Merge pull request #65 from nextflow-io/fix/multiple-samplesheetToList

Fix using `samplesheetToList` multiple times in channel operators
nextflow-io · Oct 22, 2024 · 9f6c2b2 · 9f6c2b2
2 parents b3d2cd4 + fb6a525
commit 9f6c2b2
Show file tree

Hide file tree

Showing 4 changed files with 59 additions and 20 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -4,6 +4,8 @@
 
 ## Bug fixes
 
+1. Fixed a bug in `samplesheetToList` that caused output mixing when the function was used more than once in channel operators.
+
 ## Improvements
 
 # Version 2.1.2

diff --git a/plugins/nf-schema/src/main/nextflow/validation/SamplesheetConverter.groovy b/plugins/nf-schema/src/main/nextflow/validation/SamplesheetConverter.groovy
@@ -19,16 +19,10 @@ import nextflow.Nextflow
 @CompileStatic
 class SamplesheetConverter {
 
-    private static Path samplesheetFile
-    private static Path schemaFile
     private static ValidationConfig config
-    private static Map options
 
-    SamplesheetConverter(Path samplesheetFile, Path schemaFile, ValidationConfig config, Map options) {
-        this.samplesheetFile = samplesheetFile
-        this.schemaFile = schemaFile
+    SamplesheetConverter(ValidationConfig config) {
         this.config = config
-        this.options = options
     }
 
     private static List<Map> rows = []
@@ -67,34 +61,38 @@ class SamplesheetConverter {
     /*
     Convert the samplesheet to a list of entries based on a schema
     */
-    public static List validateAndConvertToList() {
+    public static List validateAndConvertToList(
+        Path samplesheetFile,
+        Path schemaFile,
+        Map options
+    ) {
 
         def colors = Utils.logColours(config.monochromeLogs)
 
         // Some checks before validating
-        if(!this.schemaFile.exists()) {
-            def msg = "${colors.red}JSON schema file ${this.schemaFile.toString()} does not exist\n${colors.reset}\n"
+        if(!schemaFile.exists()) {
+            def msg = "${colors.red}JSON schema file ${schemaFile.toString()} does not exist\n${colors.reset}\n"
             throw new SchemaValidationException(msg)
         }
 
-        if(!this.samplesheetFile.exists()) {
-            def msg = "${colors.red}Samplesheet file ${this.samplesheetFile.toString()} does not exist\n${colors.reset}\n"
+        if(!samplesheetFile.exists()) {
+            def msg = "${colors.red}Samplesheet file ${samplesheetFile.toString()} does not exist\n${colors.reset}\n"
             throw new SchemaValidationException(msg)
         }
 
         // Validate
         final validator = new JsonSchemaValidator(config)
-        def JSONArray samplesheet = Utils.fileToJsonArray(this.samplesheetFile, this.schemaFile)
-        def List<String> validationErrors = validator.validate(samplesheet, this.schemaFile.text)
+        def JSONArray samplesheet = Utils.fileToJsonArray(samplesheetFile, schemaFile)
+        def List<String> validationErrors = validator.validate(samplesheet, schemaFile.text)
         if (validationErrors) {
-            def msg = "${colors.red}The following errors have been detected in ${this.samplesheetFile.toString()}:\n\n" + validationErrors.join('\n').trim() + "\n${colors.reset}\n"
+            def msg = "${colors.red}The following errors have been detected in ${samplesheetFile.toString()}:\n\n" + validationErrors.join('\n').trim() + "\n${colors.reset}\n"
             log.error("Validation of samplesheet failed!")
             throw new SchemaValidationException(msg, validationErrors)
         }
 
         // Convert
-        def LinkedHashMap schemaMap = new JsonSlurper().parseText(this.schemaFile.text) as LinkedHashMap
-        def List samplesheetList = Utils.fileToList(this.samplesheetFile, this.schemaFile)
+        def LinkedHashMap schemaMap = new JsonSlurper().parseText(schemaFile.text) as LinkedHashMap
+        def List samplesheetList = Utils.fileToList(samplesheetFile, schemaFile)
 
         this.rows = []
 
@@ -110,7 +108,7 @@ class SamplesheetConverter {
             }
             return result
         }
-        logUnusedHeadersWarning(this.samplesheetFile.toString())
+        logUnusedHeadersWarning(samplesheetFile.toString())
         return channelFormat
     }
 

diff --git a/plugins/nf-schema/src/main/nextflow/validation/SchemaValidator.groovy b/plugins/nf-schema/src/main/nextflow/validation/SchemaValidator.groovy
@@ -224,8 +224,9 @@ class SchemaValidator extends PluginExtensionPoint {
         final Path schema,
         final Map options = null
     ) {
-        def SamplesheetConverter converter = new SamplesheetConverter(samplesheet, schema, config, options)
-        return converter.validateAndConvertToList()
+        def SamplesheetConverter converter = new SamplesheetConverter(config)
+        def List output = converter.validateAndConvertToList(samplesheet, schema, options)
+        return output
     }
 
