dbis-functional-dependencies 1.0.1__py3-none-any.whl → 1.0.2__py3-none-any.whl

dbis_functional_dependencies/fdcheck.py

@@ -4,7 +4,7 @@ Created on 2022-06-11
 """
 import time
 from dbis_functional_dependencies.BCNF import FunctionalDependencySet
-from lodstorage.sql import SQLDB
+import sqlite3
 
 
 class FDCheck:
@@ -23,6 +23,7 @@ class FDCheck:
         self.lod = lod
         self.debug = debug
         self.entityInfo = None
+        self.conn = None
 
     def createDatabase(
         self,
@@ -52,28 +53,74 @@ class FDCheck:
                 "%s size is %d fixNone is %r fixDates is: %r"
                 % (entityName, size, fixNone, fixDates)
             )
-        self.sqlDB = SQLDB(debug=debug, errorDebug=True)
-        entityInfo = self.sqlDB.createTable(self.lod, entityName, primaryKey)
-        startTime = time.time()
-        self.sqlDB.store(self.lod, entityInfo, executeMany=executeMany, fixNone=fixNone)
-        elapsed = (
-            0.000000000001 if time.time() - startTime == 0 else time.time() - startTime
+
+        self.conn = sqlite3.connect(":memory:")
+        cursor = self.conn.cursor()
+
+        # Infer schema
+        keys = self.lod[0].keys()
+        columns = []
+        for k in keys:
+            value = self.lod[0][k]
+            sql_type = (
+                "INTEGER"
+                if isinstance(value, int)
+                else "REAL"
+                if isinstance(value, float)
+                else "TEXT"
+            )
+            col_def = f"{k} {sql_type}"
+            if k == primaryKey:
+                col_def += " PRIMARY KEY"
+            columns.append(col_def)
+        create_stmt = f"CREATE TABLE {entityName} ({', '.join(columns)});"
+        if debug:
+            print(create_stmt)
+        cursor.execute(create_stmt)
+
+        # Prepare data
+        col_names = list(keys)
+        placeholders = ", ".join(["?"] * len(col_names))
+        insert_stmt = (
+            f"INSERT INTO {entityName} ({', '.join(col_names)}) VALUES ({placeholders})"
         )
+
+        values = []
+        for row in self.lod:
+            values.append(tuple(row.get(k, None) for k in col_names))
+
+        startTime = time.time()
+        if executeMany:
+            cursor.executemany(insert_stmt, values)
+        else:
+            for v in values:
+                cursor.execute(insert_stmt, v)
+        self.conn.commit()
+
+        elapsed = max(1e-12, time.time() - startTime)
         if self.debug:
             print(
-                "adding %d %s records took %5.3f s => %5.f records/s"
-                % (size, entityName, elapsed, size / elapsed)
+                f"adding {size} {entityName} records took {elapsed:.3f} s => {size/elapsed:.0f} records/s"
             )
-        if self.debug:
-            resultList = self.sqlDB.queryAll(entityInfo, fixDates=fixDates)
+
+            cursor.execute(f"SELECT * FROM {entityName}")
+            resultList = cursor.fetchall()
             print(
-                "selecting %d %s records took %5.3f s => %5.f records/s"
-                % (len(resultList), entityName, elapsed, len(resultList) / elapsed)
+                f"selecting {len(resultList)} {entityName} records took {elapsed:.3f} s => {len(resultList)/elapsed:.0f} records/s"
             )
+
+        self.entityInfo = {
+            "name": entityName,
+            "typeMap": {k: type(self.lod[0][k]) for k in keys},
+            "fields": list(keys),
+        }
+
         if doClose:
-            self.sqlDB.close()
-        self.entityInfo = entityInfo
-        return entityInfo
+            cursor.close()
+            self.conn.close()
+            self.conn = None
+
+        return self.entityInfo
 
     def findFDs(self):
         """
@@ -83,31 +130,37 @@ class FDCheck:
         Return:
             FunctionalDependencySet: the set of functional dependencies
         """
-        if self.entityInfo is None:
-            raise Exception("createDataBase needed to supply entityInfo")
-        fields = list(self.entityInfo.typeMap.keys())
-        table_name = self.entityInfo.name
+        if self.entityInfo is None or not self.conn:
+            raise Exception("createDatabase must be called before findFDs")
+
+        table_name = self.entityInfo["name"]
+        fields = self.entityInfo["fields"]
+
         fds = FunctionalDependencySet()
         for i, field in enumerate(fields):
             attr1_var = chr(ord("A") + i)
             fds.add_attribute(attr1_var, field)
-        for i, field in enumerate(fields):
+
+        cursor = self.conn.cursor()
+        for i, field_1 in enumerate(fields):
             attr1_var = chr(ord("A") + i)
-            for j in range(0, len(fields)):
+            for j, field_2 in enumerate(fields):
                 if i == j:
                     continue
-
-                field_1 = fields[i]
-                field_2 = fields[j]
                 attr2_var = chr(ord("A") + j)
-                sql = f"SELECT {field_1}, COUNT(DISTINCT {field_2}) c FROM {table_name} GROUP BY {field_1} HAVING c > 1"
-                hits = self.sqlDB.query(sql)
+                sql = (
+                    f"SELECT {field_1}, COUNT(DISTINCT {field_2}) as c\n"
+                    f"FROM {table_name}\n"
+                    f"GROUP BY {field_1}\n"
+                    f"HAVING c > 1"
+                )
+                cursor.execute(sql)
+                hits = cursor.fetchall()
                 if self.debug:
-                    print(f"{sql}\n{hits}")
-
-                if len(hits) == 0:
-                    # Functional dependency found: it's not the case that there's more than one value (field_2)
-                    # associated with field_1
+                    print(f"{sql.strip()}\n{hits}")
+                if not hits:
                     fds.add_dependency(attr1_var, attr2_var)
+
+        cursor.close()
         self.fds = fds
         return fds

{dbis_functional_dependencies-1.0.1.dist-info → dbis_functional_dependencies-1.0.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dbis-functional-dependencies
-Version: 1.0.1
+Version: 1.0.2
 Summary: RWTH Aachen Computer Science i5/dbis assets for Lecture Datenbanken und Informationssysteme
 Author-email: DBIS i5 RWTH Aachen <dbis-vl@dbis.rwth-aachen.de>
 Project-URL: Homepage, https://git.rwth-aachen.de/i5/teaching/dbis/dbis-functional-dependencies
@@ -17,7 +17,6 @@ Requires-Dist: sphinx~=7.2
 Requires-Dist: sphinxcontrib-apidoc~=0.4
 Requires-Dist: functional-dependencies~=1.3
 Requires-Dist: build~=1.0
-Requires-Dist: pylodstorage~=0.4.11
 Provides-Extra: test
 Requires-Dist: black==23.12.1; extra == "test"
 Provides-Extra: build

dbis_functional_dependencies-1.0.2.dist-info/RECORD

@@ -0,0 +1,10 @@
+dbis_functional_dependencies/BCNF.py,sha256=7EosTQTPl07W_UIiTp41PoxM9OTWMIzwA5cdQxbrkyM,68688
+dbis_functional_dependencies/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+dbis_functional_dependencies/fdcheck.py,sha256=q_cvs--uhl5g6IZq5ral-Y_N3kwc3royti9kYb-dNV4,5238
+dbis_functional_dependencies/fds.py,sha256=CLssS0L-3NxodSCtEcOMG5lxy0elLEaWkK8cT9IUszo,24021
+dbis_functional_dependencies/fdsbase.py,sha256=wgt1tjM-EyD09v8Yf57UgoZB6ptvijIuE0WfgdKj43M,5767
+dbis_functional_dependencies-1.0.2.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+dbis_functional_dependencies-1.0.2.dist-info/METADATA,sha256=fxj0g0tIkM8QCHdkBpUVRxjgMrbx6grhvYFb6VPV5vM,6797
+dbis_functional_dependencies-1.0.2.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+dbis_functional_dependencies-1.0.2.dist-info/top_level.txt,sha256=Kxq61kOpu1YSAAigxyqE3_XCyftviQ1TW55Q1oxEP-4,29
+dbis_functional_dependencies-1.0.2.dist-info/RECORD,,

{dbis_functional_dependencies-1.0.1.dist-info → dbis_functional_dependencies-1.0.2.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (77.0.3)
+Generator: setuptools (80.9.0)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

dbis_functional_dependencies-1.0.1.dist-info/RECORD

@@ -1,10 +0,0 @@
-dbis_functional_dependencies/BCNF.py,sha256=7EosTQTPl07W_UIiTp41PoxM9OTWMIzwA5cdQxbrkyM,68688
-dbis_functional_dependencies/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-dbis_functional_dependencies/fdcheck.py,sha256=wNFJb2WZGkvQxN8Prbcd6DhuXYnQ4Zm9HsoLxc7RU7k,4082
-dbis_functional_dependencies/fds.py,sha256=CLssS0L-3NxodSCtEcOMG5lxy0elLEaWkK8cT9IUszo,24021
-dbis_functional_dependencies/fdsbase.py,sha256=wgt1tjM-EyD09v8Yf57UgoZB6ptvijIuE0WfgdKj43M,5767
-dbis_functional_dependencies-1.0.1.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-dbis_functional_dependencies-1.0.1.dist-info/METADATA,sha256=ZYgolVuBgO4i_MfRZ1bXzow3jRzVhMACwS9vJmHzQGA,6833
-dbis_functional_dependencies-1.0.1.dist-info/WHEEL,sha256=1tXe9gY0PYatrMPMDd6jXqjfpz_B-Wqm32CPfRC58XU,91
-dbis_functional_dependencies-1.0.1.dist-info/top_level.txt,sha256=Kxq61kOpu1YSAAigxyqE3_XCyftviQ1TW55Q1oxEP-4,29
-dbis_functional_dependencies-1.0.1.dist-info/RECORD,,