This package lets you find ZIP code by address in Taiwan.
The main features:
- Fast. It builds ZIP code index by tokenization.
- Gradual. It returns partial ZIP code rather than noting when address is not detailed enoguh.
- Stand-alone. It depends on nothing.
Find ZIP code gradually:
>>> import zipcodetw
>>> zipcodetw.find('臺北市')
u'1'
>>> zipcodetw.find('臺北市信義區')
u'110'
>>> zipcodetw.find('臺北市信義區市府路')
u'110'
>>> zipcodetw.find('臺北市信義區市府路1號')
u'11008'
After v0.3, you even can find ZIP code like:
>>> zipcodetw.find('松山區')
u'105'
>>> zipcodetw.find('秀山街')
u''
>>> zipcodetw.find('台北市秀山街')
u'10042'
It is available on PyPI:
$ sudo pip install zipcodetw
Just install it and have fun. :)
If you install it by pip
or python setup.py install
, a ZIP code index
will be built automatically. But if you want to use it from source code, you
have to build an index manually:
$ python -m zipcodetw.builder
The ZIP code directory is provided by Chunghwa Post, and is available from: http://www.post.gov.tw/post/internet/Download/all_list.jsp?ID=2201#dl_txt_s_A0206
- Updated to the 3+3 v2102.01 data.
- Fixed a Python 3 bug, maybe.
- A black hole ate the logs.
- Fixed the py2 py3 compatibility. Thanks the contribution from Poren Chiang and Ryan.
- Updated the data to 2014/12.
- Fixed a rarely issue that causes IndexError.
- Reverted removing insignificant tokens introduced in v0.5.4.
- It now handles insignificant tokens; and
- redundant units in the finding logic (
directory.find
). - Allowed number token ends without unit.
- Now
address.tokens
is a list.
- Fixed a gradual matching issue causing some wrong results.
- Removed the token whose unit is insignificant automatically.
- Fixed and simplified the matching logic for address tail.
- Refined the index building logic.
- Fixed the issue while it was running in multi-threaded environment.
- Added a new argument,
keep_alive
, for theDirectory
class.
- Refined the code slightly.
- It now builds a ZIP code index when you install it; so
- the package size is 12.5x smaller.
- The internal API is better now.
- It now shipped with an index compiled in SQLite; so
- initiation time is ~680x faster, i.e. ~30ms each import; and
zipcodetw.find
is ~1.9x slower, i.e. ~2ms each call; and- has bigger package size.
- All code was moved into
zipcodetw
package. zipcodetw.find
now returns unicode instead of string.
- It builds full index for middle tokens; and
- also normalizes Chinese numerals now!
zipcodetw.find
is ~1.06x faster.- But initiation time increases to ~1.7x.
zipcodetw.find
is 8x faster now!- It has a better tokenizing logic; and
- a better matching logic for sub-number now.
zipcodetw.find_zipcodes
was removed.- Internal API was changed a lot.
- The tests are better now.