app.py

# external
import pytest
# project
import textdistance
import csv
import pyarabic.araby as araby

import numpy as np
from statistics import mean

import pickle
import os

import gradio as gr

def textdistance_lcsseq(A,B):
    C = textdistance.lcsseq(A,B)
    
    str_return = ""

    i=0
    j=0
    inside=False
    grade = 0
    seq = 1
    while i<len(B) and j<len(C):
        if B[i] == C[j]:
            if inside:
                str_return += B[i]
            else:
                inside = True
                str_return += "("
                str_return += B[i]
            grade += seq
            seq += 1
            i+=1
            j+=1
        else:
            seq = 1
            grade -= 0.3
            if inside:
                str_return += ")"
                inside = False
                str_return += B[i]
                i+=1
            else:
                if C[j]==" ":
                    while not B[i] == C[j]:
                        str_return += B[i]
                        i+=1
                    j+=1
                str_return += B[i]
                i+=1
    if inside:
        str_return += ")"
    while i<len(B):
        grade -= 0.3
        str_return += B[i]
        i+=1
    
    for wordA in A.split(" "):
        for wordB in B.split(" "):
            if wordA == wordB:
                grade+=10
    
    return str_return,grade
    
def load():
    quran = []

    filename = "pickle.pkl"
    if(not os.path.exists(filename)):
        picklefile = open(filename, 'wb')
        csv_file = open('quran.csv', encoding="utf-8")
        csv_reader = csv.reader(csv_file, delimiter=',')
        for i,row in enumerate(csv_reader):
            quran.append(araby.strip_diacritics(row[2]))
        pickle.dump(quran, picklefile)
    else:
        picklefile = open(filename, 'rb')
        quran = pickle.load(picklefile)
    
    return quran

def search(query,numberOfResults):
    quran = load()
    lcsseq = []
    lengths = []


    for q in quran:
        tmp1,tmp2 = textdistance_lcsseq(query,q)
        lcsseq.append(tmp1)
        lengths.append(tmp2)

    indices =  [b[0] for b in sorted(enumerate(lengths),key=lambda i:i[1],reverse=True)]
    lengths_sorted = sorted(lengths,reverse=True)
    meanOfHead = mean(lengths_sorted[0:min(numberOfResults*3,len(lengths_sorted))])
    

    toReturn =""
    for i in range(0,min(numberOfResults,len(indices))):
        if(lengths[indices[i]] > meanOfHead):
            toReturn += "%d : %s"%(lengths[indices[i]],lcsseq[indices[i]]) + "\n"
    return toReturn
    
    
gr.Interface(fn=search, inputs=["text",gr.Slider(1, 100, value=10, step=1)], outputs=["text"]).launch()