private Dictionary<string, List<IPeople>> GroupCategories()
{
// Sort categories by projects
ReportProgress(0, "Grouping categories by projects");
var groupedCategories = this.categories.GroupBy(x => x.Project).ToList();
// Find people with more than 1 phase in each project
var projects = new Dictionary<string, List<IPeople>>();
var count = 0;
foreach (var project in groupedCategories)
{
ReportProgress(count, groupedCategories.Count, string.Format("Finding multiple phases people in {0}. Completed: {1}/{2}", project.Key, count, groupedCategories.Count), 50);
var projectPeople = new List<IPeople>();
var phasesPeople = project.Select(x => x.Contributions.Select(y => y.Contributor).Distinct().ToList()).ToList(); // People in their phases
var projectsDistinctPeople = phasesPeople.SelectMany(x => x).Distinct().ToList(); // All distinct people in the current project (from all phases)
var multiplePhasesPeople = projectsDistinctPeople.Where(x => phasesPeople.Count(y => y.Any(z => z == x)) > 1).ToList();
multiplePhasesPeople.ForEach(x =>
{
var person = new People { Name = x };
person.AddProject(project.Key);
projectPeople.Add(person);
});
projects[project.Key] = projectPeople;
}
return projects;
}
public Dictionary<string, List<string>> CountDomains()
{
var items = this.people.SelectMany(x => x.Contributions.Select(y =>
{
var person = new People
{
Name = x.Name,
URL = x.URL
};
var expert = this.specialists.FirstOrDefault(z => z.Name == x.Name && z.URL == x.URL);
if (expert != null)
{
var domains = expert.Skills.ToList();
domains.ForEach(z => person.AddSkill(z));
}
return new
{
Person = person,
Project = y.Project
};
})).GroupBy(x => x.Project)
.Select(x => new
{
Project = x.Key,
Domains = x.SelectMany(y => y.Person.Skills).Distinct().ToList() // All of each persons domain will be added
})
.ToDictionary(x => x.Project, x => x.Domains);
return items;
}
private void GetPeople(string skill, int index, int totalCount, ref List<IPeople> people)
{
var baseUrl = "https://www.quirky.com/api/v1/users/by_skill?skill={0}&paginated_options%5Busers%5D%5Bpage%5D={1}&paginated_options%5Busers%5D%5Bper_page%5D=20&paginated_options%5Busers%5D%5Border_column%5D=id&paginated_options%5Busers%5D%5Border%5D=asc";
var hasMore = true;
var page = 1;
var scrapedCount = 0;
while (hasMore)
{
var url = string.Format(baseUrl, skill, page++);
var json = Helper.GetXHRJson(url);
var jsonObj = json.FromJson<JObject>();
var stats = jsonObj["paginated_meta"]["users"];
hasMore = stats.Value<bool>("has_next_page");
var total = stats.Value<int>("total"); // The number of people with this skill
// Get followers
var users = jsonObj["data"].Value<JArray>("users");
scrapedCount += users.Count;
foreach (var user in users)
{
var personName = user.Value<string>("name");
var personUrl = string.Format(PeopleScraper.USER_URL_FORMAT, user.Value<string>("id"));
var person = people.FirstOrDefault(x => x.Name == personName && x.URL == personUrl);
if (person == null)
{
person = new People
{
Name = personName,
URL = personUrl
};
people.Add(person);
}
person.AddSkill(skill);
}
ReportProgress(index, totalCount,
string.Format("Scraping people with {0} skill... Scraped: {1}/{2}. Progress: {3}/{4}", skill, scrapedCount, total, index, totalCount));
}
ReportProgress(index + 1, totalCount,
string.Format("Completed scraping people with {0} skill. Scraped: {1}. Progress: {2}/{3}", skill, scrapedCount, index + 1, totalCount));
}
public Project(AmazonDetail amazonDetail, People inventor, IEnumerable<Category> categories)
{
AmazonDetail = amazonDetail;
Inventor = inventor;
Categories = categories;
}
private void Populate(int index, int totalCount, ref IPeople person, bool isFollower, string personId, int count)
{
var reportText = isFollower ? "followers" : "followings";
var urlpage = isFollower ? "followers" : "following";
var urlBase = "https://www.quirky.com/api/v1/user_profile/{0}/{1}?paginated_options%5Bfollows%5D%5Buse_cursor%5D=true&paginated_options%5Bfollows%5D%5Bper_page%5D=20&paginated_options%5Bfollows%5D%5Border_column%5D=created_at&paginated_options%5Bfollows%5D%5Border%5D=desc";
var baseUrl = string.Format(urlBase, personId, urlpage);
var urlCursorAddition = "&paginated_options%5Bfollows%5D%5Bcursor%5D={0}";
var hasMore = true;
var firstIteration = true;
string cursor = null;
var scrapedCount = 0;
while (hasMore)
{
var url = baseUrl;
if (firstIteration) firstIteration = false; // First iteration has no pagination cursor
else url += string.Format(urlCursorAddition, cursor);
var json = Helper.GetXHRJson(url);
var jsonObj = JsonConvert.DeserializeObject(json) as JObject;
hasMore = jsonObj["paginated_meta"]["follows"].Value<bool>("has_next_page");
var arr = jsonObj["data"].Value<JArray>("follows");
scrapedCount += arr.Count;
if (hasMore)
{
cursor = arr.Last.Value<string>("created_at");
cursor = Helper.EncodeQuirkyDate(cursor);
if (cursor == null) hasMore = false;
}
// Get followers
var users = jsonObj["data"].Value<JArray>("users");
foreach (var user in users)
{
var personName = user.Value<string>("name");
var personUrl = string.Format(PeopleScraper.USER_URL_FORMAT, user.Value<string>("id"));
var fellow = new People
{
Name = personName,
URL = personUrl
};
if (isFollower)
person.AddFollower(fellow);
else
person.AddFollowing(fellow);
}
ReportProgress(index, totalCount,
string.Format("Scraping {0}'s {1}... Scraped: {2}/{3} {1}. Progress: {4}/{5}", person.Name, reportText, scrapedCount, count, index, totalCount));
}
ReportProgress(index + 1, totalCount,
string.Format("Completed scraping {0}'s {1}. Scraped: {2}/{3} {1}. Progress: {4}/{5}", person.Name, reportText, scrapedCount, count, index + 1, totalCount));
}
public IEnumerable<object> Scrape()
{
var contributors = new List<IPeople>();
var totalCount = categories.Count;
var progress = 0;
ReportProgress(progress, "Starting people scraping...");
this.categories = this.categories.OrderBy(x => x.Project).ToList();
for (var i = 0; i < this.categories.Count; i++)
{
var category = categories[i];
ReportProgress(progress, totalCount,
string.Format("Scraping category: {0} ({1})... Contributions: {2} Progress: {3}/{4}", category.Name, category.Project, category.ContributionNum, i, this.categories.Count));
if (category.ContributionNum == 0) continue; // Nothing to do here
var addCategory = new Category
{
Name = category.Name,
Project = category.Project,
URL = category.URL
};
var contributionsString = "https://www.quirky.com/api/v1/inventions/{0}/with_build_interface_objects?with_random_contributions=true";
var projectId = Regex.Match(category.URL, "(?<=/invent/)[0-9]+(?=/)");
var json = Helper.GetXHRJson(string.Format(contributionsString, projectId));
if (json == null) continue;
var jsonObj = JsonConvert.DeserializeObject(json) as JObject;
var scrapeCount = 0;
var catDetails = jsonObj["data"]["projects"].FirstOrDefault(x => x.Value<string>("human_name") == category.Name);
if (catDetails != null)
{
var catId = catDetails.Value<long>("id");
var contributionString = "https://www.quirky.com/api/v1/contributions/for_project?parent_id={0}&parent_class=Project&paginated_options%5Bcontributions%5D%5Buse_cursor%5D=true&paginated_options%5Bcontributions%5D%5Bper_page%5D=20&paginated_options%5Bcontributions%5D%5Border_column%5D=created_at&paginated_options%5Bcontributions%5D%5Border%5D=desc";
var baseUrl = string.Format(contributionString, catId);
json = Helper.GetXHRJson(baseUrl);
var additional = "&paginated_options%5Bcontributions%5D%5Bcursor%5D={0}";
var hasMore = true;
while (hasMore)
{
jsonObj = JsonConvert.DeserializeObject(json) as JObject;
hasMore = jsonObj["paginated_meta"]["contributions"].Value<bool>("has_next_page");
var arr = jsonObj["data"].Value<JArray>("contributions");
scrapeCount += arr.Count;
var cursor = arr.Last().Value<string>("created_at");
cursor = Helper.EncodeQuirkyDate(cursor);
if (cursor == null) hasMore = false;
var url = baseUrl + string.Format(additional, cursor);
// Get contributors
var users = jsonObj["data"].Value<JArray>("users");
foreach (var user in users)
{
var personName = user.Value<string>("name");
var personUrl = string.Format(PeopleScraper.USER_URL_FORMAT, user.Value<string>("id"));
var person = contributors.FirstOrDefault(x => x.Name == personName && x.URL == personUrl);
if (person == null)
{
person = new People
{
Name = personName,
URL = personUrl
};
person.AddContribution(addCategory);
contributors.Add(person);
}
else
person.AddContribution(addCategory);
}
ReportProgress(progress, totalCount,
string.Format("Scraping category: {0} ({1})... Scraped: {2}/{3} Progress: {4}/{5}", category.Name, category.Project, scrapeCount, category.ContributionNum, i, this.categories.Count));
if (hasMore)
{
json = Helper.GetXHRJson(url);
}
}
}
ReportProgress(++progress, totalCount,
string.Format("Completed scraping category: {0} ({1}). Scraped: {2}/{3} Progress: {4}/{5}", category.Name, category.Project, scrapeCount, category.ContributionNum, i, this.categories.Count));
}
MessageBox.Show("People scraping completed...");
return contributors;
}
