/* _candidate_makeNewSpiderPage - makes a new SpiderPage out of this candidate object
*/
public SpiderPage _candidate_makeNewSpiderPage()
{
SpiderPage s = new SpiderPage(this.getUrl(), this.getLinkingToLinks(), this.getReferredByLinks(), this.getPageContent());
if (this._candidate_isAliasCandidate())
{
s.addAliasUrl(this._candidate_getUrl());
}
return(s);
}
/* spiderFetch - ThreadPool QueueUserWorkItem method, gets the links on a page and adds them to the
* candidates list.
* @args - A _spiderFetch_SpiderDataWrapper object that will be cast back from a normal object. We
* have to take normal objects as input because = requires a single
* generic object to give its worker delegate (spiderFetch) as an argument.
*/
static void spiderFetch(Object args) {
_SpiderDataWrapper_spiderFetch wrapper = (_SpiderDataWrapper_spiderFetch) args;
Spider spider_obj = wrapper.getSpiderObject();
SpiderPage current_page = wrapper.getNewPage();
// check this thread into _thread_status, a list of int[]s, where [0] is the thread ID and [1] is
// the status- 0 for not working and 1 for working. thread_index is used later to change this
// thread id's status back to not working when it's done
int thread_index = 0;
bool thread_found = false;
for (int i = 0; i < spider_obj._thread_status.Count; i++) {
if (spider_obj._thread_status.ElementAt(i)[0] == Thread.CurrentThread.ManagedThreadId) {
spider_obj._thread_status.ElementAt(i)[1] = 1;
thread_index = i;
thread_found = true;
break;
}
}
// need to make a new entry for this thread id in _thread_status...
if (!thread_found) {
// lock the thread when performing an operation that depends on _thread_status.Count, using
// a local lock object
Object lock_obj = new Object();
lock (lock_obj) {
spider_obj._thread_status.Add(new int[]{ Thread.CurrentThread.ManagedThreadId, 1 });
thread_index = spider_obj._thread_status.Count - 1;
}
}
spider_obj.writeStatus("thread id: " + Thread.CurrentThread.ManagedThreadId + ", spiderFetch(): fetching " + current_page.getUrl());
_SpiderDataWrapper_getLinks gl_wrapper = getLinks(current_page, spider_obj);
string current_page_final_url = gl_wrapper.getFinalUrl();
List<SpiderPage> current_page_links = gl_wrapper.getNewLinks();
List<string> current_page_link_strings = new List<string>();
for (int q = 0; q < current_page_links.Count; q++) {
SpiderPage qth_page = current_page_links.ElementAt(q);
spider_obj._candidate_pages.Add(qth_page);
current_page_link_strings.Add(qth_page.getUrl());
}
spider_obj._candidate_pages.Add(new SpiderPage(current_page.getUrl(), current_page_final_url, current_page.getReferencedByUrls(), current_page_link_strings));
// set this thread id's status back to not working in _thread_status
spider_obj._thread_status.ElementAt(thread_index)[1] = 0;
}
public _SpiderDataWrapper_spiderFetch(Spider spider_obj, SpiderPage new_page) {
this.spider_obj = spider_obj;
this.new_page = new_page;
}
/* getLinks() - find all the links on a given page
* @startp - the page to be scanned for links, represented as an SpiderPage object (which has a referring
* page)
* @s - the Spider object in use
*/
static _SpiderDataWrapper_getLinks getLinks(SpiderPage startp, Spider s) {
List<string> pre_pages = new List<string>();
string final_url = "";
List<SpiderPage> new_pages = new List<SpiderPage>();
StringBuilder sb = new StringBuilder();
byte[] buf = new byte[8192];
HttpWebRequest req = (HttpWebRequest)WebRequest.Create(startp.getUrl());
//req.Timeout = 1000;
HttpWebResponse resp = null;
try {
resp = (HttpWebResponse)req.GetResponse();
}
catch (Exception e) {
s.writeStatus("ERROR: " + e.Message);
s.writeStatus("\tpage - " + startp.getUrl() + "\n\t\treferred to by:");