     //

diff --git a/plugins/nf-schema/src/test/nextflow/validation/SamplesheetConverterTest.groovy b/plugins/nf-schema/src/test/nextflow/validation/SamplesheetConverterTest.groovy
@@ -537,4 +537,42 @@ class SamplesheetConverterTest extends Dsl2Spec{
         stdout.contains("[[string1:dependentRequired, string2:dependentRequired, integer1:10, integer2:10, boolean1:true, boolean2:true], string1, 25, false, [], [], [], unique2, 1, itDoesExist]")
         stdout.contains("[[string1:extraField, string2:extraField, integer1:10, integer2:10, boolean1:true, boolean2:true], string1, 25, false, ${getRootString()}/src/testResources/test.txt, ${getRootString()}/src/testResources/testDir, ${getRootString()}/src/testResources/testDir, unique3, 1, itDoesExist]" as String)
     }
+
+    def 'samplesheetToList - usage in channels' () {
+        given:
+        def SCRIPT_TEXT = '''
+            include { samplesheetToList } from 'plugin/nf-schema'
+
+            Channel.of("src/testResources/correct.csv")
+                .flatMap { it -> 
+                    samplesheetToList(it, "src/testResources/schema_input.json")
+                }
+                .map { it -> println("first: ${it}") }
+
+            Channel.of("src/testResources/correct_arrays.json")
+                .flatMap { it -> 
+                    samplesheetToList(it, "src/testResources/schema_input_with_arrays.json")
+                }
+                .map { it -> println("second: ${it}") }
+
+        '''
+
+        when:
+        dsl_eval(SCRIPT_TEXT)
+        def stdout = capture
+                .toString()
+                .readLines()
+                .findResults {it.startsWith('first') || it.startsWith('second') ? it : null }
+
+        then:
+        noExceptionThrown()
+        stdout.contains("first: [[string1:fullField, string2:fullField, integer1:10, integer2:10, boolean1:true, boolean2:true], string1, 25.12, false, ${getRootString()}/src/testResources/test.txt, ${getRootString()}/src/testResources/testDir, ${getRootString()}/src/testResources/test.txt, unique1, 1, itDoesExist]" as String)
+        stdout.contains("first: [[string1:value, string2:value, integer1:0, integer2:0, boolean1:true, boolean2:true], string1, 25.08, false, [], [], [], [], [], itDoesExist]")
+        stdout.contains("first: [[string1:dependentRequired, string2:dependentRequired, integer1:10, integer2:10, boolean1:true, boolean2:true], string1, 25, false, [], [], [], unique2, 1, itDoesExist]")
+        stdout.contains("first: [[string1:extraField, string2:extraField, integer1:10, integer2:10, boolean1:true, boolean2:true], string1, 25, false, ${getRootString()}/src/testResources/test.txt, ${getRootString()}/src/testResources/testDir, ${getRootString()}/src/testResources/testDir, unique3, 1, itDoesExist]" as String)
+        stdout.contains("second: [[array_meta:[]], [${getRootString()}/src/testResources/testDir/testFile.txt, ${getRootString()}/src/testResources/testDir2/testFile2.txt], [${getRootString()}/src/testResources/testDir, ${getRootString()}/src/testResources/testDir2], [${getRootString()}/src/testResources/testDir, ${getRootString()}/src/testResources/testDir2/testFile2.txt], [string1, string2], [25, 26], [25, 26.5], [false, true], [1, 2, 3], [true], [${getRootString()}/src/testResources/testDir/testFile.txt], [[${getRootString()}/src/testResources/testDir/testFile.txt]]]" as String)
+        stdout.contains("second: [[array_meta:[look, an, array, in, meta]], [], [], [], [string1, string2], [25, 26], [25, 26.5], [], [1, 2, 3], [false, true, false], [${getRootString()}/src/testResources/testDir/testFile.txt], [[${getRootString()}/src/testResources/testDir/testFile.txt]]]" as String)
+        stdout.contains("second: [[array_meta:[]], [], [], [], [string1, string2], [25, 26], [25, 26.5], [], [1, 2, 3], [false, true, false], [${getRootString()}/src/testResources/testDir/testFile.txt], [[${getRootString()}/src/testResources/testDir/testFile.txt], [${getRootString()}/src/testResources/testDir/testFile.txt, ${getRootString()}/src/testResources/testDir2/testFile2.txt]]]" as String)
+
+    }
 }