lib/todidnt/git_history.rb in todidnt-0.3.1 vs lib/todidnt/git_history.rb in todidnt-0.4.1

@@ -1,14 +1,18 @@
 module Todidnt
   class GitHistory
 
     attr_accessor :blames
 
-    def initialize
+    def initialize(opts)
       @history = []
       @blames = {}
       @unmatched_deletions = []
+
+      # Contributor threshold (e.g. only show as a separate contributor
+      # if they've contributed to > N% of TODOs).
+      @threshold = opts[:threshold]
     end
 
     def timeline!
       # TODO: It would probably be better/simpler to just Marshal the
       # GitHistory object itself.
@@ -63,10 +67,12 @@
       commit = nil
       seen_commits = Set.new
       count = 0
 
       command.execute! do |line|
+        line.encode!('UTF-8', 'binary', invalid: :replace, undef: :replace, replace: '')
+
         if (diff = /diff --git a\/(.*) b\/(.*)/.match(line))
           filename = diff[1]
         elsif (diff = /^\+(.*TODO.*)/.match(line))
           patch_additions << diff[1] unless filename =~ TodoLine::IGNORE
         elsif (diff = /^\-(.*TODO.*)/.match(line))
@@ -150,20 +156,26 @@
       timespan = max_commit_date - min_commit_date
 
       # Figure out what the interval should be based on the total timespan.
       if timespan > 86400 * 365 * 10 # 10+ years
         interval = 86400 * 365 # years
+        timespan_label = 'years'
       elsif timespan > 86400 * 365 * 5 # 5-10 years
         interval = 86400 * (365 / 2) # 6 months
+        timespan_label = 'months'
       elsif timespan > 86400 * 365 # 2-5 years
         interval = 86400 * (365 / 4) # 3 months
+        timespan_label = 'months'
       elsif timespan > 86400 * 30 * 6 # 6 months-3 year
         interval = 86400 * 30 # months
-      elsif timespan > 86400 * 1 # 1 month - 6 months
+        timespan_label = 'months'
+      elsif timespan > 86400 * 14 * 1 # 1/2 month - 6 months
         interval = 86400 * 7
-      else # 0 - 2 months
+        timespan_label = 'days'
+      else # 0 - 1/2 months
         interval = 86400 # days
+        timespan_label = 'days'
       end
 
       original_interval_start = Time.new(min_commit_date.year, min_commit_date.month, min_commit_date.day).to_i
       interval_start = original_interval_start
       interval_end = interval_start + interval
@@ -173,11 +185,11 @@
       current_bucket_authors = {}
       bucket_total = 0
 
       # Add the first bucket of 0
       buckets << {
-        :timestamp => Time.at(interval_start).strftime('%D'),
+        :timestamp => format_date(Time.at(interval_start), timespan_label),
         :authors => {},
         :total => 0
       }
 
       i = 0
@@ -198,11 +210,11 @@
 
         # If we're on the last slice, or the next slice would have been
         # in a new bucket, finish the current bucket.
         if i == (@history.length - 1) || @history[i + 1][:timestamp] >= interval_end
           buckets << {
-            :timestamp => ([Time.at(interval_end), max_commit_date].min).strftime('%D'),
+            :timestamp => format_date(([Time.at(interval_end), max_commit_date].min), timespan_label),
             :authors => current_bucket_authors,
             :total => bucket_total
           }
           interval_start += interval
           interval_end += interval
@@ -219,11 +231,11 @@
         significant_authors = {}
         other_count = 0
         bucket[:authors].each do |author, count|
           # Only include the author if they account for more than > 3% of
           # the TODOs in this bucket.
-          if count > bucket[:total] * 0.03
+          if count > bucket[:total] * @threshold
             significant_authors[author] = count
             authors << author
           else
             other_count += count
           end
@@ -242,7 +254,19 @@
       end
 
       [buckets, authors]
     end
 
+    private
+
+    def format_date(date, timespan_label)
+      case timespan_label
+      when 'years'
+        date.strftime('%Y')
+      when 'months'
+        date.strftime('%-m/%y')
+      when 'days'
+        date.strftime('%-m/%-d')
+      end
+    end
   end
 end