tilecloud 0.5.0.dev6

Tools for managing tiles

TileCloud

A powerful utility for generating, managing, transforming, and visualizing map tiles in multiple formats.

Introduction

TileCloud is a powerful utility for generating, managing, transforming, visualising and map tiles in multiple formats. It can create, read update, delete tiles in multiple back ends, called TileStores. Existing TileStores include:

• HTTP/REST in any layout

• WMTS

• Amazon S3 and SQS

• MBTiles

• TileJSON

• Mapnik (via mapnik2)

• Memcached

• Local file system

• Log files in any format

TileCloud is not limited to image tiles, it can also handle other tile data such as UTFGrid, or elevation data in JSON format.

TileCloud uses Python’s generators and iterators to efficiently stream tens of millions of tiles, and can handle multiple tiles in parallel using Python’s multiprocessing library.

Example tasks that TileCloud makes easy include:

• Visualize tiles stored in any TileStore with OpenLayers, Google Maps, jQuery Geo, Leaflet, Polymaps, Modest Maps, and OpenWebGlobe.

• Convert sixty million PNG tiles stored in S3 to JPEG format with different quality settings at different zoom levels.

• Transform image formats and perform arbitrary image transformations on the fly, including PNG optimization.

• Generate semi-transparent tiles with embedded tile coordinates for debugging.

• Pack multiple tile layers into a single tile on the server.

• Efficiently calculate bounding boxes and detect missing tiles in existing tile datasets.

• Simulate fast and slow tile servers.

• Efficiently delete millions of tiles in S3.

• Read JSON tiles from a tarball, compress them, and upload them.

Getting started

TileCloud depends on some Python modules. It’s easiest to install them with pip in a virtualenv:

$ virtualenv .
$ . bin/activate
$ pip install -r requirements.txt

For a quick demo, run

$ ./tc-viewer --root=3/4/2 'http://gsp2.apple.com/tile?api=1&style=slideshow&layers=default&lang=en_GB&z=%(z)d&x=%(x)d&y=%(y)d&v=9'

and point your browser at http://localhost:8080/. Type Ctrl-C to terminate tc-viewer.

Next, download an example MBTiles file from MapBox, such as Geography Class. We can quickly find out more about this tile set with the tc-info command. For example, to count the number of tiles:

$ ./tc-info -t count geography-class.mbtiles
87381

To calculate the bounding pyramid:

$ ./tc-info -t bounding-pyramid -r geography-class.mbtiles
0/0/0:+1/+1
1/0/0:+2/+2
2/0/0:+4/+4
3/0/0:+8/+8
4/0/0:+16/+16
5/0/0:+32/+32
6/0/0:+64/+64
7/0/0:+128/+128
8/0/0:+256/+256

To check for missing tiles against a bounding pyramid:

$ ./tc-info -b 0/0/0:8/*/* -t completion geography-class.mbtiles
0 1/1 (100%)
1 4/4 (100%)
2 16/16 (100%)
3 64/64 (100%)
4 256/256 (100%)
5 1024/1024 (100%)
6 4096/4096 (100%)
7 16384/16384 (100%)
8 65536/65536 (100%)

This shows, for each zoom level, the number of tiles at that zoom level, the total number of tiles expected at that zoom level for the specified bounding pyramid (0/0/0:8/*/* means all tiles from level 0 to level 8), and a percentage completion. This can be useful for checking that a tile set is complete.

Now, display this MBTiles tile set on top of the OpenStreetMap tiles and a debug tile layer:

$ ./tc-viewer tiles.openstreetmap_org geography-class.mbtiles tiles.debug.black

You’ll need to point your browser at http://localhost:8080/ and choose your favourite library.

tc-info and tc-viewer are utility programs. Normally you use TileCloud by writing short Python programs that connect the TileCloud’s modules to perform the action that you want.