List<string> curr_refs = startp.getReferencedByUrls();
for (int i = 0; i < curr_refs.Count; i++) {
s.writeStatus("\t\t\t" + curr_refs.ElementAt(i));
}
}
if (resp != null) {
// record the final Url after any redirects from this link
final_url = resp.ResponseUri.AbsoluteUri;
Stream resp_stream = resp.GetResponseStream();
string temp_string = null;
int count = 0;
do {
count = resp_stream.Read(buf, 0, buf.Length);
if (count != 0) {
temp_string = Encoding.ASCII.GetString(buf, 0, count);
sb.Append(temp_string);
}
}
while (count > 0);
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(sb.ToString());
var linksOnPage = from lnks in doc.DocumentNode.Descendants()
where lnks.Name == "a" &&
lnks.Attributes["href"] != null &&
lnks.InnerText.Trim().Length > 0
select new {
Url = lnks.Attributes["href"].Value,
};
foreach (var link in linksOnPage) {
if (link.Url.StartsWith("/")) {
if (link.Url.EndsWith("/")) {
pre_pages.Add(s.getBaseUrl() + link.Url);
}
else {
pre_pages.Add(s.getBaseUrl() + link.Url + "/");
}
}
};
List<string> distinct_pre_pages = pre_pages.Distinct().ToList();
for (int m = 0; m < distinct_pre_pages.Count; m++) {
new_pages.Add(new SpiderPage(distinct_pre_pages.ElementAt(m), startp.getUrl()));
}
}
return new _SpiderDataWrapper_getLinks(final_url, new_pages);
}
void spiderProcess() {
_thread_status.ElementAt(0)[1] = 1;
lock (this) {
int candidate_pages_count = this._candidate_pages.Count;
SpiderPage current_candidate_page = null;
SpiderPage current_master_page = null;
for (int i = 0; i < candidate_pages_count; i++) {
bool found = false;
current_candidate_page = this._candidate_pages.ElementAt(i);
if (current_candidate_page.finalUrlNeeded()) {
for (int m = 0; m < this.masterResults.Count; m++) {
current_master_page = this.masterResults.ElementAt(m);
if (!current_master_page.finalUrlNeeded()
&& current_candidate_page.getUrl() == current_master_page.getFinalUrl()) {
found = true;
List<string> current_candidate_page_ref_urls = current_candidate_page.getReferencedByUrls();
for (int g = 0; g < current_candidate_page_ref_urls.Count; g++) {
if (!current_master_page.getReferencedByUrls().Contains(current_candidate_page_ref_urls.ElementAt(g))) {
this.masterResults.ElementAt(j).addReferencedByUrl(current_candidate_page_ref_urls.ElementAt(g));
}
}
for (int j = 0; j < this.masterResults.Count; j++) {
current_master_page = this.masterResults.ElementAt(j);
if (current_master_page.finalUrlNeeded()) {
if (current_candidate_page.getUrl() == current_master_page.getUrl()) {
}
if (current_candidate_page.getFinalUrl() == this.masterResults.ElementAt(j).getFinalUrl()) {
found = true;
// add an alias entry in the masterResults if we were redirected here
if (current_candidate_page.getUrl() != current_candidate_page.getFinalUrl()) {
this.masterResults.ElementAt(j).addAliasUrl(current_candidate_page.getUrl());
}
// add all the linking URLs from the curr_page object to masterResults
List<string> current_page_link_urls = current_candidate_page.getLinkingToUrls();
for (int q = 0; q < current_page_link_urls.Count; q++) {
if (!this.masterResults.ElementAt(j).getLinkingToUrls().Contains(current_page_link_urls.ElementAt(q))) {
this.masterResults.ElementAt(j).addLinkingToUrl(current_page_link_urls.ElementAt(q));
}
}
// add all the referring URLs from the curr_page object to masterResults
List<string> current_page_ref_urls = current_candidate_page.getReferencedByUrls();
for (int g = 0; g < current_page_ref_urls.Count; g++) {
if (!this.masterResults.ElementAt(j).getReferencedByUrls().Contains(current_page_ref_urls.ElementAt(g))) {
this.masterResults.ElementAt(j).addReferencedByUrl(current_page_ref_urls.ElementAt(g));
}
}
break;
}
}
// if this is a new page...
if (!found) {
this.masterResults.Add(current_candidate_page);
ThreadPool.QueueUserWorkItem(new WaitCallback(spiderFetch), new _SpiderDataWrapper_spiderFetch(this, current_candidate_page));
}
}
// remove all the candidate pages we've just processed
_candidate_pages.RemoveRange(0, candidate_pages_count);
}
Thread.Sleep(30000);
if (this._candidate_pages.Count > 0) {
this.spiderProcess();
}
this._thread_status.ElementAt(0)[1] = 0;
}
/* _candidate_makeNewSpiderPage - makes a new SpiderPage out of this candidate object
*/
public SpiderPage _candidate_makeNewSpiderPage()
{
SpiderPage s = new SpiderPage(this.getUrl(), this.getLinkingToLinks(), this.getReferredByLinks(), this.getPageContent());
if (this._candidate_isAliasCandidate()) {
s.addAliasUrl(this._candidate_getUrl());
}
return s;
}
/* addNewPage() - adds a new page to the _master_results, i.e. a new officially vetted page
* @new_page - the SpiderPage object to add
*/
void addNewPage(SpiderPage new_page)
{
this._master_results.Add(new_page);
}
/* spiderProcess() - master spider process:
*
* PART 1: process the candidate pages that the fetchPage() threads
* crawled after PART 2 of last round, generate a list of new
* links for PART 2 (of this round)
* PART 2: make new fetchPage() threads to crawl the new candidate pages
* found in the links from PART 1
*
*/
static void spiderProcess(object o)
{
// cast our argument back to a Spider object
Spider spider_object = (Spider)o;
// loop spiderProcess() until we're done processing candidate pages
do
{
// wait for all the worker threads to be done before starting each round of spiderProcess()
bool ready = false;
do
{
ready = spider_object.checkWorkerThreads();
} while (!ready);
// all of this is dependent on _master_pages and _candidate_pages, need the spider object locked
lock (spider_object) {
// PART 1: process the candidate pages that were crawled by the worker threads created in
// the last round of spiderProcess()
// list of all the links found in the candidate pages we process
List <SpiderLink> new_links_found = new List <SpiderLink>();
// list of all the candidate page URLs that we add to the master results this round
List <string[]> added_candidate_urls = new List <string[]>();
int candidate_page_count = spider_object._candidate_pages.Count;
// iterative for-loop, don't see a better way to do this really (or why we'd want one)...
for (int i = 0; i < candidate_page_count; i++)
{
bool found = false;
_SpiderPageCandidate current_candidate_page = spider_object.getCandidatePageAtIndex(i);
// make sure this candidate page was crawled by fetchPage(), should be true for every
// candidate page that didn't return a 404 or some error, etc.
if (current_candidate_page._candidate_isDone())
{
// see if this candidate page went to the same final URL as a page that we've already
// added in this round of spiderProcess()
int already_added_candidate_index = added_candidate_urls.FindIndex(delegate(string[] s) {
return(s[0] == current_candidate_page.getUrl());
});
// two tests of whether this candidate page *could* already be in the master results: 1) if this page's
// final URL is in the already-added-list (then it's certainly in the master results), or 2) it was an
// alias candidate (i.e. a redirect to a different final url); otherwise we're guaranteed that this
// candidate page is a new page, and therefore not already in the master results, and all of this
// will be skipped
if (already_added_candidate_index > -1 || current_candidate_page._candidate_isAliasCandidate())
{
int real_page_index = -1;
if (already_added_candidate_index > -1)
{
real_page_index = Int32.Parse(added_candidate_urls.ElementAt(already_added_candidate_index)[1]);
}
else
{
real_page_index = spider_object.findPageIndex(current_candidate_page.getUrl());
}
// was it an existing page after all? if so, add any referring links that have been added to this
// candidate page (i.e. links to its alias address that were found in PART 2 of spiderProcess()
// last time), and add this alias URL to the existing page's list of alias URLs (if it was an alias,
// it's also possible that the link that generated this candidate page was found after a link that
// went to an alias of this page, in which case this one could not be an alias)
if (real_page_index > -1)
{
found = true;
SpiderPage real_page = spider_object.getPageAtIndex(real_page_index);
List <SpiderLink> current_candidate_referred_links = current_candidate_page.getReferredByLinks();
// another iterative for-loop, doesn't need to be improved really afaik?
