.4.2\", \"kb_version\": { \"monthly\": \"22.02\", \"daily\": \"22.03.25\" } } } ]}",
13+
"body" : "{ \"utils/src/main/kotlin/ArchiveUtils.kt\": [ { \"id\": \"snippet\", \"status\": \"pending\", \"lines\": \"1-240\", \"oss_lines\": \"128-367\", \"matched\": \"99%\", \"purl\": [ \"pkg:github/scanoss/ort\" ], \"vendor\": \"scanoss\", \"component\": \"ort\", \"version\": \"e654028\", \"latest\": \"b12f8ee\", \"url\": \"https://github.com/scanoss/ort\", \"release_date\": \"2021-03-18\", \"file\": \"utils/src/main/kotlin/ArchiveUtils.kt\", \"url_hash\": \"37faa38a820322fa93bf7a8fa8290bb8\", \"file_hash\": \"871fb0c5188c2f620d9b997e225b0095\", \"source_hash\": \"2e91edbe430c4eb195a977d326d6d6c0\", \"file_url\": \"https://osskb.org/api/file_contents/871fb0c5188c2f620d9b997e225b0095\", \"licenses\": [ { \"name\": \"Apache-2.0\", \"patent_hints\": \"yes\", \"copyleft\": \"no\", \"checklist_url\": \"https://www.osadl.org/fileadmin/checklists/unreflicenses/Apache-2.0.txt\", \"osadl_updated\": \"2022-03-17 13:38\", \"source\": \"file_spdx_tag\" }, { \"name\": \"Apache-2.0\", \"patent_hints\": \"yes\", \"copyleft\": \"no\", \"checklist_url\": \"https://www.osadl.org/fileadmin/checklists/unreflicenses/Apache-2.0.txt\", \"osadl_updated\": \"2022-03-17 13:38\", \"source\": \"scancode\" } ], \"server\": { \"version\": \"4.4.2\", \"kb_version\": { \"monthly\": \"22.02\", \"daily\": \"22.03.25\" } } } ], \"5530105e-0752-4750-9c07-4e4604b879a5\": [ { \"id\": \"file\", \"status\": \"pending\", \"lines\": \"all\", \"oss_lines\": \"all\", \"matched\": \"100%\", \"purl\": [ \"pkg:github/scanoss/ort\" ], \"vendor\": \"scanoss\", \"component\": \"ort\", \"version\": \"e654028\", \"latest\": \"b12f8ee\", \"url\": \"https://github.com/scanoss/ort\", \"release_date\": \"2021-03-18\", \"file\": \"scanner/src/main/kotlin/ScannerFactory.kt\", \"url_hash\": \"37faa38a820322fa93bf7a8fa8290bb8\", \"file_hash\": \"5c8ab9be40df937e46c53509481107cd\", \"source_hash\": \"5c8ab9be40df937e46c53509481107cd\", \"file_url\": \"https://osskb.org/api/file_contents/5c8ab9be40df937e46c53509481107cd\", \"licenses\": [ { \"name\": \"Apache-2.0\", \"patent_hints\": \"yes\", \"copyleft\": \"no\", \"checklist_url\": \"https://www.osadl.org/fileadmin/checklists/unreflicenses/Apache-2.0.txt\", \"osadl_updated\": \"2022-03-17 13:38\", \"source\": \"file_spdx_tag\" }, { \"name\": \"Apache-2.0\", \"patent_hints\": \"yes\", \"copyleft\": \"no\", \"checklist_url\": \"https://www.osadl.org/fileadmin/checklists/unreflicenses/Apache-2.0.txt\", \"osadl_updated\": \"2022-03-17 13:38\", \"source\": \"scancode\" } ], \"server\": { \"version\": \"4.4.2\", \"kb_version\": { \"monthly\": \"22.02\", \"daily\": \"22.03.25\" } } } ]}",
1414
"headers" : {
1515
"Server" : "nginx/1.14.2",
1616
"Date" : "Wed, 16 Mar 2022 13:07:04 GMT",

‎plugins/scanners/scanoss/src/test/kotlin/ScanOssScannerDirectoryTest.kt‎

Lines changed: 0 additions & 11 deletions
Original file line numberDiff line numberDiff line change
@@ -27,12 +27,10 @@ import io.kotest.matchers.collections.containExactlyInAnyOrder
2727
import io.kotest.matchers.collections.shouldContainExactly
2828
import io.kotest.matchers.should
2929

30-
import io.mockk.every
3130
import io.mockk.spyk
3231
import io.mockk.verify
3332

3433
import java.io.File
35-
import java.util.UUID
3634

3735
import org.ossreviewtoolkit.model.LicenseFinding
3836
import org.ossreviewtoolkit.model.PackageType
@@ -74,15 +72,6 @@ class ScanOssScannerDirectoryTest : StringSpec({
7472
}
7573

7674
"The scanner should scan a directory" {
77-
// Manipulate the UUID generation to have the same IDs as in the response.
78-
every {
79-
scanner.generateRandomUUID()
80-
} answers {
81-
UUID.fromString("5530105e-0752-4750-9c07-4e4604b879a5")
82-
} andThenAnswer {
83-
UUID.fromString("c198b884-f6cf-496f-95eb-0e7968dd2ec6")
84-
}
85-
8675
val summary = scanner.scanPath(
8776
TEST_DIRECTORY_TO_SCAN,
8877
ScanContext(labels = emptyMap(), packageType = PackageType.PACKAGE)

‎plugins/scanners/scanoss/src/test/kotlin/ScanOssScannerFileTest.kt‎

Lines changed: 0 additions & 9 deletions
Original file line numberDiff line numberDiff line change
@@ -26,12 +26,10 @@ import io.kotest.core.spec.style.StringSpec
2626
import io.kotest.matchers.collections.containExactly
2727
import io.kotest.matchers.should
2828

29-
import io.mockk.every
3029
import io.mockk.spyk
3130
import io.mockk.verify
3231

3332
import java.io.File
34-
import java.util.UUID
3533

3634
import org.ossreviewtoolkit.model.LicenseFinding
3735
import org.ossreviewtoolkit.model.PackageType
@@ -67,13 +65,6 @@ class ScanOssScannerFileTest : StringSpec({
6765
}
6866

6967
"The scanner should scan a single file" {
70-
// Manipulate the UUID generation to have the same IDs as in the response.
71-
every {
72-
scanner.generateRandomUUID()
73-
} answers {
74-
UUID.fromString("bf5401e9-03b3-4c91-906c-cadb90487b8c")
75-
}
76-
7768
val summary = scanner.scanPath(
7869
TEST_FILE_TO_SCAN,
7970
ScanContext(labels = emptyMap(), packageType = PackageType.PACKAGE)

0 commit comments

Comments
 (0)