com.conveyal.gtfs

Class GTFS

java.lang.Object

com.conveyal.gtfs.GTFS

public abstract class GTFS
extends Object

This is the public interface to the RDBMS backed functionality in gtfs-lib. To avoid coupling our other projects too closely to the internal implementation of GTFS-lib, we should only use GTFS-lib via methods in this class and the data model objects in the model package.

Constructor Summary

Constructors

Constructor and Description
GTFS()

Method Summary

Modifier and Type	Method and Description
static DataSource	createDataSource(String url, String username, String password) Create an automatically managed pool of database connections to the supplied JDBC database URL.
static void	delete(String feedId, DataSource dataSource) Deletes all tables for the specified feed.
static FeedLoadResult	export(String feedId, String outFile, DataSource dataSource, boolean fromEditor) Export a feed ID from the database to a zipped GTFS file in the specified export directory.
static FeedLoadResult	load(String filePath, DataSource dataSource) Load the GTFS data in the specified file into the given JDBC DataSource.
static void	main(String[] args) A command-line interface that lets you load GTFS feeds into a database and validate the loaded feeds.
static SnapshotResult	makeSnapshot(String feedId, DataSource dataSource) Copy all tables for a given feed ID (schema namespace) into a new namespace in the given JDBC DataSource.
static ValidationResult	validate(String feedId, DataSource dataSource) Once a feed has been loaded into the database, examine its contents looking for various problems and errors.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

GTFS

public GTFS()

Method Detail

export

public static FeedLoadResult export(String feedId,
                                    String outFile,
                                    DataSource dataSource,
                                    boolean fromEditor)

Export a feed ID from the database to a zipped GTFS file in the specified export directory.

load

public static FeedLoadResult load(String filePath,
                                  DataSource dataSource)

Load the GTFS data in the specified file into the given JDBC DataSource. About loading from an object on S3: We've got a problem with keeping the identifier on S3 the same as the identifier in the database. Question: do we really want to load the feed from S3 or do we just want to move the local file to S3 when done? i.e. user uploads to the datatools back end, The advantage of having the end users upload to s3, then is that only the files currently being processed need to be held on our backend server, not the whole queue waiting to be processed. But I think currently we have some hack where the end user gets a token to upload to S3, which seems bad. Probably we should have end users upload directly to the backend server, which then loads the feed into RDBMS, uploads the file to S3 with the unique name supplied by gtfs-lib, and deletes the local file.

Returns:

a unique identifier for the newly loaded feed in the database

makeSnapshot

public static SnapshotResult makeSnapshot(String feedId,
                                          DataSource dataSource)

Copy all tables for a given feed ID (schema namespace) into a new namespace in the given JDBC DataSource. The resulting snapshot from this operation is intended to be edited, so there are a handful of changes made to the newly copied tables: 1. The tables' id column has been modified to be auto-incrementing. 2. Primary keys may be added to certain columns/tables. 3. Additional editor-specific columns are added to certain tables.

Parameters:

feedId - feed ID (schema namespace) to copy from

dataSource - JDBC connection to existing database

Returns:

FIXME should this be a separate SnapshotResult object?

validate

public static ValidationResult validate(String feedId,
                                        DataSource dataSource)

Once a feed has been loaded into the database, examine its contents looking for various problems and errors.

delete

public static void delete(String feedId,
                          DataSource dataSource)
                   throws SQLException,
                          InvalidNamespaceException

Deletes all tables for the specified feed. Simply put, this is a "drop schema" SQL statement called on the feed's namespace.

Throws:

SQLException

InvalidNamespaceException

createDataSource

public static DataSource createDataSource(String url,
                                          String username,
                                          String password)

Create an automatically managed pool of database connections to the supplied JDBC database URL. Creating new database connections is usually an expensive operation, taking up to seconds to prepare for a transaction that only lasts milliseconds. The JDBC in Java 7 and 8 support connection pooling, which is critical for performance. We are using the standard JDBC DataSource interface, with the Apache commons-dbcp implementation. This should be roughly the same connection pooling used in the Tomcat application server. Here are some sample database URLs to serve as patterns: H2_FILE_URL = "jdbc:h2:file:~/test-db"; // H2 memory does not seem faster than H2 file SQLITE_FILE_URL = "jdbc:sqlite:/Users/abyrd/test-db"; POSTGRES_LOCAL_URL = "jdbc:postgresql://localhost/catalogue"; For local Postgres connections, you can supply a null username and password to use host-based authentication.

main

public static void main(String[] args)
                 throws IOException

A command-line interface that lets you load GTFS feeds into a database and validate the loaded feeds. It also lets you run a GraphQL API for all the feeds loaded into the database.

Throws:

IOException