Skip to content


Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
Fetching contributors…

Cannot retrieve contributors at this time

executable file 215 lines (190 sloc) 7.605 kb
#!/usr/bin/env python
# Copyright 2011 Peter Odding <>
# This program is licensed under the MIT license.
# This Python script can be used by the notes.vim plug-in to perform fast
# keyword searches in the user's notes. It has two advantages over just using
# Vim's internal :vimgrep command to search all of the user's notes:
# - Very large notes don't slow searching down so much;
# - Hundreds of notes can be searched in less than a second.
# For more information about the Vim plug-in see
# Standard library modules.
import fnmatch
import getopt
import os
import pickle
import re
import sys
import Levenshtein
levenshtein_supported = True
except ImportError:
levenshtein_supported = False
class NotesIndex:
def __init__(self):
''' Entry point to the notes search. '''
keywords = self.parse_args()
if self.dirty:
if self.keyword_filter is not None:
matches = self.search_index(keywords)
print self.encode('\n'.join(sorted(matches)))
def parse_args(self):
''' Parse the command line arguments. '''
opts, keywords = getopt.getopt(sys.argv[1:], 'l:d:n:e:h',
['list=', 'database=', 'notes=', 'encoding=', 'help'])
except getopt.GetoptError, error:
print str(error)
# Define the command line option defaults.
self.database_file = '~/.vim/misc/notes/index.pickle'
self.user_directory = '~/.vim/misc/notes/user/'
self.character_encoding = 'UTF-8'
self.keyword_filter = None
# Map command line options to variables.
for opt, arg in opts:
if opt in ('-l', '--list'):
self.keyword_filter = arg.strip().lower()
elif opt in ('-d', '--database'):
self.database_file = arg
elif opt in ('-n', '--notes'):
self.user_directory = arg
elif opt in ('-e', '--encoding'):
self.character_encoding = arg
elif opt in ('-h', '--help'):
assert False, "Unhandled option"
if self.keyword_filter is not None:
self.keyword_filter = self.decode(self.keyword_filter)
# Canonicalize pathnames, check validity.
self.database_file = self.munge_path(self.database_file)
self.user_directory = self.munge_path(self.user_directory)
if not os.path.isdir(self.user_directory):
sys.stderr.write("Notes directory %s doesn't exist!\n" % self.user_directory)
# Return tokenized keyword arguments.
return self.tokenize(' '.join(keywords))
def load_index(self):
''' Load the keyword index or start with an empty one. '''
with open(self.database_file) as handle:
self.index = pickle.load(handle)
assert self.index['version'] == 1
self.first_use = False
self.dirty = False
self.first_use = True
self.dirty = True
self.index = {'keywords': {}, 'files': {}, 'version': 1}
def save_index(self):
''' Save the keyword index to disk. '''
with open(self.database_file, 'w') as handle:
pickle.dump(self.index, handle)
def update_index(self):
''' Update the keyword index by scanning the notes directory. '''
# First we find the filenames and last modified times of the notes on disk.
notes_on_disk = {}
for filename in os.listdir(self.user_directory):
if filename != '.swp' and not fnmatch.fnmatch(filename, '.*.s??'): # (Vim swap files are ignored)
abspath = os.path.join(self.user_directory, filename)
if os.path.isfile(abspath):
notes_on_disk[abspath] = os.path.getmtime(abspath)
# Then we either scan the whole bunch or be a bit more subtle.
if self.first_use:
for filename, last_modified in notes_on_disk.iteritems():
self.add_note(filename, last_modified)
# Check for updated and/or deleted notes since the last run.
for filename, last_modified_in_db in self.index['files'].iteritems():
if filename not in notes_on_disk:
# Forget a deleted note.
# Check whether previously seen note has changed?
last_modified_on_disk = notes_on_disk[filename]
if last_modified_on_disk > last_modified_in_db:
self.add_note(filename, last_modified_on_disk)
def add_note(self, filename, last_modified):
''' Add a note to the index (assumes the note is not already indexed). '''
sys.stderr.write("Scanning %s ..\n" % filename)
self.index['files'][filename] = last_modified
with open(filename) as handle:
for kw in self.tokenize(
if kw not in self.index['keywords']:
self.index['keywords'][kw] = [filename]
self.dirty = True
def delete_note(self, filename):
''' Remove a note from the index. '''
del self.index['files'][filename]
for kw in self.index['keywords']:
filter(lambda x: x != filename, self.index['keywords'][kw])
self.dirty = True
def search_index(self, keywords):
''' Return names of files containing all of the given keywords. '''
matches = None
for kw in keywords:
filenames = self.index['keywords'].get(kw, [])
if matches is None:
matches = set(filenames)
matches &= set(filenames)
return list(matches) if matches else []
def list_keywords(self, substring, limit=25):
''' Print all (matching) keywords to standard output. '''
decorated = []
for kw, filenames in self.index['keywords'].iteritems():
if substring in kw.lower():
if levenshtein_supported:
decorated.append((Levenshtein.distance(kw.lower(), substring), -len(filenames), kw))
decorated.append((-len(filenames), kw))
selection = [d[-1] for d in decorated[:limit]]
print self.encode('\n'.join(selection))
def tokenize(self, text):
''' Tokenize a string into a list of normalized, unique keywords. '''
words = set()
text = self.decode(text).lower()
for word in re.findall(r'\w+', text, re.UNICODE):
word = word.strip()
if word != '' and not word.isspace():
return words
def encode(self, text):
''' Encode a string in the user's preferred character encoding. '''
return text.encode(self.character_encoding, 'ignore')
def decode(self, text):
''' Decode a string in the user's preferred character encoding. '''
return text.decode(self.character_encoding, 'ignore')
def munge_path(self, path):
''' Canonicalize user-defined path, making it absolute. '''
return os.path.abspath(os.path.expanduser(path))
def usage(self):
print '''
search-notes [OPTIONS] KEYWORD...
Search a directory of plain text files using a full text index,
updated automatically during each invocation of the program.
Valid options include:
-l, --list=SUBSTR list keywords matching substring
-d, --database=FILE set path to keywords index file
-n, --notes=DIR set directory with user notes
-e, --encoding=NAME set character encoding of notes
-h, --help show this message and exit
For more information see
if __name__ == '__main__':
# vim: ts=2 sw=2 et
Jump to Line
Something went wrong with that request. Please try again.