Merge pull request #10 from gimbo/master

Reduce memory consumption from 750MB to 60MB by using CSV not JSON. (Also lots of other changes.)

README.md

@@ -5,5 +5,27 @@ pytzwhere is a Python library to lookup the timezone for a given lat/lng entirel
 
 It is a port from https://github.com/mattbornski/tzwhere with a few improvements.
 
-Instructions and usage information can be seen by running:
-./tzwhere.py --help
+If used as a library, basic usage is as follows:
+
+    >>> import tzwhere
+    >>> tz = tzwhere.tzwhere()
+    Reading json input file: tz_world_compact.json
+    >>> print tz.tzNameAt(35.29, -89.66)
+    America/Chicago
+
+By default (and as shown above), the `tzwhere` class (at the heart of this library) initialises itself from a JSON file.  Note that this is very very memory hungry (about 750MB, though the file is much smaller).  You can save a lot of memory (hundred of megabytes) at the cost of another second or so initialisation time, by telling `tzwhere` to read its data in (one line at a time) from a CSV file instead:
+
+    >>> tz = tzwhere.tzwhere(input_kind='csv')
+    Reading from CSV input file: tz_world.csv
+    >>> print tz.tzNameAt(35.29, -89.66)
+    America/Chicago
+
+The module can also be run as a script, which offers some other possibilities including producing the CSV file mentioned above.  Instructions and usage information can be seen by running:
+
+    tzwhere.py --help
+
+Dependencies (both optional):
+
+  * `docopt` - if you want to use `tzwhere.py` as a script (e.g. as shown above).
+
+  * `numpy` - if you want to save about 200MB of RAM.

setup.py

@@ -11,7 +11,8 @@
     version='1.0',
     packages=['tzwhere'],
     package_data={
-        'tzwhere': ['tz_world.json', 'tz_world_compact.json', 'tz_world.pickle']
+        'tzwhere': ['tz_world.json', 'tz_world_compact.json',
+                    'tz_world.pickle', 'tz_world.csv']
         },
     include_package_data=True,
     license='MIT License',