Merge branch 'mb/extractors' into develop

16 files changed, 406 insertions, 4 deletions

diff --git a/bsie/base/extractor.py b/bsie/base/extractor.py
index d5b0922..ea43925 100644
--- a/bsie/base/extractor.py
+++ b/bsie/base/extractor.py
@@ -6,7 +6,6 @@ Author: Matthias Baumgartner, 2022
 """
 # imports
 import abc
-import collections
 import typing
 
 # inner-module imports
@@ -22,7 +21,7 @@ __all__: typing.Sequence[str] = (
 
 ## code ##
 
-class Extractor(abc.ABC, collections.abc.Iterable, collections.abc.Callable):
+class Extractor(abc.ABC):
     """Produce (node, predicate, value)-triples from some content."""
 
     # what type of content is expected (i.e. reader subclass).
diff --git a/bsie/extractor/__init__.py b/bsie/extractor/__init__.py
new file mode 100644
index 0000000..ef31343
--- /dev/null
+++ b/bsie/extractor/__init__.py
@@ -0,0 +1,15 @@
+"""Extractors produce triples from some content.
+
+Each Extractor class is linked to the Reader class whose content it requires.
+
+Part of the bsie module.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import typing
+
+# exports
+__all__: typing.Sequence[str] = []
+
+## EOF ##
diff --git a/bsie/extractor/generic/__init__.py b/bsie/extractor/generic/__init__.py
new file mode 100644
index 0000000..0cb7e7f
--- /dev/null
+++ b/bsie/extractor/generic/__init__.py
@@ -0,0 +1,16 @@
+"""Generic extractors focus on information that is typically available on all
+files. Examples include file system information (file name and size, mime type,
+etc.) and information that is independent of the actual file (constant triples,
+host platform infos, current time, etc.).
+
+Part of the bsie module.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import typing
+
+# exports
+__all__: typing.Sequence[str] = []
+
+## EOF ##
diff --git a/bsie/extractor/generic/constant.py b/bsie/extractor/generic/constant.py
new file mode 100644
index 0000000..e243131
--- /dev/null
+++ b/bsie/extractor/generic/constant.py
@@ -0,0 +1,52 @@
+"""The Constant extractor produces pre-specified triples.
+
+Part of the bsie module.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import typing
+
+# inner-module imports
+from bsie.base import extractor
+from bsie.utils.bsfs import URI
+from bsie.utils.node import Node
+
+# exports
+__all__: typing.Sequence[str] = (
+    'Constant',
+    )
+
+
+## code ##
+
+class Constant(extractor.Extractor):
+    """Extract information from file's path."""
+
+    CONTENT_READER = None
+
+    def __init__(
+            self,
+            schema: str,
+            tuples: typing.Iterable[typing.Tuple[URI, typing.Any]],
+            ):
+        self._schema = schema
+        self._tuples = tuples
+        # FIXME: use schema instance for predicate checking
+        #self._tuples = [(pred, value) for pred, value in tuples if pred in schema]
+        # FIXME: use schema instance for value checking
+
+    def schema(self) -> str:
+        return self._schema
+
+    def extract(
+            self,
+            subject: Node,
+            content: None,
+            predicates: typing.Iterable[URI],
+            ) -> typing.Iterator[typing.Tuple[Node, URI, typing.Any]]:
+        for pred, value in self._tuples:
+            if pred in predicates:
+                yield subject, pred, value
+
+## EOF ##
diff --git a/bsie/extractor/generic/path.py b/bsie/extractor/generic/path.py
new file mode 100644
index 0000000..c39bbd2
--- /dev/null
+++ b/bsie/extractor/generic/path.py
@@ -0,0 +1,70 @@
+"""
+
+Part of the bsie module.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import os
+import typing
+
+# inner-module imports
+from bsie.base import extractor
+from bsie.utils import node, ns
+from bsie.utils.bsfs import URI
+import bsie.reader.path
+
+# exports
+__all__: typing.Sequence[str] = (
+    'Path',
