harshavardhana

def clean_html(html):
    """ Remove HTML markup from the given string. """
    # remove inline JavaScript / CSS
    x = re.sub(r'(?is)<(script|style).*?>.*?(</\1>)', '', html.strip())
    # remove html comments. must be done before removing regular tags since comments can contain '>' characters.
    x = re.sub(r'(?s)<!--(.*?)-->[\n]?', '', x)
    # remove the remaining tags
    x = re.sub(r'(?s)<.*?>', ' ', x)                                                                                                          
    # remove html entities
    x = remove_entities(x)

harshavardhana

#!/usr/bin/env python
import urllib
import pycurl
import time
import sys
import json

import cStringIO

from optparse import OptionParser

harshavardhana

#!/usr/bin/env python
import sys
import logging
import chardet
import json

from optparse import OptionParser

from boilerpipy import (Extractor, isValidhtml,
                        compat_urllib_request)

harshavardhana

#!/usr/bin/env python
import urllib
import pycurl
import sys
import json

from optparse import OptionParser

parser = OptionParser(usage="%prog: [options] [URL]")
parser.add_option('-p', '--projectid', help="Project ID")

harshavardhana

package main

import (
        "path/filepath"
        "os"
        "flag"
        "fmt"
)

type fileattr struct {

harshavardhana

#!/bin/sh
# 1st sed: remove colons
# 2nd sed: replace higher level folder names with dashes
# 3rd sed: indent graph three spaces
# 4th sed: replace first dash with a vertical bar
if [ $# -eq 1 ]; then
    ls -R $1 | grep ":$" | sed -e 's/:$//' -e 's/[^-][^\/]*\//--/g' -e 's/^/  /' -e 's/-/|/'
elif [ $# -gt 1 ]; then
    echo "Only one directory at a time please!!"
    exit 255

harshavardhana

#!/bin/bash

if [ $# -eq 0 ]; then
    echo "Please provide the pcap file for sharking.."
    exit 255
fi

capinfos $1 1>/dev/null 2>/dev/null

if [ $? -eq 1 ]; then

harshavardhana

/*
 *  g++ -O3 --std=c++0x -o test_mmap test_mmap.cpp
 */

#include <vector>
#include <sys/mman.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>

harshavardhana

import re

def isvalidhostname(hostname):
    """
    Validate hostname
    """
    regex = re.compile("[^A-Z\d-]", re.IGNORECASE)

    if len(hostname) > 255:
        return False

harshavardhana

#!/bin/bash 

_init()
{
    DIRZ=$( seq 1 100 )
}

cleanup()
{
    echo "Cleaning the test"