#!/usr/bin/env python2.4

import mindex
import py
import sys
import re

extensions = ['.txt', '.rst']
filter = lambda x: not x.basename.startswith('.') # nothing that starts with a .

if len(sys.argv) != 3:
    print 'usage: %s <dbpath> <datapath>'
    print
    print '  where <dbpath> is the path where the indexes are stored, and'
    print '  <datapath> the path that contains the data to index; note that'
    print '  paths in the indexer are stored relative from <datapath>.'
    sys.exit(1)

dbpath = sys.argv[1]
datapath = sys.argv[2]
if datapath.endswith('/'):
    datapath = datapath[:-1]
qs = mindex.mindex(dbpath)
for p in py.path.local(datapath).visit(fil=filter, rec=filter):
    if not p.check(file=True) or p.ext not in extensions:
        print 'skipping', p
        continue
    else:
        relpath = str(p)[len(str(datapath)):]
        data = p.read()
        qs.index(relpath, data)
        print 'indexed', relpath, data
print 'done'