comparison extra/soundsoftware/get-apache-log-stats.rb @ 979:56a38a9f6204 cannam

Various fixes

 @projects = Hash.new
 
 parseable = 0
 unparseable = 0
 
-def known_project?(project)
+def is_public_project?(project)
   if !project
     false
   elsif @projects.key?(project)
-    true
+    @projects[project].is_public? 
   else
     pobj = Project.find_by_identifier(project)
     if pobj
       @projects[project] = pobj
-      true
+      pobj.is_public?
     else
-      print "Project not found: ", project
+      print "Project not found: ", project, "\n"
       false
     end
   end
 end
 
-ARGF.each do |line|
+STDIN.each do |line|
 
   record = parser.parse(line)
 
   # most annoyingly, the parser can't handle the comma-separated list
   # in X-Forwarded-For where it has more than one element. If it has

     record = parser.parse(filtered)
   end
 
   # discard, but count, unparseable lines
   if not record
+    print "Line not parseable: ", line, "\n"
     unparseable += 1
     next
   end
 
   # discard everything that isn't a 200 OK response

   # pull out request e.g. GET / HTTP/1.0
   request = record["%r"]
 
   # split into method, path, protocol
   if not request =~ /^[^\s]+ ([^\s]+) [^\s]+$/
+    print "Line not parseable (bad method, path, protocol): ", line, "\n"
     unparseable += 1
     next
   end
 
   # get the path e.g. /projects/weevilmatic and split on /

   components = path.split("/")
   
   # should have at least two elements unless path is "/"; first should
   # be empty (begins with /)
   if path != "/" and (components.size < 2 or components[0] != "")
+    print "Line not parseable (degenerate path): ", line, "\n"
     unparseable += 1
     next
   end
 
   if components[1] == "hg"
     
     # path is /hg/project?something or /hg/project/something
 
     project = components[2].split("?")[0]
-    if not known_project?(project)
+    if not is_public_project?(project)
       next
     end
 
     if components[2] =~ /&roots=00*$/
       clones[project] += 1

   elsif components[1] == "projects"
 
     # path is /projects/project or /projects/project/something
 
     project = components[2]
-    if not known_project?(project)
+    project = project.split("?")[0] if project
+    if not is_public_project?(project)
       next
     end
 
     project = project.split("?")[0]
     hits[project] += 1