Dealing with Version Information

About two years ago I was using a tool called zero29 to update AssemblyInfo.cs files in my C# projects. The author Mark Seemann (who's talks I can highly recommend) declined my request to add support for .nuspec files. His point was, that zero29 is a tool to manage AssemblyInfo.cs files and that's all it should ever do. While I am a big proponent of the Unix philosophy, I wished that zero29 would act as a tool to manage version information across all kinds of files. This would (in my opinion) still be in line with the Unix philosophy, as there are tools such as awk, grep or sed which do "one thing", but that "one thing" is defined in a generalized context. So I jumped down the rabbit hole and built Bumpy.

I had a lot of fun writing Bumpy. I took the time to "make it right", so I wrote several unit tests, created a fully automated build process and I even took the time to write some form of documentation, as well as an addin (an extension) for the C# build tool Cake. Inspired by editorconfig's configuration I was trying to create a config file in which I could define where to find version information in different files. I kept observing the download statistics on NuGet.org as well as Chocolatey.org daily and I was really hyped when I received my very first pull request.

The latest commit on the Bumpy project was in June 2018. I realized that while I was trying to build the project right, I was in reality not building the right project. A core concept of Bumpy was to understand different version schemas and some people even suggested that I should detect versions that are scattered among several lines of text. In the end, I decided that a different approach seemed simpler and more flexible: Why not build a "search and replace text" tool? So I wrote a prototype in Python, because its .ini library comes with a ton of interesting features. This first version of "niles" used two config files to describe where to find text snippets and to describe the desired content for these snippets. After a while, I found it cumbersome to manage two files, so I redesigned the tool, while also adding more commands. Here's the final script together which a short example:

hello.txt:

Hello! My name is Florian. How are you?

name.txt:

My name is Florian.

niles.ini:

# Python's .ini library comes with a ton of features such as referencing
# values or defined default values for other sections

[DEFAULT]
# Optional parameters. If not set, niles will use the default encoding and
# line ending defined by the operating system
_charset = utf-8
_end_of_line = crlf

[**/{hello,name}.txt]
_regex = My name is (?P<name>[A-za-z]+(\s[A-za-z]+)?)
first_name = Max
second_name = Mustermann
name = %(first_name)s %(second_name)s

Niles offers these commands:

Running niles apply would create:

Hello! My name is Max Mustermann. How are you?

and

My name is Max Mustermann.

Here's a final example of a config file which can be used to manage versions in C# projects:

niles.ini (C#):

[DEFAULT]
major = 5
minor = 10
patch = 0

[**/AssemblyInfo.cs]
_regex = ^\[assembly: Assembly(File)?Version\("(?P<assembly_version>.+)"\)\]$
assembly_version = %(major)s.%(minor)s.%(patch)s.0

[**/*.nuspec]
_regex = ^\s*<version>(?P<semantic_version>.+)<\/version>$
semantic_version = %(major)s.%(minor)s.%(patch)s

So, is this the "right" approach? To be honest, I'm not sure. At the moment I like the simplicity of a search and replace tool compared to a elaborate tool which can perform specific operations on versions.

niles.py:

#!/usr/bin/env python3

import argparse
import collections
import configparser
import glob
import re


class Config(collections.namedtuple(
        'Config', 'file_names charset end_of_line regex_list values')):
    __slots__ = ()


def expand_curly(line):
    match = re.search('{([^{]*?)}', line)
    expanded_lines = []
    if match is None:
        expanded_lines.append(line)
    else:
        snippets = match.group(1).split(',')
        for snippet in snippets:
            expanded_lines.append(
                line[:match.start()] + snippet + line[match.end():])
    return expanded_lines


def fetch_files(glob_pattern):
    expanded_patterns = expand_curly(glob_pattern)
    file_names = []
    [file_names.extend(
        glob.glob(p, recursive=True)) for p in expanded_patterns]
    return file_names


