#!/usr/bin/env python

"""
An SVN pre-commit hook which requires that commits either are marked as
whitespace cleanup commits, and contain no non-whitespace changes, or
leave whitespace alone on lines without code changes.
"""

import os
import sys
import difflib
import debug
from pysvn import Client, Transaction

prog = os.path.basename(sys.argv[0])

def die(msg):
    sys.stderr.write("\n%s: Error: %s\n" % (prog, msg))
    sys.exit(1)

if len(sys.argv) <= 1:
    die("Expected arguments: REPOSITORY TRANSACTION, found none")
    
if len(sys.argv) <= 2:
    die( "Expected arguments: REPOSITORY TRANSACTION, found only '%s'" % sys.argv[1])

repo = sys.argv[1]
txname = sys.argv[2]
tx = Transaction(repo, txname)
client = Client()

is_whitespace_cleanup = "whitespace cleanup" in tx.revpropget("svn:log").lower()

def normalize(s):
    return s.rstrip().expandtabs()    

def normalize_sequence(seq):
    o = []
    for l in seq:
        o.append(normalize(l))
    return o

def count(gen):
    return sum(1 for _ in gen)

# Function with side effects.  Acts on global varaibles
def inc_ab(flag):
    global a, b
    if   flag == ' ':
        a += 1; b += 1
    elif flag == '-':
        a += 1
    elif flag == '+':
        b += 1
    elif flag == '?':
        pass
    else:
        raise ValueError("Unexpected ndiff mark: '%s' in: %s" % (flag, plain_diff[i]))

def get_chunks(unidiff):
    if not unidiff:
        return []
    chunks = []
    chunk = []
    intro = unidiff[0:2]
    for line in unidiff[2:]:
        if line.startswith("@@"):
            if chunk:
                chunks.append(chunk)
            chunk = [line]
        else:
            chunk.append(line)
    chunks.append(chunk)
    return intro, chunks
        
changes = tx.changed()
issues = {}
context = 3
for path in changes:
    action, kind, mod, propmod = changes[path]

    # Don't try to diff added or deleted files, on ly changed text files
    if not (mod and action == "R"):
        continue

    # Don't try do diff binary files
    mimetype = tx.propget("svn:mime-type", path)
    if mimetype and not mimetype.startswith("text/"):
        continue

    new = tx.cat(path).splitlines()
    old = client.cat("file://"+os.path.join(repo,path)).splitlines()

    plain_diff = list(difflib.unified_diff(old, new, "%s (repository)"%path, "%s (commit)"%path, lineterm="" ))
    old = normalize_sequence(old)
    new = normalize_sequence(new)
    white_diff = list(difflib.unified_diff(old, new, "%s (repository)"%path, "%s (commit)"%path, lineterm=""))

    plain_count = len(plain_diff)
    white_count = len(white_diff)

#    for i in range(len(white_diff)):
#        sys.stderr.write("%-80s |  %-80s\n" % (normalize(plain_diff[i][:80]), normalize(white_diff[i][:80])))
    if white_count != plain_count and not is_whitespace_cleanup:
        intro, plain_chunks = get_chunks(plain_diff)
        intro, white_chunks = get_chunks(white_diff)
        for chunk in white_chunks:
            for i in range(len(plain_chunks)):
                if chunk == plain_chunks[i]:
                    del plain_chunks[i]
        issue = intro
        for chunk in plain_chunks:
            issue += chunk
        if  len(plain_chunks) > 1:
            are = "are"; s = "s"; an = ""
        else:
            are = "is"; s = ""; an = "an "
        issues[path] = issue
    if white_count != 0 and is_whitespace_cleanup:
        intro, white_chunks = get_chunks(white_diff)
        if  len(white_chunks) > 1:
            are = "are"; s = "s"; an = ""
        else:
            are = "is"; s = ""; an = "an "
        issues[path] = white_diff

if issues:
    if is_whitespace_cleanup:
        die("It looks as if there are non-whitespace changes in\n"
            "this commit, but it was marked as a whitespace cleanup commit.\n\n"
            "Here %s the diff chunk%s with unexpected change%s:\n\n%s\n\n"
            "Declining the commit due to a mix of code and spaces-only changes.  Please\n"
            "avoid mixing whitespace-only changes with code changes. See details above." % 
            (are, s, s, '\n\n'.join([ '\n'.join(issues[path]) for path in issues ]))
        )

    else:
        die("It looks as if there are spaces-only changes in this\n"
            "commit, but it was not marked as a whitespace cleanup commit.\n\n"
            "Here %s the diff chunk%s with unexpected change%s:\n\n%s\n\n"
            "Declining the commit due to a mix of code and spaces-only changes.  Please\n"
            "avoid mixing whitespace-only changes with code changes. See details above." % 
            (are, s, s, '\n\n'.join([ '\n'.join(issues[path]) for path in issues ]))
        )

sys.exit(0)