Merge branch 'develop'HEADv0.23.03releasemain

5 files changed, 144 insertions, 82 deletions

diff --git a/bsie/apps/__init__.py b/bsie/apps/__init__.py
index a548c3c..2fe4795 100644
--- a/bsie/apps/__init__.py
+++ b/bsie/apps/__init__.py
@@ -1,12 +1,13 @@
+#!/usr/bin/env python3
+"""BSIE tools.
 """
-
-Part of the bsie module.
-A copy of the license is provided with the project.
-Author: Matthias Baumgartner, 2022
-"""
-# imports
+# standard imports
+import argparse
 import typing
 
+# bsie imports
+import bsie
+
 # inner-module imports
 from .index import main as index
 from .info import main as info
@@ -15,6 +16,39 @@ from .info import main as info
 __all__: typing.Sequence[str] = (
     'index',
     'info',
+    'main',
     )
 
+# config
+apps = {
+    'index'     : index,
+    'info'      : info,
+    }
+
+
+## code ##
+
+def main(argv=None):
+    """Black Star File System maintenance tools."""
+    parser = argparse.ArgumentParser(description=main.__doc__, prog='bsie')
+    # version
+    parser.add_argument('--version', action='version',
+        version='%(prog)s version {}.{}.{}'.format(*bsie.version_info)) # pylint: disable=C0209
+    # application selection
+    parser.add_argument('app', choices=apps.keys(),
+        help='Select the application to run.')
+    # dangling args
+    parser.add_argument('rest', nargs=argparse.REMAINDER)
+    # parse
+    args = parser.parse_args(argv)
+    # run application
+    apps[args.app](args.rest)
+
+
+## main ##
+
+if __name__ == '__main__':
+    import sys
+    main(sys.argv[1:])
+
 ## EOF ##
diff --git a/bsie/apps/_loader.py b/bsie/apps/_loader.py
new file mode 100644
index 0000000..6411f10
--- /dev/null
+++ b/bsie/apps/_loader.py
@@ -0,0 +1,47 @@
+
+# standard imports
+import typing
+
+# external imports
+import yaml
+
+# bsie imports
+from bsie.extractor import ExtractorBuilder
+from bsie.lib import PipelineBuilder
+from bsie.lib.pipeline import Pipeline
+from bsie.reader import ReaderBuilder
+
+# constants
+DEFAULT_CONFIG_FILE = 'default_config.yaml'
+
+# exports
+__all__: typing.Sequence[str] = (
+    'DEFAULT_CONFIG_FILE',
+    'load_pipeline',
+    )
+
+
+## code ##
+
+def load_pipeline(path: str) -> Pipeline:
+    """Load a pipeline according to a config at *path*."""
+    # load config file
+    with open(path, 'rt', encoding='utf-8') as ifile:
+        cfg = yaml.safe_load(ifile)
+
+    # reader builder
+    rbuild = ReaderBuilder(cfg['ReaderBuilder'])
+    # extractor builder
+    ebuild = ExtractorBuilder(cfg['ExtractorBuilder'])
+    # pipeline builder
+    pbuild = PipelineBuilder(
+        rbuild,
+        ebuild,
+        )
+    # build pipeline
+    pipeline = pbuild.build()
+
+    # return pipeline
+    return pipeline
+
+## EOF ##
diff --git a/bsie/apps/default_config.yaml b/bsie/apps/default_config.yaml
new file mode 100644
index 0000000..a59b0f3
--- /dev/null
+++ b/bsie/apps/default_config.yaml
@@ -0,0 +1,19 @@
+
+ReaderBuilder: {}
+
+ExtractorBuilder:
+
+  - bsie.extractor.preview.Preview:
+      max_sides: [50, 100, 200,400]
+
+  - bsie.extractor.generic.path.Path: {}
+
+  - bsie.extractor.generic.stat.Stat: {}
+
+  - bsie.extractor.image.colors_spatial.ColorsSpatial:
+      width: 32
+      height: 32
+      exp: 4
+
+  - bsie.extractor.image.photometrics.Exif: {}
+
diff --git a/bsie/apps/index.py b/bsie/apps/index.py
index 1dbfdd8..d64e8c2 100644
--- a/bsie/apps/index.py
+++ b/bsie/apps/index.py
@@ -1,19 +1,15 @@
-"""
 
-Part of the bsie module.
-A copy of the license is provided with the project.
-Author: Matthias Baumgartner, 2022
-"""
-# imports
+# standard imports
 import argparse
 import os
 import typing
 
 # bsie imports
-from bsie.base import errors
-from bsie.lib import BSIE
-from bsie.tools import builder
-from bsie.utils import bsfs
+from bsie.lib import BSIE, DefaultNamingPolicy
+from bsie.utils import bsfs, errors, node as node_
+
+# inner-module imports
+from . import _loader
 
 # exports
 __all__: typing.Sequence[str] = (
@@ -26,7 +22,12 @@ __all__: typing.Sequence[str] = (
 def main(argv):
     """Index files or directories into BSFS."""
     parser = argparse.ArgumentParser(description=main.__doc__, prog='index')
-    parser.add_argument('--user', type=bsfs.URI, default=bsfs.URI('http://example.com/me'),
+    parser.add_argument('--config', type=str,
+        default=os.path.join(os.path.dirname(__file__), _loader.DEFAULT_CONFIG_FILE),
+        help='Path to the config file.')
+    parser.add_argument('--host', type=bsfs.URI, default=bsfs.URI('http://example.com'),
+        help='')
+    parser.add_argument('--user', type=str, default='me',
         help='')
     parser.add_argument('--collect', action='append', default=[],
         help='')
@@ -42,35 +43,15 @@ def main(argv):
         help='')
     args = parser.parse_args(argv)
 
-    # FIXME: Read reader/extractor configs from a config file
-    # reader builder
-    rbuild = builder.ReaderBuilder({})
-    # extractor builder
-    ebuild = builder.ExtractorBuilder([
-        {'bsie.extractor.generic.path.Path': {}},
-        {'bsie.extractor.generic.stat.Stat': {}},
-        {'bsie.extractor.generic.constant.Constant': dict(
-            tuples=[('http://bsfs.ai/schema/Entity#author', 'Me, myself, and I')],
-            schema='''
-                bse:author rdfs:subClassOf bsfs:Predicate ;
-                    rdfs:domain bsfs:Entity ;
-                    rdfs:range xsd:string ;
-                    bsfs:unique "true"^^xsd:boolean .
-                ''',
-            )},
-        ])
-    # pipeline builder
-    pbuild = builder.PipelineBuilder(
-        bsfs.Namespace(args.user + ('/' if not args.user.endswith('/') else '')),
-        rbuild,
-        ebuild,
-        )
-
     # build pipeline
-    pipeline = pbuild.build()
+    pipeline = _loader.load_pipeline(args.config)
+    # build the naming policy
+    naming_policy = DefaultNamingPolicy(
+        host=args.host,
+        user=args.user,
+        )
     # build BSIE frontend
-    bsie = BSIE(pipeline, args.collect, args.discard)
-
+    bsie = BSIE(pipeline, naming_policy, args.collect, args.discard)
 
     def walk(handle):
         """Walk through given input files."""
@@ -78,11 +59,12 @@ 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?
-        # FIXME: node renaming?
 
         # index input paths
         for path in args.input_file:
-            if os.path.isdir(path) and args.recursive:
+            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)):
@@ -105,13 +87,14 @@ def main(argv):
     store.migrate(bsie.schema)
     # process files
     def handle(node, pred, value):
+        if isinstance(value, node_.Node):
+            value = store.node(value.node_type, value.uri)
         store.node(node.node_type, node.uri).set(pred.uri, value)
     walk(handle)
     # return store
     return store
 
 
-
 ## main ##
 
 if __name__ == '__main__':
diff --git a/bsie/apps/info.py b/bsie/apps/info.py
index eaf1f71..e27b70b 100644
--- a/bsie/apps/info.py
+++ b/bsie/apps/info.py
@@ -1,18 +1,15 @@
-"""
 
-Part of the bsie module.
-A copy of the license is provided with the project.
-Author: Matthias Baumgartner, 2022
-"""
-# imports
+# standard imports
 import argparse
+import os
 import sys
 import typing
 
 # bsie imports
-from bsie.base import errors
-from bsie.tools import builder
-from bsie.utils import bsfs
+from bsie.utils import bsfs, errors
+
+# inner-module imports
+from . import _loader
 
 # exports
 __all__: typing.Sequence[str] = (
@@ -25,42 +22,24 @@ __all__: typing.Sequence[str] = (
 def main(argv):
     """Show information from BSIE."""
     parser = argparse.ArgumentParser(description=main.__doc__, prog='info')
-    parser.add_argument('what', choices=('predicates', ),
+    parser.add_argument('--config', type=str,
+        default=os.path.join(os.path.dirname(__file__), _loader.DEFAULT_CONFIG_FILE),
+        help='Path to the config file.')
+    parser.add_argument('what', choices=('predicates', 'schema'),
         help='Select what information to show.')
     args = parser.parse_args(argv)
 
-    # FIXME: Read reader/extractor configs from a config file
-    # reader builder
-    rbuild = builder.ReaderBuilder({})
-    # extractor builder
-    ebuild = builder.ExtractorBuilder([
-        {'bsie.extractor.generic.path.Path': {}},
-        {'bsie.extractor.generic.stat.Stat': {}},
-        {'bsie.extractor.generic.constant.Constant': dict(
-            tuples=[('http://bsfs.ai/schema/Entity#author', 'Me, myself, and I')],
-            schema='''
-                bse:author rdfs:subClassOf bsfs:Predicate ;
-                    rdfs:domain bsfs:Entity ;
-                    rdfs:range xsd:string ;
-                    bsfs:unique "true"^^xsd:boolean .
-                ''',
-            )},
-        ])
-    # pipeline builder
-    pbuild = builder.PipelineBuilder(
-        bsfs.Namespace('http://example.com/me/'), # not actually used
-        rbuild,
-        ebuild,
-        )
-
     # build pipeline
-    pipeline = pbuild.build()
+    pipeline = _loader.load_pipeline(args.config)
 
     # show info
     if args.what == 'predicates':
         # show predicates
         for pred in pipeline.schema.predicates():
             print(pred.uri)
+    elif args.what == 'schema':
+        # show schema
+        print(bsfs.schema.to_string(pipeline.schema))
     else:
         # args.what is already checked by argparse
         raise errors.UnreachableError()