Merge branch 'mb/diogenes' into develop

11 files changed, 252 insertions, 75 deletions

diff --git a/bsie/extractor/__init__.py b/bsie/extractor/__init__.py
index 5f385ee..36fa9ba 100644
--- a/bsie/extractor/__init__.py
+++ b/bsie/extractor/__init__.py
@@ -2,9 +2,6 @@
 
 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
 """
 # standard imports
 import typing
diff --git a/bsie/extractor/base.py b/bsie/extractor/base.py
index 89183f9..f92d7cc 100644
--- a/bsie/extractor/base.py
+++ b/bsie/extractor/base.py
@@ -1,8 +1,4 @@
 """The Extractor classes transform content into triples.
-
-Part of the bsie module.
-A copy of the license is provided with the project.
-Author: Matthias Baumgartner, 2022
 """
 # standard imports
 import abc
@@ -28,26 +24,32 @@ SCHEMA_PREAMBLE = '''
     prefix schema: <http://schema.org/>
 
     # common bsfs prefixes
-    prefix bsfs: <http://bsfs.ai/schema/>
-    prefix bse: <http://bsfs.ai/schema/Entity#>
-    prefix bsp: <http://bsfs.ai/schema/Preview#>
+    prefix bsfs: <https://schema.bsfs.io/core/>
+    prefix bsl: <https://schema.bsfs.io/core/Literal/>
+    prefix bsa: <https://schema.bsfs.io/core/Literal/Array/>
+    prefix bsd: <https://schema.bsfs.io/core/distance#>
+
+    prefix bsie: <https://schema.bsfs.io/ie/>
+    prefix bsn: <https://schema.bsfs.io/ie/Node/>
+    prefix bse: <https://schema.bsfs.io/ie/Node/Entity#>
+    prefix bsp: <https://schema.bsfs.io/ie/Node/Preview#>
 
     # default definitions
-    bsfs:Array rdfs:subClassOf bsfs:Literal .
-    bsfs:Number rdfs:subClassOf bsfs:Literal .
-    bsfs:Time rdfs:subClassOf bsfs:Literal .
-    bsfs:Feature rdfs:subClassOf bsfs:Array ;
+    bsl:Array rdfs:subClassOf bsfs:Literal .
+    bsl:Number rdfs:subClassOf bsfs:Literal .
+    bsl:Time rdfs:subClassOf bsfs:Literal .
+    bsa:Feature rdfs:subClassOf bsl:Array ;
         bsfs:dimension "1"^^xsd:integer ;
-        bsfs:dtype bsfs:f16 ;
-        bsfs:distance bsfs:euclidean .
+        bsfs:dtype <https://schema.bsfs.io/core/dtype#f16> ;
+        bsfs:distance bsd:euclidean .
 
     # essential nodes
-    bsfs:Entity rdfs:subClassOf bsfs:Node .
-    bsfs:File rdfs:subClassOf bsfs:Entity .
+    bsn:Entity rdfs:subClassOf bsfs:Node .
 
     # common definitions
     xsd:string rdfs:subClassOf bsfs:Literal .
-    xsd:integer rdfs:subClassOf bsfs:Number .
+    xsd:integer rdfs:subClassOf bsl:Number .
+    xsd:float rdfs:subClassOf bsl:Number .
 
     '''
 
@@ -93,7 +95,7 @@ class Extractor(abc.ABC):
     @property
     def principals(self) -> typing.Iterator[bsfs.schema.Predicate]:
         """Return the principal predicates, i.e., relations from/to the extraction subject."""
-        ent = self.schema.node(ns.bsfs.Entity)
+        ent = self.schema.node(ns.bsn.Entity)
         return (
             pred
             for pred
diff --git a/bsie/extractor/builder.py b/bsie/extractor/builder.py
index 0fd3685..d691b0e 100644
--- a/bsie/extractor/builder.py
+++ b/bsie/extractor/builder.py
@@ -1,9 +1,4 @@
-"""
 
-Part of the bsie module.
-A copy of the license is provided with the project.
-Author: Matthias Baumgartner, 2022
-"""
 # standard imports
 import typing
 
diff --git a/bsie/extractor/generic/__init__.py b/bsie/extractor/generic/__init__.py
index 4783949..46a4bd6 100644
--- a/bsie/extractor/generic/__init__.py
+++ b/bsie/extractor/generic/__init__.py
@@ -3,9 +3,6 @@ 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
 """
 # standard imports
 import typing
diff --git a/bsie/extractor/generic/constant.py b/bsie/extractor/generic/constant.py
index 938e20c..7acbe95 100644
--- a/bsie/extractor/generic/constant.py
+++ b/bsie/extractor/generic/constant.py
@@ -1,8 +1,4 @@
 """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
 """
 # standard imports
 import typing
diff --git a/bsie/extractor/generic/path.py b/bsie/extractor/generic/path.py
index c984515..00c1121 100644
--- a/bsie/extractor/generic/path.py
+++ b/bsie/extractor/generic/path.py
@@ -1,9 +1,4 @@
-"""
 
-Part of the bsie module.
-A copy of the license is provided with the project.
-Author: Matthias Baumgartner, 2022
-"""
 # standard imports
 import os
 import typing
@@ -31,11 +26,11 @@ class Path(base.Extractor):
     def __init__(self):
         super().__init__(bsfs.schema.from_string(base.SCHEMA_PREAMBLE + '''
             bse:filename rdfs:subClassOf bsfs:Predicate ;
-                rdfs:domain bsfs:File ;
+                rdfs:domain bsn:Entity ;
                 rdfs:range xsd:string ;
                 rdfs:label "File name"^^xsd:string ;
                 schema:description "Filename of entity in some filesystem."^^xsd:string ;
-                bsfs:unique "false"^^xsd:boolean .
+                bsfs:unique "true"^^xsd:boolean .
             '''))
         self._callmap = {
             self.schema.predicate(ns.bse.filename): self.__filename,
diff --git a/bsie/extractor/generic/stat.py b/bsie/extractor/generic/stat.py
index 9394456..92b51f3 100644
--- a/bsie/extractor/generic/stat.py
+++ b/bsie/extractor/generic/stat.py
@@ -1,8 +1,4 @@
 """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
 """
 # standard imports
 import os
@@ -33,11 +29,11 @@ class Stat(base.Extractor):
     def __init__(self):
         super().__init__(bsfs.schema.from_string(base.SCHEMA_PREAMBLE + '''
             bse:filesize rdfs:subClassOf bsfs:Predicate ;
-                rdfs:domain bsfs:File ;
+                rdfs:domain bsn:Entity ;
                 rdfs:range xsd:integer ;
                 rdfs:label "File size"^^xsd:string ;
                 schema:description "File size of entity in some filesystem."^^xsd:string ;
-                bsfs:unique "false"^^xsd:boolean .
+                bsfs:unique "true"^^xsd:boolean .
             '''))
         self._callmap = {
             self.schema.predicate(ns.bse.filesize): self.__filesize,
diff --git a/bsie/extractor/image/__init__.py b/bsie/extractor/image/__init__.py
index 75b118d..f82424a 100644
--- a/bsie/extractor/image/__init__.py
+++ b/bsie/extractor/image/__init__.py
@@ -1,9 +1,4 @@
-"""
 
-Part of the bsie module.
-A copy of the license is provided with the project.
-Author: Matthias Baumgartner, 2022
-"""
 # standard imports
 import typing
 
diff --git a/bsie/extractor/image/colors_spatial.py b/bsie/extractor/image/colors_spatial.py
index 15fd281..e6661a9 100644
--- a/bsie/extractor/image/colors_spatial.py
+++ b/bsie/extractor/image/colors_spatial.py
@@ -1,8 +1,4 @@
 """Spatial color features.
-
-Part of the bsie module.
-A copy of the license is provided with the project.
-Author: Matthias Baumgartner, 2022
 """
 # standard imports
 import typing
@@ -18,8 +14,7 @@ from bsie.utils import bsfs, node, ns
 from .. import base
 
 # constants
-FEATURE_NAME = ns.bsf + 'ColorsSpatial'
-PREDICATE_NAME = ns.bse + 'colors_spatial'
+FEATURE_NAME = ns.bsf.ColorsSpatial()
 
