return from WebPage to indicate whether a link was actually crawlable and only actually crawl it if it was

.gitignore

@@ -2,3 +2,4 @@
 venv/
 .vscode/*
 __pycache__/
+sitemap.html

README.md

@@ -1 +1,21 @@
-# Concurrent web scraper
+# Concurrent web scraper
+
+## Requirements
+
+This crawler was written in 3.7.0 to take advantage of the latest `asyncio` features.
+
+Install required modules:
+
+```bash
+pip install -r requirements.txt
+```
+
+Run:
+
+```bash
+python crawler.py -u https://urltocrawl.com
+```
+
+## Results
+
+The resulting sitemap will be output in the root of this directory as `sitemap.html`

crawler.py

@@ -0,0 +1,120 @@
+#!/usr/bin/env python
+'''
+Need a docstring.
+'''
+
+import argparse
+import jinja2
+import os
+import asyncio
+from datetime import datetime
+from utils.helpers import (UrlPool, WebPage, RobotsTxt, sanitise_url)
+
+
+def init_crawler(base_url=None, robots=None):
+    '''
+    Initialises the crawler by running the initial URL.
+    '''
+    uncrawled_urls, crawled_urls = UrlPool(), UrlPool()
+    initial_page = WebPage(url=base_url, base_url=base_url, robots=robots)
+
+    try:
+        initial_page.run()
+    except Exception as e:
+        print(e)
+
+    initial_urls = initial_page.list_urls()
+
+    # ensure the base URL isn't crawled again
+    try:
+        initial_urls.remove(base_url)
+    except KeyError:
+        pass
+    # also ensure base URL wasn't discovered with a trailing slash on the
+    # initial page scrape
+    try:
+        initial_urls.remove("".join([base_url, '/']))
+    except KeyError:
+        pass
+
+    # Add the base URL to the crawled pool
+    crawled_urls.add_to_pool(base_url)
+
+    for url in initial_urls:
+        sanitised_url = sanitise_url(url=url)
+        if sanitised_url not in crawled_urls.pool:
+            uncrawled_urls.add_to_pool(sanitised_url)
+
+    return(uncrawled_urls, crawled_urls)
+
+
+def process_pool(base_url=None, uncrawled_urls=None, crawled_urls=None, robots=None):
+    '''
+    Iterates over the pool of URLs and adds any discovered URLs.
+    '''
+    while uncrawled_urls.pool:
+        # pop url from pool
+        new_url = uncrawled_urls.remove_from_pool()
+        # create a WebPage object for the URL
+        current_page = WebPage(url=new_url, base_url=base_url, robots=robots)
+        try:
+            succeeded = current_page.run()
+        except Exception as e:
+            print(e)
+
+        if succeeded:
+            _urls = current_page.list_urls()
+            crawled_urls.add_to_pool(new_url)
+
+            for url in _urls:
+                sanitised_url = sanitise_url(url=url)
+                if sanitised_url not in crawled_urls.pool:
+                    uncrawled_urls.add_to_pool(url)
+
+        print('{0} URLs crawled, {1} remaining'.format(len(crawled_urls.pool),
+                len(uncrawled_urls.pool)))
+
+
+def render_sitemap(base_url=None, crawled_urls=None, runtime=None):
+    '''
+    Renders the sitemap as an HTML file.
+    '''
+    urlcount = len(crawled_urls)
+    sorted_urls = sorted(crawled_urls)
+
+    tmpl = jinja2.Environment(
+        loader=jinja2.FileSystemLoader('templates')
+        ).get_template('sitemap.html.j2')
+
+    rendered_html = tmpl.render(base_url=base_url, urlcount=urlcount, urls=sorted_urls, runtime=runtime)
+
+    with open('sitemap.html', 'w') as outfile:
+        outfile.write(rendered_html)
+
+    print('Sitemap available at {0}/sitemap.html'.format(os.getcwd()))
+
+
+def run(args=None):
+    '''
+    needs a docstring.
+    '''
+    starttime = datetime.now()
+
+    base_url = sanitise_url(args.url, base_url=True)
+    robots = RobotsTxt(base_url=base_url)
+
+    uncrawled_urls, crawled_urls = init_crawler(base_url, robots)
+    process_pool(base_url, uncrawled_urls, crawled_urls, robots)
+
+    runtime = int((datetime.now() - starttime).total_seconds())
+
+    render_sitemap(base_url=base_url, crawled_urls=crawled_urls.pool, runtime=runtime)
+
+
+if __name__ == '__main__':
+
+    parser = argparse.ArgumentParser(description='Recursive web crawler')
+    parser.add_argument("-u", "--url", required=True, help="Base url to crawl")
+    args = parser.parse_args()
+
+    run(args)

