#!/usr/bin/env python3

__author__ = 'Anton Vlasenko'

import os
import sys
import re

class FilesParser(object):

    def get_htm_files_list(self, root_dir):
        "Generate htm files list to process in other function"
        files_list = []
        for root, sub_folders, files in os.walk(root_dir):
            for file in files:
                if os.path.splitext(file)[1] == '.htm':
                    files_list.append(os.path.join(root,file))
        return files_list

    def replace(self, source, replacer, root_dir):
        "Function to replace text in each file given by get_htm_files_list method"
        counter = 0
        # Regexp to compiled pattern.
        pattern = re.compile(source, re.MULTILINE|re.DOTALL)
        for file_name in self.get_htm_files_list(root_dir):
	    # Open file to read it.
            file = open(file_name, "r") 
            # File content -> memory
            text = file.read()
            # Close read session.
            file.close() 
            # Open file to write into it.
            file = open(file_name, "w") 
            # Replace regexp match with text and write into a file.
            file.write(pattern.sub(replacer, text)) 
            file.close()
            # Files counter ++.
            counter += 1
        return counter
     
    def find(self, keyword, root_dir):
        files = list()
        counter = 0
        pattern = re.compile(keyword, re.MULTILINE|re.DOTALL)
        for file_name in self.get_htm_files_list(root_dir):
            file = open(file_name, "r") 
            # File content -> memory
            text = file.read()
            # Close read session.
            file.close() 
            if pattern.search(text):
                files.append(file_name)
                counter += 1
        return files, counter
    
  
if __name__ == '__main__':
    # Initializing parser
    parser = FilesParser()
    # root dir => current dir (can be read from args, but I don't need this).
    root_dir = os.getcwd()
    print ('===========!!!BACKUP YOUR FILES!!!===========')
    print ('We are going to replace text in all "htm" files in %s directory. Type "s" for search, type "exit" to exit.' % root_dir)
    regexp = input('What we should replace? (regexp):\n').strip()
    
    if regexp == 'exit':
        sys.exit(0)
    elif regexp == 's':
        default_state = False
        second_phrase = 'Okay, please write here regexp to search for ("exit" to exit):\n'
    else:
        default_state = True
        second_phrase = 'With what we should replace? ("exit" to exit).\n'

    replacer = input(second_phrase).strip()
    if replacer == 'exit':
         sys.exit(0)
    if default_state:
        third_phrase = 'Continue? You\'re going to replace %s with %s in all "htm" files in %s and recursive. \
            This can\'t be undone ("y" or "n", "yes" or "no"): \n ' % (regexp, replacer, root_dir)
    else:
        third_phrase = 'Continue with the search of %s in %s? (("y" or "n", "yes" or "no"): \n' % (replacer, root_dir)
    
    okay = input(third_phrase).strip()
    if okay in ('y', 'yes'):
        if default_state:
            counter = parser.replace(regexp, replacer, root_dir)
            if counter > 0:
                print ("Done! \n%s files were processed." % str(counter))
            else:
                print ("No files were processed.")
        else:
            files_list, counter = parser.find(replacer, root_dir)
            if files_list:
                print ('Requested pattern %s was found in %s "htm" files:' % (replacer, counter))
                for file in files_list:
                    print (file)
            else:
                print ('Sorry, nothing was found')
      
    print ('Bye!')