 # exports
 __all__: typing.Sequence[str] = (
@@ -62,16 +57,17 @@ class ColorsSpatial(base.Extractor):
             'exp': exp,
             })
         # determine symbol names
-        instance_name = FEATURE_NAME[uuid]
-        predicate_name = PREDICATE_NAME[uuid]
+        instance_name = getattr(FEATURE_NAME, uuid)
+        predicate_name = getattr(ns.bse, 'colors_spatial_' + uuid)
         # get vector dimension
         dimension = self.dimension(width, height, exp)
         # initialize parent with the schema
         super().__init__(bsfs.schema.from_string(base.SCHEMA_PREAMBLE + f'''
-            <{FEATURE_NAME}> rdfs:subClassOf bsfs:Feature ;
+            <{FEATURE_NAME}> rdfs:subClassOf bsa:Feature ;
                 # annotations
                 rdfs:label "Spatially dominant colors"^^xsd:string ;
                 schema:description "Domiant colors of subregions in an image."^^xsd:string ;
+                bsfs:distance <https://schema.bsfs.io/core/distance#euclidean> ;
                 bsfs:dtype xsd:integer .
 
             <{instance_name}> rdfs:subClassOf <{FEATURE_NAME}> ;
@@ -82,7 +78,7 @@ class ColorsSpatial(base.Extractor):
                 <{FEATURE_NAME}/args#exp> "{exp}"^^xsd:float .
 
             <{predicate_name}> rdfs:subClassOf bsfs:Predicate ;
-                rdfs:domain bsfs:File ;
+                rdfs:domain bsn:Entity ;
                 rdfs:range <{instance_name}> ;
                 bsfs:unique "true"^^xsd:boolean .
 
