home · contact · privacy
Enhance BaseModel comparisons by hashing versioned and relations attributes.
[plomtask] / plomtask / db.py
index 7962eabeffd28964c0892b87f7ce35e6052a2f3e..a47dff15917651940483afe83f41152399bedf7f 100644 (file)
@@ -6,8 +6,9 @@ from difflib import Differ
 from sqlite3 import connect as sql_connect, Cursor, Row
 from typing import Any, Self, TypeVar, Generic
 from plomtask.exceptions import HandledException, NotFoundException
+from plomtask.dating import valid_date
 
-EXPECTED_DB_VERSION = 1
+EXPECTED_DB_VERSION = 5
 MIGRATIONS_DIR = 'migrations'
 FILENAME_DB_SCHEMA = f'init_{EXPECTED_DB_VERSION}.sql'
 PATH_DB_SCHEMA = f'{MIGRATIONS_DIR}/{FILENAME_DB_SCHEMA}'
@@ -17,8 +18,9 @@ class UnmigratedDbException(HandledException):
     """To identify case of unmigrated DB file."""
 
 
-class DatabaseFile:  # pylint: disable=too-few-public-methods
+class DatabaseFile:
     """Represents the sqlite3 database's file."""
+    # pylint: disable=too-few-public-methods
 
     def __init__(self, path: str) -> None:
         self.path = path
@@ -37,7 +39,7 @@ class DatabaseFile:  # pylint: disable=too-few-public-methods
     def migrate(cls, path: str) -> DatabaseFile:
         """Apply migrations from_version to EXPECTED_DB_VERSION."""
         migrations = cls._available_migrations()
-        from_version = cls.get_version_of_db(path)
+        from_version = cls._get_version_of_db(path)
         migrations_todo = migrations[from_version+1:]
         for j, filename in enumerate(migrations_todo):
             with sql_connect(path) as conn:
@@ -53,7 +55,7 @@ class DatabaseFile:  # pylint: disable=too-few-public-methods
         """Check file exists, and is of proper DB version and schema."""
         if not isfile(self.path):
             raise NotFoundException
-        if self.user_version != EXPECTED_DB_VERSION:
+        if self._user_version != EXPECTED_DB_VERSION:
             raise UnmigratedDbException()
         self._validate_schema()
 
@@ -85,7 +87,7 @@ class DatabaseFile:  # pylint: disable=too-few-public-methods
         return migrations_list
 
     @staticmethod
-    def get_version_of_db(path: str) -> int:
+    def _get_version_of_db(path: str) -> int:
         """Get DB user_version, fail if outside expected range."""
         sql_for_db_version = 'PRAGMA user_version'
         with sql_connect(path) as conn:
@@ -98,9 +100,11 @@ class DatabaseFile:  # pylint: disable=too-few-public-methods
         return db_version
 
     @property
-    def user_version(self) -> int:
+    def _user_version(self) -> int:
         """Get DB user_version."""
-        return self.__class__.get_version_of_db(self.path)
+        # pylint: disable=protected-access
+        # (since we remain within class)
+        return self.__class__._get_version_of_db(self.path)
 
     def _validate_schema(self) -> None:
         """Compare found schema with what's stored at PATH_DB_SCHEMA."""
@@ -131,6 +135,9 @@ class DatabaseFile:  # pylint: disable=too-few-public-methods
                         new_row += [f'    {segment}']
                 new_row[0] = new_row[0].lstrip()
                 new_row[-1] = new_row[-1].lstrip()
+                if new_row[-1] != ')' and new_row[-3][-1] != ',':
+                    new_row[-3] = new_row[-3] + ','
+                    new_row[-2:] = ['    ' + new_row[-1][:-1]] + [')']
                 new_rows += ['\n'.join(new_row)]
             return new_rows
 
@@ -152,8 +159,7 @@ class DatabaseConnection:
     """A single connection to the database."""
 
     def __init__(self, db_file: DatabaseFile) -> None:
-        self.file = db_file
-        self.conn = sql_connect(self.file.path)
+        self.conn = sql_connect(db_file.path)
 
     def commit(self) -> None:
         """Commit SQL transaction."""
@@ -163,18 +169,28 @@ class DatabaseConnection:
         """Add commands to SQL transaction."""
         return self.conn.execute(code, inputs)
 
+    def exec_on_vals(self, code: str, inputs: tuple[Any, ...]) -> Cursor:
+        """Wrapper around .exec appending adequate " (?, …)" to code."""
+        q_marks_from_values = '(' + ','.join(['?'] * len(inputs)) + ')'
+        return self.exec(f'{code} {q_marks_from_values}', inputs)
+
     def close(self) -> None:
         """Close DB connection."""
         self.conn.close()
 
     def rewrite_relations(self, table_name: str, key: str, target: int | str,
-                          rows: list[list[Any]]) -> None:
-        """Rewrite relations in table_name to target, with rows values."""
+                          rows: list[list[Any]], key_index: int = 0) -> None:
+        # pylint: disable=too-many-arguments
+        """Rewrite relations in table_name to target, with rows values.
+
+        Note that single rows are expected without the column and value
+        identified by key and target, which are inserted inside the function
+        at key_index.
+        """
         self.delete_where(table_name, key, target)
         for row in rows:
-            values = tuple([target] + row)
-            q_marks = self.__class__.q_marks_from_values(values)
-            self.exec(f'INSERT INTO {table_name} VALUES {q_marks}', values)
+            values = tuple(row[:key_index] + [target] + row[key_index:])
+            self.exec_on_vals(f'INSERT INTO {table_name} VALUES', values)
 
     def row_where(self, table_name: str, key: str,
                   target: int | str) -> list[Row]:
@@ -182,6 +198,17 @@ class DatabaseConnection:
         return list(self.exec(f'SELECT * FROM {table_name} WHERE {key} = ?',
                               (target,)))
 
+    # def column_where_pattern(self,
+    #                          table_name: str,
+    #                          column: str,
+    #                          pattern: str,
+    #                          keys: list[str]) -> list[Any]:
+    #     """Return column of rows where one of keys matches pattern."""
+    #     targets = tuple([f'%{pattern}%'] * len(keys))
+    #     haystack = ' OR '.join([f'{k} LIKE ?' for k in keys])
+    #     sql = f'SELECT {column} FROM {table_name} WHERE {haystack}'
+    #     return [row[0] for row in self.exec(sql, targets)]
+
     def column_where(self, table_name: str, column: str, key: str,
                      target: int | str) -> list[Any]:
         """Return column of table where key == target."""
@@ -199,11 +226,6 @@ class DatabaseConnection:
         """Delete from table where key == target."""
         self.exec(f'DELETE FROM {table_name} WHERE {key} = ?', (target,))
 
-    @staticmethod
-    def q_marks_from_values(values: tuple[Any]) -> str:
-        """Return placeholder to insert values into SQL code."""
-        return '(' + ','.join(['?'] * len(values)) + ')'
-
 
 BaseModelId = TypeVar('BaseModelId', int, str)
 BaseModelInstance = TypeVar('BaseModelInstance', bound='BaseModel[Any]')
@@ -214,24 +236,33 @@ class BaseModel(Generic[BaseModelId]):
     table_name = ''
     to_save: list[str] = []
     to_save_versioned: list[str] = []
-    to_save_relations: list[tuple[str, str, str]] = []
+    to_save_relations: list[tuple[str, str, str, int]] = []
     id_: None | BaseModelId
     cache_: dict[BaseModelId, Self]
+    to_search: list[str] = []
 
     def __init__(self, id_: BaseModelId | None) -> None:
         if isinstance(id_, int) and id_ < 1:
             msg = f'illegal {self.__class__.__name__} ID, must be >=1: {id_}'
             raise HandledException(msg)
+        if isinstance(id_, str) and "" == id_:
+            msg = f'illegal {self.__class__.__name__} ID, must be non-empty'
+            raise HandledException(msg)
         self.id_ = id_
 
+    def __hash__(self) -> int:
+        hashable = [self.id_] + [getattr(self, name) for name in self.to_save]
+        for definition in self.to_save_relations:
+            attr = getattr(self, definition[2])
+            hashable += [tuple(rel.id_ for rel in attr)]
+        for name in self.to_save_versioned:
+            hashable += [hash(getattr(self, name))]
+        return hash(tuple(hashable))
+
     def __eq__(self, other: object) -> bool:
         if not isinstance(other, self.__class__):
             return False
-        to_hash_me = tuple([self.id_] +
-                           [getattr(self, name) for name in self.to_save])
-        to_hash_other = tuple([other.id_] +
-                              [getattr(other, name) for name in other.to_save])
-        return hash(to_hash_me) == hash(to_hash_other)
+        return hash(self) == hash(other)
 
     def __lt__(self, other: Any) -> bool:
         if not isinstance(other, self.__class__):
@@ -339,6 +370,49 @@ class BaseModel(Generic[BaseModelId]):
                 items[item.id_] = item
         return list(items.values())
 
+    @classmethod
+    def by_date_range_with_limits(cls: type[BaseModelInstance],
+                                  db_conn: DatabaseConnection,
+                                  date_range: tuple[str, str],
+                                  date_col: str = 'day'
+                                  ) -> tuple[list[BaseModelInstance], str,
+                                             str]:
+        """Return list of items in database within (open) date_range interval.
+
+        If no range values provided, defaults them to 'yesterday' and
+        'tomorrow'. Knows to properly interpret these and 'today' as value.
+        """
+        start_str = date_range[0] if date_range[0] else 'yesterday'
+        end_str = date_range[1] if date_range[1] else 'tomorrow'
+        start_date = valid_date(start_str)
+        end_date = valid_date(end_str)
+        items = []
+        sql = f'SELECT id FROM {cls.table_name} '
+        sql += f'WHERE {date_col} >= ? AND {date_col} <= ?'
+        for row in db_conn.exec(sql, (start_date, end_date)):
+            items += [cls.by_id(db_conn, row[0])]
+        return items, start_date, end_date
+
+    @classmethod
+    def matching(cls: type[BaseModelInstance], db_conn: DatabaseConnection,
+                 pattern: str) -> list[BaseModelInstance]:
+        """Return all objects whose .to_search match pattern."""
+        items = cls.all(db_conn)
+        if pattern:
+            filtered = []
+            for item in items:
+                for attr_name in cls.to_search:
+                    toks = attr_name.split('.')
+                    parent = item
+                    for tok in toks:
+                        attr = getattr(parent, tok)
+                        parent = attr
+                    if pattern in attr:
+                        filtered += [item]
+                        break
+            return filtered
+        return items
+
     def save(self, db_conn: DatabaseConnection) -> None:
         """Write self to DB and cache and ensure .id_.
 
@@ -352,20 +426,19 @@ class BaseModel(Generic[BaseModelId]):
         """
         values = tuple([self.id_] + [getattr(self, key)
                                      for key in self.to_save])
-        q_marks = DatabaseConnection.q_marks_from_values(values)
         table_name = self.table_name
-        cursor = db_conn.exec(f'REPLACE INTO {table_name} VALUES {q_marks}',
-                              values)
+        cursor = db_conn.exec_on_vals(f'REPLACE INTO {table_name} VALUES',
+                                      values)
         if not isinstance(self.id_, str):
             self.id_ = cursor.lastrowid  # type: ignore[assignment]
         self.cache()
         for attr_name in self.to_save_versioned:
             getattr(self, attr_name).save(db_conn)
-        for table, column, attr_name in self.to_save_relations:
+        for table, column, attr_name, key_index in self.to_save_relations:
             assert isinstance(self.id_, (int, str))
             db_conn.rewrite_relations(table, column, self.id_,
                                       [[i.id_] for i
-                                       in getattr(self, attr_name)])
+                                       in getattr(self, attr_name)], key_index)
 
     def remove(self, db_conn: DatabaseConnection) -> None:
         """Remove from DB and cache, including dependencies."""
@@ -373,7 +446,7 @@ class BaseModel(Generic[BaseModelId]):
             raise HandledException('cannot remove unsaved item')
         for attr_name in self.to_save_versioned:
             getattr(self, attr_name).remove(db_conn)
-        for table, column, attr_name in self.to_save_relations:
+        for table, column, attr_name, _ in self.to_save_relations:
             db_conn.delete_where(table, column, self.id_)
         self.uncache()
         db_conn.delete_where(self.table_name, 'id', self.id_)