Merge branch 'mb/iptc' into develop

10 files changed, 157 insertions, 29 deletions

diff --git a/bsie/apps/_loader.py b/bsie/apps/_loader.py
index 6411f10..d9ea9bb 100644
--- a/bsie/apps/_loader.py
+++ b/bsie/apps/_loader.py
@@ -1,5 +1,6 @@
 
 # standard imports
+import os
 import typing
 
 # external imports
@@ -12,8 +13,7 @@ from bsie.lib.pipeline import Pipeline
 from bsie.reader import ReaderBuilder
 
 # constants
-DEFAULT_CONFIG_FILE = 'default_config.yaml'
-
+DEFAULT_CONFIG_FILE = os.path.join(os.path.dirname(__file__), 'default_config.yaml')
 # exports
 __all__: typing.Sequence[str] = (
     'DEFAULT_CONFIG_FILE',
@@ -23,7 +23,7 @@ __all__: typing.Sequence[str] = (
 
 ## code ##
 
-def load_pipeline(path: str) -> Pipeline:
+def load_pipeline(path: str = DEFAULT_CONFIG_FILE) -> Pipeline:
     """Load a pipeline according to a config at *path*."""
     # load config file
     with open(path, 'rt', encoding='utf-8') as ifile:
diff --git a/bsie/apps/index.py b/bsie/apps/index.py
index d64e8c2..7dda6f4 100644
--- a/bsie/apps/index.py
+++ b/bsie/apps/index.py
@@ -6,7 +6,7 @@ import typing
 
 # bsie imports
 from bsie.lib import BSIE, DefaultNamingPolicy
-from bsie.utils import bsfs, errors, node as node_
+from bsie.utils import bsfs, errors, node as node_, list_files
 
 # inner-module imports
 from . import _loader
@@ -23,7 +23,7 @@ def main(argv):
     """Index files or directories into BSFS."""
     parser = argparse.ArgumentParser(description=main.__doc__, prog='index')
     parser.add_argument('--config', type=str,
-        default=os.path.join(os.path.dirname(__file__), _loader.DEFAULT_CONFIG_FILE),
+        default=_loader.DEFAULT_CONFIG_FILE,
         help='Path to the config file.')
     parser.add_argument('--host', type=bsfs.URI, default=bsfs.URI('http://example.com'),
         help='')
@@ -59,22 +59,9 @@ def main(argv):
         # FIXME: simplify code (below but maybe also above)
         # FIXME: How to handle dependencies between data?
         #        E.g. do I still want to link to a tag despite not being permitted to set its label?
-
-        # index input paths
-        for path in args.input_file:
-            if not os.path.exists(path):
-                pass # FIXME: notify the user
-            elif os.path.isdir(path) and args.recursive:
-                for dirpath, _, filenames in os.walk(path, topdown=True, followlinks=args.follow):
-                    for filename in filenames:
-                        for node, pred, value in bsie.from_file(os.path.join(dirpath, filename)):
-                            handle(node, pred, value)
-            elif os.path.isfile(path):
-                for node, pred, value in bsie.from_file(path):
-                    handle(node, pred, value)
-            else:
-                raise errors.UnreachableError()
-
+        for path in list_files(args.input_file, args.recursive, args.follow):
+            for node, pred, value in bsie.from_file(path):
+                handle(node, pred, value)
 
     if args.print:
         walk(print)
diff --git a/bsie/apps/info.py b/bsie/apps/info.py
index e27b70b..b6494da 100644
--- a/bsie/apps/info.py
+++ b/bsie/apps/info.py
@@ -23,7 +23,7 @@ def main(argv):
     """Show information from BSIE."""
     parser = argparse.ArgumentParser(description=main.__doc__, prog='info')
     parser.add_argument('--config', type=str,
-        default=os.path.join(os.path.dirname(__file__), _loader.DEFAULT_CONFIG_FILE),
+        default=_loader.DEFAULT_CONFIG_FILE,
         help='Path to the config file.')
     parser.add_argument('what', choices=('predicates', 'schema'),
         help='Select what information to show.')
diff --git a/bsie/extractor/builder.py b/bsie/extractor/builder.py
index d691b0e..8353a93 100644
--- a/bsie/extractor/builder.py
+++ b/bsie/extractor/builder.py
@@ -67,6 +67,7 @@ class ExtractorBuilder():
             return cls(**kwargs)
 
         except Exception as err:
-            raise errors.BuilderError(f'failed to build extractor {name} due to {bsfs.typename(err)}: {err}') from err
+            raise errors.BuilderError(
+                f'failed to build extractor {name} due to {bsfs.typename(err)}: {err}') from err
 