for (int k = 0; k < current_candidate_referred_links.Count; k++)
{
real_page.addReferredByLink(current_candidate_referred_links.ElementAt(k));
}
if (current_candidate_page._candidate_isAliasCandidate())
{
real_page.addAliasUrl(current_candidate_page._candidate_getUrl());
}
}
}
// this candidate page was a real new page- add it to the master results, add its links to the
// new links found this round, and add it to the list of pages added this round
if (!found)
{
SpiderPage new_page = current_candidate_page._candidate_makeNewSpiderPage();
new_links_found.AddRange(new_page.getLinkingToLinks());
spider_object.addNewPage(new_page);
added_candidate_urls.Add(new string[] { new_page.getUrl(), spider_object.getLastPageIndex().ToString() });
}
// this candidate page is done being processed- remove it from the list
spider_object._candidate_pages.RemoveAt(i);
candidate_page_count--;
i--;
}
}
// PART 2: make new candidate pages from the new links that go to pages we haven't seen before,
// create new fetchPage() worker threads to crawl them
List <_SpiderPageCandidate> new_candidate_pages = new List <_SpiderPageCandidate>();
for (int j = 0; j < new_links_found.Count; j++)
{
SpiderLink current_link = new_links_found.ElementAt(j);
if (current_link.isLegalLink())
{
// see if we've made a new candidate page for this link already
int link_index = -1;
// for-loop being used for search, DEFINITELY can be improved with some
// better data-structure etc.
for (int y = 0; y < new_candidate_pages.Count; y++)
{
if (new_candidate_pages.ElementAt(y)._candidate_getUrl() == current_link.getNormalizedUrl())
{
link_index = y;
break;
}
}
// if we have made a new candidate page already, just add a referred-by link to the
// candidate page we already made
if (link_index > -1)
{
new_candidate_pages.ElementAt(link_index).addReferredByLink(current_link);
}
// otherwise, search the master results to see if we need to create a new candidate
// page or not
else
{
int real_page_index = spider_object.findPageIndex(current_link.getNormalizedUrl());
// if this link's URL exists in the master results already, just add a referred-by link
if (real_page_index > -1)
{
SpiderPage real_page = spider_object.getPageAtIndex(real_page_index);
real_page.addReferredByLink(current_link);
}
// otherwise, make a new candidate page from this link
else
{
new_candidate_pages.Add(new _SpiderPageCandidate(current_link));
}
}
}
}
// create a new fetchPage() worker thread for every new candidate page we made
// iterative for-loop, seems fine...
for (int p = 0; p < new_candidate_pages.Count; p++)
{
spider_object._candidate_pages.Add(new_candidate_pages.ElementAt(p));
spider_object.addThreadStatus();
ThreadPool.QueueUserWorkItem(new WaitCallback(fetchPage),
new _SpiderWorkItemDataWrapper(spider_object, spider_object._candidate_pages.Count - 1));
}
}
}
// loop spiderProcess() until there are either no candidate pages in the list or there are only
// error candidate pages left
while (spider_object._candidate_pages.Count > 0 &&
spider_object._candidate_pages.Any(delegate(_SpiderPageCandidate spc) { return(!spc._candidate_isError()); }));
// we're done spidering now, clear our _thread_status (the 0-index in _thread_status is reserved for
// spiderProcess(), worker threads are indices > 0)
spider_object._thread_status.RemoveAt(0);
}
/* addNewPage() - adds a new page to the _master_results, i.e. a new officially vetted page
* @new_page - the SpiderPage object to add
*/
void addNewPage(SpiderPage new_page)
{
this._master_results.Add(new_page);
}
