Implement some basic workflow linting.

gxformat2/lint.py

@@ -0,0 +1,60 @@
+import sys
+
+from gxformat2._yaml import ordered_load
+
+EXIT_CODE_SUCCESS = 0
+EXIT_CODE_LINT_FAILED = 1
+EXIT_CODE_FORMAT_ERROR = 2
+EXIT_CODE_FILE_PARSE_FAILED = 3
+
+
+def lint_ga(workflow_dict):
+    if workflow_dict.get("format-version") != "0.1":
+        return EXIT_CODE_FORMAT_ERROR
+    if workflow_dict.get("a_galaxy_workflow") != "true":
+        return EXIT_CODE_FORMAT_ERROR
+
+    native_steps = workflow_dict.get("steps")
+    found_outputs = False
+    found_output_without_label = False
+
+    for step in native_steps.values():
+        for workflow_output in step.get("workflow_outputs", []):
+            found_outputs = True
+
+            if not workflow_output.get("label"):
+                found_output_without_label = True
+
+    if not found_outputs:
+        return EXIT_CODE_LINT_FAILED
+
+    if found_output_without_label:
+        return EXIT_CODE_LINT_FAILED
+
+    return EXIT_CODE_SUCCESS
+
+
+def lint_format2(workflow_dict):
+    # Lint for outputs...
+    if not workflow_dict.get("outputs", None):
+        return EXIT_CODE_LINT_FAILED
+    return EXIT_CODE_SUCCESS
+
+
+def main(argv):
+    path = argv[1]
+    with open(path, "r") as f:
+        try:
+            workflow_dict = ordered_load(f)
+        except Exception:
+            return EXIT_CODE_FILE_PARSE_FAILED
+    workflow_class = workflow_dict.get("class")
+    if workflow_class == "GalaxyWorkflow":
+        exit_code = lint_format2(workflow_dict)
+    else:
+        exit_code = lint_ga(workflow_dict)
+    return exit_code
+
+
+if __name__ == "__main__":
+    sys.exit(main(sys.argv))

tests/_helpers.py

@@ -0,0 +1,28 @@
+from gxformat2.converter import yaml_to_workflow
+from gxformat2.interface import ImporterGalaxyInterface
+
+
+TEST_GOOD_WORKFLOW = """
+"""
+
+
+def to_native(has_yaml, **kwds):
+    return yaml_to_workflow(has_yaml, MockGalaxyInterface(), None, **kwds)
+
+
+def assert_valid_native(as_dict_native):
+    assert as_dict_native["a_galaxy_workflow"] == "true"
+    assert as_dict_native["format-version"] == "0.1"
+    assert "steps" in as_dict_native
+    step_count = 0
+    for key, value in as_dict_native["steps"].items():
+        assert key == str(step_count)
+        step_count += 1
+        assert "type" in value
+        assert value["type"] in ["data_input", "data_collection_input", "tool", "subworkflow"]
+
+
+class MockGalaxyInterface(ImporterGalaxyInterface):
+
+    def import_workflow(self, workflow, **kwds):
+        pass

tests/test_basic.py

@@ -1,6 +1,6 @@
-from gxformat2.converter import ImportOptions, yaml_to_workflow
+from gxformat2.converter import ImportOptions
 from gxformat2.export import from_galaxy_native
-from gxformat2.interface import ImporterGalaxyInterface
+from ._helpers import to_native, assert_valid_native
 
 
 def test_import_export():
@@ -199,28 +199,6 @@ def from_native(native_as_dict):
     return from_galaxy_native(native_as_dict, None)
 
