Pass a PathMapper to the function that computes checksums, so it can resolve files downloaded for instance

Author: kinow

cwltool/command_line_tool.py

@@ -826,6 +826,7 @@ def job(
             _checksum = partial(
                 compute_checksums,
                 runtimeContext.make_fs_access(runtimeContext.basedir),
+                cachebuilder,
             )
             visit_class(
                 [cachebuilder.files, cachebuilder.bindings],
@@ -1252,7 +1253,7 @@ def collect_output_ports(
                 )
 
                 if compute_checksum:
-                    adjustFileObjs(ret, partial(compute_checksums, fs_access))
+                    adjustFileObjs(ret, partial(compute_checksums, fs_access, builder))
             expected_schema = cast(Schema, self.names.get_name("outputs_record_schema", None))
             validate_ex(
                 expected_schema,

cwltool/process.py

@@ -391,7 +391,7 @@ def _check_adjust(a_file: CWLObjectType) -> CWLObjectType:
     visit_class(outputObj, ("File", "Directory"), _check_adjust)
 
     if compute_checksum:
-        visit_class(outputObj, ("File",), functools.partial(compute_checksums, fs_access))
+        visit_class(outputObj, ("File",), functools.partial(compute_checksums, fs_access, None))
     return outputObj
 
 
@@ -1341,14 +1341,18 @@ def scandeps(
     return r
 
 
-def compute_checksums(fs_access: StdFsAccess, fileobj: CWLObjectType) -> None:
+def compute_checksums(fs_access: StdFsAccess, builder: Builder, fileobj: CWLObjectType) -> None:
     if "checksum" not in fileobj:
         checksum = hashlib.sha1()  # nosec
-        location = cast(str, fileobj["location"])
-        with fs_access.open(location, "rb") as f:
+        location = file_path = cast(str, fileobj["location"])
+        if builder:
+            if not builder.pathmapper:
+                raise ValueError("Do not call compute_checksums using a builder that doesn't have a pathmapper.")
+            file_path = builder.pathmapper.mapper(location)[0]
+        with fs_access.open(file_path, "rb") as f:
             contents = f.read(1024 * 1024)
             while contents != b"":
                 checksum.update(contents)
                 contents = f.read(1024 * 1024)
         fileobj["checksum"] = "sha1$%s" % checksum.hexdigest()
-        fileobj["size"] = fs_access.size(location)
+        fileobj["size"] = fs_access.size(file_path)