Validation of output tarballs

docs/requirements.txt

@@ -2,4 +2,3 @@ m2r
 sphinx_rtd_theme
 docutils<0.18
 mistune<2.0.0
-

examples/Getting_started.ipynb

@@ -310,7 +310,7 @@
    "outputs": [],
    "source": [
     "outdir = \"example_output\"\n",
-    "submission.create_files(outdir)"
+    "submission.create_files(outdir,remove_old=True)"
    ]
   },
   {

examples/combine_limits.ipynb

@@ -276,7 +276,7 @@
     "table.add_variable(obs)\n",
     "table.add_variable(exp)\n",
     "submission.add_table(table)\n",
-    "submission.create_files(\"example_output\")"
+    "submission.create_files(\"example_output\",remove_old=True)"
    ]
   },
   {

examples/correlation.ipynb

@@ -282,7 +282,7 @@
     "# Create the submission object and write output\n",
     "sub = Submission()\n",
     "sub.add_table(table)\n",
-    "sub.create_files(\"./output/\")\n",
+    "sub.create_files(\"./output/\",remove_old=True)\n",
     "\n",
     "!ls -l submission.tar.gz"
    ]

examples/read_c_file.ipynb

@@ -152,7 +152,7 @@
     "table.add_variable(obs)\n",
     "table.add_variable(exp)\n",
     "submission.add_table(table)\n",
-    "submission.create_files(\"example_output\")"
+    "submission.create_files(\"example_output\",remove_old=True)"
    ]
   },
   {

examples/reading_histograms.ipynb

@@ -258,7 +258,7 @@
     "\n",
     "submission.add_table(table)\n",
     "\n",
-    "submission.create_files(\"example_output\")"
+    "submission.create_files(\"example_output\",remove_old=True)"
    ]
   },
   {
@@ -311,7 +311,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.6.5"
+   "version": "3.8.10"
   }
  },
  "nbformat": 4,

hepdata_lib/__init__.py

@@ -11,14 +11,14 @@
 import numpy as np
 import yaml
 from future.utils import raise_from
-
 # try to use LibYAML bindings if possible
 try:
     from yaml import CLoader as Loader, CSafeDumper as Dumper
 except ImportError:
     from yaml import Loader, SafeDumper as Dumper
 from yaml.representer import SafeRepresenter
 
+from hepdata_validator.full_submission_validator import FullSubmissionValidator
 from hepdata_lib import helpers
 from hepdata_lib.root_utils import RootFileReader
 
@@ -538,13 +538,21 @@ def files_to_copy_nested(self):
             files = files + table.files_to_copy
         return files
 
-    def create_files(self, outdir="."):
+    def create_files(self, outdir=".", validate=True, remove_old=True):
         """
         Create the output files.
 
         Implicitly triggers file creation for all tables that have been added to the submission,
         all variables associated to the tables and all uncertainties associated to the variables.
+
+        If `validate` is True, the hepdata-validator package will be used to validate the
+        output tar ball.
+
+        If `remove_old` is True, the output directory will be deleted before recreation.
         """
+        if remove_old and os.path.exists(outdir):
+            shutil.rmtree(outdir)
+
         if not os.path.exists(outdir):
             os.makedirs(outdir)
 
@@ -580,13 +588,21 @@ def create_files(self, outdir="."):
         files_to_add.extend(
             [os.path.join(outdir, os.path.basename(x)) for x in  self.files_to_copy_nested()]
         )
-        with tarfile.open("submission.tar.gz", "w:gz") as tar:
+        tarfile_path = "submission.tar.gz"
+        with tarfile.open(tarfile_path, "w:gz") as tar:
             for filepath in files_to_add:
                 tar.add(
                         filepath,
                         arcname=os.path.basename(filepath)
                         )
 
+        if validate:
+            full_submission_validator = FullSubmissionValidator()
+            is_archive_valid = full_submission_validator.validate(archive=tarfile_path)
+            if not is_archive_valid:
+                for filename in full_submission_validator.get_messages():
+                    full_submission_validator.print_errors(filename)
+            assert is_archive_valid, "The tar ball is not valid"
 
 class Uncertainty(object):
     """

requirements.txt

@@ -0,0 +1,5 @@
+numpy
+PyYAML>4.*
+future
+six
+hepdata-validator>=0.3.2

setup.py

@@ -8,7 +8,8 @@
 except ImportError:
     print("ROOT is required by this library.")
 
-DEPS = ['numpy', 'PyYAML>4.*', 'future', 'six']
+with open("requirements.txt","r") as f:
+    DEPS = f.readlines()
 
 HERE = path.abspath(path.dirname(__file__))
 
@@ -27,8 +28,6 @@
     classifiers=[
         'Development Status :: 4 - Beta',
         'License :: OSI Approved :: MIT License',
-        'Programming Language :: Python :: 2',
-        'Programming Language :: Python :: 2.7',
         'Programming Language :: Python :: 3',
         'Programming Language :: Python :: 3.6',
         'Programming Language :: Python :: 3.7',
@@ -39,7 +38,6 @@
     install_requires=DEPS,
     setup_requires=['pytest-runner', 'pytest-cov'],
     tests_require=['pytest', 'papermill', 'six',
-                   'pylint==1.9.5; python_version<"3"',
                    'pylint==2.9.6; python_version>="3"',
                   ],
     project_urls={

tests/test_output.py

@@ -2,18 +2,20 @@
 # -*- coding:utf-8 -*-
 """Test Output."""
 from __future__ import print_function
-import os
 from collections import defaultdict
 from unittest import TestCase
 import shutil
+import os
 import yaml
+from test_utilities import tmp_directory_name
 from hepdata_lib import Submission, Table, Variable
 
 class TestOutput(TestCase):
     """Test output"""
 
     def test_yaml_output(self):
         """Test yaml dump"""
+        tmp_dir = tmp_directory_name()
 