diff --git a/bsie/extractor/image/photometrics.py b/bsie/extractor/image/photometrics.py
new file mode 100644
index 0000000..42eb3c8
--- /dev/null
+++ b/bsie/extractor/image/photometrics.py
@@ -0,0 +1,211 @@
+
+# standard imports
+from fractions import Fraction
+import typing
+
+# bsie imports
+from bsie.utils import bsfs, node, ns
+
+# inner-module imports
+from .. import base
+
+# exports
+__all__: typing.Sequence[str] = (
+    'Exif',
+    )
+
+
+## code ##
+
+def _gps_to_dec(coords: typing.Tuple[float, float, float]) -> float:
+    """Convert GPS coordinates from exif to float."""
+    # unpack args
+    deg, min, sec = coords # pylint: disable=redefined-builtin # min
+    # convert to float
+    deg = float(Fraction(deg))
+    min = float(Fraction(min))
+    sec = float(Fraction(sec))
+
+    if float(sec) > 0:
+        # format is deg+min+sec
+        return (float(deg) * 3600 + float(min) * 60 + float(sec)) / 3600
+    # format is deg+min
+    return float(deg) + float(min) / 60
+
+
+class Exif(base.Extractor):
+    """Extract information from EXIF/IPTC tags of an image file."""
+
+    CONTENT_READER = 'bsie.reader.exif.Exif'
+
+    def __init__(self):
+        super().__init__(bsfs.schema.from_string(base.SCHEMA_PREAMBLE + '''
+            #bse:t_capture rdfs:subClassOf bsfs:Predicate ;
+            #    rdfs:domain bsn:Entity ;
+            #    rdfs:range xsd:float ;
+            #    bsfs:unique "true"^^xsd:boolean .
+            bse:exposure rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsn:Entity ;
+                rdfs:range xsd:float ;
+                bsfs:unique "true"^^xsd:boolean .
+            bse:aperture rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsn:Entity ;
+                rdfs:range xsd:float ;
+                bsfs:unique "true"^^xsd:boolean .
+            bse:iso rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsn:Entity ;
+                rdfs:range xsd:integer ;
+                bsfs:unique "true"^^xsd:boolean .
+            bse:focal_length rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsn:Entity ;
+                rdfs:range xsd:float ;
+                bsfs:unique "true"^^xsd:boolean .
+            bse:width rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsn:Entity ;
+                rdfs:range xsd:integer ;
+                bsfs:unique "true"^^xsd:boolean .
+            bse:height rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsn:Entity ;
+                rdfs:range xsd:integer ;
+                bsfs:unique "true"^^xsd:boolean .
+            bse:orientation rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsn:Entity ;
+                rdfs:range xsd:integer ;
+                bsfs:unique "true"^^xsd:boolean .
+            bse:orientation_label rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsn:Entity ;
+                rdfs:range xsd:string ;
+                bsfs:unique "true"^^xsd:boolean .
+            bse:altitude rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsn:Entity ;
+                rdfs:range xsd:float ;
+                bsfs:unique "true"^^xsd:boolean .
+            bse:latitude rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsn:Entity ;
+                rdfs:range xsd:float ;
+                bsfs:unique "true"^^xsd:boolean .
+            bse:longitude rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsn:Entity ;
+                rdfs:range xsd:float ;
+                bsfs:unique "true"^^xsd:boolean .
+            '''))
+        # initialize mapping from predicate to callback
+        self._callmap = {
+            #self.schema.predicate(ns.bse.t_capture):         self._date,
+            self.schema.predicate(ns.bse.exposure):          self._exposure,
+            self.schema.predicate(ns.bse.aperture):          self._aperture,
+            self.schema.predicate(ns.bse.iso):               self._iso,
+            self.schema.predicate(ns.bse.focal_length):      self._focal_length,
+            self.schema.predicate(ns.bse.width):             self._width,
+            self.schema.predicate(ns.bse.height):            self._height,
+            self.schema.predicate(ns.bse.orientation):       self._orientation,
+            self.schema.predicate(ns.bse.orientation_label): self._orientation_label,
+            self.schema.predicate(ns.bse.altitude):          self._altitude,
+            self.schema.predicate(ns.bse.latitude):          self._latitude,
+            self.schema.predicate(ns.bse.longitude):         self._longitude,
+            }
+
+    def extract(
+            self,
+            subject: node.Node,
+            content: dict,
+            principals: typing.Iterable[bsfs.schema.Predicate],
+            ) -> typing.Iterator[typing.Tuple[node.Node, bsfs.schema.Predicate, typing.Any]]:
+        for pred in principals:
+            # 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 _date(self, content: dict): # FIXME: Return type annotation
+    #    date_keys = (
+    #        'Exif.Photo.DateTimeOriginal',
+    #        'Exif.Photo.DateTimeDigitized',
+    #        'Exif.Image.DateTime',
+    #        )
+    #    for key in date_keys:
+    #        if key in content:
+    #            dt = content[key].value
+    #            if dt.tzinfo is None:
+    #                dt = dt.replace(tzinfo=ttime.NoTimeZone)
+    #            return dt
+    #    return None
+
+
+    ## photometrics
+
+    def _exposure(self, content: dict) -> typing.Optional[float]:
+        if 'Exif.Photo.ExposureTime' in content:
+            return 1.0 / float(Fraction(content['Exif.Photo.ExposureTime']))
+        return None
+
+    def _aperture(self, content: dict) -> typing.Optional[float]:
+        if 'Exif.Photo.FNumber' in content:
+            return float(Fraction(content['Exif.Photo.FNumber']))
+        return None
+
+    def _iso(self, content: dict) -> typing.Optional[int]:
+        if 'Exif.Photo.ISOSpeedRatings' in content:
+            return int(content['Exif.Photo.ISOSpeedRatings'])
+        return None
+
+    def _focal_length(self, content: dict) -> typing.Optional[float]:
+        if 'Exif.Photo.FocalLength' in content:
+            return float(Fraction(content['Exif.Photo.FocalLength']))
+        return None
+
+
+    ## image dimensions
+
+    def _width(self, content: dict) -> typing.Optional[int]:
+        # FIXME: consider orientation!
+        if 'Exif.Photo.PixelXDimension' in content:
+            return int(content['Exif.Photo.PixelXDimension'])
+        return None
+
+    def _height(self, content: dict) -> typing.Optional[int]:
+        # FIXME: consider orientation!
+        if 'Exif.Photo.PixelYDimension' in content:
+            return int(content['Exif.Photo.PixelYDimension'])
+        return None
+
+    def _orientation(self, content: dict) -> typing.Optional[int]:
+        if 'Exif.Image.Orientation' in content:
+            return int(content['Exif.Image.Orientation'])
+        return None
+
+    def _orientation_label(self, content: dict) -> typing.Optional[str]:
+        width = self._width(content)
+        height = self._height(content)
+        ori = self._orientation(content)
+        if width is not None and height is not None and ori is not None:
+            if ori <= 4:
+                return 'landscape' if width >= height else 'portrait'
+            return 'portrait' if width >= height else 'landscape'
+        return None
+
+
+    ## location
+
+    def _altitude(self, content: dict) -> typing.Optional[float]:
+        if 'Exif.GPSInfo.GPSAltitude' in content:
+            return float(Fraction(content['Exif.GPSInfo.GPSAltitude']))
+        return None
+
+    def _latitude(self, content: dict) -> typing.Optional[float]:
+        if 'Exif.GPSInfo.GPSLatitude' in content:
+            return _gps_to_dec(content['Exif.GPSInfo.GPSLatitude'].split())
+        return None
+
+    def _longitude(self, content: dict) -> typing.Optional[float]:
+        if 'Exif.GPSInfo.GPSLongitude' in content:
+            return _gps_to_dec(content['Exif.GPSInfo.GPSLongitude'].split())
+        return None
+
+## EOF ##
diff --git a/bsie/extractor/preview.py b/bsie/extractor/preview.py
index 1531d62..145a01a 100644
--- a/bsie/extractor/preview.py
+++ b/bsie/extractor/preview.py
@@ -1,9 +1,4 @@
-"""
 
