new-lines-check.py

import urllib2
import re

test_diff = """
diff --git a/aaaa.asc b/aaaa.asc
index e6ec784..81c9e30 100644
--- a/aaaa.asc
+++ b/aaaa.asc

-A aaa. B bbb. C ccc.
+A aaa.
+B bbb.
+C ccc.

diff --git a/README.asc b/README.asc
index e6ec784..81c9e30 100644
--- a/README.asc
+++ b/README.asc
@@ -6,19 +6,24 @@ You can find this book online at: http://git-scm.com/book

 You can find the current builds on http://git-scm.com/book[] and more information about the builds available at https://progit.org[].

-The other way to generate e-book files is to do so manually with Asciidoctor. If you run the following you _may_ actually get HTML, Epub, Mobi and PDF output files:
+The other way to generate e-book files is to do so manually with Asciidoctor.
+If you run the following you _may_ actually get HTML, Epub, Mobi and PDF output files:

diff --git a/11111.asc b/11111.asc
index e6ec784..81c9e30 100644
--- a/11111.asc
+++ b/11111.asc

-1 111. 2 22 2. 3 3 333 33 3 3.
+1 111.
+2 22 2.
+3 3 333 33 3 3.
"""


def remove_minus_sign_from_removed_line(removed):
    """
    '-A aaa. B bbb. C ccc.\n'  --> 'A aaa. B bbb. C ccc.\n'

    There is always only one removed line per chunk.
    """
    assert removed[0] == '-'

    return removed[1:]


def remove_plus_sign_from_each_line(added):
    """
    '+A aaa.\n+B bbb.\n+C ccc.\n'  --> 'A aaa.\nB bbb.\nC ccc.\n'
    """

    splitted = []
    for line in added.split('\n'):
        assert line == '' or line[0] == '+'
        splitted.append(line[1:])

    return "\n".join(splitted)


def concat_splited_lines(lines):
    """
    'A aaa.\nB bbb.\nC ccc.\n'  --> 'A aaa. B bbb. C ccc.\n'
    """
    return lines.replace('\n', ' ')[:-1] + '\n'


def assert_diff_only_change_newlines(diff):
    pattern = r'^(-[^--].*\n)((?:\+[^\+\+].*\n)*)'

    for match in re.findall(pattern, diff, flags=re.M):
        # print match[0]
        # print match[1]

        removed = match[0]
        added = match[1]

        removed = remove_minus_sign_from_removed_line(removed)
        added = remove_plus_sign_from_each_line(added)

        concated = concat_splited_lines(added)

        assert concated == removed

        # problems = 0
        # if concated != removed:
        #     problems += 1
        #
        #     print "%d - #####################" % problems
        #     print ">%s<" % removed
        #     print ">%s<" % concated

    print "No asserts means, that everything is ok :)"


if __name__ == '__main__':

    #assert_diff_only_change_newlines(test_diff)

    ##### get diff from downloaded repository
    # from git import Repo
    # repo_url = 'git@github.com:noisy/progit2-pl.git'
    # branch = 'origin/new-lines'
    #
    # print "Clonning repo %s ..." % repo_url
    # Repo.clone_from(repo_url, '/tmp/new_lines/')
    #
    # repo = Repo('/tmp/new_lines/')
    #
    # diff = repo.git.diff(branch + '~1', branch)


    ##### get diff from direct commit from github
    f = urllib2.urlopen("https://github.com/noisy/steem-bluepaper-pl/commit/4037affa78348d178633c8159c83862602aaa1c5.diff")
    diff = f.read()

    assert_diff_only_change_newlines(diff)