+    )
+
+
+## code ##
+
+class Path(extractor.Extractor):
+    """Extract information from file's path."""
+
+    CONTENT_READER = bsie.reader.path.Path
+
+    def __init__(self):
+        self.__callmap = {
+            ns.bse.filename:    self.__filename,
+            }
+
+    def schema(self) -> str:
+        return '''
+            bse:filename a bsfs:Predicate ;
+                rdfs:domain bsfs:Entity ;
+                rdfs:range xsd:string ;
+                rdf:label "File name"^^xsd:string ;
+                schema:description "Filename of entity in some filesystem."^^xsd:string ;
+                owl:maxCardinality "INF"^^xsd:number .
+            '''
+
+    def extract(
+            self,
+            subject: node.Node,
+            content: CONTENT_READER.CONTENT_TYPE,
+            predicates: typing.Iterable[URI],
+            ) -> typing.Iterator[typing.Tuple[node.Node, URI, typing.Any]]:
+        for pred in predicates:
+            # find callback
+            clbk = self.__callmap.get(pred)
+            if clbk is None:
+                continue
+            # get value
+            value = clbk(content)
+            if value is None:
+                continue
+            # produce triple
+            yield subject, pred, value
+
+    def __filename(self, path: str) -> str:
+        try:
+            return os.path.basename(path)
+        except Exception:
+            # FIXME: some kind of error reporting (e.g. logging)
+            return None
+
+## EOF ##
diff --git a/bsie/extractor/generic/stat.py b/bsie/extractor/generic/stat.py
new file mode 100644
index 0000000..d74369c
--- /dev/null
+++ b/bsie/extractor/generic/stat.py
@@ -0,0 +1,71 @@
+"""Extract information from the file system, such as filesize.
+
+Part of the bsie module.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import typing
+
+# inner-module imports
+from bsie.base import extractor
+from bsie.utils import node, ns
+from bsie.utils.bsfs import URI
+import bsie.reader.stat
+
+
+# exports
+__all__: typing.Sequence[str] = (
+    'Stat',
+    )
+
+
+## code ##
+
+class Stat(extractor.Extractor):
+    """Extract information from the file system."""
+
+    CONTENT_READER = bsie.reader.stat.Stat
+
+    def __init__(self):
+        self.__callmap = {
+            ns.bse.filesize:    self.__filesize,
+            }
+
+    def schema(self) -> str:
+        return '''
+            bse:filesize a bsfs:Predicate ;
+                rdfs:domain bsfs:Entity ;
+                rdfs:range xsd:integer ;
+                rdf:label "File size"^^xsd:string ;
+                schema:description "File size of entity in some filesystem."^^xsd:string ;
+                owl:maxCardinality "INF"^^xsd:number .
+            '''
+
+    def extract(
+            self,
+            subject: node.Node,
+            content: CONTENT_READER.CONTENT_TYPE,
+            predicates: typing.Iterable[URI],
+            ) -> typing.Iterator[typing.Tuple[node.Node, URI, typing.Any]]:
+        for pred in predicates:
+            # find callback
+            clbk = self.__callmap.get(pred)
+            if clbk is None:
+                continue
+            # get value
+            value = clbk(content)
+            if value is None:
+                continue
+            # produce triple
+            yield subject, pred, value
+
+    def __filesize(self, content: CONTENT_READER.CONTENT_TYPE) -> int:
+        """Return the file size."""
+        try:
+            return content.st_size
+        except Exception:
+            # FIXME: some kind of error reporting (e.g. logging)
+            return None
+
+## EOF ##
diff --git a/bsie/utils/__init__.py b/bsie/utils/__init__.py
index 1137187..bd22236 100644
--- a/bsie/utils/__init__.py
+++ b/bsie/utils/__init__.py
@@ -9,12 +9,14 @@ import typing
 
