Prevent TaskPath leakage from typed record outputs

modules/nextflow/src/main/groovy/nextflow/processor/TaskOutputResolver.groovy

@@ -16,6 +16,8 @@
 
 package nextflow.processor
 
+import java.io.UncheckedIOException
+import java.lang.reflect.Modifier
 import java.nio.file.Path
 
 import groovy.transform.CompileDynamic
@@ -25,7 +27,10 @@ import groovy.util.logging.Slf4j
 import nextflow.exception.IllegalArityException
 import nextflow.exception.MissingFileException
 import nextflow.exception.MissingValueException
+import nextflow.extension.Bolts
 import nextflow.script.params.v2.ProcessFileOutput
+import nextflow.script.types.Record
+import nextflow.util.RecordMap
 import org.codehaus.groovy.runtime.InvokerHelper
 /**
  * Implements the resolution of task outputs.
@@ -49,6 +54,75 @@ class TaskOutputResolver implements Map<String,Object> {
         this.delegate = task.context
     }
 
+    /**
+     * Resolve and normalize an output expression before it is emitted.
+     *
+     * Values from the task context may contain {@link TaskPath}, which is a
+     * task-local view of an input file. It is valid for script interpolation,
+     * but it must not escape through output channels because downstream tasks
+     * need durable source/work-directory paths for hashing and staging.
+     *
+     * @param value
+     *      A lazy output expression, such as a closure or GString
+     * @return
+     *      The resolved output value with nested TaskPath instances converted
+     *      back to durable Path values
+     */
+    Object resolveOutput(Object value) {
+        return normalizeOutputValue(Bolts.resolveLazy(this, value))
+    }
+
+    static Object normalizeOutputValue(Object value) {
+        if( value instanceof TaskPath ) {
+            try {
+                return value.toRealPath()
+            }
+            catch( IOException e ) {
+                throw new UncheckedIOException(e)
+            }
+        }
+
+        if( value instanceof RecordMap ) {
+            final normalized = new LinkedHashMap<String,Object>()
+            for( final entry : value.entrySet() )
+                normalized.put(entry.key, normalizeOutputValue(entry.value))
+            return new RecordMap(normalized)
+        }
+
+        if( value instanceof Map ) {
+            final normalized = new LinkedHashMap<Object,Object>()
+            for( final entry : value.entrySet() )
+                normalized.put(entry.key, normalizeOutputValue(entry.value))
+            return normalized
+        }
+
+        if( value instanceof Set ) {
+            final normalized = new LinkedHashSet<Object>()
+            for( final item : value )
+                normalized.add(normalizeOutputValue(item))
+            return normalized
+        }
+
+        if( value instanceof Collection ) {
+            final normalized = new ArrayList<Object>()
+            for( final item : value )
+                normalized.add(normalizeOutputValue(item))
+            return normalized
+        }
+
+        if( value instanceof Record ) {
+            final fields = value.getClass().getFields()
+                .findAll { field -> !Modifier.isStatic(field.modifiers) && !field.synthetic }
+                .sort { it.name }
+            final normalized = new LinkedHashMap<String,Object>()
+            for( final field : fields )
+                normalized.put(field.name, normalizeOutputValue(field.get(value)))
+            return new RecordMap(normalized)
+        }
+
+        return value
+    }
+
     /**
      * Get an environment variable from the task environment.
      *

modules/nextflow/src/main/groovy/nextflow/processor/TaskProcessor.groovy

@@ -1421,12 +1421,12 @@ class TaskProcessor {
         final resolver = new TaskOutputResolver(declaredOutputs.getFiles(), task)
 
         for( final param : declaredOutputs.getParams() ) {
-            final value = resolver.resolveLazy(param.getLazyValue())
+            final value = resolver.resolveOutput(param.getLazyValue())
             task.setOutput(param, value)
         }
 
         for( final topic : declaredOutputs.getTopics() ) {
-            final value = resolver.resolveLazy(topic.getLazyValue())
+            final value = resolver.resolveOutput(topic.getLazyValue())
             topic.getChannel().bind(value)
         }
 

modules/nextflow/src/test/groovy/nextflow/processor/TaskOutputResolverTest.groovy

@@ -22,6 +22,7 @@ import nextflow.exception.IllegalArityException
 import nextflow.exception.MissingFileException
 import nextflow.exception.MissingValueException
 import nextflow.script.params.v2.ProcessFileOutput
+import nextflow.util.RecordMap
 import spock.lang.Specification
 import spock.lang.TempDir
 