 ## EOF ##
diff --git a/bsie/extractor/image/iptc.py b/bsie/extractor/image/iptc.py
new file mode 100644
index 0000000..195eff7
--- /dev/null
+++ b/bsie/extractor/image/iptc.py
@@ -0,0 +1,70 @@
+
+# standard imports
+import typing
+
+# bsie imports
+from bsie.utils import bsfs, node, ns
+
+# inner-module imports
+from .. import base
+
+# exports
+__all__: typing.Sequence[str] = (
+    'Iptc',
+    )
+
+
+## code ##
+
+class Iptc(base.Extractor):
+    """Turn IPTC keywords into tags."""
+
+    CONTENT_READER = 'bsie.reader.exif.Iptc'
+
+    def __init__(self):
+        super().__init__(bsfs.schema.from_string(base.SCHEMA_PREAMBLE + '''
+            bsn:Tag rdfs:subClassOf bsfs:Node .
+
+            bse:tag rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsn:Entity ;
+                rdfs:range bsn:Tag .
+
+            <https://schema.bsfs.io/ie/Node/Tag#label> rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsn:Tag ;
+                rdfs:range xsd:string ;
+                bsfs:unique "true"^^xsd:boolean .
+
+            '''))
+        self._callmap = {
+            self.schema.predicate(ns.bse.tag): self._keywords,
+            }
+
+    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
+            # produce triples
+            yield from clbk(subject, content)
+
+    def _keywords(
+            self,
+            subject: node.Node,
+            content: dict,
+            ) -> typing.Iterator[typing.Tuple[node.Node, bsfs.schema.Predicate, typing.Any]]:
+        if 'Iptc.Application2.Keywords' not in content:
+            return
+        for keyword in content['Iptc.Application2.Keywords']:
+            tag = node.Node(ns.bsn.Tag, label=keyword)
+            yield subject, self.schema.predicate(ns.bse.tag), tag
+            yield tag, self.schema.predicate(ns.bst.label), keyword
+
+
+
+## EOF ##
diff --git a/bsie/lib/naming_policy.py b/bsie/lib/naming_policy.py
index 9b9a45d..ffef7d9 100644
--- a/bsie/lib/naming_policy.py
+++ b/bsie/lib/naming_policy.py
@@ -4,6 +4,9 @@ import abc
 import os
 import typing
 
+# external imports
+import urllib.parse
+
 # bsie imports
 from bsie.utils import bsfs, errors, ns
 from bsie.utils.node import Node
@@ -80,14 +83,16 @@ class DefaultNamingPolicy(NamingPolicy):
     def handle_node(self, node: Node) -> Node:
         if node.uri is not None:
             return node
-        if node.node_type == ns.bsn.Entity :
-            return self.name_file(node)
+        if node.node_type == ns.bsn.Entity:
+            return self.name_entity(node)
         if node.node_type == ns.bsn.Preview:
             return self.name_preview(node)
-        raise errors.ProgrammingError('no naming policy available for {node.node_type}')
+        if node.node_type == ns.bsn.Tag:
+            return self.name_tag(node)
+        raise errors.ProgrammingError(f'no naming policy available for {node.node_type}')
 
-    def name_file(self, node: Node) -> Node:
-        """Set a bsfs:File node's uri fragment to its ucid."""
+    def name_entity(self, node: Node) -> Node:
+        """Set a bsn:Entity node's uri fragment to its ucid."""
         if 'ucid' in node.hints: # content id
             fragment = node.hints['ucid']
         else: # random name
@@ -96,7 +101,7 @@ class DefaultNamingPolicy(NamingPolicy):
         return node
 
     def name_preview(self, node: Node) -> Node:
-        """Set a bsfs:Preview node's uri fragment to its ucid.
+        """Set a bsn:Preview node's uri fragment to its ucid.
         Uses its source fragment as fallback. Appends the size if provided.
         """
         fragment = None
@@ -112,4 +117,14 @@ class DefaultNamingPolicy(NamingPolicy):
         node.uri = getattr(self._prefix.preview(), fragment)
         return node
 
