Madis is a extensible relational database system built upon the SQLite database and with extensions written in Python (via APSW SQLite wrapper). Its is developed at:
http://madis.googlecode.com Due to Madis’ SQLite core, the database format of Madis is exactly the same as SQLite’s one. This means that all SQLite databases are directly usable with Madis. Madis' enables to quickly develop (in Python) and test new relational functions and virtual tables. SQL syntax extensions were also created to simplify queries which use these functions. Madis' main goal is to promote the handling of data related tasks within an extended relational model. In doing so, it promotes the database, from a support role (storing and retrieving data), to being a full data processing system on its own. Madis already includes functions for file import/export, keyword analysis, data mining tasks, fast indexing, pivoting, statistics and workflow execution. Some of the functionality of Madis is: - Programmable (in Python) row functions (via APSW): mterm> select detectlang('Il en est des livres comme du feu de nos foyers'); french - Programmable (in Python) aggregate functions (via APSW): mterm> select concatterms(a) from (select "term1+term2" as a UNION select "term2 term3" as a); term1+term2 term2 term3 - Programmable (in Python) virtual tables (via APSW): mterm> select * from file('./demo/continents.tsv') limit 2; Asia|AF Europe|AL - Multisets (functions that return multiple rows/columns): mterm> select * from table1; a |b ------------------------------------------- 'car wood bike'| 'first group' 'car car wood' |'first group' 'car wood' |'first group' 'car wood ice' |'first group' 'ice' |'second group' 'car ice' |'second group' 'car cream' |'second group' 'icecream ice car' |'second group' mterm> select b, freqitemsets(a, 'threshold:2', 'noautothres:1', 'maxlen:2') from table1 group by b b | itemset_id | itemset_length | itemset_frequency | item --------------------------------------------------------------------- first group | 1 | 1 | 4 | wood first group | 2 | 1 | 4 | car first group | 3 | 2 | 4 | car first group | 3 | 2 | 4 | wood second group | 1 | 1 | 3 | ice second group | 2 | 1 | 3 | car second group | 3 | 2 | 2 | car second group | 3 | 2 | 2 | ice - On the fly multidimensional indexing (the cache virtual table): The index is based on kdtrees and is extremely fast with queries involving multiple constraints. mterm> select country.c2, continent.c1 from file('countries.tsv') as country, file('continents.tsv') as continent where country.c1=continent.c2; Aruba|Americas Antigua and Barbuda|Americas United Arab Emirates|Asia Afghanistan|Asia . . . . . . . . . Query executed in 0 min 2 sec 40 msec mterm> select country.c2, continent.c1 from file('countries.tsv') as country, (CACHE file 'continents.tsv') as continent where country.c1=continent.c2; Aruba|Americas Antigua and Barbuda|Americas United Arab Emirates|Asia Afghanistan|Asia . . . . . . . . . Query executed in 0 min 0 sec 71 msec - Workflows: mterm> exec flow file 'workflow.sql'; Above query uses Madis' SQL inverted syntax. - Pivoting: http://madis.googlecode.com/svn/publish/row.html#pivoting -- All above functionality has been created via row/aggregate/vtable Python extensions (APSW offers these in very nice API), and the aforementioned SQL syntax extensions. In practise Madis has been proven to be very fast in data analysis tasks and in the development of data processing workflows. A little note: The high quality of APSW's and SQLite's code, has helped immensely in developing Madis. We have strained both of these projects as much as we could, and they coped beautifully. We literally had queries that spanned multiple pages, which executed in seconds. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
