view collection_analysis/chord_sequence_mining/spmf.py @ 0:e34cf1b6fe09 tip

commit

# Part of DML (Digital Music Laboratory)
# Copyright 2014-2015 Daniel Wolff, City University
 
# This program is free software; you can redistribute it and/or
# modify it under the terms of the GNU General Public License
# as published by the Free Software Foundation; either version 2
# of the License, or (at your option) any later version.
# 
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
# 
# You should have received a copy of the GNU General Public
# License along with this library; if not, write to the Free Software
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA

#!/usr/bin/python
# -*- coding: utf-8 -*-
#
# This is a data conversion wrapper for the spmf toolkit.
# The toolkit has been released under GPL3 at www.philippe-fournier-viger.com/spmf

__author__="Daniel Wolff"

import chord2function as c2f
import csv
import re

# takes a dictionary of chords for one or multiple files 
# in the form of dict[clipid] = [ (time,key,mode,fun,typ,bfun) ]
# and converts it into spmf
def folder2spmf(folderin = 'D:/mirg/Chord_Analysis20141216/', fileout = 'D:/mirg/Chord_Analysis20141216/Beethoven.spmf'):

    # get chords for all files                    
    output  = c2f.folder2functions(folderin)

    # open log
    logfile = fileout + '.dic'
    csvfile = open(logfile, "w+b") #opens the file for updating
    w = csv.writer(csvfile)
    w.writerow(["track","key","mode","sequence length"])
    
    # open spmf file
    fspmf = open(fileout,'w')
    # ---
    # this is writing the spmf format
    for track,trackdata in output.iteritems():
        # write chord sequence as one line in spmf file
        for (time,key,mode,fun,typ,bfun) in trackdata:
            chord = c2f.fun2num(fun,typ,bfun,mode)
            
            # -1 is the spearator of items or itemsets
            fspmf.write(str(chord) + ' -1 ') 
            
        # the sequence is closed with -2
        fspmf.write('-2\n') 
        w.writerow([track, str(key), str(mode),str(len(trackdata))])
            
    fspmf.close()
    csvfile.close()

# read an spmf file
# def parsespmf(filein = 'D:/mirg/Chord_Analysis20141216/Beethoven.txt'):

# string sourcefile path to the source spmf file with chords from records
# string patternfile path to the pattern spmf file
# matches each of the patterns in patternfile
#  to the chord sequences in sourcefile
def match(sourcefile = 'D:/mirg/Chord_Analysis20141216/Beethoven.spmf',sourcedict = 'D:/mirg/Chord_Analysis20141216/Beethoven.spmf.dic', patternfile = 'D:/mirg/Chord_Analysis20141216/Beethoven_70.txt'):
    
    # define regular expressions for matching
    # closed sequence
    
    # ---
    # we here assume that there are more files than patterns,
    # as display of patterns is somehow limited
    # therefore parallelisation will be 1 pattern/multiple files
    # per instance
    # ---
    
    patterns = []
    patterns_raw = []
    # read all patterns
    f = open(patternfile, 'r')
    for line in f:
        # a line looks like this:
        # 1120401 -1 1120101 -1 #SUP: 916
       

        # save pattern
        #patterns.append(pattern)
        #numeric? or just regex?
        # we'll use string, so any representation works
        
        pattern,support = readPattern(line)
        patterns.append(pattern)
        
        # here's the regex
        # first the spacer
        #spacer = '((\s-1\s)|((\s-1\s)*[0-9]+\s-1\s)+)'
        #repattern = r'(' + spacer + '*' + spacer.join(pattern) + spacer + '*' + '.*)'
        #print repattern
        #patterns.append(re.compile(repattern))
        
    # ---
    # now for the input sequences
    # ---
    # first: read track dictionary and get the input sequence names
    tracks = getClipDict(sourcedict)
       
    # read the input sequences
    source = open(sourcefile, 'r')
    patterns_tracks = dict()
    tracks_patterns = dict()
    
    # iterate over all tracks - to be parallelised
    for track,count in tracks.iteritems():
        sequence = readSequence(next(source))
        print track
        for p in range(0,len(patterns)):
            # match open or closed pattern
            if openPatternInSequence(sequence,patterns[p]):
                if patterns_tracks.has_key(p):
                    patterns_tracks[p].append(track)
                else:
                    patterns_tracks[p] = [track]

                if tracks_patterns.has_key(track):
                    tracks_patterns[track].append(p)
                else:
                    tracks_patterns[track] = [p]  
    
    # write clip index to files
    writeAllPatternsForClips('D:/mirg/Chord_Analysis20141216/',tracks_patterns)
    #print patterns_tracks[p]

# writes results to disk per key
def writeAllPatternsForClips(path = 'D:/mirg/Chord_Analysis20141216/',tracks_patterns = dict()):

    for name, contents in tracks_patterns.iteritems():
        # create new file
        csvfile = open(path + '/' + name + '_patterns.csv', "w+b") #opens the file for updating
        w = csv.writer(csvfile)
   
        # compress pattern data ?
        # e.g. 2 columns from-to for the long series of atomic increments
        
        w.writerow(contents)
        csvfile.close()
        

# @param line: reads a line in the spmf output file with frequent patterns
# returns list of strings "pattern" and int "support"
def readPattern(line):
    # locate support 
    suploc = line.find('#SUP:')
    support = int(line[suploc+5:-1])

    # extract pattern
    pattern = line[:suploc].split(' -1 ')[:-1]
    return (pattern,support)
   
# @param line: reads a line in the spmf input file with chord sequence
# returns list of strings "pattern" and int "support"
def readSequence(line):
    # locate support 
    suploc = line.find('-2')

    # extract pattern
    sequence = line[:suploc].split(' -1 ')[:-1] 
    return sequence

# finds open pattern in sequences
# @param [string] sequence input sequence
# @param [string] pattern pattern to be found
def openPatternInSequence(sequence,pattern):
    patidx = 0
    for item in sequence:
        if item == pattern[patidx]:
            patidx +=1
            
            # did we complet the pattern?
            if patidx >= (len(pattern)-1):
                # could also return the start index
                return 1
    # finished the sequence before finishing pattern
    return 0
            
# finds closed pattern in sequences
# @param [string] sequence input sequence
# @param [string] pattern pattern to be found
def closedPatternInSequence(sequence,pattern):
    # alternatively use KnuthMorrisPratt with unsplit string
    return ''.join(map(str, pattern)) in ''.join(map(str, sequence))  
            
# reads all track names from the dictionary created by folder2spmf
# @param sourcedict path to dictionary
def getClipDict(sourcedict):
    
    f = open(sourcedict, 'rt')
    reader = csv.reader(f)
    
    # skip first roow that contains legend
    next(reader)
    
    # get following rows
    tracks = dict()
    for (track,key,mode,seqlen) in reader:
        tracks[track]= (key,mode,seqlen)
        #tracks.append((track,count))
    
    f.close()
    return tracks 
    
            
# run spmf afterwards with java -jar spmf.jar run CM-SPADE Beethoven.spmf output.txt 50% 3
if __name__ == "__main__":
    #folder2spmf()
    match()