X-Git-Url: http://git.veekun.com/zzz-pokedex.git/blobdiff_plain/b3b3285a703226e551e34a89e59a0c6c4e97b16f..1136f6cf582e59265f6ce096b269c7403c918296:/pokedex/__init__.py diff --git a/pokedex/__init__.py b/pokedex/__init__.py index 1e6c6d1..393979c 100644 --- a/pokedex/__init__.py +++ b/pokedex/__init__.py @@ -1,19 +1,26 @@ # encoding: utf8 +from optparse import OptionParser +import os import sys -from sqlalchemy.exc import IntegrityError -import sqlalchemy.types - -from .db import connect, metadata, tables as tables_module -from pokedex.lookup import lookup as pokedex_lookup +import pokedex.db +import pokedex.db.load +import pokedex.db.tables +import pokedex.lookup +from pokedex import defaults 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 or 'utf8' + args = [_.decode(enc) for _ in args] + # Find the command as a function in this file func = globals().get("command_%s" % command, None) if func: @@ -22,188 +29,254 @@ def main(): command_help() -def command_csvimport(engine_uri, directory='.'): - import csv - - from sqlalchemy.orm.attributes import instrumentation_registry - - session = connect(engine_uri) - - 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) - - 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 - session = connect(engine_uri) - - results, exact = pokedex_lookup(session, name) - if exact: +def get_parser(verbose=True): + """Returns an OptionParser prepopulated with the global options. + + `verbose` is whether or not the options should be verbose by default. + """ + parser = OptionParser() + parser.add_option('-e', '--engine', dest='engine_uri', default=None) + parser.add_option('-i', '--index', dest='index_dir', default=None) + parser.add_option('-q', '--quiet', dest='verbose', default=verbose, action='store_false') + parser.add_option('-v', '--verbose', dest='verbose', default=verbose, action='store_true') + return parser + +def get_session(options): + """Given a parsed options object, connects to the database and returns a + session. + """ + + engine_uri = options.engine_uri + got_from = 'command line' + + if engine_uri is None: + engine_uri, got_from = defaults.get_default_db_uri_with_origin() + + session = pokedex.db.connect(engine_uri) + + if options.verbose: + print "Connected to database {engine} (from {got_from})" \ + .format(engine=session.bind.url, got_from=got_from) + + return session + +def get_lookup(options, session=None, recreate=False): + """Given a parsed options object, opens the whoosh index and returns a + PokedexLookup object. + """ + + if recreate and not session: + raise ValueError("get_lookup() needs an explicit session to regen the index") + + index_dir = options.index_dir + got_from = 'command line' + + if index_dir is None: + index_dir, got_from = defaults.get_default_index_dir_with_origin() + + if options.verbose: + print "Opened lookup index {index_dir} (from {got_from})" \ + .format(index_dir=index_dir, got_from=got_from) + + lookup = pokedex.lookup.PokedexLookup(index_dir, session=session) + + if recreate: + lookup.rebuild_index() + + return lookup + +def get_csv_directory(options): + """Prints and returns the csv directory we're about to use.""" + + if not options.verbose: + return + + csvdir = options.directory + got_from = 'command line' + + if csvdir is None: + csvdir, got_from = defaults.get_default_csv_dir_with_origin() + + print "Using CSV directory {csvdir} (from {got_from})" \ + .format(csvdir=csvdir, got_from=got_from) + + return csvdir + + +### Plumbing commands + +def command_dump(*args): + parser = get_parser(verbose=True) + parser.add_option('-d', '--directory', dest='directory', default=None) + options, tables = parser.parse_args(list(args)) + + session = get_session(options) + get_csv_directory(options) + + pokedex.db.load.dump(session, directory=options.directory, + tables=tables, + verbose=options.verbose) + +def command_load(*args): + parser = get_parser(verbose=True) + parser.add_option('-d', '--directory', dest='directory', default=None) + parser.add_option('-D', '--drop-tables', dest='drop_tables', default=False, action='store_true') + options, tables = parser.parse_args(list(args)) + + if not options.engine_uri: + print "WARNING: You're reloading the default database, but not the lookup index. They" + print " might get out of sync, and pokedex commands may not work