X-Git-Url: http://git.veekun.com/zzz-pokedex.git/blobdiff_plain/638bc0eb815f3cb8bff0bfc027d4d0d931ab7273..204a43b3b15946a64b971966706fba713065b7ea:/pokedex/__init__.py?ds=sidebyside diff --git a/pokedex/__init__.py b/pokedex/__init__.py index 02b8ce5..ccffaf2 100644 --- a/pokedex/__init__.py +++ b/pokedex/__init__.py @@ -1,176 +1,109 @@ # 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 +import pokedex.lookup def main(): if len(sys.argv) <= 1: - help() + command_help() command = sys.argv[1] args = sys.argv[2:] + # XXX there must be a better way to get Unicode argv + # XXX this doesn't work on Windows durp + enc = sys.stdin.encoding + args = [_.decode(enc) for _ in args] + # Find the command as a function in this file - func = globals().get(command, None) - if func and callable(func) and command != 'main': + func = globals().get("command_%s" % command, None) + if func: func(*args) else: - help() - - -def csvimport(engine_uri, directory='.'): - import csv - - from sqlalchemy.orm.attributes import instrumentation_registry - - # Use autocommit in case rows fail due to foreign key incest - session = connect(engine_uri, autocommit=True, autoflush=False) - - 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. We'll keep - # a running list of these and try inserting them again after the - # rest are done - failed_rows = [] - - 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) - - try: - session.add(row) - session.flush() - except IntegrityError as e: - failed_rows.append(row) - - # Loop over the failed rows and keep trying to insert them. If a loop - # doesn't manage to insert any rows, bail. - do_another_loop = True - while failed_rows and do_another_loop: - do_another_loop = False - - for i, row in enumerate(failed_rows): - try: - session.add(row) - session.flush() - - # Success! - del failed_rows[i] - do_another_loop = True - except IntegrityError as e: - pass - - if failed_rows: - print len(failed_rows), "rows failed" - else: - print 'loaded' - -def csvexport(engine_uri, directory='.'): - import csv - session = connect(engine_uri) - - 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) - - for row in session.query(table).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') + command_help() + + +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) + parser.add_option('-q', '--quiet', dest='verbose', default=True, action='store_false') + options, _ = parser.parse_args(list(args)) + + session = connect(options.engine_uri) + pokedex.db.load.dump(session, directory=options.directory, + verbose=options.verbose) + +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') + parser.add_option('-q', '--quiet', dest='verbose', default=True, action='store_false') + options, _ = parser.parse_args(list(args)) + + session = connect(options.engine_uri) + + pokedex.db.load.load(session, directory=options.directory, + drop_tables=options.drop_tables, + verbose=options.verbose) + +def command_setup(*args): + session = connect() + pokedex.db.load.load(session, verbose=False, drop_tables=True) + pokedex.lookup.open_index(session=session, recreate=True) + + +def command_lookup(name): + results = pokedex.lookup.lookup(name) + if not results: + print "No matches." + elif results[0].exact: + print "Matched:" + else: + print "Fuzzy-matched:" - csvs.append(val) + for result in results: + if hasattr(result.object, 'full_name'): + name = result.object.full_name + else: + name = result.object.name - writer.writerow(csvs) + print "%s: %s" % (result.object.__tablename__, name), + if result.language: + print "(%s in %s)" % (result.name, result.language) + else: + print -def help(): +def command_help(): print u"""pokedex -- a command-line Pokédex interface - - help Displays this message. - - 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. + setup Loads Pokédex data into the right place and creates a + lookup index in the right place. No options or output. + This will blow away the default database and index! + +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. + -q|--quiet Turn off any unnecessary status output from dump/load. +""".encode(sys.getdefaultencoding(), 'replace') sys.exit(0)