/// <summary>
/// Collects the posts out of a quest based on the quest's configuration.
/// </summary>
/// <param name="quest">The quest to read.</param>
/// <param name="token">The cancellation token.</param>
/// <returns>Returns a list of posts extracted from the quest.</returns>
public async Task <List <PostComponents> > ReadQuestAsync(IQuest quest, CancellationToken token)
{
IForumAdapter adapter = await GetForumAdapterAsync(quest, token).ConfigureAwait(false);
if (adapter == null)
{
throw new InvalidOperationException("Unable to acquire forum adapter for the quest.");
}
ThreadRangeInfo rangeInfo = await GetStartInfoAsync(quest, adapter, token).ConfigureAwait(false);
if (rangeInfo == null)
{
throw new InvalidOperationException("Unable to determine post range for the quest.");
}
List <Task <HtmlDocument> > loadedPages = await LoadQuestPagesAsync(quest, adapter, rangeInfo, token).ConfigureAwait(false);
if (loadedPages == null)
{
throw new InvalidOperationException("Unable to load pages for the quest.");
}
List <PostComponents> posts = await GetPostsFromPagesAsync(quest, adapter, rangeInfo, loadedPages, token).ConfigureAwait(false);
if (posts == null)
{
throw new InvalidOperationException("Unable to extract posts from quest pages.");
}
return(posts);
}
/// <summary>
/// Loads the HTML pages that are relevant to a quest's tally.
/// </summary>
/// <param name="quest">The quest being loaded.</param>
/// <param name="adapter">The quest's forum adapter, used to forum the URLs to load.</param>
/// <param name="threadRangeInfo">The range info that determines which pages to load.</param>
/// <param name="token">The cancellation token.</param>
/// <returns>Returns a list of tasks that are handling the async loading of the requested pages.</returns>
private async Task <List <Task <HtmlDocument> > > LoadQuestPagesAsync(
IQuest quest, IForumAdapter adapter, ThreadRangeInfo threadRangeInfo, CancellationToken token)
{
var scanInfo = await GetPagesToScanAsync(quest, adapter, threadRangeInfo, token).ConfigureAwait(false);
int firstPageNumber = scanInfo.Item1;
int lastPageNumber = scanInfo.Item2;
int pagesToScan = scanInfo.Item3;
// We will store the loaded pages in a new List.
List <Task <HtmlDocument> > pages = new List <Task <HtmlDocument> >();
IPageProvider pageProvider = ViewModels.ViewModelService.MainViewModel.PageProvider;
// Initiate the async tasks to load the pages
if (pagesToScan > 0)
{
// Initiate tasks for all pages other than the first page (which we already loaded)
var results = from pageNum in Enumerable.Range(firstPageNumber, pagesToScan)
let pageUrl = adapter.GetUrlForPage(pageNum, quest.PostsPerPage)
let shouldCache = (pageNum == lastPageNumber) ? ShouldCache.No : ShouldCache.Yes
select pageProvider.GetPage(pageUrl, $"Page {pageNum}", CachingMode.UseCache, shouldCache, SuppressNotifications.No, token);
pages.AddRange(results.ToList());
}
return(pages);
}
/// <summary>
/// Get the last page number of the tally. This may be determined solely
/// from the thread range info, or might require information from the
/// provided first page, where we can extract how many pages are in the thread.
/// </summary>
/// <param name="quest">The quest being tallied.</param>
/// <param name="adapter">The forum adapter that handles the quest's thread.</param>
/// <param name="threadRangeInfo">The range of posts that are wanted in the tally.</param>
/// <param name="firstPage">The first page of the tally, from which we can get the page range of the thread.</param>
/// <returns>Returns the last page number of the tally.</returns>
private async Task <int> GetLastPageNumber(IQuest quest, IForumAdapter2 adapter,
ThreadRangeInfo threadRangeInfo, Task <HtmlDocument?> firstPage)
{
// Check for quick results first.
if (threadRangeInfo.Pages > 0)
{
// If the page range has already been determined, use that.
return(threadRangeInfo.Pages);
}
if (!quest.ReadToEndOfThread && !threadRangeInfo.IsThreadmarkSearchResult)
{
// If we're not reading to the end of the thread, just calculate
// what the last page number will be. Pages to scan will be the
// difference in pages +1.
return(ThreadInfo.GetPageNumberOfPost(quest.EndPost, quest));
}
// If we're reading to the end of the thread (end post 0, or based on a threadmark),
// then we need to load the first page to find out how many pages there are in the thread.
var page = await firstPage.ConfigureAwait(false);
if (page == null)
{
throw new InvalidOperationException($"Unable to load first page of {quest.ThreadName}");
}
return(adapter.GetThreadInfo(page).Pages);
}
/// <summary>
/// Acquire a list of page loading tasks for the pages that are intended
/// to be tallied.