 
-def to_native(has_yaml, **kwds):
-    return yaml_to_workflow(has_yaml, MockGalaxyInterface(), None, **kwds)
-
-
 def assert_valid_format2(as_dict_format2):
     assert as_dict_format2["class"] == "GalaxyWorkflow"
     assert "steps" in as_dict_format2
-
-
-def assert_valid_native(as_dict_native):
-    assert as_dict_native["a_galaxy_workflow"] == "true"
-    assert as_dict_native["format-version"] == "0.1"
-    assert "steps" in as_dict_native
-    step_count = 0
-    for key, value in as_dict_native["steps"].items():
-        assert key == str(step_count)
-        step_count += 1
-        assert "type" in value
-        assert value["type"] in ["data_input", "data_collection_input", "tool", "subworkflow"]
-
-
-class MockGalaxyInterface(ImporterGalaxyInterface):
-
-    def import_workflow(self, workflow, **kwds):
-        pass

tests/test_lint.py

@@ -0,0 +1,150 @@
+import copy
+import json
+import os
+import tempfile
+
+from gxformat2.lint import main
+from gxformat2._yaml import ordered_dump, ordered_load
+from ._helpers import assert_valid_native, to_native
+
+
+TEST_PATH = os.path.abspath(os.path.dirname(__file__))
+TEST_EXAMPLES = os.path.join(TEST_PATH, "examples")
+
+BASIC_WORKFLOW = """
+class: GalaxyWorkflow
+doc: |
+  Simple workflow that no-op cats a file and then selects 10 random lines.
+inputs:
+  the_input:
+    type: File
+    doc: input doc
+outputs:
+  the_output:
+    outputSource: cat/out_file1
+steps:
+  cat:
+    tool_id: cat1
+    doc: cat doc
+    in:
+      input1: the_input
+"""
+
+WORKFLOW_WITH_REPEAT = """
+class: GalaxyWorkflow
+inputs:
+  input1: data
+outputs:
+  out1:
+    outputSource: first_cat/out_file1
+steps:
+  first_cat:
+    tool_id: cat
+    in:
+      input1: input1
+      queries_0|input2: input1
+      queries_1|input2: input1
+"""
+
+
+RULES_TOOL = """
+class: GalaxyWorkflow
+inputs:
+  input_c: collection
+outputs:
+  out1:
+    outputSource: random_lines/out_file1
+steps:
+  apply:
+    tool_id: __APPLY_RULES__
+    state:
+      input:
+        $link: input_c
+      rules:
+        rules:
+          - type: add_column_metadata
+            value: identifier0
+          - type: add_column_metadata
+            value: identifier0
+        mapping:
+          - type: list_identifiers
+            columns: [0, 1]
+  random_lines:
+    tool_id: random_lines1
+    state:
+      num_lines: 1
+      input:
+        $link: apply#output
+      seed_source:
+        seed_source_selector: set_seed
+        seed: asdf
+"""
+
+
+def setup_module(module):
+    # Setup an examples directory with examples we want to correspond to what exit codes,
+    # do this so we can run same tests in Java.
+    green_format2 = ordered_load(BASIC_WORKFLOW)
+    _dump_with_exit_code(green_format2, 0, "basic_format2")
+    green_native = to_native(BASIC_WORKFLOW)
