Merge pull request #9 from marqh/prefcont

marqh · web-flow · commit 70f82d302008 · 2016-10-18T13:01:08.000Z
Prefcont
diff --git a/.travis.yml b/.travis.yml
@@ -0,0 +1,23 @@
+language: python
+python:
+  - "2.7"
+install:
+  - wget https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh
+  - bash Miniconda3-latest-Linux-x86_64.sh -b -p $HOME/miniconda
+  - export PATH="$HOME/miniconda/bin:$PATH"
+
+  - conda config --add channels conda-forge
+  - conda config --add channels bioconda
+  - conda config --set always_yes yes --set changeps1 no
+  - conda config --set show_channel_urls True
+
+  - ENV_NAME='testing'
+  - conda create --quiet -n $ENV_NAME python=$TRAVIS_PYTHON_VERSION
+  - source activate $ENV_NAME
+  - conda install --quiet --file conda-requirements.txt
+  - conda list
+  - conda info -a
+  - python setup.py --quiet install
+
+script:
+  python -m unittest discover -s bald.tests -v
diff --git a/README.md b/README.md
@@ -1,3 +1,5 @@
-# bald
+# Binary Array Linked Data: bald
+
 [![Build Status](https://api.travis-ci.org/repositories/binary-array-ld/bald.svg?branch=master)](http://travis-ci.org/binary-array-ld/bald/branches)
-Python library for validating and managing binary array linked data files.
+
+A Python library for validating and managing binary array linked data files.
diff --git a/conda-requirements.txt b/conda-requirements.txt
@@ -0,0 +1,5 @@
+numpy
+h5py
+netCDF4
+requests
+rdflib
diff --git a/lib/bald/__init__.py b/lib/bald/__init__.py
@@ -7,6 +7,7 @@
 
 import bald.validation as bv
 
+__version__ = '0.2'
 
 class HttpCache(object):
     """
@@ -15,9 +16,12 @@ class HttpCache(object):
     def __init__(self):
         self.cache = {}
 
+    def is_http_uri(self, item):
+        return item.startswith('http://') or item.startswith('https://')
+
     def __getitem__(self, item):
 
-        if not item.startswith('http://') or item.startswith('https://'):
+        if not self.is_http_uri(item):
             raise ValueError('{} is not a HTTP URI.'.format(item))
         if item not in self.cache:
             headers = {'Accept': 'text/turtle'}
@@ -33,23 +37,29 @@ def check_uri(self, uri):
 
 
 class Subject(object):
-    def __init__(self, attrs=None):
+    def __init__(self, attrs=None, prefixes=None, aliases=None):
         """
         A subject of metadata statements.
 
         attrs: an dictionary of key value pair attributes
         """
         if attrs is None:
-            attrs = []
+            attrs = {}
+        if prefixes is None:
+            prefixes = {}
+        if aliases is None:
+            aliases = {}
         self.attrs = attrs
+        self.aliases = aliases
+        self._prefixes = prefixes
         self._prefix_suffix = re.compile('(^(?:(?!__).)*)__((?!.*__).*$)')
         _http_p = 'http[s]?://.*'
         self._http_uri = re.compile('{}'.format(_http_p))
         self._http_uri_prefix = re.compile('{}/|#'.format(_http_p))
 
     def prefixes(self):
         prefixes = {}
-        for key, value in self.attrs.iteritems():
+        for key, value in self._prefixes.iteritems():
             if key.endswith('__') and self._http_uri_prefix.match(value):
                 pref = key.rstrip('__')
                 if pref in prefixes:
@@ -66,6 +76,8 @@ def unpack_uri(self, astring):
                 if self._http_uri.match(self.prefixes()[prefix]):
                     result = astring.replace('{}__'.format(prefix),
                                              self.prefixes()[prefix])
+        elif astring in self.aliases:
+            result = self.aliases[astring]
         return result
 
