public static void AnalyzePostTimes(string storiesFilePath, string postTimeStatsFilePath)
{
//Get posts and convert to date
var posts = JsonNetUtils.DeserializeSequenceFromJson <JObject>(storiesFilePath)
.SelectMany(responseJson => responseJson["hits"])
.Select(hitJson => new StoryRating
{
CreatedOn = Utils.FromUnixTime(hitJson["created_at_i"].Value <long>()).ToLocalTime(),
Points = hitJson["points"].Value <int>(),
Comments = hitJson["num_comments"].Value <int>()
}).ToList();
var postsPerMinutes = posts
.GroupBy(post => post.CreatedOn.ToString("ddd hh:mm tt"))
.OrderBy(g => g.Key);
using (var outputFile = File.CreateText(postTimeStatsFilePath))
{
outputFile.WriteLine("Time\tPoints\tComments");
foreach (var postsPerMinute in postsPerMinutes)
{
outputFile.WriteLine("{0}\t{1}\t{2}".FormatEx(postsPerMinute.Key,
postsPerMinute.Select(p => p.Points.ToStringInvariant()).ToDelimitedString(","),
postsPerMinute.Select(p => p.Comments.ToStringInvariant()).ToDelimitedString(",")
));
}
}
}
private void buttongetStats_Click(object sender, EventArgs e)
{
foreach (var filePath in new[] { textBoxStoriesFilePath.Text, textBoxCommentsFilePath.Text })
{
var createDatesUnix = JsonNetUtils.DeserializeSequenceFromJson <JObject>(filePath)
.SelectMany(responseJson => responseJson["hits"])
.Select(hitJson => hitJson["created_at_i"].Value <long>());
long minDateUnix = long.MaxValue, maxDateUnix = long.MinValue, count = 0;
foreach (var createDateUnix in createDatesUnix)
{
minDateUnix = Math.Min(minDateUnix, createDateUnix);
maxDateUnix = Math.Max(maxDateUnix, createDateUnix);
count++;
}
MessageBox.Show("{3}\n{0} items from {1} to {2}".FormatEx(count, Utils.FromUnixTime(minDateUnix).ToString("r"), Utils.FromUnixTime(maxDateUnix).ToString("r"), filePath));
}
}
private void buttonAnalyzeUniqueUrls_Click(object sender, EventArgs e)
{
var storyUrlsStats = new Dictionary <string, StoryUrlStats>();
int storyCount = 0, blankUrls = 0, badUrls = 0;
//Consider stories upto 3 years old for this analysis
var minUnixDate = DateTime.UtcNow.Subtract(TimeSpan.FromDays(365 * 3)).ToUnixTime();
foreach (var responseJson in JsonNetUtils.DeserializeSequenceFromJson <JObject>(textBoxStoriesFilePath.Text))
{
try
{
foreach (var hitJson in responseJson["hits"])
{
storyCount++;
if (hitJson["created_at_i"].Value <long>() < minUnixDate)
{
goto endFileReading;
}
var urlString = hitJson["url"].ToString();
if (string.IsNullOrWhiteSpace(urlString))
{
blankUrls++;
continue;
}
try
{
var uri = new Uri(urlString);
var host = uri.Host;
var stats = storyUrlsStats.AddOrGetValue(host, () => StoryUrlStats.Create(urlString));
stats.AddStory(hitJson);
}
catch (UriFormatException)
{
badUrls++;
}
}
}
catch {}
}
endFileReading:
Task.WaitAll(storyUrlsStats.Values.Select(s => s.AnalyzeWpTask).ToArray());
using (var urlStatsFile = File.CreateText(textBoxUrlAnalysisFilePath.Text))
{
foreach (var urlStatsKvp in storyUrlsStats)
{
if (string.IsNullOrWhiteSpace(urlStatsKvp.Value.ThemeNames) && string.IsNullOrWhiteSpace(urlStatsKvp.Value.PluginNames))
{
continue;
}
//Tab seperated Column Header
//Host URLSample PointsSum ThemeNames PluginNames StoryCount CommentsSum MaxDate MinDate
var values = new string[] { urlStatsKvp.Key, urlStatsKvp.Value.Url, urlStatsKvp.Value.PointsSum.ToStringInvariant(),
urlStatsKvp.Value.ThemeNames, urlStatsKvp.Value.PluginNames,
urlStatsKvp.Value.StoryCount.ToStringInvariant(), urlStatsKvp.Value.CommentsSum.ToStringInvariant(),
urlStatsKvp.Value.CreatedUnixTimeMax.ToString(), urlStatsKvp.Value.CreatedUnixTimeMin.ToString() };
urlStatsFile.WriteLine(values.ToDelimitedString("\t"));
}
}
MessageBox.Show("Total: {0}, Blank: {1}, Bad: {2}".FormatEx(storyCount, blankUrls, badUrls));
}