notes.md

@@ -1,9 +1,56 @@
 ## Thoughts
 
-###### for each URL, do the following:
- * mark it as crawled
- * get page content
-   * if that fails, mark the link as invalid
- * find all links in the content
-   * check each link for dupes
-   * add to pool or discard
+  * ~~strip hashes and everything following (as they're in-page anchors)~~
+  * ~~strip args~~
+  * ~~use `pop()` on the set instead of `.remove()`~~
+    * ~~return false once the set is empty~~
+  * ~~`WebPage.parse_urls()` needs to compare startswith to base url~~
+  * ~~ignore any links which aren't to pages~~
+  * ~~better url checking to get bare domain~~ #wontfix
+  * ~~remove trailing slash from any discovered url~~
+  * ~~investigate lxml parser~~
+  * ~~remove base url from initial urls with and without trailing slash~~
+  * ~~investigate using [tldextract](https://github.com/john-kurkowski/tldextract) to match urls~~ #wontfix
+  * ~~implement parsing of [robots.txt](http://docs.w3cub.com/python~3.6/library/urllib.robotparser/)~~
+  * ~~investigate [gzip encoding](https://stackoverflow.com/questions/36383227/avoid-downloading-images-using-beautifulsoup-and-urllib-request)~~
+  * ~~implement some kind of progress display~~
+  * async
+  * better exception handling
+  * randomise output filename
+
+### Async bits
+
+in `__main__`:
+
+```python
+loop = asyncio.get_event_loop()
+try:
+    loop.run_until_complete(main())
+finally:
+    loop.close()
+```
+
+  * initialises loop and runs it to completion
+  * needs to handle errors (try/except/finally)
+
+```python
+async def run(args=None):
+    tasks = []
+
+    for url in pool:
+        tasks.append(url)
+    # for i in range(10):
+        # tasks.append(asyncio.ensure_future(myCoroutine(i)))
+
+    # gather completed tasks
+    await asyncio.gather(*tasks)
+```
+
+Getting the contents of the page needs to be async too
+
+```python
+async def get_source():
+    blah
+    blah
+    await urlopen(url)
+```

requirements.txt

@@ -3,5 +3,6 @@ bs4==0.0.1
 certifi==2018.8.13
 chardet==3.0.4
 idna==2.7
-requests==2.19.1
-urllib3==1.23
+Jinja2==2.10
+lxml==4.2.4
+MarkupSafe==1.0

scraper.py

@@ -1,88 +0,0 @@
-#!/usr/bin/env python
-
-import re
-import argparse
-import urllib.request
-from bs4 import BeautifulSoup
-
-class WebPage(object):
-
-    headers = {'User-Agent': 'Mozilla/5.0 (X11; Linux x86_64; rv:61.0) Gecko/20100101 Firefox/61.0'}
-
-    def __init__(self, args):
-        self.url = args['url']
-
-
-    def get_source(self, args=None):
-        request = urllib.request.Request(self.url, headers=headers)
-        page = urllib.request.urlopen(request)
-        self.source = page.read()
-
-
-    def find_links(self, args=None, source=None):
-        soup = BeautifulSoup(self.source, 'html.parser')
-        links = soup.find_all('a')
-        hrefs = []
-
-        for link in links:
-            if link['href'].startswith('/'):
-                hrefs.append("".join([self.url, link['href']]))
-            else:
-                hrefs.append(link['href'])
-
-        return hrefs
-
-
-def run(args=None):
-    source = get_source(args)
-    urls = find_links(args, source)
-    local_urls = parse_urls(args, urls)
-
-    print(local_urls)
-
-def get_source(args=None):
-    url = args.url
-    useragent = 'Mozilla/5.0 (X11; Linux x86_64; rv:61.0) Gecko/20100101 Firefox/61.0'
-    headers = {'User-Agent': useragent}
-    request = urllib.request.Request(url, headers=headers)
-    page = urllib.request.urlopen(request)
-    source = page.read()
-
-    return source
-
-
-def find_links(args=None, source=None):
-    soup = BeautifulSoup(source, 'html.parser')
-    links = soup.find_all('a')
-    hrefs = []
-
-    for link in links:
-        if link['href'].startswith('/'):
-            hrefs.append("".join([args.url, link['href']]))
-        else:
-            hrefs.append(link['href'])
-
-    return hrefs
-
-
-def parse_urls(args=None, urls=None):
-    local_urls = []
-
-    for url in urls:
-        if url.startswith(args.url):
-            local_urls.append(url)
-
-
-    return local_urls
-
-
-if __name__ == "__main__":
-
-    parser = argparse.ArgumentParser(description='Recursive web scraper')
-    parser.add_argument("-u", "--url", required=True, help="Base url to scrape")
-    args = parser.parse_args()
-
-    if not args.url.startswith('http'):
-        raise SystemExit('URL must start with a protocol (http(s)).')
-
-    run(args)

