CodeSearch query

csquery.py

import argparse
import requests
import sys

from collections import namedtuple
from bs4 import BeautifulSoup

CSResult = namedtuple('CSResult', ['pkg', 'version', 'file', 'href', 'lineno'])

def fetch_page(query):
    r = requests.get('http://codesearch.debian.net/search?q={}'.format(query))
    return r.text, r.status_code

def parse_result_data(data):
    bs = BeautifulSoup(data)
    results = bs.select('html body div#content ul#results li a')
    if len(results) == 0:
        return None
    for result in results:
        href = result.attrs.get('href')
        rline = result.text
        pkg, *rem  = rline.split('_')
        version, *file_line = rem[0].split('/')
        path, lineno = file_line[0].split(':')
        yield CSResult(pkg, version, path, href, lineno)

def display_result(results):
    for r in results:
        print("%-6s : %20s", 'Package', r.pkg)
        print("%-6s : %20s", 'Version', r.version)
        print("%-6s : %20s", 'File Match', r.file)
        print("%-6s : %20s", 'Line No', r.lineno)
        print("%-6s : %20s", 'URL', r.href)
    

def setup_options():
    parser = argparse.ArgumentParser(
        description="Command line interface to codesearch.debian.net",
        epilog="For more information please check \
        http://codesearch.debian.net/faq")
    
    # query
    parser.add_argument("-q", "--query", action="append", 
                        help="Query to be executed, can be used more than once",
                        required=True)

    # options to code search
    parser.add_argument("-f", "--file-type",
                        help="Search only in specific file type",
                        choices=["c", "c++", "perl", "python", "go", "java",
                                 "ruby", "shell", "vala"])
    parser.add_argument("-P", "--package", help="Search in only given package")
    parser.add_argument("-p", "--path", help="Search only in given path")

    return parser


def main():
    parser = setup_options()
    args = parser.parse_args()

    query = []
    query.append(*args.query)
    if args.file_type:
        query.append('filetype:' + args.file_type)

    if args.package:
        query.append('package:' + args.package)
        
    if args.path:
        query.append('path:' + args.path)
    
    page, scode = fetch_page(' '.join(query))
    if scode != 200:
        printf("Something went wrong with server, code: {}".format(scode),
               file=sys.stderr)
        return 1
    else:
        rgen = parse_result_data(page)
        if rgen is None:
            printf("No match found for your query", file=sys.stderr)
            return 1
        
        display_result(rgen)

        
        
    

if __name__ == "__main__":
    main()