As a first example, run the following:

$ PYTHONPATH=. examples/download.py

This will download a few tiles from OpenStreetMap and save them in a local MBTiles file called local.mbtiles. Look at the source code to examples/download.py to see how it works. If there are problems with the download, just interrupt it with Ctrl-C and re-run it: the program will automatically resume where it left off.

Once you have downloaded a few tiles, you can view them directly with tc-viewer:

$ ./tc-viewer --root=4/8/5 local.mbtiles tiles.debug.black

Point your browser at http://localhost:8080 as usual. The --root option to tc-viewer instructs the viewer to start at a defined tile, rather than at 0/0/0, so you don’t have to zoom in to find the tiles that you downloaded.

Tile coordinates, tile layouts, tile grids and bounding pyramids

TileCloud always represents tile coordinates as strings like z/x/y. TileCloud primarily works in tile coordinates, although geographic coordinates can be used in some places.

Tile layouts convert tile coordinates to and from strings for use in paths, URLs, keys, etc.

Tile grids are used to convert tile coordinates to and from geographic coordinates, and to relate tiles with different z values.

Bounding pyramids represent a range of tiles in the x, y and z directions. The format is basically minz/minx/miny:maxz/maxx/maxy but maxz is optional and maxz, maxx and maxy can be prefixed with an + sign to indicate that they are relative to the corresponding min value. This is probably best demonstrated by a few examples:

4/10/20:15/25

This corresponds to a range of tiles with z=4, x=10..15 and y=20..25

4/10/20:+5/+5

This is the same range (z=4, x=10..15, y=20..25) but expressed using relative sizes.

4/10/20:5/15/20

This is the same range of tiles above, but also includes all the tiles at level z=5 which overlap the above range. TileCloud uses the tile grid to calculate which tiles from level z=5 to include.

4/10/20:+1/+5/+5

This represents the range same as the previous example using a relative maxz.

Quick tile generation

The tc-copy command can be used to copy tiles between different TileStores. If a TileStore has the side effect of generating tiles, then it functions as a quick tile generation utility. First, some quick examples.

To convert from one tile format to another, just copy from source to destination. For example, to convert an MBTiles file in to a ZIP file, just run:

$ ./tc-copy geography-class.mbtiles geography-class.zip

You can check this worked with unzip:

$ unzip -t geography-class.zip

Equally, tc-copy can be used to download multiple tiles:

$ ./tc-copy --bounding-pyramid 4/0/0:0/16/16 tiles.openstreetmap_org osm-up-to-z4.mbtiles

Here we downloaded zoom levels 0 to 4 of the OpenStreetMap tiles into a local MBTiles file. The --bounding-pyramid option is required because otherwise we would download all OpenStreetMap tiles – which might take some time (and also contravene OpenStreetMap’s tile usage policy). Note that, by default, tc-copy won’t overwrite tiles if they already exist in the destination. This means that you can interrupt the above command and restart it, and it will resume where it was interrupted. If you want to overwrite tiles in the destination then pass the --overwrite option to tc-copy.

In the same way, tc-copy can also be used to upload tiles. For example, to upload an MBTiles file to S3, just use:

$ ./tc-copy osm-up-to-z4.mbtiles s3://bucket/prefix/%(z)d/%(x)d/%(y)d.jpg

bucket is the name of your S3 bucket. You’ll need to have set the AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY environment variables to have permission to upload to S3. The rest of the destination (prefix/%(z)d/%(x)d/%(y)d.jpg) is a template describing the layout of the tiles in S3. It’s a normal Python format string: %(x)d means substitute the tile’s x coordinate as a decimal integer.

You can pass the same s3:// URL to tc-viewer. This allows you to visualise your tiles stored in S3 with your favourite mapping library. For example:

$ ./tc-viewer s3://bucket/prefix/%(z)d/%(x)d/%(y)d.jpg

Here, tc-viewer is acting as a proxy, serving tiles stored in S3 over HTTP, bypassing any caches or access controls (assuming you have the correct credentials, of course). This allows you to visualize the exact tiles that you’ve stored.