@@ -30,6 +31,12 @@ import spock.lang.TempDir
  */
 class TaskOutputResolverTest extends Specification {
 
+    static class SampleRecord implements nextflow.script.types.Record {
+        public String id
+        public Path path
+        public List<Path> paths
+    }
+
     @TempDir
     Path tempDir
 
@@ -290,4 +297,47 @@ class TaskOutputResolverTest extends Specification {
         def e = thrown(MissingValueException)
         e.message.contains('Missing variable in process output')
     }
+
+    def 'should normalize task paths in output values'() {
+        given:
+        def source = tempDir.resolve('input.txt')
+        def taskPath = new TaskPath(source, 'input.txt')
+        def record = new SampleRecord(id: 'alpha', path: taskPath, paths: [taskPath])
+        def value = new RecordMap([
+            id: 'alpha',
+            path: taskPath,
+            nested: [path: taskPath],
+            paths: [taskPath] as Set,
+            record: record,
+        ])
+
+        when:
+        def result = TaskOutputResolver.normalizeOutputValue(value)
+
+        then:
+        result instanceof RecordMap
+        result.path == source
+        !(result.path instanceof TaskPath)
+        result.nested.path == source
+        !(result.nested.path instanceof TaskPath)
+        result.paths == [source] as Set
+        result.record instanceof RecordMap
+        result.record.path == source
+        result.record.paths == [source]
+    }
+
+    def 'should normalize resolved lazy output values'() {
+        given:
+        def source = tempDir.resolve('input.txt')
+        def task = makeTask([sample: new RecordMap(path: new TaskPath(source, 'input.txt'))])
+        def resolver = new TaskOutputResolver([:], task)
+
+        when:
+        def result = resolver.resolveOutput({ -> sample })
+
+        then:
+        result instanceof RecordMap
+        result.path == source
+        !(result.path instanceof TaskPath)
+    }
 }

tests/checks/.IGNORE-PARSER-V2

@@ -9,6 +9,7 @@ params-dsl.nf
 record-types.nf
 records.nf
 task-ext-block.nf
+typed-record-output-taskpath.nf
 topic-channel-typed.nf
 type-annotations.nf
 workflow-oncomplete-v2.nf

tests/checks/typed-record-output-taskpath.nf/.checks

@@ -0,0 +1,20 @@
+set -e
+
+export NXF_SYNTAX_PARSER=v2
+
+$NXF_RUN | tee stdout
+[[ $(grep INFO .nextflow.log | grep -c 'Submitted process > P1') == 1 ]]
+[[ $(grep INFO .nextflow.log | grep -c 'Submitted process > P2') == 1 ]]
+grep 'test1' stdout
+grep 'test2' stdout
+
+TASK_DIR=$($NXF_CMD log last -F "process == 'P2'")
+grep "cat 'input.txt' 'test2.txt' > combined.txt" "$TASK_DIR/.command.sh"
+test -e "$TASK_DIR/input.txt"
+test -e "$TASK_DIR/test2.txt"
+
+$NXF_RUN -resume | tee stdout
+[[ $(grep INFO .nextflow.log | grep -c 'Cached process > P1') == 1 ]]
+[[ $(grep INFO .nextflow.log | grep -c 'Cached process > P2') == 1 ]]
+grep 'test1' stdout
+grep 'test2' stdout

tests/typed-record-output-taskpath.nf

@@ -0,0 +1,63 @@
+#!/usr/bin/env nextflow
+/*
+ * Copyright 2013-2026, Seqera Labs
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+nextflow.enable.types = true
+
+workflow {
+    sample = record(
+        id: 'alpha',
+        file: file("${projectDir}/typed-record-output-taskpath/input.txt")
+    )
+
+    result = P2(P1(channel.of(sample)))
+    result.view { it -> it.text.trim() }
+}
+
+process P1 {
+    input:
+    sample: SampleRecord
+
+    output:
+    sample + record(file2: file('test2.txt'))
+
+    script:
+    """
+    echo 'test2' > test2.txt
+    """
+}
+
+process P2 {
+    input:
+    sample: SampleRecord
+
+    output:
+    file('combined.txt')
+
+    script:
+    """
+    test '${sample.file}' = 'input.txt'
+    test '${sample.file2}' = 'test2.txt'
+
+    cat '${sample.file}' '${sample.file2}' > combined.txt
+    """
+}
+
+record SampleRecord {
+    id: String
+    file: Path
+    file2: Path?
+}

tests/typed-record-output-taskpath/input.txt

@@ -0,0 +1 @@
+test1