Fix AzureRepositoryProvider readBytes operation (#6286) [ci fast]

Changed Azure API call from JSON-embedded content to direct binary download
using 'download: true' parameter. This prevents binary data corruption that
occurred when converting JSON-escaped strings back to bytes.

Signed-off-by: Paolo Di Tommaso <[email protected]>
Co-authored-by: Nathan Johnson <[email protected]>
Co-authored-by: Paolo Di Tommaso <[email protected]>
Co-authored-by: Claude <[email protected]>

Author: 4 people

modules/nextflow/src/main/groovy/nextflow/scm/AzureRepositoryProvider.groovy

@@ -183,12 +183,35 @@ final class AzureRepositoryProvider extends RepositoryProvider {
         "${config.server}/${urlPath}"
     }
 
-    /** {@inheritDoc} */
     @Override
-    byte[] readBytes(String path) {
+    String readText( String path ) {
         final url = getContentUrl(path)
         final response = invokeAndParseResponse(url)
-        return response.get('content')?.toString()?.getBytes()
+        return response.get('content')?.toString()
+    }
+
+    /** {@inheritDoc} */
+    @Override
+    byte[] readBytes(String path) {
+        // For binary content, use direct download instead of JSON embedding
+        final queryParams = [
+                'download': true,
+                'includeContent': false,
+                'includeContentMetadata': false,
+                "api-version": 6.0,
+                'path': path
+        ] as Map<String,Object>
+        
+        if( revision ) {
+            queryParams['versionDescriptor.version'] = revision
+            if( COMMIT_REGEX.matcher(revision).matches() )
+                queryParams['versionDescriptor.versionType'] = 'commit'
+        }
+        
+        final queryString = queryParams.collect({ "$it.key=$it.value"}).join('&')
+        final url = "$endpointUrl/items?$queryString"
+        // Use invokeBytes for direct binary content download
+        return invokeBytes(url)
     }
 
 }

modules/nextflow/src/test/groovy/nextflow/cli/CmdConfigTest.groovy

@@ -24,6 +24,7 @@ import nextflow.extension.FilesEx
 import nextflow.plugin.Plugins
 import nextflow.secret.SecretsLoader
 import spock.lang.IgnoreIf
+import spock.lang.Requires
 import spock.lang.Specification
 /**
  *
@@ -384,10 +385,12 @@ class CmdConfigTest extends Specification {
 
     }
 
-
     @IgnoreIf({System.getenv('NXF_SMOKE')})
+    @Requires({System.getenv('NXF_GITHUB_ACCESS_TOKEN')})
     def 'should resolve remote config' () {
         given:
+        SysEnv.push(GITHUB_TOKEN: System.getenv('NXF_GITHUB_ACCESS_TOKEN'))
+        and:
         def buffer = new ByteArrayOutputStream()
         def cmd = new CmdConfig(
                 args: ['https://github.com/nextflow-io/hello'],
@@ -404,6 +407,9 @@ class CmdConfigTest extends Specification {
             }
             '''
             .stripIndent()
+
+        cleanup:
+        SysEnv.pop()
     }
 
     @IgnoreIf({System.getenv('NXF_SMOKE')})

modules/nextflow/src/test/groovy/nextflow/scm/AzureRepositoryProviderTest.groovy

@@ -135,6 +135,24 @@ class AzureRepositoryProviderTest extends Specification {
         result == 'println "Hello from Azure repos!"'
     }
 
+    @IgnoreIf({System.getenv('NXF_SMOKE')})
+    @Requires({System.getenv('NXF_AZURE_REPOS_TOKEN')})
+    def 'should read bytes file content'() {
+        given:
+        def token = System.getenv('NXF_AZURE_REPOS_TOKEN')
+        def config = new ProviderConfig('azurerepos').setAuth(token)
+
+        when:
+        // uses repo at https://pditommaso.visualstudio.com/nf-azure-repo/_git/nf-azure-repo
+        def repo = new AzureRepositoryProvider('pditommaso/nf-azure-repo', config)
+        def result = repo.readBytes('/docs/images/nf-core-rnaseq_logo_light.png')
+
+        then:
+        result.length == 22915
+        result.sha256() == '7a396344498750f614155f6e4f38b7d6ca98ced45daf0921b64acf73b18efaf4'
+    }
+
+
     @IgnoreIf({System.getenv('NXF_SMOKE')})
     @Requires({System.getenv('NXF_AZURE_REPOS_TOKEN')})
     def 'should fetch repo tags'() {
@@ -170,7 +188,7 @@ class AzureRepositoryProviderTest extends Specification {
         result == [
                 new RepositoryProvider.BranchInfo('dev', 'cc0ca18640a5c995231e22d91f1527d5155d024b'),
                 new RepositoryProvider.BranchInfo('feature-x', '13456a001ba5a27d643755614ab8e814d94ef888'),
-                new RepositoryProvider.BranchInfo('master', 'f84130388714582e20f0e2ff9a44b41978ec8929'),
+                new RepositoryProvider.BranchInfo('master', 'a207636e419f18c4b8c8586b00e329ab4788a7f5'),
         ]
     }
 

modules/nextflow/src/test/groovy/nextflow/scm/GiteaRepositoryProviderTest.groovy

@@ -106,4 +106,19 @@ class GiteaRepositoryProviderTest extends Specification {
         result == DATA
     }
 
+    @IgnoreIf({System.getenv('NXF_SMOKE')})
+    @Requires({System.getenv('NXF_GITEA_ACCESS_TOKEN')})
+    def 'should read bytes file content'() {
+        given:
+        def token =  System.getenv('NXF_GITEA_ACCESS_TOKEN')
+        def config = new ProviderConfig('gitea', new ConfigSlurper().parse(CONFIG).providers.mygitea as ConfigObject).setAuth(token)
+
+        when:
+        def repo = new GiteaRepositoryProvider('pditommaso/test-hello', config)
+        def result = repo.readBytes('docs/images/nf-core-rnaseq_logo_light.png')
+
+        then:
+        result.length == 22915
+        result.sha256() == '7a396344498750f614155f6e4f38b7d6ca98ced45daf0921b64acf73b18efaf4'
+    }
 }