Rendering the World

At FOSS4G-NA, MapBox presented an excellent strategy for rendering the world. TileCloud supports the subdivision strategy. To run the demo, execute:

$ python examples/renderingtheworld.py

This will generate tiles from a WMTS tile server and save them in a local MBTiles tiles. When the above command is complete, you can see the bounding pyramid of the generated tiles:

$ ./tc-info -t bounding-pyramid -r medford_buildings.mbtiles
0/0/0:+1/+1
1/0/0:+1/+1
2/0/1:+1/+1
3/1/2:+1/+1
4/2/5:+1/+1
5/5/11:+1/+1
6/10/23:+1/+1
7/20/47:+1/+1
8/40/94:+2/+2
9/80/189:+2/+1
10/162/378:+1/+2
11/324/757:+2/+2
12/649/1514:+3/+3
13/1299/3028:+4/+5
14/2598/6057:+7/+8
15/5196/12114:+13/+15

You can look at these tiles (which show buildings in Medford, OR) with the command:

./tc-viewer --root=7/20/47 tiles.openstreetmap_org medford_buildings.mbtiles

A cheap-and-cheerful tile server

tc-viewer can be used as a lightweight tile server, which can be useful for development, debugging and off-line demos. The TileStores passed as arguments to tc-viewer are available at the URL:

http://localhost:8080/tiles/{index}/tiles/{z}/{x}/{y}

where {index} is the index of the TileStore on the command line (starting from 0 for the first tile store), and {z}, {x} and {y} are the components of the tile coordinate. The second tiles in the URL is present to work around assumptions made by OpenWebGlobe. This layout is directly usable by most mapping libraries, see the code in views/*.tpl for examples. The host and port can be set with the --host and --port command line options, respectively.

Note that there is no file extension. tc-viewer will automatically set the correct content type and content encoding headers if it can determine them, and, failing this, most browsers will figure it out for themselves.

For convenience, tc-viewer serves everything in the static directory under the URL /static. This can be used to serve your favourite mapping library and/or application code directly for testing purposes.

By default, tc-viewer will use Tornado as a web server, if it is available, otherwise it will fall back to WSGIRef. Tornado has reasonably good performance, and is adequate for local development and off-line demos, especially when used with a MBTiles TileStore. WSGIRef has very poor performance (it handles only one request at a time) and as such can be used as a “slow” tile server, ideal for debugging tile loading code or testing how your web application performs over a slow network connection. tc-viewer is particularly slow when used to proxy tiles being served by a remote server. You can set the server explicitly with the --server option.

tc-viewer sets the Access-Control-Allow-Origin header to * for all the tiles it serves, this allows the tiles to be used as textures for WebGL applications running on different hosts/ports. For more information, see Cross-Domain Textures.

tc-viewer is designed as a development tool, and the power that it offers comes at the expense of fragility. It makes many assumptions - including the benevolence of the user - that make it entirely unsuitable as a generic tile server. It should only be used in development or demonstration environments.

Comparing mapping libraries

tc-viewer supports most popular web mapping libraries out-of-the box. This can be very useful for quick, practical comparisons. If your favourite mapping library is missing, please submit an issue, or, even better, a pull request.

Contributing

Please report bugs and feature requests using the GitHub issue tracker.

If you’d like to contribute to TileCloud, please install the development requirements:

$ pip install -r dev-requirements.txt

TileCloud comes with unit tests in the tilecloud/tests directory. You can run these with the command:

$ make test

This is equivalent to:

$ python setup.py nosetests

For pull requests, it is very much appreciated if your code passes pep8 and pyflakes without warnings, with the exception of pep8 warning “E501 line too long”, which is allowed. You can run pep8 and pyflakes on the codebase with the command:

$ make pep8 pyflakes

License

Copyright (c) 2012, Tom Payne [e-mail] All rights reserved.

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:

• Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.

• Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

vim: set filetype=rst spell spelllang=en textwidth=0: