CLI directory input

chanzuckerberg · Sep 4, 2020 · d2e4327 · d2e4327
1 parent 2e4ee08
commit d2e4327
Show file tree

Hide file tree

Showing 4 changed files with 58 additions and 7 deletions.
diff --git a/WDL/CLI.py b/WDL/CLI.py
@@ -1000,6 +1000,20 @@ def runner_input_value(s_value, ty, file_found, root):
         elif not (file_found and file_found(fn)):  # maybe URI
             raise Error.InputError("File not found: " + fn)
         return Value.File(fn)
+    if isinstance(ty, Type.Directory):
+        dn = os.path.expanduser(s_value)
+        if os.path.isdir(dn):
+            dn = os.path.abspath(dn)
+            if not path_really_within(dn, root):
+                raise Error.InputError(
+                    f"all input paths must be located within the configured `file_io.root' directory `{root}' "
+                    f"unlike `{dn}'"
+                )
+            # TODO: courtesy check for symlinks that have absolute paths or relatively point
+            # outside the directory
+        else:  # TODO: relax for URIs
+            raise Error.InputError("Directory not found: " + dn)
+        return Value.Directory(dn)
     if isinstance(ty, Type.Boolean):
         if s_value == "true":
             return Value.Boolean(True)
@@ -1241,7 +1255,7 @@ def localize(
             doc = load(wdlfile, path or [], check_quant=check_quant, read_source=read_source)
 
             def file_found(fn):
-                return runtime.download.able(cfg, fn) or os.path.isfile(fn)
+                return runtime.download.able(cfg, fn) or os.path.exists(fn)
 
             try:
                 target, input_env, input_json = runner_input(

diff --git a/WDL/Value.py b/WDL/Value.py
@@ -186,11 +186,7 @@ class Directory(String):
 
     def coerce(self, desired_type: Optional[Type.Base] = None) -> Base:
         ""
-        if self.value is None:
-            if isinstance(desired_type, Type.Directory) and desired_type.optional:
-                return Null(self.expr)
-            else:
-                raise FileNotFoundError()
+        # TODO: similar coercion logic for Directory? outputs when we support those
         return super().coerce(desired_type)
 
 

diff --git a/tests/runner.t b/tests/runner.t
@@ -11,7 +11,7 @@ source tests/bash-tap/bash-tap-bootstrap
 export PYTHONPATH="$SOURCE_DIR:$PYTHONPATH"
 miniwdl="python3 -m WDL"
 
-plan tests 53
+plan tests 55
 
 $miniwdl run_self_test
 is "$?" "0" "run_self_test"
@@ -246,6 +246,41 @@ $miniwdl run --copy-input-files mv_input_file.wdl file=quick
 is "$?" "0" "copy input files"
 is "$(basename `jq -r '.["mv_input_file.xxx"]' _LAST/outputs.json`)" "xxx" "updated _LAST"
 
+cat << 'EOF' > dir_io.wdl
+version development
+workflow w {
+    input {
+        Directory d
+    }
+    call t {
+        input:
+        d = d
+    }
+    output {
+        Int dsz = round(size(t.files))
+    }
+}
+task t {
+    input {
+        Directory d
+    }
+    command <<<
+        mkdir outdir
+        find ~{d} -type f | xargs -i{} cp {} outdir/
+    >>>
+    output {
+        Array[File] files = glob("outdir/*")
+    }
+}
+EOF
+
+mkdir -p indir/subdir
+echo alice > indir/alice.txt
+echo bob > indir/subdir/bob.txt
+miniwdl run dir_io.wdl d=indir
+is "$?" "0" "directory input"
+is `jq -r '.["w.dsz"]' _LAST/outputs.json` "10" "use of directory input"
+
 cat << 'EOF' > uri_inputs.json
 {"my_workflow.files": ["https://google.com/robots.txt", "https://raw.githubusercontent.com/chanzuckerberg/miniwdl/main/tests/alyssa_ben.txt"]}
 EOF

diff --git a/tests/test_7runner.py b/tests/test_7runner.py
@@ -57,6 +57,12 @@ def _run(self, wdl:str, inputs = None, expected_exception: Exception = None, cfg
         return WDL.values_to_json(outputs)
 
 class TestDirectoryIO(RunnerTestCase):
+    def test_coercion(self):
+        assert WDL.Type.Directory().coerces(WDL.Type.String())
+        d = WDL.Value.String("foo").coerce(WDL.Type.Directory())
+        assert isinstance(d, WDL.Value.Directory)
+        assert d.value == "foo"
+
     def test_basic_directory(self):
         wdl = R"""
         version development