templates/sitemap.html.j2

@@ -0,0 +1,14 @@
+<html>
+<head>
+ <title>Sitemap for {{ base_url }}</title>
+</head>
+<body>
+<p>
+Crawled {{ urlcount }} URLs on {{ base_url }} in ~{{ runtime }} seconds.
+<ul>
+{% for url in urls %}
+  <li><a href="{{ url }}">{{ url }}</a></li>
+{% endfor %}
+</ul>
+</body>
+</html>

test_helpers.py

@@ -1,37 +1,35 @@
 #!/usr/bin/env python
 
 import unittest
-from utils.helpers import (clean_base_url)
+from utils.helpers import (sanitise_url)
 
 class TestUrls(unittest.TestCase):
 
-    base_url = "github.com"
-
     base_url_list = (('eu.httpbin.org', 'http://eu.httpbin.org'),
                      ('www.simonweald.com', 'http://www.simonweald.com'),
                      ('http://www.github.com/', 'http://www.github.com'),
                      ('https://www.github.com', 'https://www.github.com'))
 
-    valid_urls = ["https://www.github.com", "http://www.github.com",
-                  "github.com", "/some/url/", "index.html"]
+    urls_to_clean = (('https://www.github.com/', 'https://www.github.com/'),
+                     ('https://github.com/?foo=bar', 'https://github.com/'),
+                     ('https://github.com/#anchor', 'https://github.com/'))
 
 
-    def test_clean_base_url(self):
+    def test_sanitise_base_url(self):
         '''
         Tests whether a URL's protocol can be discovered if not provided.
         '''
         for url, target in self.base_url_list:
-            result = clean_base_url(url)
+            result = sanitise_url(url, base_url=True)
             self.assertEqual(result, target)
 
-    # def test_url_validation(self):
-    #     '''
-    #     Passes when given a valid URL. A valid URL is qualified
-    #     by being local to the domain to be crawled.
-    #     '''
-    #     for url in self.valid_urls:
-    #         result = url_validation(self.base_url, url)
-    #         self.assertTrue(result)
+    def test_sanitise_url(self):
+        '''
+        Tests whether a URL's protocol can be discovered if not provided.
+        '''
+        for url, target in self.urls_to_clean:
+            result = sanitise_url(url)
+            self.assertEqual(result, target)
 
 
 if __name__ == '__main__':

utils/helpers.py

@@ -3,56 +3,202 @@
 Utilities to provide various misc functions.
 '''
 
-from urllib.parse import urljoin
+from bs4 import BeautifulSoup
+import aiohttp
+import urllib.request
+import urllib.robotparser
+import urllib.error
+import gzip
+from urllib.parse import (urljoin, urlsplit)
 
-class UrlPoolManager(object):
+
+class UrlPool(object):
     '''
-    Object to manage the lifecycle of a pool of URLs.
+    Object to manage a pool of URLs.
     '''
 
     def __init__(self):
-        self.url_pool = set()
+        self.pool = set()
 
     def check_duplicate(self, new_url):
         '''
         Checks if a URL exists in the current pool.
         '''
-        if new_url in self.url_pool:
+        if new_url in self.pool:
             return True
         else:
             return False
 
-    def invalidate_url(self, url):
-        self.url_pool.remove(url)
+    def remove_from_pool(self):
+        '''
+        Remove a URL from the pool and return it to be crawled.
+        '''
+        return(self.pool.pop())
 
-    def add_to_list(self, url):
-        self.url_pool.add(url)
+    def add_to_pool(self, url):
+        self.pool.add(url)
+
+    def list_pool(self):
+        pool = self.pool
+        return pool
 
 
-def clean_base_url(url):
+class WebPage(object):
     '''
-    Standardise the URL to be scraped to ensure it
-    is added to relative URLs in a consistent manner.
+    Object to manage common operations required to return
+    the data from each individual page.
     '''
-    protocol = 'http://'
 
-    if url.startswith('http'):
-        base_url = url
+    # set a sane user-agent and request compression if available.
+    headers = {'Accept-Encoding': 'gzip, deflate',
+               'User-Agent': 'Mozilla/5.0 (X11; Linux x86_64; rv:61.0) Gecko/20100101 Firefox/61.0'}
+
+    def __init__(self, url=None, base_url=None, robots=None):
+        self.url = url
+        self.base_url = base_url
+        self.robots = robots
+        self.source = None
+        self.urls_to_crawl = set()
+
+
+    def get_source(self):
+        '''
+        Retrieve a page's source.
+        '''
+
+        request = urllib.request.Request(self.url, headers=self.headers)
+        page = urllib.request.urlopen(request, timeout=5)
+
+        # handle the content encoding in case it needs decompressing.
+        if 'text/html' in page.info().get('Content-Type'):
+            if page.info().get('Content-Encoding'):
+                if page.info().get('Content-Encoding') == 'gzip':
+                    self.source = gzip.decompress(page.read())
+                elif page.info().get('Content-Encoding') == 'deflate':
+                    self.source = page.read()
+            else:
+                self.source = page.read()
+
+
+    def find_links(self):
+        '''
+        Find all URLs on a page and ensure they are absolute. If they are
+        relative then they will be appended to the base URL.
+        '''
+        hrefs = set()
+
+        soup = BeautifulSoup(self.source, 'lxml')
+        links = soup.find_all('a', href=True)
+
+        for link in links:
+            if link['href'].startswith('/'):
+                hrefs.add(urljoin(self.url, link['href']))
+            else:
+                hrefs.add(link['href'])
+
+        self.discovered_hrefs = hrefs
+
+
+    def parse_urls(self):
+        '''
+        Iterate through the list of discovered URLs and add them to the
+        pool if they start with the base URL.
+        '''
+        for url in self.discovered_hrefs:
+            if url.startswith(self.base_url) and self.robots.check(url):
+                sanitised_url = sanitise_url(url=url)
+                self.urls_to_crawl.add(sanitised_url)
+
+
+    def list_urls(self):
+        '''
+        Returns all valid discovered URLs.
+        '''
+
+        return self.urls_to_crawl
+
+
+    def run(self):
+        '''
+        Attempt to get the page's source and if successful, iterate through it
+        to find any links we can crawl.
+        '''
+        try:
+            self.get_source()
+        except Exception:
+            # skip if we didn't retrieve the source.
+            pass
+
+        if self.source:
+            self.find_links()
+            self.parse_urls()
+            return True
+        else:
+            return False
+
+
+class RobotsTxt(object):
+    '''
+    needs a docstring
+    '''
+
+    def __init__(self, base_url=None):
+        '''
+        Manually retrieve robots.txt to allow us to set the user-agent.
+        '''
+        self.base_url = base_url
+        self.headers = {'User-Agent': 'Mozilla/5.0 (X11; Linux x86_64; rv:61.0) Gecko/20100101 Firefox/61.0'}
+
+        robots_url = urljoin(self.base_url, 'robots.txt')
+        request = urllib.request.Request(robots_url, headers=self.headers)
+
+        robots = urllib.robotparser.RobotFileParser()
+        robots.set_url(robots_url)
+
+        try:
+            response = urllib.request.urlopen(request, timeout=5)
+        except urllib.error.HTTPError:
+            robots.allow_all = True
+        else:
+            data = response.read()
+            decoded_data = data.decode("utf-8").splitlines()
+            robots.parse(decoded_data)
+
+        self.robots = robots
+
+
+    def check(self, url):
+        '''
+        Test if robots allows us to crawl that URL.
+        '''
+        return self.robots.can_fetch("*", url)
+
+
+def sanitise_url(url, base_url=False):
+    '''
+    If `base_url` is True, we attempt to standardise `url` to ensure it can be
+    prepended to relative URLs. If no scheme has been provided then we default
+    to http as any sane https-only site should 301 redirect http > https.
+
+    If `base_url` is False, we sanitise URLs to strip queries and fragments (we
+    don't want to scrape in-page anchors etc).
+
+    Returns a sanitised URL as a string.
+    '''
+    default_proto = 'http'
+    delim = '://'
+
+    split_url = urlsplit(url)
+
+    if base_url:
+        # This will sanitise the initial url for the initial page crawl.
+        if split_url.scheme and split_url.scheme.startswith('http'):
+            sanitised_url = "".join([split_url.scheme, delim, split_url.netloc])
+        elif (split_url.path and not split_url.scheme and not split_url.netloc):
+            sanitised_url = "".join([default_proto, delim, split_url.path])
     else:
-        # otherwise assume HTTP as any sane site should upgrade
-        # to HTTPS via a 301 redirect.
-        base_url = "".join([protocol, url])
+        # Sanitise discovered URLs. We already expect them in the format
+        # protocol://base_url/path
+        sanitised_url = "".join([split_url.scheme, delim, split_url.netloc, split_url.path])
 
-    return base_url
-
-
-def get_url_validation(base_url=None, url=None):
-    '''
-    Ensure any URLs discovered are absolute. If relative,
-    they will be appended to the base URL.
-    '''
-
-    if url.startswith('/'):
-        return urljoin(base_url, url)
-    if url.startswith(base_url):
-        return url
+    return sanitised_url