 # inner-module imports
 from . import bsfs
+from . import namespaces as ns
 from . import node
 
 # exports
 __all__: typing.Sequence[str] = (
     'bsfs',
     'node',
+    'ns',
     )
 
 ## EOF ##
diff --git a/bsie/utils/bsfs.py b/bsie/utils/bsfs.py
index 33eb178..1ae657c 100644
--- a/bsie/utils/bsfs.py
+++ b/bsie/utils/bsfs.py
@@ -8,11 +8,12 @@ Author: Matthias Baumgartner, 2022
 import typing
 
 # bsfs imports
-from bsfs.utils import URI
-from bsfs.utils import typename
+from bsfs.namespace import Namespace
+from bsfs.utils import URI, typename
 
 # exports
 __all__: typing.Sequence[str] = (
+    'Namespace',
     'URI',
     'typename',
     )
diff --git a/bsie/utils/namespaces.py b/bsie/utils/namespaces.py
new file mode 100644
index 0000000..67ccc71
--- /dev/null
+++ b/bsie/utils/namespaces.py
@@ -0,0 +1,25 @@
+"""Default namespaces used throughout BSIE.
+
+Part of the bsie module.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import typing
+
+# bsie imports
+from . import bsfs as _bsfs
+
+# constants
+bse = _bsfs.Namespace('http://bsfs.ai/schema/Entity#')
+bsfs = _bsfs.Namespace('http://bsfs.ai/schema/')
+bsm = _bsfs.Namespace('http://bsfs.ai/schema/meta#')
+
+# export
+__all__: typing.Sequence[str] = (
+    'bse',
+    'bsfs',
+    'bsm',
+    )
+
+## EOF ##
diff --git a/test/__init__.py b/test/__init__.py
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/test/__init__.py
diff --git a/test/extractor/__init__.py b/test/extractor/__init__.py
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/test/extractor/__init__.py
diff --git a/test/extractor/generic/__init__.py b/test/extractor/generic/__init__.py
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/test/extractor/generic/__init__.py
diff --git a/test/extractor/generic/test_constant.py b/test/extractor/generic/test_constant.py
new file mode 100644
index 0000000..f3ab0a3
--- /dev/null
+++ b/test/extractor/generic/test_constant.py
@@ -0,0 +1,63 @@
+"""
+
+Part of the bsie test suite.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import unittest
+
+# bsie imports
+from bsie.utils import ns
+from bsie.utils.node import Node
+
+# objects to test
+from bsie.extractor.generic.constant import Constant
+
+
+## code ##
+
+class TestConstant(unittest.TestCase):
+    def test_extract(self):
+        schema = '''
+            bse:author a bsfs:Predicate ;
+                rdfs:domain bsfs:Entity ;
+                rdfs:range xsd:string ;
+                owl:maxCardinality "1"^^xsd:number .
+
+            bse:comment a bsfs:Predicate ;
+                rdfs:domain bsfs:Entity ;
+                rdfs:range xsd:string ;
+                owl:maxCardinality "INF"^^xsd:number .
+
+            '''
+        tuples = [
+            (ns.bse.author, 'Me, myself, and I'),
+            (ns.bse.comment, 'the quick brown fox jumps over the lazy dog.'),
+            ]
+        node = Node(ns.bsfs.Entity, '') # Blank node
+        predicates = (ns.bse.author, ns.bse.comment)
+        ext = Constant(schema, tuples)
+        # baseline
+        self.assertSetEqual(set(ext.extract(node, None, predicates)),
+            {(node, pred, value) for pred, value in tuples})
+        # predicates is respected
+        self.assertSetEqual(set(ext.extract(node, None, (ns.bse.author, ns.bse.foobar))),
+            {(node, ns.bse.author, 'Me, myself, and I')})
+        self.assertSetEqual(set(ext.extract(node, None, (ns.bse.comment, ns.bse.foobar))),
