WIP tool to "analyze" dokuwiki contents (currently prints out folders, pages and page revisions)

attic.py

import os
import datetime


TOPFOLDER = r'..\wiki\data\attic'


class Folder():
    _page_blacklist = ['sidebar']
    blacklist = ['tag', 'testi', 'wiki']

    def __init__(self, name, parent=None):
        self.name = name if name else '<root>'
        self.parent = parent
        self.subfolders = []
        self._pages = []

    def __str__(self):
        return self.name
    def __repr__(self):
        return self.name

    def getpage(self, pagename):
        page = None
        for p in self._pages:
            if p.name == pagename:
                page = p
                break
        return page

    def addpage(self, page):
        if page not in self._pages and page.name not in Folder._page_blacklist:
            self._pages.append(page)

    def printpages(self):
        print(self._pages)

    def getpages(self):
        for p in self._pages:
            yield p

    def printfiles(self, pagename):
        page = self.getpage(pagename)
        print(page.files)

    def getfolder(self, foldername):
        folder = self
        for f in self.subfolders:
            if f.name == foldername:
                folder = f
                break
        return folder

    def printall(self):
        if self.parent:
            print("".join([self.parent.name, ':', self.name]))
        else:
            print(self)
        self.printpages()
        for f in self.subfolders:
            f.printall()

    def findfolder(self, names):
        # names = ['web-ohjelmointi', 'javascript']
        for f in self.subfolders:
            if f.name == names[0]:
                names = names[1:]
                if names:
                    return f.findfolder(names)
                else:
                    return f


class Page():
    def __init__(self, name, folder=None):
        self.name = name
        self.folder = folder
        self.files = []

    def __str__(self):
        return "".join([self.name, ' (', str(len(self.files)), ')'])
    def __repr__(self):
        return "".join([self.name, ' (', str(len(self.files)), ')'])

class File():
    def __init__(self, name, date=None, page=None):
        self.name = name
        self.date = date
        self.page = page

    def __str__(self):
        return self.name
    def __repr__(self):
        return self.name


if __name__ == '__main__':

    first = None
    rootfolder = None

    for root, dirs, files in os.walk(TOPFOLDER):

        foldername = root[len(TOPFOLDER)+1:]

        if foldername in Folder.blacklist:
            continue

        if first:
            rootfolder = first.findfolder(foldername.split('\\'))
        else:
            rootfolder = Folder(foldername)
            first = rootfolder

        for d in dirs:
            if d in Folder.blacklist:
                continue
            folder = Folder(d, parent=rootfolder)
            rootfolder.subfolders.append(folder)

        for f in files:
            if not f.endswith('.txt.gz'):
                continue
            pagename = f[:f.index('.')]
            page = rootfolder.getpage(pagename)
            if page:
                file = File(f, page=page)
                page.files.append(file)
            else:
                folder = rootfolder.getfolder(pagename)
                page = Page(pagename, folder=folder)
            rootfolder.addpage(page)

    first.printall()