From 6af0c42ed6234d8db295b8bf3811fbacac75dbc6 Mon Sep 17 00:00:00 2001 From: Eevee Date: Tue, 18 Aug 2009 18:02:53 -0700 Subject: [PATCH] Vastly improved the pokedex import/export UI. csvimport is now load; csvexport is now dump. Both take an optional -e switch to specify an engine, but will happily use a default SQLite database in the pokedex package directory. Additionally, the CSV directory is now controlled by the optional -d switch, and defaults to Doing The Right Thing. So `pokedex load` now does exactly what you'd expect: loads the data from the right files into a consistently-located database. --- pokedex/__init__.py | 206 ++++++++++--------------------------------------- pokedex/db/__init__.py | 13 +++- pokedex/db/load.py | 197 ++++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 249 insertions(+), 167 deletions(-) create mode 100644 pokedex/db/load.py diff --git a/pokedex/__init__.py b/pokedex/__init__.py index 1e6c6d1..df995ab 100644 --- a/pokedex/__init__.py +++ b/pokedex/__init__.py @@ -1,15 +1,14 @@ # encoding: utf8 +from optparse import OptionParser import sys -from sqlalchemy.exc import IntegrityError -import sqlalchemy.types - -from .db import connect, metadata, tables as tables_module +from .db import connect, metadata +import pokedex.db.load from pokedex.lookup import lookup as pokedex_lookup def main(): if len(sys.argv) <= 1: - help() + command_help() command = sys.argv[1] args = sys.argv[2:] @@ -22,161 +21,27 @@ def main(): command_help() -def command_csvimport(engine_uri, directory='.'): - import csv +def command_dump(*args): + parser = OptionParser() + parser.add_option('-e', '--engine', dest='engine_uri', default=None) + parser.add_option('-d', '--directory', dest='directory', default=None) + options, _ = parser.parse_args(list(args)) - from sqlalchemy.orm.attributes import instrumentation_registry + session = connect(options.engine_uri) + pokedex.db.load.dump(session, directory=options.directory) - session = connect(engine_uri) +def command_load(*args): + parser = OptionParser() + parser.add_option('-e', '--engine', dest='engine_uri', default=None) + parser.add_option('-d', '--directory', dest='directory', default=None) + parser.add_option('-D', '--drop-tables', dest='drop_tables', default=False, action='store_true') + options, _ = parser.parse_args(list(args)) - metadata.create_all() - - # SQLAlchemy is retarded and there is no way for me to get a list of ORM - # classes besides to inspect the module they all happen to live in for - # things that look right. - table_base = tables_module.TableBase - orm_classes = {} # table object => table class - - for name in dir(tables_module): - # dir() returns strings! How /convenient/. - thingy = getattr(tables_module, name) - - if not isinstance(thingy, type): - # Not a class; bail - continue - elif not issubclass(thingy, table_base): - # Not a declarative table; bail - continue - elif thingy == table_base: - # Declarative table base, so not a real table; bail - continue - - # thingy is definitely a table class! Hallelujah. - orm_classes[thingy.__table__] = thingy - - # Okay, run through the tables and actually load the data now - for table_obj in metadata.sorted_tables: - table_class = orm_classes[table_obj] - table_name = table_obj.name - - # Print the table name but leave the cursor in a fixed column - print table_name + '...', ' ' * (40 - len(table_name)), - sys.stdout.flush() - - try: - csvfile = open("%s/%s.csv" % (directory, table_name), 'rb') - except IOError: - # File doesn't exist; don't load anything! - print 'no data!' - continue - - reader = csv.reader(csvfile, lineterminator='\n') - column_names = [unicode(column) for column in reader.next()] - - # Self-referential tables may contain rows with foreign keys of other - # rows in the same table that do not yet exist. Pull these out and add - # them to the session last - # ASSUMPTION: Self-referential tables have a single PK called "id" - deferred_rows = [] # ( row referring to id, [foreign ids we need] ) - seen_ids = {} # primary key we've seen => 1 - - # Fetch foreign key columns that point at this table, if any - self_ref_columns = [] - for column in table_obj.c: - if any(_.references(table_obj) for _ in column.foreign_keys): - self_ref_columns.append(column) - - for csvs in reader: - row = table_class() - - for column_name, value in zip(column_names, csvs): - column = table_obj.c[column_name] - if column.nullable and value == '': - # Empty string in a nullable column really means NULL - value = None - elif isinstance(column.type, sqlalchemy.types.Boolean): - # Boolean values are stored as string values 0/1, but both - # of those evaluate as true; SQLA wants True/False - if value == '0': - value = False - else: - value = True - else: - # Otherwise, unflatten from bytes - value = value.decode('utf-8') - - setattr(row, column_name, value) - - # May need to stash this row and add it later if it refers to a - # later row in this table - if self_ref_columns: - foreign_ids = [getattr(row, _.name) for _ in self_ref_columns] - foreign_ids = [_ for _ in foreign_ids if _] # remove NULL ids - - if not foreign_ids: - # NULL key. Remember this row and add as usual. - seen_ids[row.id] = 1 - - elif all(_ in seen_ids for _ in foreign_ids): - # Non-NULL key we've already seen. Remember it and commit - # so we know the old row exists when we add the new one - session.commit() - seen_ids[row.id] = 1 - - else: - # Non-NULL future id. Save this and insert it later! - deferred_rows.append((row, foreign_ids)) - continue - - session.add(row) - - session.commit() - - # Attempt to add any spare rows we've collected - for row, foreign_ids in deferred_rows: - if not all(_ in seen_ids for _ in foreign_ids): - # Could happen if row A refers to B which refers to C. - # This is ridiculous and doesn't happen in my data so far - raise ValueError("Too many levels of self-reference! " - "Row was: " + str(row.__dict__)) - - session.add(row) - seen_ids[row.id] = 1 - session.commit() - - print 'loaded' - -def command_csvexport(engine_uri, directory='.'): - import csv - session = connect(engine_uri) + session = connect(options.engine_uri) + + pokedex.db.load.load(session, directory=options.directory, + drop_tables=options.drop_tables) - for table_name in sorted(metadata.tables.keys()): - print table_name - table = metadata.tables[table_name] - - writer = csv.writer(open("%s/%s.csv" % (directory, table_name), 'wb'), - lineterminator='\n') - columns = [col.name for col in table.columns] - writer.writerow(columns) - - primary_key = table.primary_key - for row in session.query(table).order_by(*primary_key).all(): - csvs = [] - for col in columns: - # Convert Pythony values to something more universal - val = getattr(row, col) - if val == None: - val = '' - elif val == True: - val = '1' - elif val == False: - val = '0' - else: - val = unicode(val).encode('utf-8') - - csvs.append(val) - - writer.writerow(csvs) def command_lookup(engine_uri, name): # XXX don't require uri! somehow @@ -194,16 +59,25 @@ def command_lookup(engine_uri, name): def command_help(): print u"""pokedex -- a command-line Pokédex interface - - help Displays this message. - lookup {uri} [name] Look up something in the Pokédex. - - These commands are only useful for developers: - csvimport {uri} [dir] Import data from a set of CSVs to the database - given by the URI. - csvexport {uri} [dir] Export data from the database given by the URI - to a set of CSVs. - Directory defaults to cwd. +usage: pokedex {command} [options...] +Run `pokedex setup` first, or nothing will work! + +Commands: + help Displays this message. + lookup [thing] Look up something in the Pokédex. + +System commands: + load Load Pokédex data into a database from CSV files. + dump Dump Pokédex data from a database into CSV files. + +Options: + -d|--directory By default, load and dump will use the CSV files in the + pokedex install directory. Use this option to specify + a different directory. + -D|--drop-tables With load, drop all tables before loading data. + -e|--engine=URI By default, all commands try to use a SQLite database + in the pokedex install directory. Use this option to + specify an alternate database. """.encode(sys.getdefaultencoding(), 'replace') sys.exit(0) diff --git a/pokedex/db/__init__.py b/pokedex/db/__init__.py index ced519b..99e058b 100644 --- a/pokedex/db/__init__.py +++ b/pokedex/db/__init__.py @@ -1,13 +1,24 @@ +import pkg_resources + from sqlalchemy import MetaData, Table, create_engine, orm from .tables import metadata -def connect(uri, **kwargs): +def connect(uri=None, **kwargs): """Connects to the requested URI. Returns a session object. + With the URI omitted, attempts to connect to a default SQLite database + contained within the package directory. + Calling this function also binds the metadata object to the created engine. """ + # Default to a URI within the package, which was hopefully created at some point + if not uri: + sqlite_path = pkg_resources.resource_filename('pokedex', + 'data/pokedex.sqlite') + uri = 'sqlite:///' + sqlite_path + ### Do some fixery for MySQL if uri[0:5] == 'mysql': # MySQL uses latin1 for connections by default even if the server is diff --git a/pokedex/db/load.py b/pokedex/db/load.py new file mode 100644 index 0000000..1cff39d --- /dev/null +++ b/pokedex/db/load.py @@ -0,0 +1,197 @@ +"""CSV to database or vice versa.""" +import csv +import pkg_resources +import sys + +from sqlalchemy.orm.attributes import instrumentation_registry +import sqlalchemy.types + +from pokedex.db import metadata +import pokedex.db.tables as tables + + +def load(session, directory=None, drop_tables=False): + """Load data from CSV files into the given database session. + + Tables are created automatically. + + `session` + SQLAlchemy session to use. + + `directory` + Directory the CSV files reside in. Defaults to the `pokedex` data + directory. + + `drop_tables` + If set to True, existing `pokedex`-related tables