+    def name_tag(self, node: Node) -> Node:
+        # NOTE: Must ensure to produce the same name for that tags with the same label.
+        if 'label' in node.hints: # tag label
+            fragment = urllib.parse.quote(node.hints['label'])
+        else: # random name
+            fragment = self._uuid()
+        # FIXME: match to existing tags in bsfs storage!
+        node.uri = getattr(self._prefix.tag(), fragment)
+        return node
+
 ## EOF ##
diff --git a/bsie/reader/exif.py b/bsie/reader/exif.py
index 2d0428b..7ec7574 100644
--- a/bsie/reader/exif.py
+++ b/bsie/reader/exif.py
@@ -17,6 +17,7 @@ MATCH_RULE = 'mime=image/jpeg'
 # exports
 __all__: typing.Sequence[str] = (
     'Exif',
+    'Iptc',
     )
 
 
@@ -41,4 +42,24 @@ class Exif(base.Reader):
         except (TypeError, OSError, RuntimeError) as err:
             raise errors.ReaderError(path) from err
 
+
+class Iptc(base.Reader):
+    """Use pyexiv2 to read iptc metadata from image files."""
+
+    def __init__(self):
+        self._match = filematcher.parse(MATCH_RULE)
+
+    def __call__(self, path: str) -> dict:
+        # perform quick checks first
+        if not self._match(path):
+            raise errors.UnsupportedFileFormatError(path)
+
+        try:
+            # open the file
+            img = pyexiv2.Image(path)
+            # read metadata
+            return img.read_iptc()
+        except (TypeError, OSError, RuntimeError) as err:
+            raise errors.ReaderError(path) from err
+
 ## EOF ##
diff --git a/bsie/utils/__init__.py b/bsie/utils/__init__.py
index 18c8db7..4f08604 100644
--- a/bsie/utils/__init__.py
+++ b/bsie/utils/__init__.py
@@ -8,6 +8,7 @@ from . import bsfs
 from . import filematcher
 from . import namespaces as ns
 from . import node
+from .filewalker import list_files
 from .loading import safe_load, unpack_qualified_name
 
 # exports
diff --git a/bsie/utils/filewalker.py b/bsie/utils/filewalker.py
new file mode 100644
index 0000000..3c36926
--- /dev/null
+++ b/bsie/utils/filewalker.py
@@ -0,0 +1,31 @@
+
+# standard imports
+import os
+import typing
+
+# exports
+__all__: typing.Sequence[str] = (
+    'list_files',
+    )
+
+
+## code ##
+
+def list_files(
+        roots: typing.Iterable[str],
+        recursive: bool = True,
+        follow_symlinks: bool = True,
+        ) -> typing.Iterator[str]:
+    """Iterate over all files in *roots*, recursively by default."""
+    # index input paths
+    for path in roots:
+        if not os.path.exists(path):
+            continue
+        elif os.path.isdir(path) and recursive:
+            for dirpath, _, filenames in os.walk(path, topdown=True, followlinks=follow_symlinks):
+                for filename in filenames:
+                    yield os.path.join(dirpath, filename)
+        elif os.path.isfile(path):
+            yield path
+
+## EOF ##
diff --git a/bsie/utils/namespaces.py b/bsie/utils/namespaces.py
index 4a66048..9357253 100644
--- a/bsie/utils/namespaces.py
+++ b/bsie/utils/namespaces.py
@@ -20,6 +20,7 @@ bsf = bsie.Literal.Array.Feature
 bsl = bsfs.Literal
 bsn = bsie.Node
 bsp = bsie.Node.Preview()
+bst = bsie.Node.Tag()
 
 # export
 __all__: typing.Sequence[str] = (
@@ -32,6 +33,7 @@ __all__: typing.Sequence[str] = (
     'bsl',
     'bsn',
     'bsp',
+    'bst',
     'xsd',
     )