def parse_config(text):
    config_parser = configparser.ConfigParser()
    config_parser.read_string(text)
    config_list = []
    for section in config_parser.sections():
        file_names = fetch_files(section)
        section_data = config_parser[section]
        charset = section_data.get('_charset')
        end_of_line_text = section_data.get('_end_of_line')
        if end_of_line_text is None:
            end_of_line = None
        elif end_of_line_text == 'lf':
            end_of_line = '\n'
        elif end_of_line_text == 'crlf':
            end_of_line = '\r\n'
        else:
            raise ValueError(
                "Supported end of line characters: 'lf' and 'crlf'")
        regex_list_text = section_data.get('_regex', '').splitlines()
        regex_list = [re.compile(r) for r in regex_list_text]
        config_list.append(Config(file_names, charset, end_of_line,
                                  regex_list, dict(config_parser[section])))
    return tuple(config_list)


def load_config():
    with open('niles.ini', 'r') as f:
        return parse_config(f.read())


def read_file_lines(file_name, charset, end_of_line):
    with open(file_name, 'r', encoding=charset, newline=end_of_line) as f:
        return f.read().splitlines()


def write_file_lines(file_name, lines, charset, end_of_line):
    with open(file_name, 'w', encoding=charset, newline=end_of_line) as f:
        for line in lines:
            f.write(line + '\n')


def replace_line(regex, line, values):
    match = regex.search(line)
    replaced_line = line
    line_changed = False
    offset = 0
    if match is not None:
        group_dict = match.groupdict()
        for group_name in group_dict.keys():
            if group_name in values:
                old_value_len = len(match.group(group_name))
                new_value = values[group_name]
                start_index = offset + match.start(group_name)
                end_index = start_index + old_value_len
                start = replaced_line[:start_index]
                end = replaced_line[end_index:]
                replaced_line = start + new_value + end
                line_changed = line != replaced_line
                offset += len(new_value) - old_value_len
    return line_changed, replaced_line


def list_values(config, read_file, log):
    for entry in config:
        for file_name in entry.file_names:
            for line in read_file(file_name, entry.charset, entry.end_of_line):
                for regex in entry.regex_list:
                    match = regex.search(line)
                    if match is not None:
                        for group_name in match.groupdict().keys():
                            log("{}: {}".format(
                                file_name, match.group(group_name)))


def apply_changes(config, read_file, write_file, log):
    for entry in config:
        for file_name in entry.file_names:
            new_lines = []
            file_dirty = False
            for line in read_file(file_name, entry.charset, entry.end_of_line):
                new_line = line
                for regex in entry.regex_list:
                    line_changed, new_line = replace_line(
                        regex, new_line, entry.values)
                    file_dirty |= line_changed
                new_lines.append(new_line)
            if file_dirty:
                write_file(
                    file_name, new_lines, entry.charset, entry.end_of_line)
                print("Wrote file '{}'".format(file_name))
            else:
                print("Skipped file '{}'".format(file_name))


def list_files(config, log):
    for entry in config:
        for file_name in entry.file_names:
            log(file_name)


def list_values_cmd():
    list_values(load_config(), read_file_lines, print)


def apply_changes_cmd():
    apply_changes(load_config(), read_file_lines, write_file_lines, print)


def list_files_cmd():
    list_files(load_config(), print)


def main():
    arg_parser = argparse.ArgumentParser(
        description='Search and replace text in files')
    subparsers = arg_parser.add_subparsers()
    values_parser = subparsers.add_parser('values', help='List all values')
    values_parser.set_defaults(func=list_values_cmd)
    apply_parser = subparsers.add_parser('apply', help='Writes values to files')
    apply_parser.set_defaults(func=apply_changes_cmd)
    files_parser = subparsers.add_parser('files', help='List all files')
    files_parser.set_defaults(func=list_files_cmd)
    args = arg_parser.parse_args()
    if hasattr(args, 'func'):
        args.func()
    else:
        arg_parser.print_help()


if __name__ == '__main__':
    main()

Published: 2019-11-01