 
@@ -105,17 +117,25 @@ def validate_netcdf(afilepath):
 
     with load(afilepath) as fhandle:
         sval = bv.StoredValidation()
+        prefix_group = fhandle[fhandle.bald__isPrefixedBy] if hasattr(fhandle, 'bald__isPrefixedBy') else {}
+        prefixes = {}
+        if prefix_group:
+            prefixes = dict([(prefix, getattr(prefix_group, prefix)) for prefix in prefix_group.ncattrs()])
+        else:
+            for k in fhandle.ncattrs():
+                if k.endswith('__'):
+                    prefixes[k] = getattr(fhandle, k)            
         attrs = {}
         for k in fhandle.ncattrs():
             attrs[k] = getattr(fhandle, k)
-        root_container = Subject(attrs)
+        root_container = Subject(attrs, prefixes=prefixes)
         root_val = bv.ContainerValidation(subject=root_container,
                                           fhandle=fhandle)
         sval.stored_exceptions += root_val.exceptions()
         for name in fhandle.variables:
             sattrs = fhandle.__dict__.copy()
             sattrs.update(fhandle.variables[name].__dict__.copy())
-            var = Subject(sattrs)
+            var = Subject(sattrs, prefixes=prefixes)
             var_val = bv.ArrayValidation(name, fhandle.variables[name], fhandle=fhandle,
                                          subject=var)
             sval.stored_exceptions += var_val.exceptions()
@@ -132,7 +152,15 @@ def validate_hdf5(afilepath):
     with load(afilepath) as fhandle:
         sval = bv.StoredValidation()
         cache = {}
-        root_container = Subject(fhandle.attrs)
+        prefix_group = fhandle.attrs.get('bald__isPrefixedBy')
+        prefixes = {}
+        if prefix_group:
+            prefixes = fhandle[prefix_group].attrs
+        alias_group = fhandle.attrs.get('bald__isAliasedBy')
+        aliases = {}
+        if alias_group:
+            aliases = dict(fhandle[alias_group].attrs.iteritems())
+        root_container = Subject(fhandle.attrs, prefixes=prefixes, aliases=aliases)
         root_val = bv.ContainerValidation(subject=root_container,
                                           fhandle=fhandle)
         sval.stored_exceptions += root_val.exceptions()
@@ -144,7 +172,7 @@ def validate_hdf5(afilepath):
             # #
             sattrs = dict(fhandle.attrs).copy()
             sattrs.update(dataset.attrs)
-            dset = Subject(sattrs)
+            dset = Subject(sattrs, prefixes, aliases)
             dset_val = bv.ArrayValidation(name, dataset, fhandle=fhandle,
                                             subject=dset)
             sval.stored_exceptions += dset_val.exceptions()
diff --git a/lib/bald/tests/integration/test_aliases.py b/lib/bald/tests/integration/test_aliases.py
@@ -0,0 +1,59 @@
+import unittest
+
+import h5py
+import numpy as np
+
+import bald
+from bald.tests import BaldTestCase
+
+def _fattrs(f):
+    f.attrs['rdf__type'] = 'bald__Container'
+    group_pref = f.create_group('bald__prefix_list')
+    group_pref.attrs['bald__'] = 'http://binary-array-ld.net/latest/'
+    group_pref.attrs['rdf__'] = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#'
+    f.attrs['bald__isPrefixedBy'] = group_pref.ref
+    return f
+
+def _create_parent_child(f, pshape, cshape):
+    dsetp = f.create_dataset("parent_dataset", pshape, dtype='i')
+    dsetc = f.create_dataset("child_dataset", cshape, dtype='i')
+    dsetp.attrs['rdf__type'] = 'bald__Array'
+    dsetp.attrs['bald__references'] = dsetc.ref
+    dsetc.attrs['rdf__type'] = 'bald__Array'
+    dsetc.attrs['rdf__type'] = 'bald__Reference'
+    dsetc.attrs['bald__array'] = dsetc.ref
+    return f
+
+
+class Test(BaldTestCase):
+        
+    def test_valid_uri(self):
+        with self.temp_filename('.hdf') as tfile:
+            f = h5py.File(tfile, "w")
+            f = _fattrs(f)
+            group_alias = f.create_group('bald__alias_list')
+            f.attrs['bald__isAliasedBy'] = group_alias.ref
+            group_alias.attrs['skosPrefLabel'] = 'http://www.w3.org/2004/02/skos/core#prefLabel'
+            dsetp = f.create_dataset("parent_dataset", (11, 17), dtype='i')
+            dsetp.attrs['skosPrefLabel'] = 'alabel'
+            f.close()
+            validation = bald.validate_hdf5(tfile)
+            self.assertTrue(validation.is_valid())
+
+    def test_invalid_uri(self):
+        with self.temp_filename('.hdf') as tfile:
+            f = h5py.File(tfile, "w")
+            f = _fattrs(f)
+            f.attrs['bald__turtle'] = 'bald__walnut'
+            group_alias = f.create_group('bald__alias_list')
+            f.attrs['bald__isAliasedBy'] = group_alias.ref            
+            group_alias.attrs['skosPrefLabel'] = 'http://www.w3.org/2004/02/skos/core#notThisPrefLabel'
+            dsetp = f.create_dataset("parent_dataset", (11, 17), dtype='i')
+            dsetp.attrs['skosPrefLabel'] = 'alabel'
+            f.close()
+            validation = bald.validate_hdf5(tfile)
+            self.assertFalse(validation.is_valid())
+
+
+if __name__ == '__main__':
+    unittest.main()
diff --git a/lib/bald/tests/integration/test_netcdf.py b/lib/bald/tests/integration/test_netcdf.py
@@ -8,9 +8,11 @@
 from bald.tests import BaldTestCase
 
