/**
+ ** 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?) -- seems ok now?
+
+
+
+
needs to scan 2 things
a) our mailfort email database
point it at the top directory, containing YEAR/MONTH/DAY.... directories.
public static int opt_limit = -1;
+ 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;
{ "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 },
+ { "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 }
};
// 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);
return 0;
}
- strip.scan_dir(opt_path);
+ strip.scan_dir(opt_path, "");
// initialize it with known data..
// that should wipe out dupes.
+ var matches = this.execute("SELECT count(id) FROM Attachment WHERE id = %d".printf(
+ int.parse(sid)));
+
+ if (matches=="0") {
+ GLib.error("Failed to find id :%s", sid);
+ return;
+ }
+
+
+ // initialize it with known data..
+ // that should wipe out dupes.
+ var filesize = this.execute("SELECT filesize FROM Attachment WHERE id = %d".printf(
+ int.parse(sid)));
+
+ if (filesize=="") {
+ GLib.debug("Ignoring record id (missing in database) :%s", sid);
+ return;
+ }
+ if (int.parse(filesize) < 1) {
+ GLib.debug("Could not get filesize from id :%s = %s", sid,filesize);
+ Posix.exit(0);
+ return;
+ }
+ var chksum = this.query("SELECT checksum FROM Attachment WHERE id = %d".printf(
+ int.parse(sid)
+ ));
+ var mime_filename = this.query("SELECT mime_filename FROM Attachment WHERE id = %d".printf(
+ int.parse(sid)));
+
this.query("""
- SELECT
-
- attachment_init(
- msgid,
- checksum,
- mime_filename,
- filesize
- ) as id
- FROM
- attachment
- WHERE
- id = %d
-
+ SELECT
+ attachment_init(
+ '%s', '%s', '%s', %d
+ ) as id
+
""".printf(
- int.parse(sid)
+ this.mysql_escape(this.active_message_exim_id),
+ this.mysql_escape(chksum),
+ this.mysql_escape(mime_filename),
+ int.parse(filesize)
));
this.query("""
- SELECT attachment_update(
- %d, -- in_id INT(11),
- '%s', -- in_mime_type varchar(255),
- %d, -- in_mime_size int(11),
- '%s', -- in_created DATETIME,
- '%s' -- in_mailfort_sig varchar(64)
-
- )""
-
-
+ SELECT attachment_update(
+ %d, -- in_id INT(11),
+ '%s', -- in_mime_type varchar(255),
+ '%s', -- in_created DATETIME,
+ '%s' -- in_mailfort_sig varchar(64)
+ )
""".printf(
int.parse(sid),
"", // this will be ignored..
- 0, // this will be ingored..
this.created_date,
this.mysql_escape(this.active_message_x_mailfort_sig)
)
);
-
-
+ this.mysql.store_result();
+
+
}
var c = attachment.get_content_object();
- var filename = attachment.get_filename().replace("/", "-");
+ var filename = attachment.get_filename().replace("/", "-").replace("\n", "").replace("\t", " ");
var fn = GLib.Environment.get_tmp_dir() +
"/"+ this.active_name + "."+ filename;
// at this point we have to do our database magic...
//filesize / name / date / checksum / mimetype -- into mailfort should be OK.
- this.query("""
+ var file_id = this.query("""
SELECT
attachment_init(
this.mysql_escape( attachment.get_filename() ), // what is thsi is invalid?
file_size)
);
- var file_id = "0";
- var rs = mysql.use_result();
-
- var row = rs.fetch_row();
-
- var target_fn = "";
-
- //var create_dir = this.created_dir;
-
+
- if (rs == null || rs.num_rows() < 1) {
+ if (file_id.length < 1) {
GLib.debug("ERROR - CALL to attachment_init failed");
Posix.unlink(fn);
return;
}
- file_id = row[0];
+
if (int.parse(file_id) < 1) {
GLib.debug("ERROR - CALL to attachment_init failed - returned 0?");
Posix.unlink(fn);
return;
}
-
+
GLib.debug("fn = %s, m5=%s, id= %s", filename, mime_type, this.active_message_id);
this.query("""
%d, -- in_id INT(11),
'%s', -- in_mime_type varchar(255),
'%s', -- in_created DATETIME,
- '%s', -- in_mailfort_sig varchar(64)
+ '%s' -- in_mailfort_sig varchar(64)
) as result
""".printf(
this.created_date,
this.mysql_escape(this.active_message_x_mailfort_sig)
));
-
+ this.mysql.store_result();
this.used_space_after += file_size;
-
+ var target_fn = "";
if (StripApplication.opt_is_extracting) {
target_fn = StripApplication.opt_target_path + "/" + this.created_dir +"/"+ file_id + "-" + filename;
int.parse(file_id),
this.mysql_escape( stored)
));
-
-
+
var rep = new GMime.Part.with_type("text","html");
// we have to set up a redirect server - to redirect hpasite... to their internal service..
rep.set_filename(filename);
}
GLib.debug("Creating file %s", target_fn);
if (!FileUtils.test (target_fn, FileTest.EXISTS)) {
- Posix.link(fn, target_fn);
+ var from = File.new_for_path (fn);
+ var to = File.new_for_path (target_fn);
+ from.copy(to, 0, null);
+
}
} else {
GLib.debug("Skipping extraction %s", target_fn);
}
-
- public int query(string str)
+ public string query(string str)
+ {
+ return this.real_query(true, str);
+ }
+ public string execute(string str)
+ {
+ return this.real_query(false, str);
+ }
+ public string real_query(bool need_return, string str)
{
+ GLib.debug("Before Query : %u : %s\n", this.mysql.errno(), this.mysql.error());
+
+
if (StripApplication.opt_debug_sql) {
GLib.debug("SQL: %s\n", str);
}
+
+
var rc= this.mysql.query(str);
if ( rc != 0 ) {
GLib.debug("ERROR %u: Query failed: %s\n", this.mysql.errno(), this.mysql.error());
+ Posix.exit(1);
}
- return rc;
+
+
+ var rs = mysql.use_result();
+
+ var got_row = false;
+ string[] row;
+ string ret = "";
+ while( (row = rs.fetch_row()) != null) {
+ got_row = true;
+ ret = row[0];
+
+ }
+ if (!need_return) {
+ return got_row ? "" : ret;
+ }
+ if (!got_row) {
+ GLib.debug("ERROR : no rows returned");
+ Posix.exit(1);
+ return "";
+ }
+ GLib.debug("got %s", ret);
+ return ret;
+
+
}
public string mysql_escape(string str)
this.active_name = name;
this.active_message_id = "";
- this.created_dir = this.active_path.substring(this.base_dir.length + 1 );
- this.created_date = this.created_dir.replace("/", "-");
+ var mailtime = new DateTime.now_local();
+ if (StripApplication.opt_scan_mailfort) {
+ this.created_dir = this.active_path.substring(this.base_dir.length + 1 );
+ this.created_date = this.created_dir.replace("/", "-");
+ var bits = this.created_date.split("-");
+ 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_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_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;
+ }
+
+ }
+
var fileinfo = File.new_for_path(path +"/" + name)
.query_info(GLib.FileAttribute.STANDARD_SIZE+","+GLib.FileAttribute.TIME_MODIFIED
,GLib.FileQueryInfoFlags.NONE,null);
var file_size = (int) fileinfo.get_size();
- var mod_time = fileinfo.get_modification_time();
+ var mod_time = fileinfo.get_modification_time();
+
+
+
+ if (!StripApplication.opt_scan_mailfort) {
+
+ // 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_date = mailtime.format("%Y-%m-%d %H:%M:%S");
+
+ }
+ // check on age of file...
+
+
+
+
this.used_space_before += file_size;
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];
+ 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");
+ }
+ } else {
+ GLib.debug("Could not read time from headers?");
+ }
+ }
+
}
}
+
+ var oldest = new DateTime.now_local();
+ 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(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;
+ }
+
+
+
/*
GLib.debug("Message DATA:\n mid: %s\nmailfort: %s \nexim_id: %s",
this.active_message_id,
GLib.debug("finished writing output %d", file_size);
//
+ outstream = null;
this.used_space_after += file_size;
if (StripApplication.opt_is_replacing) {
Posix.unlink(path +"/" + name);
GLib.debug("copy tmp file %s to %s" , tmpfile, path +"/" + name);
- Posix.link(tmpfile, path +"/" + name);
- Posix.unlink(tmpfile);
- var nf = File.new_for_path(path +"/" + name);
+
+ // link will not work, as we are doing it accross file systems
+ var from = File.new_for_path (tmpfile);
+ var nf = File.new_for_path (path +"/" + name);
+ from.copy(nf, 0, null);
+
+
var newfileinfo = nf.query_info(GLib.FileAttribute.TIME_MODIFIED,GLib.FileQueryInfoFlags.NONE,null);
newfileinfo.set_modification_time(mod_time);
nf.set_attributes_from_info(newfileinfo,FileQueryInfoFlags.NONE);
+ Posix.unlink(tmpfile);
}
this.processed++;
}
- 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 {
if (next_file == null) {
break;
}
-
-
+
+
+ if (next_file.get_is_symlink()) {
+ next_file = null;
+ continue;
+ }
+
+ var ds = next_file.get_display_name();
if (next_file.get_file_type() != FileType.DIRECTORY) {
- if (next_file.get_display_name()[0] == ',') {
+
+
+ if (ds[0] == ',') {
continue;
}
-
- this.scan_file(path , next_file.get_display_name());
+ // other files to ignore?
+ if (Regex.match_simple (".tgz$", ds)) {
+ continue;
+ }
+ this.scan_file(basepath + subpath , ds);
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 , ds));
}
continue;
}
// next_file.get_display_name(), next_file.get_file_type());
- if (next_file.get_is_symlink()) {
- next_file = null;
- continue;
- }
-
- var ds = next_file.get_display_name();
+
+
+ // not really needed?? - we are storing attachments in a seperate location now...
if (ds[0] == '.') {
next_file = null;
continue;
}
- 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);
}