Wavelet Matrix/Tree succinct data structure for full text search (using shellinford C++ library)
Project description
shellinford
Shellinford is an implementation of a Wavelet Matrix/Tree succinct data structure for document retrieval.
It is based on shellinford C++ library.
NOTE: This module requires C++11 compiler
Installation
$ pip install shellinford
Usage
Create a new FM-index instance
>>> import shellinford
>>> fm = shellinford.FMIndex()
shellinford.Shellinford([use_wavelet_tree=True, filename=None])
When given a filename, Shellinford loads FM-index data from the file
Build FM-index
>>> fm.build(['Milky Holmes', 'Sherlock "Sheryl" Shellingford', 'Milky'], 'milky.fm')
build([docs, filename])
When given a filename, Shellinford stores FM-index data to the file
Search word from FM-index
>>> for doc in fm.search('Milky'):
>>> print('doc_id:', doc.doc_id)
>>> print('count:', doc.count)
>>> print('text:', doc.text)
doc_id: 0
count: [1]
text: Milky Holmes
doc_id: 2
count: [1]
text: Milky
>>> for doc in fm.search(['Milky', 'Holmes']):
>>> print('doc_id:', doc.doc_id)
>>> print('count:', doc.count)
>>> print('text:', doc.text)
doc_id: 1
count: [1]
text: Milky Holmes
search(query, [_or=False, ignores=[]])
If _or = True, then “OR” search is executed, else “AND” search
Given ignores, “NOT” search is also executed
NOTE: The search function is available after FM-index is built or loaded
Count word from FM-index
>>> fm.count('Milky'):
2
>>> fm.count(['Milky', 'Holmes']):
1
count(query, [_or=False])
If _or = True, then “OR” search is executed, else “AND” search
NOTE: The count function is available after FM-index is built or loaded
This function is slightly faster than the search function
Add a document
>>> fm.push_back('Baritsu')
push_back(doc)
NOTE: A document added by this method is not available to search until build
Read FM-index from a binary file
>>> fm.read('milky_holmes.fm')
read(path)
Write FM-index binary to a file
>>> fm.write('milky_holmes.fm')
write(path)
Check Whether FM-Index contains string
>>> 'baritsu' in fm
License
Wrapper code is licensed under the New BSD License.
Bundled shellinford C++ library (c) 2012 echizen_tm is licensed under the New BSD License.
CHANGES
0.4.0 (2018-09-30)
FMIndex.count() is added
No longer support Python 2.6
bug fix
0.3.5 (2018-09-05)
FMIndex.build() and FMIndex.pushback() ignore empty string
FMIndex supports “in” operator. (e.g., ‘a’ in fm)
Support Python 3.5, 3.6 and 3.7
0.3.4 (2016-10-28)
FMIndex.search() returns list
0.3 (2014-11-24)
“OR” search and “NOT” search are available in FMIndex.search().
FMIndex.size and FMIndex.docsize are available as property
0.2 (2014-03-28)
“AND” search is available by giving Sequence (list, tuple, etc.) FMIndex.search()
0.1 (2014-03-11)
First release.
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.