+            {(node, ns.bse.comment, 'the quick brown fox jumps over the lazy dog.')})
+        self.assertSetEqual(set(ext.extract(node, None, (ns.bse.foobar, ns.bse.barfoo))), set())
+
+        # FIXME: should change!
+        # for now: no schema compliance
+        ext = Constant('', tuples)
+        self.assertSetEqual(set(ext.extract(node, None, predicates)),
+            {(node, pred, value) for pred, value in tuples})
+
+
+## main ##
+
+if __name__ == '__main__':
+    unittest.main()
+
+## EOF ##
diff --git a/test/extractor/generic/test_path.py b/test/extractor/generic/test_path.py
new file mode 100644
index 0000000..8623490
--- /dev/null
+++ b/test/extractor/generic/test_path.py
@@ -0,0 +1,45 @@
+"""
+
+Part of the bsie test suite.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import unittest
+
+# bsie imports
+from bsie.utils import ns
+from bsie.utils.node import Node
+
+# objects to test
+from bsie.extractor.generic.path import Path
+
+
+## code ##
+
+class TestPath(unittest.TestCase):
+    def test_extract(self):
+        node = Node(ns.bsfs.Entity, '') # Blank node
+        ext = Path()
+
+        # baseline
+        self.assertSetEqual(set(ext.extract(node, '/tmp/foo/bar', (ns.bse.filename, ))),
+            {(node, ns.bse.filename, 'bar')})
+        # predicates parameter is respected
+        self.assertSetEqual(set(ext.extract(node, '/tmp/foo/bar', (ns.bse.filename, ns.bse.foo))),
+            {(node, ns.bse.filename, 'bar')})
+        self.assertSetEqual(set(ext.extract(node, '/tmp/foo/bar', (ns.bse.foo, ))), set())
+        # path variations
+        self.assertSetEqual(set(ext.extract(node, 'bar', (ns.bse.filename, ))),
+            {(node, ns.bse.filename, 'bar')})
+        self.assertSetEqual(set(ext.extract(node, '', (ns.bse.filename, ))),
+            {(node, ns.bse.filename, '')})
+        self.assertSetEqual(set(ext.extract(node, None, (ns.bse.filename, ))), set())
+
+
+## main ##
+
+if __name__ == '__main__':
+    unittest.main()
+
+## EOF ##
diff --git a/test/extractor/generic/test_stat.py b/test/extractor/generic/test_stat.py
new file mode 100644
index 0000000..f89b053
--- /dev/null
+++ b/test/extractor/generic/test_stat.py
@@ -0,0 +1,43 @@
+"""
+
+Part of the bsie test suite.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import os
+import unittest
+
+# bsie imports
+from bsie.utils import ns
+from bsie.utils.node import Node
+
+# objects to test
+from bsie.extractor.generic.stat import Stat
+
+
+## code ##
+
+class TestConstant(unittest.TestCase):
+    def test_extract(self):
+        node = Node(ns.bsfs.Entity, '') # Blank node
+        content = os.stat(__file__)
+        ext = Stat()
+
+        # baseline
+        self.assertSetEqual(set(ext.extract(node, content, (ns.bse.filesize, ))),
+            {(node, ns.bse.filesize, content.st_size)})
+        # predicates parameter is respected
+        self.assertSetEqual(set(ext.extract(node, content, (ns.bse.filesize, ns.bse.foo))),
+            {(node, ns.bse.filesize, content.st_size)})
+        self.assertSetEqual(set(ext.extract(node, content, (ns.bse.foo, ))), set())
+        # content variations
+        self.assertSetEqual(set(ext.extract(node, None, (ns.bse.filesize, ))), set())
+
+
+## main ##
+
+if __name__ == '__main__':
+    unittest.main()
+
+## EOF ##
diff --git a/test/reader/__init__.py b/test/reader/__init__.py
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/test/reader/__init__.py