-Part of the bsie module.
-A copy of the license is provided with the project.
-Author: Matthias Baumgartner, 2022
-"""
 # imports
 import io
 import typing
@@ -33,28 +28,30 @@ class Preview(base.Extractor):
     def __init__(self, max_sides: typing.Iterable[int]):
         super().__init__(bsfs.schema.from_string(base.SCHEMA_PREAMBLE + '''
 
-            bsfs:Preview rdfs:subClassOf bsfs:Node .
-            bsfs:BinaryBlob rdfs:subClassOf bsfs:Literal .
-            bsfs:JPEG rdfs:subClassOf bsfs:BinaryBlob .
+
+
+            bsn:Preview rdfs:subClassOf bsfs:Node .
+            bsl:BinaryBlob rdfs:subClassOf bsfs:Literal .
+            <https://schema.bsfs.io/ie/Literal/BinaryBlob/JPEG> rdfs:subClassOf bsl:BinaryBlob .
 
             bse:preview rdfs:subClassOf bsfs:Predicate ;
-                rdfs:domain bsfs:File ;
-                rdfs:range bsfs:Preview ;
+                rdfs:domain bsn:Entity ;
+                rdfs:range bsn:Preview ;
                 bsfs:unique "false"^^xsd:boolean .
 
             bsp:width rdfs:subClassOf bsfs:Predicate ;
-                rdfs:domain bsfs:Preview ;
+                rdfs:domain bsn:Preview ;
                 rdfs:range xsd:integer ;
                 bsfs:unique "true"^^xsd:boolean .
 
             bsp:height rdfs:subClassOf bsfs:Predicate ;
-                rdfs:domain bsfs:Preview ;
+                rdfs:domain bsn:Preview ;
                 rdfs:range xsd:integer ;
                 bsfs:unique "true"^^xsd:boolean .
 
             bsp:asset rdfs:subClassOf bsfs:Predicate ;
-                rdfs:domain bsfs:Preview ;
-                rdfs:range bsfs:JPEG ;
+                rdfs:domain bsn:Preview ;
+                rdfs:range <https://schema.bsfs.io/ie/Literal/BinaryBlob/JPEG> ;
                 bsfs:unique "true"^^xsd:boolean .
 
             '''))
@@ -85,7 +82,7 @@ class Preview(base.Extractor):
             buffer = io.BytesIO()
             img.save(buffer, format='jpeg')
             # create a preview node
-            preview = node.Node(ns.bsfs.Preview,
+            preview = node.Node(ns.bsn.Preview,
                 ucid=bsfs.uuid.UCID.from_bytes(buffer.getvalue()),
                 size=max_side,
                 source=subject,