will be dropped. + """ + + if not directory: + directory = pkg_resources.resource_filename('pokedex', 'data/csv') + + # Drop all tables if requested + if options.drop_tables: + print 'Dropping tables...' + metadata.drop_all() + + metadata.create_all() + + # SQLAlchemy is retarded and there is no way for me to get a list of ORM + # classes besides to inspect the module they all happen to live in for + # things that look right. + table_base = tables.TableBase + orm_classes = {} # table object => table class + + for name in dir(tables): + # dir() returns strings! How /convenient/. + thingy = getattr(tables, name) + + if not isinstance(thingy, type): + # Not a class; bail + continue + elif not issubclass(thingy, table_base): + # Not a declarative table; bail + continue + elif thingy == table_base: + # Declarative table base, so not a real table; bail + continue + + # thingy is definitely a table class! Hallelujah. + orm_classes[thingy.__table__] = thingy + + # Okay, run through the tables and actually load the data now + for table_obj in metadata.sorted_tables: + table_class = orm_classes[table_obj] + table_name = table_obj.name + + # Print the table name but leave the cursor in a fixed column + print table_name + '...', ' ' * (40 - len(table_name)), + sys.stdout.flush() + + try: + csvfile = open("%s/%s.csv" % (directory, table_name), 'rb') + except IOError: + # File doesn't exist; don't load anything! + print 'no data!' + continue + + reader = csv.reader(csvfile, lineterminator='\n') + column_names = [unicode(column) for column in reader.next()] + + # Self-referential tables may contain rows with foreign keys of other + # rows in the same table that do not yet exist. Pull these out and add + # them to the session last + # ASSUMPTION: Self-referential tables have a single PK called "id" + deferred_rows = [] # ( row referring to id, [foreign ids we need] ) + seen_ids = {} # primary key we've seen => 1 + + # Fetch foreign key columns that point at this table, if any + self_ref_columns = [] + for column in table_obj.c: + if any(_.references(table_obj) for _ in column.foreign_keys): + self_ref_columns.append(column) + + for csvs in reader: + row = table_class() + + for column_name, value in zip(column_names, csvs): + column = table_obj.c[column_name] + if column.nullable and value == '': + # Empty string in a nullable column really means NULL + value = None + elif isinstance(column.type, sqlalchemy.types.Boolean): + # Boolean values are stored as string values 0/1, but both + # of those evaluate as true; SQLA wants True/False + if value == '0': + value = False + else: + value = True + else: + # Otherwise, unflatten from bytes + value = value.decode('utf-8') + + setattr(row, column_name, value) + + # May need to stash this row and add it later if it refers to a + # later row in this table + if self_ref_columns: + foreign_ids = [getattr(row, _.name) for _ in self_ref_columns] + foreign_ids = [_ for _ in foreign_ids if _] # remove NULL ids + + if not foreign_ids: + # NULL key. Remember this row and add as usual. + seen_ids[row.id] = 1 + + elif all(_ in seen_ids for _ in foreign_ids): + # Non-NULL key we've already seen. Remember it and commit + # so we know the old row exists when we add the new one + session.commit() + seen_ids[row.id] = 1 + + else: + # Non-NULL future id. Save this and insert it later! + deferred_rows.append((row, foreign_ids)) + continue + + session.add(row) + + session.commit() + + # Attempt to add any spare rows we've collected + for row, foreign_ids in deferred_rows: + if not all(_ in seen_ids for _ in foreign_ids): + # Could happen if row A refers to B which refers to C. + # This is ridiculous and doesn't happen in my data so far + raise ValueError("Too many levels of self-reference! " + "Row was: " + str(row.__dict__)) + + session.add(row) + seen_ids[row.id] = 1 + session.commit() + + print 'loaded' + + + +def dump(session, directory=None): + """Dumps the contents of a database to a set of CSV files. Probably not + useful to anyone besides a developer. + + `session` + SQLAlchemy session to use. + + `directory` + Directory the CSV files should be put in. Defaults to the `pokedex` + data directory. + """ + + if not directory: + directory = pkg_resources.resource_filename('pokedex', 'data/csv') + + for table_name in sorted(metadata.tables.keys()): + print table_name + table = metadata.tables[table_name] + + writer = csv.writer(open("%s/%s.csv" % (directory, table_name), 'wb'), + lineterminator='\n') + columns = [col.name for col in table.columns] + writer.writerow(columns) + + primary_key = table.primary_key + for row in session.query(table).order_by(*primary_key).all(): + csvs = [] + for col in columns: + # Convert Pythony values to something more universal + val = getattr(row, col) + if val == None: + val = '' + elif val == True: + val = '1' + elif val == False: + val = '0' + else: + val = unicode(val).encode('utf-8') + + csvs.append(val) + + writer.writerow(csvs) -- 2.7.4