me-suzy

import os
import re
from collections import defaultdict

# Define directories
base_dir = r'e:\Carte\BB\17 - Site Leadership\alte\Ionel Balauta\Aryeht\Task 1 - Traduce tot site-ul\Doar Google Web\Andreea\Meditatii\2023\Intersection\test'
ro_dir = os.path.join(base_dir, 'ro')
en_dir = os.path.join(base_dir, 'en')

# Function to extract canonical from HTML content

me-suzy

import os
import re
from collections import defaultdict

# Define directories
base_dir = r'e:\Carte\BB\17 - Site Leadership\Principal 2022'
ro_dir = os.path.join(base_dir, 'ro')
en_dir = os.path.join(base_dir, 'en')

# Function to extract canonical from HTML content

me-suzy

import os
import re
from collections import defaultdict

# Define directories
base_dir = r'e:\Carte\BB\17 - Site Leadership\Principal 2022'
ro_dir = os.path.join(base_dir, 'ro')
en_dir = os.path.join(base_dir, 'en')

# Function to extract canonical from HTML content

me-suzy

<?php
$to_be_replaced = 'wxyz'; // exactly what it wants replaced
$nr_start = 1; // from which no start counting
$path_files = getcwd();
// echo $path_files;
$excluded_files = array(
	'.htaccess',
	'robots.txt',
	'.ftpquota',
	'dezabonare.html',

me-suzy

<?php
$de_inlocuit = 'wxyz'; // exact ce se doreste inlocuit
$nr_start = 1; // de la ce nr sa inceapa numaratoarea
$cale_fisiere = getcwd();
// echo $cale_fisiere;
$fisiere_excluse = array(
	'.htaccess',
	'robots.txt',
	'.ftpquota',
	'dezabonare.html',

me-suzy

import os
import re

def fix_specific_cross_reference():
    """Repară cross-reference-ul specific pentru exemplul tău"""
    
    # Fișierul EN care trebuie reparat
    en_file = r'e:\Carte\BB\17 - Site Leadership\Principal 2022\en\the-eyes-see-the-world-the-heart-feels-it-deep-down.html'
    correct_ro_filename = 'ochii-vad-lumea-inima-o-simte-pana-in-adanc.html'
    

me-suzy

import os
import re

def test_one_file():
    """Test pe un singur fișier să văd exact ce se întâmplă"""

    # Testez fișierul pe care l-ai menționat
    test_file = r'e:\Carte\BB\17 - Site Leadership\Principal 2022\ro\ochii-vad-lumea-inima-o-simte-pana-in-adanc.html'

    if not os.path.exists(test_file):

me-suzy

import os
import re
from pathlib import Path

ro_directory = r'e:\Carte\BB\17 - Site Leadership\Principal\ro'
en_directory = r'e:\Carte\BB\17 - Site Leadership\Principal\en'

def read_file_with_fallback_encoding(file_path):
    encodings = ['utf-8', 'latin1', 'cp1252', 'iso-8859-1']
    for encoding in encodings:

me-suzy

import os
import re
from pathlib import Path

ro_directory = r'e:\Carte\BB\17 - Site Leadership\Principal\ro'
en_directory = r'e:\Carte\BB\17 - Site Leadership\Principal\en'

def read_file_with_fallback_encoding(file_path):
    encodings = ['utf-8', 'latin1', 'cp1252', 'iso-8859-1']
    for encoding in encodings:

me-suzy

import os
import re
from pathlib import Path

ro_directory = r'e:\Carte\BB\17 - Site Leadership\Principal\ro'
en_directory = r'e:\Carte\BB\17 - Site Leadership\Principal\en'

def read_file_with_fallback_encoding(file_path):
    encodings = ['utf-8', 'latin1', 'cp1252', 'iso-8859-1']
    for encoding in encodings: