src/strip.vala
[app.mailtrimmer] / src / strip.vala
index a5332df..ee3132e 100644 (file)
@@ -1,5 +1,13 @@
 /**
 
+ ** check left to do:  
+  - range scans on maildir
+  - see how replacing the links works in the resulting email via thunderbird etc..
+  - some checksum issues (see dupelicates?? suspect 0byte issues?)  
+
+
+
   needs to scan 2 things
   a) our mailfort email database
        point it at the top directory, containing YEAR/MONTH/DAY.... directories.
@@ -82,12 +90,13 @@ public class StripApplication : GLib.Application {
 
        public static int    opt_limit = -1;
 
-       public static int    opt_min_age = 6;
+       public static int    opt_age_newest = 1;
+       public static int    opt_age_oldest = 6;
 
 
        public static bool      opt_is_extracting = false;
        public static bool      opt_is_replacing = false;
-       public static bool      opt_scan_imap  = false; 
+       public static bool      opt_scan_maildir  = false; 
        public static bool      opt_scan_mailfort  = false;     
        public static bool              opt_dump = false;       
        public static bool              opt_debug = false; 
@@ -118,9 +127,10 @@ public class StripApplication : GLib.Application {
                { "dump", 0, 0, OptionArg.NONE, ref opt_dump, "Print the replaced mail contents to stdout", null },         
 
                { "limit", 0, 0, OptionArg.INT, ref opt_limit, "stop after X number of messages with attachments have been processed", null },         
-               { "min-age", 0, 0, OptionArg.INT, ref opt_min_age, "minimum age of messages (default is 6 months)", null },
+               { "newest", 0, 0, OptionArg.INT, ref opt_age_newest, "do not replace messages newer that X months (default is 1 months)", null },
+               { "oldest", 0, 0, OptionArg.INT, ref opt_age_oldest, "do not replace messages older than X (default is 6 months)", null },
 
-               { "scan-imap", 0, 0, OptionArg.NONE, ref opt_scan_imap, "scan an imap tree", null },
+               { "scan-maildir", 0, 0, OptionArg.NONE, ref opt_scan_maildir, "scan an maildir tree", null },
                { "scan-mailfort", 0, 0, OptionArg.NONE, ref opt_scan_mailfort, "scan a mailfort tree", null },  
                { null }       
        };         
@@ -144,7 +154,7 @@ public class StripApplication : GLib.Application {
  
                            
              // options that have to be set.. bee or hive... (or stop all)
-            if ((!opt_scan_mailfort && !opt_scan_imap) || (opt_scan_mailfort && opt_scan_imap))  {
+            if ((!opt_scan_mailfort && !opt_scan_maildir) || (opt_scan_mailfort && opt_scan_maildir))  {
                stdout.printf ("You must specify the type of directory tree to scan - either imap or mailfort\n%s",
                    opt_context.get_help(true, null));
                GLib.Process.exit(Posix.EXIT_FAILURE);
@@ -228,7 +238,7 @@ public class StripApplication : GLib.Application {
                return 0;
         }
 
-               strip.scan_dir(opt_path);
+               strip.scan_dir(opt_path, "");
         
 
         
@@ -656,11 +666,19 @@ public class Strip : GLib.Object {
                        mailtime = new DateTime.local(int.parse(bits[0]),int.parse(bits[1]),int.parse(bits[2]),0,0,0);
                        
                        var oldest = new  DateTime.now_local();
-                       oldest = oldest.add_months(-1 * StripApplication.opt_min_age);
+                       oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
                        var tspan = mailtime.difference(oldest) / GLib.TimeSpan.DAY;
 
                        if (tspan < 0) {
-                               GLib.debug("skip file is %d days older than %d months", (int)tspan, StripApplication.opt_min_age);
+                               GLib.debug("skip file is %d days older than %d months", (int)tspan, StripApplication.opt_age_oldest);
+                               return;
+                       }
+                       
+                       var newest = new  DateTime.now_local();
+                       newest = newest.add_months(-1 * StripApplication.opt_age_newest);
+                       tspan = mailtime.difference(newest) / GLib.TimeSpan.DAY;
+                       if (tspan > 0) {
+                               GLib.debug("skip file is %d days newer than %d months", (int)tspan, StripApplication.opt_age_newest);
                                return;
                        }
                        
@@ -680,7 +698,7 @@ public class Strip : GLib.Object {
                // it's a mail directory...
                // use the last modification time? as the default...
                 mailtime = new DateTime.from_timeval_utc(mod_time);
-                this.created_dir = mailtime.format("/%Y/%m/%d");
+                this.created_dir = mailtime.format("%Y/%m/%d");
                         this.created_date =  mailtime.format("%Y-%m-%d %H:%M:%S");
  
        }
@@ -711,26 +729,29 @@ public class Strip : GLib.Object {
                var recvd = message.get_header("received");
                this.active_message_exim_id = "";
                if (recvd != null && recvd.length > 1) {
-                       // GLib.debug("RECV: %s", recvd);
+                       GLib.debug("RECV: %s", recvd);
                        var lines = recvd.split("\t");
                        for (var i = 0; i < lines.length;i++) {
                                var bits = lines[i].strip().split(" ");
                                if (bits[0] == "id") {
                                        this.active_message_exim_id = bits[1].replace(";","");
 
-                                       
                                }
+                               
                                if (lines[i].contains(";")) {
                                        var dbits = lines[i].strip().split(";");                                
+                                       GLib.debug("Reading time from : %s", dbits[1]);
                                        var timez = GMime.utils_header_decode_date(dbits[1], null);
                                        if (timez != 0) {
                                                mailtime = new DateTime.from_unix_utc(timez);
                                                this.created_date = mailtime.format("%Y-%m-%d %H:%M:%S");
+                                               GLib.debug("Time is %s",this.created_date);
                                                // if it's not mailfort we can use that date to determine where to store it...
                                                if (!StripApplication.opt_scan_mailfort) {
-                                                       this.created_dir = mailtime.format("/%Y/%m/%d");
+                                                       this.created_dir = mailtime.format("%Y/%m/%d");
                                                }
+                                       } else {
+                                               GLib.debug("Could not read time from headers?");
                                        }
                                }
 
@@ -738,14 +759,21 @@ public class Strip : GLib.Object {
                }
                
                var oldest = new  DateTime.now_local();
-               oldest.add_months(-1 * StripApplication.opt_min_age);
+               oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
                var rtspan = mailtime.difference(oldest) / GLib.TimeSpan.DAY;
-
+               GLib.debug("Checking oldest %d days difference", (int)rtspan   );
                if (rtspan < 0) {
-                       GLib.debug("skip file is %d days older than %d months", (int)rtspan, StripApplication.opt_min_age);
+                       GLib.debug("skip(2) file is %d days older than %d months", (int)rtspan, StripApplication.opt_age_oldest);
+                       return;
+               }
+               var newest = new  DateTime.now_local();
+               newest = newest.add_months(-1 * StripApplication.opt_age_newest);
+               rtspan = mailtime.difference(newest) / GLib.TimeSpan.DAY;
+               if (rtspan > 0) {
+                       GLib.debug("skip(2) file is %d days newer than %d months : %s", (int)rtspan, StripApplication.opt_age_newest,
+                               mailtime.format("%Y-%m-%d %H:%M:%S"));
                        return;
                }
-
                
                
                
@@ -842,9 +870,58 @@ public class Strip : GLib.Object {
     }
     
     
-    public void scan_dir(string path)
+    public void scan_dir(string basepath, string subpath)
     {
-        var f = File.new_for_path(path);
+        
+        
+        // determine if path is to old to scan..
+        if (subpath.length > 0 && StripApplication.opt_scan_mailfort) {
+                       var year =  int.parse(subpath.substring(1,4));  // "/2000"
+                       var month = subpath.length > 5 ? int.parse(subpath.substring(6,2)) : 999; // "/2000/12"                 
+                       var day = subpath.length > 8 ? int.parse(subpath.substring(9,2)) : 999; // "/2000/12/01"                        
+               
+               var oldest = new  DateTime.now_local();
+                       oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
+                       
+                       //GLib.debug("Checking directory %s is older than min: %d/%d/%d", subpath, oldest.get_year() , oldest.get_month(), oldest.get_day_of_month() );                                 
+                       
+                       if (year < oldest.get_year()) {
+                               GLib.debug("Skip directory %s is older than min year: %d", subpath, oldest.get_year());
+                               return;
+                       }
+                       if (year == oldest.get_year() &&  month < oldest.get_month()) {
+                               GLib.debug("Skip directory %s is older than min month: %d/%d", subpath, oldest.get_year() , oldest.get_month() );
+                               return;
+                       }
+               if (year == oldest.get_year() &&  month == oldest.get_month() && day < oldest.get_day_of_month()) {
+                               GLib.debug("Skip directory %s is older than min day: %d/%d/%d", subpath, oldest.get_year() , oldest.get_month(), oldest.get_day_of_month() );           
+                               return;
+                       }
+               
+               var newest = new  DateTime.now_local();
+                       newest = newest.add_months(-1 * StripApplication.opt_age_newest);
+                       
+                       //GLib.debug("Checking directory %s is newer than max: %d/%d/%d", subpath, newest.get_year() , newest.get_month(), newest.get_day_of_month() );                                 
+                       
+                       if (year > newest.get_year()) {
+                               GLib.debug("Skip directory %s is newer than max year: %d", subpath, newest.get_year());
+                               return;
+                       }
+                       if (year == newest.get_year() &&  month != 999 && month > newest.get_month()) {
+                               GLib.debug("Skip directory %s is newer than max month: %d/%d", subpath, newest.get_year() , newest.get_month() );
+                               return;
+                       }
+               if (year == newest.get_year() &&  month == newest.get_month() &&  day != 999 && day > newest.get_day_of_month()) {
+                               GLib.debug("Skip directory %s is newer than max day: %d/%d/%d", subpath, newest.get_year() , newest.get_month(), newest.get_day_of_month() );           
+                               return;
+                       }
+               
+               
+               
+        }
+        
+        
+        var f = File.new_for_path(basepath + subpath);
                FileEnumerator file_enum;
         var cancellable = new Cancellable ();
         try {      
@@ -886,9 +963,9 @@ public class Strip : GLib.Object {
                }
                // other files to ignore?
                 
-                this.scan_file(path , next_file.get_display_name());
+                this.scan_file(basepath + subpath , next_file.get_display_name());
                                if(this.has_replaced) {
-                                this.report_state("After scanning %s/%s".printf(path , next_file.get_display_name()));
+                                this.report_state("After scanning %s/%s".printf(basepath + subpath , next_file.get_display_name()));
                        }
                 continue;
             }
@@ -910,14 +987,14 @@ public class Strip : GLib.Object {
                }
             
             
-            var sp = path+"/"+next_file.get_display_name();
+            var sp = subpath+"/"+next_file.get_display_name();
             // skip modules.
             //print("got a file : " + sp);
          
             next_file = null;
             
             
-            this.scan_dir(sp);
+            this.scan_dir(basepath,sp);
             
         }