#!/usr/bin/env python
# coding: utf-8

# # Difference between feature 'word' and 'unicode'

# The following small script shows the differences between feature 'word' and 'unicode'.

# In[1]:


# Following variables should contain the relative path and name of the two files to compare
WordFile="../tf/0.4/word.tf"
UnicodeFile="../tf/0.4/unicode.tf"
# How many difference to show
NumberExamples = 10


# In[2]:


import os

def compare_files(file1_path, file2_path):
    FoundDifferences=0
    with open(file1_path, 'r', encoding='utf-8') as file1, open(file2_path, 'r',encoding='utf-8') as file2:
        lines_file1 = file1.readlines()
        lines_file2 = file2.readlines()

        # Check the number of lines in both files
        if len(lines_file1) != len(lines_file2):
            print("Files have different numbers of lines.")
            return

        # Compare content line by line
        for line_num, (line1, line2) in enumerate(zip(lines_file1, lines_file2)):
            line1 = line1.strip()
            line2 = line2.strip()

            # Compare the lines and print any differences
            if line1 != line2:
                
                if line1.startswith("@"):
                    continue  # Skip lines that start with "@"
                if line1.startswith("\n"):
                    continue  # Skip lines that start with " "
                
                print(f"Line {line_num + 1} differs:")
                print(f"File {file1_path}: {line1}")
                print(f"File {file2_path}: {line2}")
                print()
                FoundDifferences+=1
            if FoundDifferences==NumberExamples: 
                print ('Stoped comparing after ', FoundDifferences,'differences')
                break

        # If no differences found, print a message
        print("Finished.")

# main part
#First check if the file exist, then check its content
if os.path.exists(WordFile):
    if os.path.exists(UnicodeFile):
        print ("Comparing file ",WordFile," with ",UnicodeFile,"\n\nResult:\n\n",end="")  
        compare_files(WordFile, UnicodeFile)
    else:
        print (f"Could not find file {UnicodeFile}.")
else:
    print(f"Could not find file {WordFile}.")


# In[ ]: