public void GoThroughEachPostAndGetTheCommentsOhMyGodThisWillDestroyMyLaptop()
{
const int LastScrapeAmount = 0;
int i = 0;
AllResponse <ScrapedPost> posts = PostScraper.All(new SortField[] { new SortField {
Field = "created_time", Order = SortOrder.Descending
} });
foreach (ScrapedPost post in posts.Data)
{
i++;
if (post.CreatedTime < new DateTime(2017, 04, 01))
{
continue;
}
if (i > LastScrapeAmount)
{
List <ScrapedComment> comments = CommentScraper.Scrape(post).ToList();
Console.WriteLine($"{i}/{posts.TotalCount}: {post.Id}; {comments.Count}");
}
else
{
Console.WriteLine($"{i}/{posts.TotalCount}: {post.Id}; Already scraped.");
}
}
}
public PostScrapeHistory ScrapePosts([FromBody] PostScrapeRequest request)
{
Debug.Assert(request != null);
Console.WriteLine("Started Scraping");
// If no specific pages were specified, scrape them all.
PageMetadata[] pages;
if (request.Pages == null)
{
pages = PageMetadataRepository.All().Data.ToArray();
}
else
{
pages = request.Pages.Select(p => PageMetadataRepository.Get(p)).ToArray();
}
int numberOfComments = 0;
ScrapedPost[] posts = PostScraper.Scrape(pages, request.Since, request.Until).ToArray();
Console.WriteLine($"Started scraping comments for {posts.Length} posts");
foreach (ScrapedPost post in posts)
{
ScrapedComment[] comments = CommentScraper.Scrape(post).ToArray();
numberOfComments += comments.Length;
Console.WriteLine(numberOfComments);
}
Console.WriteLine($"Done scraping {pages.Length} pages. Scraped {posts.Length} posts with {numberOfComments} comments");
var postScrape = new PostScrapeHistory
{
Id = Guid.NewGuid().ToString(),
Since = request.Since,
Until = request.Until,
ImportStart = posts.FirstOrDefault()?.Scraped ?? DateTime.Now,
ImportEnd = DateTime.Now,
NumberOfPosts = posts.Length,
NumberOfComments = numberOfComments,
Pages = pages
};
return(PostScrapeHistoryRepository.Save(postScrape));
}
public IEnumerable <ScrapedComment> ScrapeComments([FromBody] CommentScrapeRequest request)
{
return(CommentScraper.Scrape(PostScraper.Get(request?.PostId)));
}