/// </summary>
/// <param name="quest">The quest for which the tally is being run.</param>
/// <param name="adapter">The forum adapter that handles the quest's thread.</param>
/// <param name="threadRangeInfo">The range of posts that are wanted in the tally.</param>
/// <param name="token">A cancellation token.</param>
/// <returns>Returns a list of page loading tasks.</returns>
private async Task <List <Task <HtmlDocument?> > > LoadQuestPagesAsync(
IQuest quest, IForumAdapter2 adapter, ThreadRangeInfo threadRangeInfo, IPageProvider pageProvider, CancellationToken token)
{
int firstPageNumber = threadRangeInfo.GetStartPage(quest);
// Get the first page in order to find out how many pages are in the thread
// Keep it as a task.
Task <HtmlDocument?> firstPage = GetFirstPage(firstPageNumber, quest, adapter, pageProvider, token);
// Get the last page number.
int lastPageNumber = await GetLastPageNumber(quest, adapter, threadRangeInfo, firstPage).ConfigureAwait(false);
// Initiate tasks for any remaining pages
IEnumerable <Task <HtmlDocument?> > remainingPages =
GetRemainingPages(firstPageNumber, lastPageNumber, quest, adapter, pageProvider, token);
// Collect all the page load tasks (including the finished first page) to return to caller.
List <Task <HtmlDocument?> > pagesToLoad = new List <Task <HtmlDocument?> >()
{
firstPage
};
pagesToLoad.AddRange(remainingPages);
return(pagesToLoad);
}
/// <summary>
/// Collects the posts out of a quest based on the quest's configuration.
/// </summary>
/// <param name="quest">The quest to read.</param>
/// <param name="pageProvider">The page provider to use to read this quest.</param>
/// <param name="token">The cancellation token.</param>
/// <returns>Returns a list of posts extracted from the quest.</returns>
private async Task <(string threadTitle, List <Post> posts)> ReadQuestAsyncImpl(
IQuest quest, IPageProvider pageProvider, CancellationToken token)
{
logger.LogDebug($"Reading quest {quest.DisplayName} with ForumReader.");
IForumAdapter2 adapter = await forumAdapterFactory.CreateForumAdapterAsync(quest, pageProvider, token).ConfigureAwait(false);
logger.LogDebug($"Forum adapter created for {quest.DisplayName}.");
SyncQuestWithForumAdapter(quest, adapter);
logger.LogDebug($"Quest {quest.DisplayName} synced with forum adapter.");
ThreadRangeInfo rangeInfo = await GetStartInfoAsync(quest, adapter, pageProvider, token).ConfigureAwait(false);
logger.LogDebug($"Range info acquired for {quest.DisplayName}. ({rangeInfo})");
List <Task <HtmlDocument?> > loadingPages = await LoadQuestPagesAsync(quest, adapter, rangeInfo, pageProvider, token).ConfigureAwait(false);
logger.LogDebug($"Got {loadingPages.Count} pages loading {quest.DisplayName}.");
var(threadInfo, posts2) = await GetPostsFromPagesAsync(loadingPages, quest, adapter, rangeInfo).ConfigureAwait(false);
logger.LogDebug($"Got {posts2.Count} posts for quest {quest.DisplayName}.");
List <Post> filteredPosts = FilterPosts(posts2, quest, threadInfo, rangeInfo);
logger.LogDebug($"Filtered to {filteredPosts.Count} posts for quest {quest.DisplayName}.");
return(threadInfo.Title, filteredPosts);
}
/// <summary>
/// Gets the thread range info (page and post numbers) based on the quest configuration.
/// May load pages (such as for checking threadmarks), so will use the ViewModel's page provider.
/// </summary>
/// <param name="quest">The quest we're getting thread info for.</param>
/// <param name="adapter">The quest's forum adapter.</param>
/// <param name="token">The cancellation token.</param>
/// <returns>Returns the quest's thread range info.</returns>
private async Task <ThreadRangeInfo> GetStartInfoAsync(IQuest quest, IForumAdapter adapter, CancellationToken token)
{
IPageProvider pageProvider = ViewModels.ViewModelService.MainViewModel.PageProvider;
ThreadRangeInfo rangeInfo = await adapter.GetStartingPostNumberAsync(quest, pageProvider, token).ConfigureAwait(false);
return(rangeInfo);
}
/// <summary>
/// Gets a list of posts from the provided pages from a quest.