         # Create test dictionary
         testlist = [("x", 1.2), ("x", 2.2), ("y", 0.12), ("y", 0.22)]
@@ -23,7 +25,6 @@ def test_yaml_output(self):
 
         # Create test submission
         test_submission = Submission()
-        testdir = "./output"
         test_table = Table("TestTable")
         x_variable = Variable("X", is_independent=True, is_binned=False)
         x_variable.values = testdict['x']
@@ -32,11 +33,12 @@ def test_yaml_output(self):
         test_table.add_variable(x_variable)
         test_table.add_variable(y_variable)
         test_submission.add_table(test_table)
-        test_submission.create_files(testdir)
+        test_submission.create_files(tmp_dir)
 
         # Test read yaml file
+        table_file = os.path.join(tmp_dir, "testtable.yaml")
         try:
-            with open("output/testtable.yaml", 'r') as testfile:
+            with open(table_file, 'r') as testfile:
                 testyaml = yaml.safe_load(testfile)
         except yaml.YAMLError as exc:
             print(exc)
@@ -45,10 +47,10 @@ def test_yaml_output(self):
         testtxt = ("dependent_variables:\n- header:\n    name: Y\n  values:\n" +
                    "  - value: 0.12\n  - value: 0.22\nindependent_variables:\n" +
                    "- header:\n    name: X\n  values:\n  - value: 1.2\n  - value: 2.2\n")
-        with open("output/testtable.yaml", 'r') as testfile:
+        with open(table_file, 'r') as testfile:
             testyaml = testfile.read()
 
         self.assertEqual(str(testyaml), testtxt)
         self.addCleanup(os.remove, "submission.tar.gz")
-        self.addCleanup(shutil.rmtree, testdir)
+        self.addCleanup(shutil.rmtree, tmp_dir)
         self.doCleanups()

tests/test_submission.py

@@ -7,6 +7,7 @@
 from builtins import bytes
 from unittest import TestCase
 import tarfile
+from test_utilities import tmp_directory_name
 from hepdata_lib import Submission, Table, Variable, Uncertainty
 
 class TestSubmission(TestCase):
@@ -71,15 +72,45 @@ def test_additional_resource_size(self):
     def test_create_files(self):
         """Test create_files() for Submission."""
 
-        testdir = "test_output"
+        testdir = tmp_directory_name()
         test_submission = Submission()
-        self.addCleanup(os.remove, "submission.tar.gz")
-        self.addCleanup(shutil.rmtree, testdir)
-
+        tab = Table("test")
+        test_submission.add_table(tab)
         test_submission.create_files(testdir)
 
         self.doCleanups()
 
+    def test_create_files_with_removal(self):
+        """Test the removal of old files in create_files()"""
+        testdir = tmp_directory_name()
+
+        # Step 1: Create test directory containing random file
+        os.makedirs(testdir)
+        self.addCleanup(shutil.rmtree, testdir)
+        testfile = os.path.join(testdir, "test.txt")
+        with open(testfile, "w") as f:
+            f.write("test")
+        self.assertTrue(os.path.isfile(testfile))
+
+        # Step 2: Create submission and write output to test directory
+        # Without overwriting of files
+        test_submission = Submission()
+        tab = Table("test")
+        test_submission.add_table(tab)
+        test_submission.create_files(testdir, remove_old=False)
+
+        # Test file should still exist
+        self.assertTrue(os.path.isfile(testfile))
+
+        # Step 3: Recreate submission files with removal
+        test_submission.create_files(testdir, remove_old=True)
+
+        # Test file should no longer exist
+        self.assertFalse(os.path.isfile(testfile))
+
+
+
+
     def test_read_abstract(self):
         """Test read_abstract function."""
         some_string = string.ascii_lowercase

tests/test_table.py

@@ -5,6 +5,7 @@
 import shutil
 from unittest import TestCase
 
+from test_utilities import tmp_directory_name
 from hepdata_lib import Table, Variable, Uncertainty
 
 class TestTable(TestCase):
@@ -61,7 +62,7 @@ def test_write_yaml(self):
         test_table = Table("Some Table")
         test_variable = Variable("Some Variable")
         test_table.add_variable(test_variable)
-        testdir = "test_output"
+        testdir = tmp_directory_name()
         self.addCleanup(shutil.rmtree, testdir)
         try:
             test_table.write_yaml(testdir)
@@ -110,7 +111,7 @@ def test_write_images(self):
 
         # This should work fine
         test_table.add_image(some_pdf)
-        testdir = "test_output"
+        testdir = tmp_directory_name()
         self.addCleanup(shutil.rmtree, testdir)
         try:
             test_table.write_images(testdir)
@@ -182,7 +183,7 @@ def test_copy_files(self):
         """Test the copy_files function."""
         test_table = Table("Some Table")
         some_pdf = "%s/minimal.pdf" % os.path.dirname(__file__)
-        testdir = "test_output"
+        testdir = tmp_directory_name()
         self.addCleanup(shutil.rmtree, testdir)
         os.makedirs(testdir)
 

tests/test_utilities.py

@@ -62,6 +62,17 @@ def get_random_id(length=12):
     return "".join(random.sample(string.ascii_uppercase+string.digits, length))
 
 
+def tmp_directory_name():
+    """
+    Generate a random directory name for testing.
+
+    Guaranteed to not exist.
+    """
+    tmp_name = "/tmp/hepdata_lib_test_" + get_random_id()
+    if os.path.exists(tmp_name):
+        return tmp_directory_name()
+    return tmp_name
+
 def remove_if_exist(path_to_file):
     """Remove file if it exists."""
     if os.path.exists(path_to_file):