tracwiki - Dump the current version of each page in a Trac wiki

tracwiki.py

#!/usr/bin/env python3
"""Dump the current version of each Trac wiki page."""

__RCS__ = '$Id: tracwiki 1975 2020-03-19 14:37:50Z darren $'
__initialdate__ = 'January 2020'
__author__ = 'Darren Paul Griffith <http://madphilosopher.ca/>'

import sqlite3
import argparse

DATABASE = '/home/trac/darren/db/trac.db'

KILLFILE = ('CamelCase',
            'PageTemplates',
            'InterMapTxt',
            'InterTrac',
            'InterWiki',
            'RecentChanges',
            'SandBox',
            )

if __name__ == '__main__':

    # get command-line arguments
    parser = argparse.ArgumentParser()
    parser.add_argument('-t', '--titles', help='Show wiki page names', action='store_true')
    args = parser.parse_args()

    # connect to the Trac database
    conn = sqlite3.connect(DATABASE)
    c = conn.cursor()

    # setup before main loop
    version_table = {}          # to store the latest version number of each page
    text_table = {}             # to store the text of the page
    max_name = 0                # length of the longest page name; used for formatting name column

    # get all the versions of all the wiki pages
    for row in c.execute('SELECT name,version,text from wiki;'):
        name, version, text = row
        if (name[0:4] != 'Trac') and (name[0:4] != 'Wiki') and (name not in KILLFILE):    # ignore Trac wiki pages
            version_table[name] = version
            if not args.titles:
                text_table[(name, version)] = text
                if len(name) > max_name:
                    max_name = len(name)

    # main output begins here
    if args.titles:
        # print all titles (wiki page names)
        for name in sorted(version_table.keys()):
            print(name)

    else:
        # output all wiki pages with the page name in the left-hand column
        for name in sorted(version_table.keys()):
            version = version_table[name]
            key = (name, version)
            text = text_table[key]

            for line in text.splitlines():
                print("{0:{1}} | {2}".format(name, max_name, line))

            print()
            print()