/// </summary>
/// <param name="quest">The quest being tallied.</param>
/// <param name="adapter">The quest's forum adapter.</param>
/// <param name="rangeInfo">The thread range info for the tally.</param>
/// <param name="pages">The pages that are being loaded.</param>
/// <param name="token">The cancellation token.</param>
/// <returns>Returns a list of PostComponents comprising the posts from the threads that fall within the specified range.</returns>
private async Task <List <PostComponents> > GetPostsFromPagesAsync(
IQuest quest, IForumAdapter adapter, ThreadRangeInfo rangeInfo, List <Task <HtmlDocument> > pages, CancellationToken token)
{
List <PostComponents> postsList = new List <PostComponents>();
var firstPageTask = pages.First();
while (pages.Any())
{
var finishedPage = await Task.WhenAny(pages).ConfigureAwait(false);
pages.Remove(finishedPage);
if (finishedPage.IsCanceled)
{
throw new OperationCanceledException();
}
// This will throw any pending exceptions that occurred while trying to load the page.
// This removes the need to check for finishedPage.IsFaulted.
var page = await finishedPage.ConfigureAwait(false);
if (page == null)
{
Exception ae = new Exception("Not all pages loaded. Rerun tally.");
ae.Data["Application"] = true;
throw ae;
}
var posts = from post in adapter.GetPosts(page, quest)
where post != null && post.IsVote && post.IsAfterStart(rangeInfo) &&
(quest.ReadToEndOfThread || rangeInfo.IsThreadmarkSearchResult || post.Number <= quest.EndPost)
select post;
postsList.AddRange(posts);
}
var firstPage = firstPageTask.Result;
ThreadInfo threadInfo = adapter.GetThreadInfo(firstPage);
ViewModelService.MainViewModel.VoteCounter.Title = threadInfo.Title;
// Get all posts that are not filtered out, either explicitly, or (for the thread author) implicity.
postsList = postsList
.Where(p => (
(quest.UseCustomUsernameFilters && !quest.UsernameFilter.Match(p.Author)) || (!quest.UseCustomUsernameFilters && p.Author != threadInfo.Author)) &&
(!quest.UseCustomPostFilters || !(quest.PostsToFilter.Contains(p.Number) || quest.PostsToFilter.Contains(p.IDValue))
)
)
.Distinct().OrderBy(p => p.Number).ToList();
return(postsList);
}
/// <summary>
/// Determines the page number range that will be loaded for the quest.
/// Returns a tuple of first page number, last page number, and pages to scan.
/// </summary>
/// <param name="quest">The quest being tallied.</param>
/// <param name="adapter">The forum adapter for the quest.</param>
/// <param name="threadRangeInfo">The thread range info, as provided by the adapter.</param>
/// <param name="token">The cancellation token.</param>
/// <returns>Returns a tuple of the page number info that was determined.</returns>
private async Task <Tuple <int, int, int> > GetPagesToScanAsync(
IQuest quest, IForumAdapter adapter, ThreadRangeInfo threadRangeInfo, CancellationToken token)
{
IPageProvider pageProvider = ViewModels.ViewModelService.MainViewModel.PageProvider;
int firstPageNumber = threadRangeInfo.GetStartPage(quest);
int lastPageNumber = 0;
int pagesToScan = 0;
if (threadRangeInfo.Pages > 0)
{
// If the startInfo obtained the thread pages info, just use that.
lastPageNumber = threadRangeInfo.Pages;
}
else if (quest.ReadToEndOfThread || threadRangeInfo.IsThreadmarkSearchResult)
{
// If we're reading to the end of the thread (end post 0, or based on a threadmark),
// then we need to load the first page to find out how many pages there are in the thread.
// Make sure to bypass the cache, since it may have changed since the last load.
string firstPageUrl = adapter.GetUrlForPage(firstPageNumber, quest.PostsPerPage);
HtmlDocument page = await pageProvider.GetPage(firstPageUrl, $"Page {firstPageNumber}",
CachingMode.BypassCache, ShouldCache.Yes, SuppressNotifications.No, token)
.ConfigureAwait(false);
if (page == null)
{
throw new InvalidOperationException($"Unable to load web page: {firstPageUrl}");
}
lastPageNumber = adapter.GetThreadInfo(page).Pages;
}
else
{
// If we're not reading to the end of the thread, just calculate
// what the last page number will be. Pages to scan will be the
// difference in pages +1.
lastPageNumber = quest.GetPageNumberOf(quest.EndPost);
}
pagesToScan = lastPageNumber - firstPageNumber + 1;
Tuple <int, int, int> result = new Tuple <int, int, int>(firstPageNumber, lastPageNumber, pagesToScan);
return(result);
}
/// <summary>
/// Gets all posts from the provided pages list.
/// </summary>
/// <param name="loadingPages">The pages that are being loaded for the tally.</param>
/// <param name="quest">The quest being tallied.</param>
/// <param name="adapter">The forum adapter that handles the quest's thread.</param>
/// <returns>Returns all posts extracted from all pages provided,
/// and the thread title.</returns>
private async Task <(ThreadInfo threadInfo, List <Post> posts)> GetPostsFromPagesAsync(
List <Task <HtmlDocument?> > loadingPages,
IQuest quest, IForumAdapter2 adapter,
ThreadRangeInfo threadRangeInfo)
{
ThreadInfo? threadInfo = null;
List <Post> postsList = new List <Post>();
int pageNumber = threadRangeInfo.GetStartPage(quest) - 1;
bool incomplete = false;
foreach (var loadingPage in loadingPages)
{
var page = await loadingPage.ConfigureAwait(false);
pageNumber++;
if (page == null)
{
incomplete = true;
continue;
}
if (threadInfo == null)
{
threadInfo = adapter.GetThreadInfo(page);
}
postsList.AddRange(adapter.GetPosts(page, quest, pageNumber));
}
if (incomplete)
{
InvalidOperationException e = new InvalidOperationException("Unable to load all pages.");
e.Data["Application"] = true;
throw e;
}
if (threadInfo == null)
{
threadInfo = new ThreadInfo("Unknown", "Unknown", 0);
}
return(threadInfo, postsList);
}
/// <summary>
/// Run the provided post list through the various filters, as set
/// by quest options and post numbers, etc.
/// </summary>
/// <param name="postsList">The list of posts to filter.</param>
/// <param name="quest">The quest with relevant options.</param>
/// <param name="threadInfo">Thread info provides the thread author.</param>
/// <param name="rangeInfo">Range info provides information on the range of valid posts.</param>
/// <returns>Returns a list of posts that satisfy the filtering criteria.</returns>
private List <Post> FilterPosts(List <Post> postsList,
IQuest quest, ThreadInfo threadInfo, ThreadRangeInfo rangeInfo)
{
// Remove any posts that are not votes, that aren't in the valid post range, or that
// hit any filters the quest has set up. Then do a grouping to get distinct results.
var filtered = from post in postsList
where post.HasVote &&
(PostIsAfterStart(post, rangeInfo) && PostIsBeforeEnd(post, quest, rangeInfo)) &&
((quest.UseCustomUsernameFilters && !quest.UsernameFilter.Match(post.Origin.Author)) ||
(!quest.UseCustomUsernameFilters && !string.Equals(post.Origin.Author, threadInfo.Author, StringComparison.Ordinal))) &&
(!quest.UseCustomPostFilters ||
!(quest.PostsToFilter.Contains(post.Origin.ThreadPostNumber) ||
quest.PostsToFilter.Contains(post.Origin.ID.Value)))
// Group to deal with sticky posts that should only be processed once.
group post by post.Origin.ThreadPostNumber into postNumGroup
orderby postNumGroup.Key
select postNumGroup.First();
return(filtered.ToList());
}
/// <summary>
/// Check whether the given post is before the endpoint of the tally.
/// </summary>
/// <param name="post">The post to check.</param>
/// <param name="quest">Quest options.</param>
/// <param name="rangeInfo">Specific range information.</param>
/// <returns>Returns true if the post comes before the end of the tally.</returns>
private bool PostIsBeforeEnd(Post post, IQuest quest, ThreadRangeInfo rangeInfo)
{
return(quest.ReadToEndOfThread || rangeInfo.IsThreadmarkSearchResult || post.Origin.ThreadPostNumber <= quest.EndPost);
}
/// <summary>
/// Check whether the given post is after the startpoint of the tally.
/// </summary>
/// <param name="post">The post to check.</param>
/// <param name="rangeInfo">The range which shows where the tally starts.</param>
/// <returns>Returns true if the post comes after the start of the tally.</returns>
private bool PostIsAfterStart(Post post, ThreadRangeInfo rangeInfo)
{
return((rangeInfo.ByNumber && post.Origin.ThreadPostNumber >= rangeInfo.Number) || (!rangeInfo.ByNumber && post.Origin.ID > rangeInfo.ID));
}
/// <summary>
/// Gets the thread range info (page and post numbers) based on the quest configuration.
/// May load pages (such as for checking threadmarks), so will use the ViewModel's page provider.
/// </summary>
/// <param name="quest">The quest we're getting thread info for.</param>
/// <param name="adapter">The quest's forum adapter.</param>
/// <param name="token">The cancellation token.</param>
/// <returns>Returns the quest's thread range info.</returns>
private async Task <ThreadRangeInfo> GetStartInfoAsync(IQuest quest, IForumAdapter2 adapter, IPageProvider pageProvider, CancellationToken token)
{
ThreadRangeInfo rangeInfo = await adapter.GetQuestRangeInfoAsync(quest, pageProvider, token).ConfigureAwait(false);
return(rangeInfo);
}