+    assert_valid_native(green_native)
+    _dump_with_exit_code(green_native, 0, "basic_native")
+
+    invalid_format2_no_format_dict = copy.deepcopy(green_format2)
+    del invalid_format2_no_format_dict["class"]
+    _dump_with_exit_code(invalid_format2_no_format_dict, 2, "format2_no_class")
+
+    invalid_ga_no_format_dict = copy.deepcopy(green_native)
+    del invalid_ga_no_format_dict["a_galaxy_workflow"]
+    _dump_with_exit_code(invalid_ga_no_format_dict, 2, "native_no_class")
+
+    red_ga_no_outputs = copy.deepcopy(green_native)
+    red_ga_no_outputs_steps = red_ga_no_outputs.get("steps")
+    for step in red_ga_no_outputs_steps.values():
+        step.pop("workflow_outputs", None)
+    _dump_with_exit_code(red_ga_no_outputs, 1, "native_no_outputs")
+
+    red_ga_no_output_labels = copy.deepcopy(green_native)
+    red_ga_no_output_labels_steps = red_ga_no_output_labels.get("steps")
+    for step in red_ga_no_output_labels_steps.values():
+        for workflow_output in step.get("workflow_outputs", []):
+            workflow_output["label"] = None
+    _dump_with_exit_code(red_ga_no_outputs, 1, "native_no_output_labels")
+
+    red_format2_no_outputs = copy.deepcopy(green_format2)
+    del red_format2_no_outputs["outputs"]
+    _dump_with_exit_code(red_format2_no_outputs, 1, "format2_no_output")
+
+    green_format2_rules = ordered_load(RULES_TOOL)
+    _dump_with_exit_code(green_format2_rules, 0, "format2_rules")
+
+    green_native_rules = to_native(RULES_TOOL)
+    _dump_with_exit_code(green_native_rules, 0, "native_format")
+
+    green_format2_repeat = ordered_load(WORKFLOW_WITH_REPEAT)
+    _dump_with_exit_code(green_format2_repeat, 0, "format2_repeat")
+    green_native_rules = to_native(WORKFLOW_WITH_REPEAT)
+    _dump_with_exit_code(green_native_rules, 0, "native_repeat")
+
+
+def test_lint_ga_basic():
+    assert main(["lint", os.path.join(TEST_PATH, "wf3-shed-tools-raw.ga")]) == 1  # no outputs
+
+
+def test_lint_examples():
+    for file_name in os.listdir(TEST_EXAMPLES):
+        file_path = os.path.join(TEST_EXAMPLES, file_name)
+        expected_exit_code = int(file_name[0])
+        actual_exit_code = main(["lint", file_path])
+        if actual_exit_code != expected_exit_code:
+            contents = open(file_path, "r").read()
+            template = "File [%s] didn't lint properly - expected exit code [%d], got [%d]. Contents:\n%s"
+            raise AssertionError(template % (file_name, expected_exit_code, actual_exit_code, contents))
+
+
+def _dump_with_exit_code(as_dict, exit_code, description):
+    if not os.path.exists(TEST_EXAMPLES):
+        os.makedirs(TEST_EXAMPLES)
+    with open(os.path.join(TEST_EXAMPLES, "%d_%s.yml" % (exit_code, description)), "w") as fd:
+        ordered_dump(as_dict, fd)
+        fd.flush()

