Archive for the ‘python’ Category
Using TileCache on IIS
I’ve previously written about using PyISAPie to run Python under IIS – this allows Python scripts to run faster than using CGI. Rather than starting up the Python interpreter each time a request is made to the web server, PyISAPIe starts the interpreter once and only needs to run the script at each request.
TileCache provides a Python-based WMS-C/TMS server – this is allows you create a local disk-based cache of any WMS server (in my case MapServer), and then connect to it through a WMS-C client such as OpenLayers. Using a cache means that rather than have MapServer create an image for every single request for a tile, the first time a tile is requested the image is saved in a folder on the server. On all subsequent requests for the same data TileCache can retrieve this file without needing MapServer. Performance increase is incredible – however watch out for filling up disk space, and remember to empty the cache if you change symbology or update your data.
As mod_python is now officially dead (and has been moved in Apache’s attic), it is yet another reason if you are using Windows servers to run everything through IIS rather than the rather clumsy Apache on Windows. Why pay for Windows, and then install a web server port, thereby losing all the tools, security, and utilities within Windows to manage and administer the web server?
The aim of this post is to describe the set up of TileCache, running through IIS. Read the rest of this entry »
Visualising GeoJSON in 15 Minutes
The objective of this post is to save a GeoJSON feature collected from the web to an image file, and was inspired by the Shapely manual. This will be done using Windows, Python 2.5 (although 2.6 should be almost exactly the same steps), and some additional Python packages. This example uses GeoJSON data from a CloudMade geocoding service.
The greatest difficulty when trying to do things in Python is getting the right packages to match your Python version, and making sure all the package dependencies are in place.
Read the rest of this entry »
GeoJSON Where are You?
I was looking around the web to try and find some services that returned GeoJSON to try out Shapely. It was surprisingly hard to find servers or web services given the amount of tools that have already been built to use GeoJSON.
Eventually I found the CloudMade’s geocoding developer samples. CloudMade is a company that builds tools and services on top of OpenStreetMap. Even though many of the people involved in both CloudMade and OSM are the same, using the two brands interchangeably has not been without controversy.
Once you’ve signed up for a free account, you can then register for an API key that allows you to call their web services. See here for differences between free and paid accounts. The service I was interested in testing was the geocoding service that returns GeoJSON.
Depending on the type of feature you are looking for you can get different geometry types – the standard point, lines, and polygons, all as GeoJSON objects.
The API uses URLs and GET requests to return data. Calls can be made to either a find.html page which returns the GeoJSON object displayed on a map, as shown in the query for “england” on the left. Alternatively the query can be made to find.geojs to return just the GeoJSON itself. There is also a find.js which can be used to return geometry as JSON, although this seems a little confusing – even more so as the coordinate order is reversed from the standard GeoJSON [lon,lat].
Compiling a 64 bit Version of PyISAPIe
Welcome to a veritable novella. I’ll jump straight to the end – there is no happy outcome, as yet, to this post there is a happy outcome! I have compiled a 64 bit PyISAPIe DLL (a program that runs Python at high speed under IIS), and it is up and running on Windows Server 2008 and IIS7. Many thanks to Phillip Stabon, the creator of PyISAPIe, for additional help, and for the latest version of PyISAPIe which makes compilation for different set ups much easier.
Some background information – I’ve not done any C++ for 10 years (and even then my most advanced program was to finish off a code sample on a hotel register for pirates – of the traditional kind). Also I’ve never used Subversion. Hence there are lots of pictures, and every step is detailed. If there’s anything missing let me know.
I’m not sure why I started doing this – the performance of Python using 32-bit version is fine, and since I started working on this Phillip Stabon, has mentioned he will be making a 64 bit DLL himself. However as there are many different versions of Python around, and different GIS tools rely on different versions it seemed a good idea to document how to do this from scratch.
In this example I’m using Visual Studio 2008, and a number of other tools that are detailed as I go along. I’ve also got the 64 bit compilation tools installed. There are a few notes from Phillip on compiling it for Windows 64-bit in the discussion lists (see quoted text underneath the post). Since this post was first published there is a newer discussion in the forums.
Anyway “How to Compile PyISAPIe for a 64-bit Server in 5 Easy Steps!”…
Read the rest of this entry »
Automated Diagrams and Documentation
As I previously wrote, one of the ways to get through the task of documenting a project is to turn the documentation process into a coding project. Documentation sometimes seems pointless as the likelihood of anyone other than yourself reading it are low to nil.
To increase the chances pretty pictures always help. If they are also useful then the documentation starts to have real value. Up until now I’ve been using Microsoft’s Visio, but recently I tried out a couple of online tools.
Class Diagrams
The MapServer Mapscript documentation currently uses of diagrams made from ASCII text to illustrate relationships between classes. I’m not sure if these were generated automatically, but the images now appear a little dated. They do however have the advantage over an image of being able to copy and paste class names.
+-------+ 0..* 1 +-------+ | Class | <--------> | Layer | +-------+ +-------+
I tried to recreate these diagrams using the online service http://yuml.me/. On the site you enter your model description using a fairly basic syntax. From then on the same diagram can be accessed using different URLs, which are used to control display and formatting options.
The diagram above was created using the syntax below. You can play around with this sample by clicking here.
[Hashtable]1..1->1[Class{bg:yellow}],
[Label ]0..1->1[Class],
[Style]0..*->1[Class],
[Class]0..*->1[Layer{bg:green}]
Reboot to Refresh Environment Variables
I wasn’t aware that if you change Windows environment variables that are used by IIS then you need to reboot the machine for the changes to take effect. As described by David Wong
This is because NT services like IIS inherit their environment from services.exe, which does not get updated when you change system environment until you reboot.
This knowledge may save a couple of hours of frustration if you are running Python scripts on IIS through PyISAPIe.
Setting Up Python on IIS7
I had already configured Python to run through the Apache webserver on my development server, but after a few issues on the production server (Apache freezing / crashing) I wanted to test running Python scripts with IIS7. The principle aim was to run TileCache through IIS rather than requiring Apache.
Why Not CGI?
While IIS 7 has Fast-CGI installed (see this IIS forum), even better performance can be achieved using ISAPI. This answer from ServerFault has a good summary on why ISAPI should be preferred over than CGI. Not only performance should be considered – maintenance should also be taken into account. If a web master or hosting service is familiar with IIS then they are also likely to be familiar with ISAPI.
From the PyISAPIe site:
The reason ISAPI applications have the capability of being better than CGI or FastCGI applications is mostly due to its tight integration with the web server environment. Instead of initializing an entire program from scratch (in this case, the Python interpreter) every time a request is made for a page, an ISAPI extension only has to provide a function that is called upon every request. For interpreting Python scripts on a per-request basis, this means that the interpreter can be initialized once and used many times, creating a very noticeable performance gain.