added persistent storage module for large collections

autoimport/__main__.py

@@ -28,6 +28,7 @@ Options:
 
 from .commands import copy, move, placeholders, select
 from .tmpdb import get_temporary_db
+from .db import get_persistent_db
 
 args = docopt.docopt(usage)
 
@@ -40,11 +41,20 @@ logging.debug("ARGUMENTS:")
 for k,v in args.items():
     logging.debug("\t{}: \t{}".format(k,v))
 
+if(not args["--storage"]):
     try:
         db = get_temporary_db(args["--implementation"])
     except Exception as e:
         print(e)
         sys.exit(1)
+    stored = False
+else:
+    try:
+        db = get_persistent_db(args["--storage"])
+    except Exception as e:
+        print(e)
+        sys.exit(1)
+    stored = True
 
 if(args["placeholders"]):
     result = placeholders()
@@ -56,7 +66,8 @@ elif(args["copy"]):
             , not args["--no-select-stop-on-error"]
             , args["--walk"]
             , args["--postfix"]
-            , args["--dry-run"])
+            , args["--dry-run"]
+            , stored)
 
 elif(args["move"]):
     result = move(db
@@ -66,14 +77,16 @@ elif(args["move"]):
             , not args["--no-select-stop-on-error"]
             , args["--walk"]
             , args["--postfix"]
-            , args["--dry-run"])
+            , args["--dry-run"]
+            , stored)
 elif(args["select"]):
     result = select(db
             , args["SRC_PATH"]
             , not args["--no-select-stop-on-error"]
             , args["--walk"]
             , args["--postfix"]
-            , args["--dry-run"])
+            , args["--dry-run"]
+            , stored)
 
 
 

autoimport/commands.py

@@ -13,7 +13,13 @@ def placeholders():
 
     return 0
 
-def select(db, src_path, stop_on_error, walk, postfix, dryrun):
+def select( db
+            , src_path
+            , stop_on_error
+            , walk
+            , postfix
+            , dryrun
+            , stored):
     logger = logging.getLogger(__name__)
     extensions = postfix.split(",")
     try:
@@ -51,7 +57,8 @@ def copy(db
         , stop_on_error
         , walk
         , postfix
-        , dryrun):
+        , dryrun
+        , stored):
 
     return do_copy_or_move(db
         , src_path
@@ -61,7 +68,8 @@ def copy(db
         , walk
         , postfix
         , dryrun
-        , False)
+        , False
+        , stored)
 
 def move(db
         , src_path
@@ -70,7 +78,8 @@ def move(db
         , stop_on_error
         , walk
         , postfix
-        , dryrun):
+        , dryrun
+        , stored):
 
     return do_copy_or_move(db
         , src_path
@@ -80,7 +89,8 @@ def move(db
         , walk
         , postfix
         , dryrun
-        , True)
+        , True
+        , stored)
 
 
 def do_copy_or_move(db
@@ -91,10 +101,12 @@ def do_copy_or_move(db
         , walk
         , postfix
         , dryrun
-        , move):
+        , move
+        , stored):
 
     logger = logging.getLogger(__name__)
     extensions = postfix.split(",")
+    if(not stored):
         try:
             findall(src_path, walk, extensions, db, stop_on_error)
         except Exception as e:

autoimport/db.py

@@ -0,0 +1,88 @@
+"""
+This module provides a way to construct (persistent) databases
+used by ``autoimport``. In normal mode this module is replaced
+by ``tmpdb``, however it might be useful to keep the data produced
+by ``autoimport``.
+"""
+
+import sqlite3
+import abc
+import os
+
+class AbstractDatabase(abc.ABC):
+    """
+    Abstract base class for all ``TemporaryDatabase``
+    implementations. 
+
+    **Note**: ``__init__`` must set ``self._db`` to an
+          open sqlite3 connection.
+    """
+    def __init__(self):
+        abc.ABC.__init__(self)
+        self._db = None
+
+    @abc.abstractmethod
+    def close(self):
+        pass
+
+    def cursor(self):
+        return self._db.cursor()
+
+    def dump_db(self, file):
+        for line in self._db.iterdump():
+            file.write("{}\n".format(line))
+    def commit(self):
+        return self._db.commit()
+
+class PersistentDatabase(AbstractDatabase):
+    def __init__(self, database_path):
+        AbstractDatabase.__init__(self)
+        self._database_path = database_path
+        self._db = sqlite3.connect(database_path)
+
+    def close(self):
+        self._db.close()
+
+def initialize_database(db):
+    cursor = db.cursor()
+
+    cursor.execute(
+        '''CREATE TABLE FILES(
+            name TEXT,
+            DateTime TEXT,
+            DateTimeDigitized TEXT,
+            DateTimeOriginal TEXT,
+            Model TEXT,
+            Make TEXT,
+            Software TEXT)'''
+        )
+
+    cursor.execute(
+        '''CREATE TABLE DIRECTORIES(
+            name TEXT)'''
+        )
+
+    cursor.execute(
+        '''CREATE TABLE ASSOCIATIONS(file_id INTEGER,
+            directory_id INTEGER)'''
+        )
+
+    cursor.execute(
+        '''CREATE TABLE KV(key TEXT,
+            value TEXT)'''
+        )
+    cursor.execute(
+        '''CREATE TABLE EXTENSIONS_SEARCHED(extension TEXT)'''
+        )
+    db.commit()
+
+
+def get_persistent_db(path):
+    if(not os.path.exists(path)):
+        if(not os.path.dirname(path)):
+            db = PersistentDatabase(path)
+            initialize_database(db)
+            return db
+        if(not os.path.exists(os.path.dirname(path))):
+            raise IOError("path '{}' does not exist".format(os.path.dirname(path)))
+    return PersistentDatabase(path)