tests/wf3-shed-tools-raw.ga

@@ -0,0 +1 @@
+{"uuid": "eb7fc9c5-992b-495a-9baa-4a9bb67d6262", "tags": [], "format-version": "0.1", "name": "Workflow constructed from history 'Cell paper'", "steps": {"0": {"tool_id": null, "tool_version": null, "outputs": [], "workflow_outputs": [], "input_connections": {}, "tool_state": "{\"name\": \"ftp://ftp.ncbi.nlm.nih.gov/geo/series/GSE37nnn/GSE37268/suppl/GSE37268%5Fmof3%2Eout%2Ehpeak%2Etxt%2E\"}", "id": 0, "uuid": "4216f74b-f994-432a-b5be-2a6b981ac16a", "errors": null, "name": "Input dataset", "label": "ftp://ftp.ncbi.nlm.nih.gov/geo/series/GSE37nnn/GSE37268/suppl/GSE37268%5Fmof3%2Eout%2Ehpeak%2Etxt%2E", "inputs": [{"name": "ftp://ftp.ncbi.nlm.nih.gov/geo/series/GSE37nnn/GSE37268/suppl/GSE37268%5Fmof3%2Eout%2Ehpeak%2Etxt%2E", "description": ""}], "position": {"top": 335, "left": 200}, "annotation": "", "content_id": null, "type": "data_input"}, "1": {"tool_id": null, "tool_version": null, "outputs": [], "workflow_outputs": [], "input_connections": {}, "tool_state": "{\"name\": \"Genes\"}", "id": 1, "uuid": "0d2e344b-fd58-462c-bd21-c60acd045d9d", "errors": null, "name": "Input dataset", "label": "Genes", "inputs": [{"name": "Genes", "description": ""}], "position": {"top": 455, "left": 200}, "annotation": "", "content_id": null, "type": "data_input"}, "2": {"tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_replace_in_column/1.1.0", "tool_version": "1.1.0", "outputs": [{"type": "input", "name": "outfile"}], "workflow_outputs": [], "input_connections": {"infile": {"output_name": "output", "id": 0}}, "tool_state": "{\"__page__\": null, \"find_pattern\": \"\\\"[0-9]+\\\"\", \"column\": \"\\\"1\\\"\", \"chromInfo\": \"\\\"/usr/local/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"__rerun_remap_job_id__\": null, \"replace_pattern\": \"\\\"chr&\\\"\", \"infile\": \"{\\\"__class__\\\": \\\"RuntimeValue\\\"}\"}", "id": 2, "tool_shed_repository": {"owner": "bgruening", "changeset_revision": "20344ce0c811", "name": "text_processing", "tool_shed": "toolshed.g2.bx.psu.edu"}, "uuid": "d40b3d69-2a99-4b00-bf75-7ded995dfbd7", "errors": null, "name": "Replace Text", "post_job_actions": {}, "label": null, "inputs": [{"name": "infile", "description": "runtime parameter for tool Replace Text"}], "position": {"top": 335, "left": 420}, "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_replace_in_column/1.1.0", "type": "tool"}, "3": {"tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/get_flanks/get_flanks1/1.0.0", "tool_version": "1.0.0", "outputs": [{"type": "input", "name": "out_file1"}], "workflow_outputs": [], "input_connections": {"input": {"output_name": "output", "id": 1}}, "tool_state": "{\"__page__\": null, \"direction\": \"\\\"Upstream\\\"\", \"region\": \"\\\"start\\\"\", \"offset\": \"\\\"10000\\\"\", \"input\": \"{\\\"__class__\\\": \\\"RuntimeValue\\\"}\", \"chromInfo\": \"\\\"/usr/local/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/mm9.len\\\"\", \"__rerun_remap_job_id__\": null, \"size\": \"\\\"12000\\\"\"}", "id": 3, "tool_shed_repository": {"owner": "devteam", "changeset_revision": "077f404ae1bb", "name": "get_flanks", "tool_shed": "toolshed.g2.bx.psu.edu"}, "uuid": "652af7e8-710b-4476-952f-484d6a2e4a7a", "errors": null, "name": "Get flanks", "post_job_actions": {}, "label": null, "inputs": [{"name": "input", "description": "runtime parameter for tool Get flanks"}], "position": {"top": 475, "left": 415}, "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/devteam/get_flanks/get_flanks1/1.0.0", "type": "tool"}, "4": {"tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_replace_in_column/1.1.0", "tool_version": "1.1.0", "outputs": [{"type": "input", "name": "outfile"}], "workflow_outputs": [], "input_connections": {"infile": {"output_name": "outfile", "id": 2}}, "tool_state": "{\"__page__\": null, \"find_pattern\": \"\\\"chr20\\\"\", \"column\": \"\\\"1\\\"\", \"chromInfo\": \"\\\"/usr/local/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"__rerun_remap_job_id__\": null, \"replace_pattern\": \"\\\"chrX\\\"\", \"infile\": \"{\\\"__class__\\\": \\\"RuntimeValue\\\"}\"}", "id": 4, "tool_shed_repository": {"owner": "bgruening", "changeset_revision": "20344ce0c811", "name": "text_processing", "tool_shed": "toolshed.g2.bx.psu.edu"}, "uuid": "37f15259-3b03-4461-8da3-6e8ee34fd2f5", "errors": null, "name": "Replace Text", "post_job_actions": {}, "label": null, "inputs": [{"name": "infile", "description": "runtime parameter for tool Replace Text"}], "position": {"top": 335, "left": 640}, "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_replace_in_column/1.1.0", "type": "tool"}, "5": {"tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_replace_in_column/1.1.0", "tool_version": "1.1.0", "outputs": [{"type": "input", "name": "outfile"}], "workflow_outputs": [], "input_connections": {"infile": {"output_name": "outfile", "id": 4}}, "tool_state": "{\"__page__\": null, \"find_pattern\": \"\\\"chr21\\\"\", \"column\": \"\\\"1\\\"\", \"chromInfo\": \"\\\"/usr/local/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"__rerun_remap_job_id__\": null, \"replace_pattern\": \"\\\"chrY\\\"\", \"infile\": \"{\\\"__class__\\\": \\\"RuntimeValue\\\"}\"}", "id": 5, "tool_shed_repository": {"owner": "bgruening", "changeset_revision": "20344ce0c811", "name": "text_processing", "tool_shed": "toolshed.g2.bx.psu.edu"}, "uuid": "8ce00200-94b7-4daf-a0e5-e3b8d6d87eeb", "errors": null, "name": "Replace Text", "post_job_actions": {}, "label": null, "inputs": [{"name": "infile", "description": "runtime parameter for tool Replace Text"}], "position": {"top": 335, "left": 860}, "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_replace_in_column/1.1.0", "type": "tool"}, "6": {"tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/bedtools/bedtools_intersectbed/2.27.0.1", "tool_version": "2.27.0.1", "outputs": [{"type": "input", "name": "output"}], "workflow_outputs": [], "input_connections": {"inputA": {"output_name": "out_file1", "id": 3}, "reduce_or_iterate|inputB": {"output_name": "outfile", "id": 5}}, "tool_state": "{\"count\": \"\\\"false\\\"\", \"__page__\": null, \"reciprocal\": \"\\\"false\\\"\", \"overlap_mode\": \"\\\"-wa\\\"\", \"invert\": \"\\\"false\\\"\", \"header\": \"\\\"false\\\"\", \"inputA\": \"{\\\"values\\\": [{\\\"src\\\": \\\"hda\\\", \\\"id\\\": 6112359}]}\", \"reduce_or_iterate\": \"{\\\"inputB\\\": {\\\"values\\\": [{\\\"src\\\": \\\"hda\\\", \\\"id\\\": 6112359}]}, \\\"reduce_or_iterate_selector\\\": \\\"iterate\\\", \\\"__current_case__\\\": 0}\", \"split\": \"\\\"false\\\"\", \"fraction\": \"\\\"\\\"\", \"__rerun_remap_job_id__\": null, \"strand\": \"\\\"\\\"\", \"once\": \"\\\"false\\\"\"}", "id": 6, "tool_shed_repository": {"owner": "iuc", "changeset_revision": "33c3ddea63c5", "name": "bedtools", "tool_shed": "toolshed.g2.bx.psu.edu"}, "uuid": "c65fd1f5-9fc2-4acf-96c6-94ea61906042", "errors": null, "name": "Intersect intervals", "post_job_actions": {}, "label": null, "inputs": [], "position": {"top": 655, "left": 1022}, "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/bedtools/bedtools_intersectbed/2.27.0.1", "type": "tool"}, "7": {"tool_id": "Grouping1", "tool_version": "2.1.1", "outputs": [{"type": "tabular", "name": "out_file1"}], "workflow_outputs": [], "input_connections": {"input1": {"output_name": "output", "id": 6}}, "tool_state": "{\"operations\": \"[{\\\"opcol\\\": \\\"1\\\", \\\"__index__\\\": 0, \\\"optype\\\": \\\"length\\\", \\\"opround\\\": \\\"no\\\"}]\", \"__page__\": null, \"input1\": \"{\\\"__class__\\\": \\\"RuntimeValue\\\"}\", \"ignorelines\": \"null\", \"groupcol\": \"\\\"1\\\"\", \"__rerun_remap_job_id__\": null, \"ignorecase\": \"\\\"false\\\"\", \"chromInfo\": \"\\\"/usr/local/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/mm9.len\\\"\"}", "id": 7, "uuid": "be264526-e498-4050-b6cd-f7a842268d4a", "errors": null, "name": "Group", "post_job_actions": {}, "label": null, "inputs": [{"name": "input1", "description": "runtime parameter for tool Group"}], "position": {"top": 335, "left": 1300}, "annotation": "", "content_id": "Grouping1", "type": "tool"}}, "annotation": "", "a_galaxy_workflow": "true"}