|
| 1 | +#! /usr/bin/env python |
| 2 | +# |
| 3 | +# Script to conservatively convert initial spaces to tabs |
| 4 | +# |
| 5 | +# The script first guesses the number of spaces used (e.g. 2,3,4, or 8). |
| 6 | +# It then passes this to the unexpand utility for the actual conversion. |
| 7 | +import sys,os |
| 8 | +import optparse |
| 9 | +import subprocess |
| 10 | + |
| 11 | +def space_distribution(filename): |
| 12 | + """Reads the specified file and creates a histogram of the number of initial |
| 13 | + spaces in each line of the file. |
| 14 | +
|
| 15 | + Returns a dictionary mapping n to the number of lines with n initial spaces |
| 16 | + """ |
| 17 | + spaces = {} |
| 18 | + with open(filename,'r') as file: |
| 19 | + for line in file.readlines(): |
| 20 | + s = 0 |
| 21 | + for c in line: |
| 22 | + if c == ' ': |
| 23 | + s += 1 |
| 24 | + elif c == '*': |
| 25 | + s -= 1 |
| 26 | + break |
| 27 | + else: |
| 28 | + break |
| 29 | + spaces.setdefault(s,0) |
| 30 | + spaces[s] += 1 |
| 31 | + return spaces |
| 32 | + |
| 33 | +def max_compatible(distribution, threshold=.8, possible=[8,4,3,2]): |
| 34 | + """Guess the most likely number of spaces per tab for a file |
| 35 | +
|
| 36 | + distribution: a dictionary giving the number of lines with a particular |
| 37 | + number of initial spaces |
| 38 | + threshold: Fraction of lines that must be correctly spaced |
| 39 | + possible: list of possible tabstops to consider |
| 40 | +
|
| 41 | + Returns the highest tabstop such that at least threshold-fraction of the |
| 42 | + lines contain initial spaces that are a multiple of that tabstop. |
| 43 | + Returns None if no possiblities meet the threshold, or if the file |
| 44 | + did not contain sufficient lines to determine the tabstop |
| 45 | + unambiguously. |
| 46 | + """ |
| 47 | + for poss in sorted(possible,reverse=True): |
| 48 | + compat = 0 |
| 49 | + incompat = 0 |
| 50 | + for spaces, count in distribution.items(): |
| 51 | + # Only assess lines with spaces |
| 52 | + if spaces == 0: |
| 53 | + continue |
| 54 | + if spaces%poss == 0: |
| 55 | + compat += count |
| 56 | + else: |
| 57 | + incompat += count |
| 58 | + # require sufficient lines with spaces |
| 59 | + if threshold < 1 and (compat+incompat)*(1-threshold) < 2: |
| 60 | + return None |
| 61 | + if compat >= (compat+incompat)*threshold: |
| 62 | + return poss |
| 63 | + |
| 64 | +def unexpand(filename, spaces): |
| 65 | + """Calls the unexpand command to convert initial spaces into tabs |
| 66 | + """ |
| 67 | + args = ["unexpand","--first-only","-t",str(spaces),filename] |
| 68 | + print " ".join(args) |
| 69 | + bits = subprocess.check_output(args) |
| 70 | + with open(filename,'wb') as out: |
| 71 | + out.write(bits) |
| 72 | + |
| 73 | + |
| 74 | +if __name__ == "__main__": |
| 75 | + parser = optparse.OptionParser( usage="usage: python %prog [options] filenames..." ) |
| 76 | + parser.add_option("-t","--threshold",help="fraction of lines with compatible spacing", |
| 77 | + dest="threshold",default=0.8, type='float',action='store') |
| 78 | + parser.add_option("-x","--unexpand",help="Unexpand spaces in the input files", |
| 79 | + dest="unexpand",default=False, action='store_true') |
| 80 | + (options,args) = parser.parse_args() |
| 81 | + |
| 82 | + if len(args) < 1: |
| 83 | + parser.print_usage() |
| 84 | + parser.exit("Error. No files") |
| 85 | + |
| 86 | + |
| 87 | + for f in args: |
| 88 | + dist = space_distribution(f) |
| 89 | + # skip tabbed files |
| 90 | + if dist.keys() == [0]: |
| 91 | + continue |
| 92 | + c = max_compatible(dist,options.threshold) |
| 93 | + if c is not None: |
| 94 | + if options.unexpand: |
| 95 | + unexpand(f,c) |
| 96 | + else: |
| 97 | + print "Match\t%s\t%d\t%s"%(f,c,dist) |
| 98 | + else: |
| 99 | + print "Unclear\t%s\t?\t%s"%(f,dist) |
| 100 | + |
0 commit comments