 def _fattrs(f):
-    f.bald__ = 'http://binary-array-ld.net/latest/'
-    f.rdf__ = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#'
     f.rdf__type =  'bald__Container'
+    group_pref = f.createGroup('bald__prefix_list')
+    group_pref.bald__ = 'http://binary-array-ld.net/latest/'
+    group_pref.rdf__ = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#'
+    f.bald__isPrefixedBy = 'bald__prefix_list'
     return f
 
 def _create_parent_child(f, pshape, cshape):
diff --git a/lib/bald/tests/integration/test_netcdf4_classic.py b/lib/bald/tests/integration/test_netcdf4_classic.py
@@ -0,0 +1,78 @@
+import unittest
+
+import h5py
+import netCDF4
+import numpy as np
+
+import bald
+from bald.tests import BaldTestCase
+
+def _fattrs(f):
+    f.bald__ = 'http://binary-array-ld.net/latest/'
+    f.rdf__ = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#'
+    f.rdf__type =  'bald__Container'
+    return f
+
+def _create_parent_child(f, pshape, cshape):
+    for i, pdimsize in enumerate(pshape):
+        f.createDimension("pdim{}".format(str(i)), pdimsize)
+    for i, cdimsize in enumerate(cshape):
+        f.createDimension("cdim{}".format(str(i)), cdimsize)
+    varp = f.createVariable("parent_variable", 'i4', tuple(["pdim{}".format(str(i)) for i, _ in enumerate(pshape)]))
+    varc = f.createVariable("child_variable", 'i4', tuple(["cdim{}".format(str(i)) for i, _ in enumerate(cshape)]))
+    varp.rdf__type = 'bald__Array'
+    varp.bald__references = "child_variable"
+    varc.rdf__type = 'bald__Array'
+    varc.rdf__type = 'bald__Reference'
+    varc.bald__array = "child_variable"
+    return f
+
+
+class Test(BaldTestCase):
+        
+    def test_valid_uri(self):
+        with self.temp_filename('.nc') as tfile:
+            f = netCDF4.Dataset(tfile, "w", format="NETCDF4_CLASSIC")
+
+            f = _fattrs(f)
+            f.close()
+            validation = bald.validate_netcdf(tfile)
+            self.assertTrue(validation.is_valid())
+
+    def test_invalid_uri(self):
+        with self.temp_filename('.nc') as tfile:
+            f = netCDF4.Dataset(tfile, "w", format="NETCDF4_CLASSIC")
+
+            f = _fattrs(f)
+            setattr(f, 'bald__turtle', 'bald__walnut')
+            f.close()
+            validation = bald.validate_netcdf(tfile)
+            self.assertFalse(validation.is_valid())
+
+
+class TestArrayReference(BaldTestCase):
+    def test_match(self):
+        with self.temp_filename('.nc') as tfile:
+            f = netCDF4.Dataset(tfile, "w", format="NETCDF4_CLASSIC")
+            f = _fattrs(f)
+            f = _create_parent_child(f, (11, 17), (11, 17))
+            f.close()
+            validation = bald.validate_netcdf(tfile)
+            self.assertTrue(validation.is_valid())
+
+    def test_mismatch_zeroth(self):
+        with self.temp_filename('.nc') as tfile:
+            f = netCDF4.Dataset(tfile, "w", format="NETCDF4_CLASSIC")
+            f = _fattrs(f)
+            f = _create_parent_child(f, (11, 17), (11, 13))
+            f.close()
+            validation = bald.validate_netcdf(tfile)
+            self.assertFalse(validation.is_valid())
+
+
+if __name__ == '__main__':
+    unittest.main()
+
+
+
+
diff --git a/lib/bald/tests/integration/test_validation.py b/lib/bald/tests/integration/test_validation.py
@@ -7,9 +7,11 @@
 from bald.tests import BaldTestCase
 
 def _fattrs(f):
-    f.attrs['bald__'] = 'http://binary-array-ld.net/latest/'
-    f.attrs['rdf__'] = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#'
     f.attrs['rdf__type'] = 'bald__Container'
+    group_pref = f.create_group('bald__prefix_list')
+    group_pref.attrs['bald__'] = 'http://binary-array-ld.net/latest/'
+    group_pref.attrs['rdf__'] = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#'
+    f.attrs['bald__isPrefixedBy'] = group_pref.ref
     return f
 
 def _create_parent_child(f, pshape, cshape):
diff --git a/lib/bald/tests/unit/test_HttpCache.py b/lib/bald/tests/unit/test_HttpCache.py
@@ -4,16 +4,19 @@
 import numpy as np
 
 from bald.tests import BaldTestCase
-from bald import validation
+import bald
+
+class TestHttpCache(unittest.TestCase):
+    def setUp(self):
+        self.cache = bald.HttpCache()
 
-class Test(unittest.TestCase):
     def test_check_uri_200(self):
         auri = 'http://binary-array-ld.net/experimental'
-        self.assertTrue(validation.check_uri(auri))
+        self.assertTrue(self.cache.check_uri(auri))
 
     def test_check_uri_404(self):
         notauri = 'http://binary-array-ld.net/experimentalish'
-        self.assertFalse(validation.check_uri(notauri))
+        self.assertFalse(self.cache.check_uri(notauri))
 
 
 if __name__ == '__main__':
diff --git a/lib/bald/validation.py b/lib/bald/validation.py
@@ -91,12 +91,16 @@ def _check_uri(uri, exceptions):
         for pref, uri in self.subject.prefixes().iteritems():
             exceptions = _check_uri(self.subject.unpack_uri(uri),
                                     exceptions)
+        for alias, uri in self.subject.aliases.iteritems():
+            exceptions = _check_uri(self.subject.unpack_uri(uri),
+                                    exceptions)
         for attr, value in self.subject.attrs.iteritems():
             exceptions = _check_uri(self.subject.unpack_uri(attr),
                                     exceptions)
             if isinstance(value, str):
-                exceptions = _check_uri(self.subject.unpack_uri(value),
-                                        exceptions)
+                val = self.subject.unpack_uri(value)
+                if self.cache.is_http_uri(val):
+                    exceptions = _check_uri(val, exceptions)
         return exceptions
 
     def check_attr_domain_range(self, exceptions):
@@ -106,7 +110,11 @@ def check_attr_domain_range(self, exceptions):
                 # thus we have a payload
                 # go rdf
                 g = rdflib.Graph()
-                g.parse(data=self.cache[uri].text, format="n3")
+                data=self.cache[uri].text
+                try:
+                    g.parse(data=self.cache[uri].text, format="n3")
+                except Exception:
+                    g.parse(data=self.cache[uri].text, format="xml")
                 query = ('SELECT ?s \n'
                          '(GROUP_CONCAT(?domain; SEPARATOR=" | ") AS ?domains)'
                          ' \n'
diff --git a/setup.py b/setup.py

-Original file line number
+Diff line change
@@ @@ -0,0 +1,5 @@ @@
 +numpy
 +h5py
 +netCDF4
 +requests
 +rdflib