autoimport/order/order.py

@@ -63,6 +63,7 @@ def order(db, path_specifier):
         path_id = get_path_id(db, this_path)
 
         cursor.execute("INSERT INTO ASSOCIATIONS(file_id, directory_id) VALUES(?, ?)", (rowid, path_id))
+    db.commit()
 
 
 
@@ -76,5 +77,6 @@ def get_path_id(db, path):
         return result[0]
     
     cursor.execute("INSERT INTO DIRECTORIES(name) VALUES(?)", (path,))
+    db.commit()
     return cursor.lastrowid
 

autoimport/select/select.py

@@ -11,6 +11,7 @@ def findall_this_directory(directory, files, extensions, db, stop_on_error):
         if(filename.split(".")[-1] in extensions):
             filename = os.path.join(directory, filename)
             insert_file_into_db(filename, db, stop_on_error)
+    db.commit()
 
 
 def insert_file_into_db(filename, db, stop_on_error):
@@ -47,8 +48,21 @@ def insert_file_into_db(filename, db, stop_on_error):
             , data)
 
 def findall(directory, walk, extensions, db, stop_on_error):
+    cursor = db.cursor()
+    cursor.execute("SELECT extension FROM EXTENSIONS_SEARCHED")
+    ext_already_searched = {i[0] for i in cursor.fetchall()}
+    ext_to_search = set(extensions) - ext_already_searched
+    ext_omit = set(extensions) - ext_to_search
+
+    if(ext_omit):
+        module_logger.warn("Omitting the extensions {} as they are already in the database.".format(ext_omit))
+
+    extensions = list(ext_to_search)
     for dir_, paths, files in os.walk(directory):
         findall_this_directory(dir_, files, extensions, db, stop_on_error)
         if(not walk):
             break
 
+    cursor.executemany("INSERT INTO EXTENSIONS_SEARCHED(extension) VALUES(?)", [(i,) for i in extensions])
+    db.commit()
+

autoimport/tmpdb.py

@@ -14,6 +14,9 @@ import sqlite3
 import tempfile
 import abc
 
+from .db import AbstractDatabase, initialize_database
+
+
 def _open_db_mem():
     return (sqlite3.connect(":memory:"), None)
 
@@ -22,28 +25,9 @@ def _open_db_disk():
     db = sqlite3.connect(file.name)
     return (db, file)
 
-class AbstractTemporaryDatabase(abc.ABC):
-    """
-    Abstract base class for all ``TemporaryDatabase``
-    implementations. 
-
-    **Note**: ``__init__`` must set ``self._db`` to an
-          open sqlite3 connection.
-    """
-    def __init__(self):
-        abc.ABC.__init__(self)
-        self._db = None
-    @abc.abstractmethod
-    def close(self):
+class AbstractTemporaryDatabase(AbstractDatabase):
     pass
 
-    def cursor(self):
-        return self._db.cursor()
-
-    def dump_db(self, file):
-        for line in self._db.iterdump():
-            file.write("{}\n".format(line))
-
     
 class MemoryTemporaryDatabase(AbstractTemporaryDatabase):
     def __init__(self):
@@ -80,33 +64,6 @@ def get_temporary_db(type_):
 
     impl = implementations[type_]
     instance = impl()
-
-    cursor = instance.cursor()
-
-    cursor.execute(
-        '''CREATE TABLE FILES(
-            name TEXT,
-            DateTime TEXT,
-            DateTimeDigitized TEXT,
-            DateTimeOriginal TEXT,
-            Model TEXT,
-            Make TEXT,
-            Software TEXT)'''
-        )
-
-    cursor.execute(
-        '''CREATE TABLE DIRECTORIES(
-            name TEXT)'''
-        )
-
-    cursor.execute(
-        '''CREATE TABLE ASSOCIATIONS(file_id INTEGER,
-            directory_id INTEGER)'''
-        )
-
-    cursor.execute(
-        '''CREATE TABLE KV(key TEXT,
-            value TEXT)'''
-        )
+    initialize_database(instance) 
 
     return instance