correctly!" + print "To fix this, run `pokedex reindex` when this command finishes. Or, just use" + print "`pokedex setup` to do both at once." + print + + session = get_session(options) + get_csv_directory(options) + + pokedex.db.load.load(session, directory=options.directory, + drop_tables=options.drop_tables, + tables=tables, + verbose=options.verbose) + +def command_reindex(*args): + parser = get_parser(verbose=True) + options, _ = parser.parse_args(list(args)) + + session = get_session(options) + lookup = get_lookup(options, session=session, recreate=True) + + print "Recreated lookup index." + + +def command_setup(*args): + parser = get_parser(verbose=False) + options, _ = parser.parse_args(list(args)) + + options.directory = None + + session = get_session(options) + get_csv_directory(options) + pokedex.db.load.load(session, directory=None, drop_tables=True, + verbose=options.verbose) + + lookup = get_lookup(options, session=session, recreate=True) + + print "Recreated lookup index." + + +def command_status(*args): + parser = get_parser(verbose=True) + options, _ = parser.parse_args(list(args)) + options.verbose = True + options.directory = None + + # Database, and a lame check for whether it's been inited at least once + session = get_session(options) + print " - OK! Connected successfully." + + if pokedex.db.tables.Pokemon.__table__.exists(session.bind): + print " - OK! Database seems to contain some data." + else: + print " - WARNING: Database appears to be empty." + + # CSV; simple checks that the dir exists + csvdir = get_csv_directory(options) + if not os.path.exists(csvdir): + print " - ERROR: No such directory!" + elif not os.path.isdir(csvdir): + print " - ERROR: Not a directory!" + else: + print " - OK! Directory exists." + + if os.access(csvdir, os.R_OK): + print " - OK! Can read from directory." + else: + print " - ERROR: Can't read from directory!" + + if os.access(csvdir, os.W_OK): + print " - OK! Can write to directory." + else: + print " - WARNING: Can't write to directory! " \ + "`dump` will not work. You may need to sudo." + + # Index; the PokedexLookup constructor covers most tests and will + # cheerfully bomb if they fail + lookup = get_lookup(options, recreate=False) + print " - OK! Opened successfully." + + +### User-facing commands + +def command_lookup(*args): + parser = get_parser(verbose=False) + options, words = parser.parse_args(list(args)) + + name = u' '.join(words) + + session = get_session(options) + lookup = get_lookup(options, session=session, recreate=False) + + results = lookup.lookup(name) + if not results: + print "No matches." + elif results[0].exact: print "Matched:" else: print "Fuzzy-matched:" - for object in results: - print object.__tablename__, object.name + for result in results: + if hasattr(result.object, 'full_name'): + name = result.object.full_name + else: + name = result.object.name + + print "%s: %s" % (result.object.__tablename__, name), + if result.language: + print "(%s in %s)" % (result.name, result.language) + else: + print 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! +See http://bugs.veekun.com/projects/pokedex/wiki/CLI for more documentation. + +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. + reindex Rebuilds the lookup index from the database. + setup Combines load and reindex. + status No effect, but prints which engine, index, and csv + directory would be used for other commands. + +Global options: + -e|--engine=URI By default, all commands try to use a SQLite database + in the pokedex install directory. Use this option (or + a POKEDEX_DB_ENGINE environment variable) to specify an + alternate database. + -i|--index=DIR By default, all commands try to put the lookup index in + the pokedex install directory. Use this option (or a + POKEDEX_INDEX_DIR environment variable) to specify an + alternate loction. + -q|--quiet Don't print system output. This is the default for + non-system commands and setup. + -v|--verbose Print system output. This is the default for system + commands, except setup. + +System options: + -d|--directory=DIR 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. + + Additionally, load and dump accept a list of table names (possibly with + wildcards) and/or csv fileames as an argument list. """.encode(sys.getdefaultencoding(), 'replace') sys.exit(0)