Below you see the product of my first baby-steps in programming. The purpose of the script is twofold: 1. Take html input of a specific website, process it, and return relevant info such as document id, text, and headline. 2. Generate a count of the words in all the articles.

The script is working and does what it is supposed to, however, I cannot help but feel that I'm missing a lot in terms of performance.

import re
import pandas as pd
from urllib.request import urlopen as uReq
from sklearn.feature_extraction.text import CountVectorizer

TAG_RE = re.compile(r'<[^>]+>')
def RemoveTags(text):
     """Remove all html tags"""
     return TAG_RE.sub('', text)

ESCAPES_RE = re.compile(r'\\.')
def RemoveEscapes(text):
    """Remove extra escape characters from encoding"""
    return ESCAPES_RE.sub('', text)

def ReadFromLink(link):
    """Read html from link and return raw html"""
    with uReq(link) as response:
        html = response.read()#
        html = str(html).lower()
        return html.lower()

def ArticleRaw(html):
    """Find articles in html"""
    article = re.findall(r'<doc>.*?</doc>', html)
    return article

def GetDocID(html):
    """Find document ids in html"""
    docid = re.findall(r'<docid>(.*?)</docid>', html)
    docid = [docid.strip() for docid in docid]
    docid = [int(docid) for docid in docid] 
    return docid

def GetHeadline(html):
    """Find headlines in html"""
    headline = re.findall(r'<headline>(.*?)</headline>', html)
    headline = [RemoveTags(headline) for headline in headline]
    headline = [RemoveEscapes(headline) for headline in headline]
    return headline 

def GetMainText(html):
    """Find maintext in html"""
    maintext = re.findall(r'<text>(.*?)</text>', html)
    maintext = [RemoveTags(maintext) for maintext in maintext]
    maintext = [RemoveEscapes(maintext) for maintext in maintext]
    maintext = [' '.join(maintext.split()) for maintext in maintext]
    return maintext

link = link
html = ReadFromLink(link)

ArticlesDict = {
        "docid": GetDocID(html), 
        "raw_article": ArticleRaw(html), 
        "headline": GetHeadline(html), 
        "maintext": GetMainText(html)

def CountFeatures(text):
    documents = ArticlesDict['maintext']
    # Stem first?
    vector = CountVectorizer()
    x = vector.fit_transform(documents)
    df_features = pd.DataFrame(x.toarray(), columns = vector.get_feature_names())
    return df_features

df_features = CountFeatures(df_articles['maintext'])

If I may suggest, using a tool like Beautiful Soup can greatly help you get around html elements in a simple way


Here you have a very brief example on how it operates

from bs4 import BeautifulSoup
import requests

r  = requests.get("http://any_url_you_want.com")

data = r.text

soup = BeautifulSoup(data)

for text in soup.find_all('text'):
    # Here you do whatever you want with text

You can adapt your methods to use the functions depending on the tags, or however you want

Check also this article, explains quite well what you can do with it and is accessible for beginners


  • 1
    \$\begingroup\$ There is now also github.com/kennethreitz/requests-html :) \$\endgroup\$ – hjpotter92 Mar 3 '18 at 2:06
  • \$\begingroup\$ I've tried it out and it works! Thank you very much for the feedback. \$\endgroup\$ – Daniel Hansen Mar 7 '18 at 8:57
  • \$\begingroup\$ Very happy to hear :) I will also take a look at requests-html, thanks hjpotter92 for the suggestion \$\endgroup\$ – A. Romeu Mar 7 '18 at 9:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.