import gradio as gr
import os
import textstat
import urllib.request
import re
from bs4 import BeautifulSoup

CLEANR = re.compile('<.*?>')

DEFAULT_SYSTEM_PROMPT = """ 
Algorithm for this site is based on Readability Wiki - https://en.wikipedia.org/wiki/Readability 
The following links are the detailed references:
1. https://en.wikipedia.org/wiki/Flesch%E2%80%93Kincaid_readability_tests#Flesch_reading_ease
2. https://en.wikipedia.org/wiki/SMOG
3. https://en.wikipedia.org/wiki/Coleman%E2%80%93Liau_index
4. https://en.wikipedia.org/wiki/Dale%E2%80%93Chall_readability_formula
"""

def measure_readability(message,history):
    if "https://" in message:
        response = urllib.request.urlopen(message)
        html = response.read().decode('utf8')
        cleantext = BeautifulSoup(html).text
        #cleantext = BeautifulSoup(html,'lxml').text
        text = re.sub(CLEANR,'', cleantext)
    else:
        text = message
        
    vline1 = "==== Content Info ==== " + os.linesep
    vline2 = "Character Count "+str(textstat.char_count(text, ignore_spaces=True)) + os.linesep
    vline3 = "Lexicon Count "+str(textstat.lexicon_count(text, removepunct=True)) + os.linesep
    vline4 = "Syllable Count "+str(textstat.syllable_count(text)) + os.linesep
    vline5 = "Sentence Count "+str(textstat.sentence_count(text)) + os.linesep
    vline6 = " " + os.linesep
    vline7 = "==== Result ==== " + os.linesep
    vline8 = "Flesch Reading Ease (0-29 = Very confusing to read, 90-100 = Easy to read) is "+str(textstat.flesch_reading_ease(text)) + os.linesep
    #print("Flesch-Kincaid Grade Level is "+str(textstat.flesch_reading_ease(text)))
    vline9 = "Smog Index (Years of education before a reader understand) is "+str(textstat.smog_index(text)) + os.linesep
    vline10 = "Coleman Liau Index (Grade level before a reader understand) is "+str(textstat.coleman_liau_index(text)) + os.linesep
    #print("Automated Readability Index (Grade level before a reader understand) is "+str(textstat.automated_readability_index(text)))
    vline11 = "Dale-Chall Readability(1-5 = can be understood by 4th grader, 8-9 = can be undestood by 11th to 15th grade student) Score "+str(textstat.dale_chall_readability_score(text)) + os.linesep
    vline12 = "Gunning Fog Index (Years of formal education before a reader understand) is "+str(textstat.gunning_fog(text)) + os.linesep
    #print("Grade Level Comprehension is "+str(textstat.automated_readability_index(text)))
    #vline13 = "Difficult Words "+str(textstat.difficult_words(text)) + os.linesep
    vline14 = "Reading Time is "+str(textstat.reading_time(text, ms_per_char=14.69))+" seconds"+ os.linesep
    answer = vline1+vline2+vline3+vline4+vline5+vline6+vline7+vline8+vline9+vline10+vline11+vline12+vline14
    #answer = "Flesch Reading Ease (90-100 = Easy to read, 0-29 = Very confusing to read) is "+str(textstat.flesch_reading_ease(text))
    return answer


    
Conversing = gr.ChatInterface(measure_readability, chatbot=gr.Chatbot(height=400,label = "Enter URL or String to evaluate"), retry_btn=None,theme=gr.themes.Monochrome(),
                              title = 'Ecommerce Content Readability Tool', description = DEFAULT_SYSTEM_PROMPT ,undo_btn = None, clear_btn = None, css='footer {visibility: hidden}').launch()
#"Algorithm for this site is based on Readability Wiki - https://en.wikipedia.org/wiki/Readability "


if __name__ == "__main__":
    Conversing.launch()