[Archive] replace column 'bbox' with 'geometry'

environment-dev.yml

@@ -17,6 +17,7 @@ dependencies:
   - pillow
   - lxml
   - packaging
+  - libspatialite>=5.1.0
   - coveralls
   - coverage
   - pytest

environment.yml

@@ -17,3 +17,4 @@ dependencies:
   - pillow
   - lxml
   - packaging
+  - libspatialite>=5.1.0

pyroSAR/drivers.py

@@ -2285,10 +2285,14 @@ class Archive(object):
     def __init__(self, dbfile, custom_fields=None, postgres=False, user='postgres',
                  password='1234', host='localhost', port=5432, cleanup=True,
                  legacy=False):
+
+        if dbfile.endswith('.csv'):
+            raise RuntimeError("Please create a new Archive database and import the"
+                               "CSV file using db.import_outdated('<file>.csv').")
         # check for driver, if postgres then check if server is reachable
         if not postgres:
             self.driver = 'sqlite'
-            dirname = os.path.dirname(dbfile)
+            dirname = os.path.dirname(os.path.abspath(dbfile))
             w_ok = os.access(dirname, os.W_OK)
             if not w_ok:
                 raise RuntimeError('cannot write to directory {}'.format(dirname))
@@ -2328,7 +2332,7 @@ def __init__(self, dbfile, custom_fields=None, postgres=False, user='postgres',
         self.engine = create_engine(url=self.url, echo=False,
                                     connect_args=connect_args)
 
-        # call to ____load_spatialite() for sqlite, to load mod_spatialite via event handler listen()
+        # call to __load_spatialite() for sqlite, to load mod_spatialite via event handler listen()
         if self.driver == 'sqlite':
             log.debug('loading spatialite extension')
             listen(target=self.engine, identifier='connect', fn=self.__load_spatialite)
@@ -2362,11 +2366,15 @@ def __init__(self, dbfile, custom_fields=None, postgres=False, user='postgres',
         self.__init_data_table()
         self.__init_duplicates_table()
 
+        msg = ("the 'data' table is missing {}. Please create a new database "
+               "and import the old one opened in legacy mode using "
+               "Archive.import_outdated.")
         pk = sql_inspect(self.data_schema).primary_key
         if 'product' not in pk.columns.keys() and not legacy:
-            raise RuntimeError("the 'data' table is missing a primary key 'product'. "
-                               "Please create a new database and import the old one "
-                               "opened in legacy mode using Archive.import_outdated.")
+            raise RuntimeError(msg.format("a primary key 'product'"))
+
+        if 'geometry' not in self.get_colnames() and not legacy:
+            raise RuntimeError(msg.format("the 'geometry' column"))
 
         self.Base = automap_base(metadata=self.meta)
         self.Base.prepare(self.engine, reflect=True)
@@ -2387,7 +2395,7 @@ def add_tables(self, tables):
         .. note::
 
             Columns using Geometry must have setting management=True for SQLite,
-            for example: ``bbox = Column(Geometry('POLYGON', management=True, srid=4326))``
+            for example: ``geometry = Column(Geometry('POLYGON', management=True, srid=4326))``
 
         Parameters
         ----------
@@ -2437,8 +2445,8 @@ def __init_data_table(self):
                                  Column('vv', Integer),
                                  Column('hv', Integer),
                                  Column('vh', Integer),
-                                 Column('bbox', Geometry(geometry_type='POLYGON',
-                                                         management=True, srid=4326)))
+                                 Column('geometry', Geometry(geometry_type='POLYGON',
+                                                             management=True, srid=4326)))
         # add custom fields
         if self.custom_fields is not None:
             for key, val in self.custom_fields.items():
@@ -2513,13 +2521,13 @@ def __prepare_insertion(self, scene):
         insertion = self.Data()
         colnames = self.get_colnames()
         for attribute in colnames:
-            if attribute == 'bbox':
-                geom = id.bbox()
+            if attribute == 'geometry':
+                geom = id.geometry()
                 geom.reproject(4326)
                 geom = geom.convert2wkt(set3D=False)[0]
                 geom = 'SRID=4326;' + str(geom)
                 # set attributes of the Data object according to input
-                setattr(insertion, 'bbox', geom)
+                setattr(insertion, 'geometry', geom)
             elif attribute in ['hh', 'vv', 'hv', 'vh']:
                 setattr(insertion, attribute, int(attribute in pols))
             else:
@@ -2873,9 +2881,9 @@ def import_outdated(self, dbfile):
                     scenes.append(row['scene'])
                 self.insert(scenes)
         elif isinstance(dbfile, Archive):
-            select = dbfile.conn.execute('SELECT * from data')
-            self.conn.execute(insert(self.Data).values(*select))
-            # duplicates in older databases may fit into the new data table
+            scenes = dbfile.conn.execute('SELECT scene from data')
+            scenes = [s.scene for s in scenes]
+            self.insert(scenes)
             reinsert = dbfile.select_duplicates(value='scene')
             if reinsert is not None:
                 self.insert(reinsert)
@@ -3030,11 +3038,11 @@ def select(self, vectorobject=None, mindate=None, maxdate=None, date_strict=True
                     site_geom = vec.convert2wkt(set3D=False)[0]
                 # postgres has a different way to store geometries
                 if self.driver == 'postgresql':
-                    arg_format.append("st_intersects(bbox, 'SRID=4326; {}')".format(
+                    arg_format.append("st_intersects(geometry, 'SRID=4326; {}')".format(
                         site_geom
                     ))
                 else:
-                    arg_format.append('st_intersects(GeomFromText(?, 4326), bbox) = 1')
+                    arg_format.append('st_intersects(GeomFromText(?, 4326), geometry) = 1')
                     vals.append(site_geom)
             else:
                 log.info('WARNING: argument vectorobject is ignored, must be of type spatialist.vector.Vector')

tests/conftest.py

@@ -39,7 +39,8 @@ def testdata(testdir):
         # used in test_osv
         's1_orbit': os.path.join(testdir, 'S1A_IW_GRDH_1SDV_20210119T031653_20210119T031718_036201_043ED0_8255.zip'),
         'tif': os.path.join(testdir, 'S1A__IW___A_20150309T173017_VV_grd_mli_geo_norm_db.tif'),
-        'archive_old': os.path.join(testdir, 'archive_outdated.csv'),
+        'archive_old_csv': os.path.join(testdir, 'archive_outdated.csv'),
+        'archive_old_bbox': os.path.join(testdir, 'archive_outdated_bbox.db'),
         'dempar': os.path.join(testdir, 'dem.par'),
         'mlipar': os.path.join(testdir, 'mli.par')
     }

tests/test_drivers.py

@@ -201,9 +201,26 @@ def test_archive2(tmpdir, testdata):
     os.remove(dbfile)
     assert not os.path.isfile(dbfile)
     assert Vector(shp).nfeatures == 1
+
     with pytest.raises(OSError):
         with pyroSAR.Archive(dbfile) as db:
-            db.import_outdated(testdata['archive_old'])
+            db.import_outdated(testdata['archive_old_csv'])
+
+    # the archive_old_bbox database contains a relative file name for the scene
+    # so that it can be reimported into the new database. The working directory
+    # is changed temporarily so that the scene can be found.
+    cwd = os.getcwd()
+    folder = os.path.dirname(os.path.realpath(__file__))
+    os.chdir(os.path.join(folder, 'data'))
+    with pyroSAR.Archive(dbfile) as db:
+        with pyroSAR.Archive(testdata['archive_old_bbox'], legacy=True) as db_old:
+            db.import_outdated(db_old)
+    os.chdir(cwd)
+
+    with pytest.raises(RuntimeError):
+        db = pyroSAR.Archive(testdata['archive_old_csv'])
+    with pytest.raises(RuntimeError):
+        db = pyroSAR.Archive(testdata['archive_old_bbox'])
 
 
 def test_archive_postgres(tmpdir, testdata):
@@ -237,16 +254,38 @@ def test_archive_postgres(tmpdir, testdata):
     with pytest.raises(TypeError):
         db.filter_scenelist([1])
     db.close()
-    with pyroSAR.Archive('test', postgres=True, port=pgport, user=pguser, password=pgpassword) as db:
+    with pyroSAR.Archive('test', postgres=True, port=pgport,
+                         user=pguser, password=pgpassword) as db:
         assert db.size == (1, 0)
         shp = os.path.join(str(tmpdir), 'db.shp')
         db.export2shp(shp)
         pyroSAR.drop_archive(db)
     assert Vector(shp).nfeatures == 1
 
-    with pyroSAR.Archive('test', postgres=True, port=pgport, user=pguser, password=pgpassword) as db:
+    with pyroSAR.Archive('test', postgres=True, port=pgport,
+                         user=pguser, password=pgpassword) as db:
         with pytest.raises(OSError):
-            db.import_outdated(testdata['archive_old'])
+            db.import_outdated(testdata['archive_old_csv'])
+        pyroSAR.drop_archive(db)
+
+    # the archive_old_bbox database contains a relative file name for the scene
+    # so that it can be reimported into the new database. The working directory
+    # is changed temporarily so that the scene can be found.
+    cwd = os.getcwd()
+    folder = os.path.dirname(os.path.realpath(__file__))
+    os.chdir(os.path.join(folder, 'data'))
+    with pyroSAR.Archive('test', postgres=True, port=pgport,
+                         user=pguser, password=pgpassword) as db:
+        with pyroSAR.Archive(testdata['archive_old_bbox'], legacy=True) as db_old:
+            db.import_outdated(db_old)
+        pyroSAR.drop_archive(db)
+    os.chdir(cwd)
+
+    dbfile = os.path.join(str(tmpdir), 'scenes.db')
+    with pyroSAR.Archive('test', postgres=True, port=pgport,
+                         user=pguser, password=pgpassword) as db:
+        with pyroSAR.Archive(dbfile, legacy=True) as db_sqlite:
+            db.import_outdated(db_sqlite)
         pyroSAR.drop_archive(db)
 
     with pytest.raises(SystemExit) as pytest_wrapped_e: