PCap

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
# vim:ts=4:sw=4:softtabstop=4:smarttab:expandtab

# target script

import sys
import re

def roman_siecle(bk):
    data =""
    for (id, href) in bk.text_iter():

        data = bk.readfile(id)

        reg = re.compile(r'(?P<b1>(<span class="(([a-zA-Z0-9_-]+) |)CharOverride-10)(| ([a-zA-Z0-9_-]+))">)(?P<roman>([xiv]+))(?P<b2>(</span>))')

        search = reg.search(data)
        print(search)
        if search != None:
            balise1  = search.group('b1')
            balise2 = search.group('b2')
            roman = search.group('roman')

            sub_str = balise1+"<span style=\"font-size:83%\">"+roman.upper()+"</span>"+balise2

            test2 = data.replace(balise1+roman+balise2, sub_str)
            data.replace(balise1+roman+balise2, sub_str)
            data2=re.sub(reg, sub_str, data)

            #effacer le fichier ancien
            bk.deletefile(id)
            basename = id+".xhtml"
            mt = "application/xhtml+xml"
            uid1 = id

            #ajouter un nouveau fichier
            bk.addfile(uid1, basename, data2, mt)

def PCapH2(bk):
    data =""
    for (id, href) in bk.text_iter():
        data = bk.readfile(id)
        reg = re.compile(r'<h2 class="T3" xml:lang="fr-CH"><span class="CharOverride-2">(?P<content>(.+))</span></h2>')
        search = reg.search(data)
        print(search)
        if search != None:
            Content = search.group('content')
            minReg = re.compile(r'(?P<minuscule>[a-zàâäéèêëûüôöîïÿ]+)')
            minSearch = minReg.search(Content)
            Content2=""
            while minSearch:
                if minSearch != None:
                    minuscul = minSearch.group('minuscule')
                    a_remp = "<span style=\"font-size:83%\">"+minuscul.upper()+"</span>"
                Content = re.sub(minReg, a_remp, Content)
                    #print(Content)

                bk.deletefile(id)
                basename = id+".xhtml"
                mt = "application/xhtml+xml"
                uid1 = id

                #ajouter un nouveau fichier
                bk.addfile(uid1, basename, Content2, mt)

                sub_str = "<h2 class=\"T3\" xml:lang=\"fr-CH\"><span class=\"CharOverride-2\">"+Content+"</span></h2>"
                print(sub_str)
            #test2 = data.replace(Content, sub_str)
            #data.replace(balise1+roman+balise2, sub_str)
                data2 = re.sub(reg, sub_str, data)

            #effacer le fichier ancien
                bk.deletefile(id)
                basename = id+".xhtml"
                mt = "application/xhtml+xml"
                uid1 = id

            #ajouter un nouveau fichier
                bk.addfile(uid1, basename, data2, mt)

def run(bk):
    #roman_siecle(bk)
    PCapH2(bk)


def main():
    print("I reached main when I should not have\n")
    return -1

if __name__ == "__main__":
    sys.exit(main())