/**
+ ** 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.
{ "oldest", 0, 0, OptionArg.INT, ref opt_age_oldest, "do not replace messages older than X (default is 6 months)", 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 },
+ { "scan-mailfort", 0, 0, OptionArg.NONE, ref opt_scan_mailfort, "scan a mailfort tree", null },
+ { "scan-mailfort-only", 0, 0, OptionArg.NONE, ref opt_scan_mailfort, "scan a mailfort Year/month eg. 2010/08", null },
{ null }
};
public StripApplication( string[] args )
return;
}
+ // initialize it with known data..
+ // that should wipe out dupes.
+ var matches = this.query("SELECT count(id) FROM Attachment WHERE id = %d".printf(
+ int.parse(sid)));
+
+
+ if (matches == "0") {
+ // our old mailfort code deleted the crap out of old records...
+ // if this occurs we will need to create the record again..
+ this.fix_deleted_attachment_db(int.parse(sid),attachment);
+ return;
+
+ }
+
+
// initialize it with known data..
// that should wipe out dupes.
var filesize = this.query("SELECT filesize FROM Attachment WHERE id = %d".printf(
- int.parse(sid)));
+ int.parse(sid)));
+
+ if (filesize=="") {
+ GLib.error("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);
}
+ public void fix_deleted_attachment_db(int id, GMime.Part attachment)
+ {
+
+ var filename = attachment.get_header("X-strip-content-name");
+ var file_path = attachment.get_header("X-strip-path");
+ var fn = StripApplication.opt_target_path + "/" + file_path;
+
+
+ if (!FileUtils.test (fn, FileTest.EXISTS)) {
+ GLib.debug("SKIP -- file does not exist");
+ return;
+ }
+
+ var chksum = this.md5_file(fn);
+ var mime_type = attachment.get_header("X-strip-content-type");
+
+ var fileinfo = File.new_for_path(fn)
+ .query_info(GLib.FileAttribute.STANDARD_SIZE+","+GLib.FileAttribute.TIME_MODIFIED
+ ,GLib.FileQueryInfoFlags.NONE,null);
+ var file_size = (int) fileinfo.get_size();
+
+
+ this.real_query(-1, """
+
+
+ INSERT INTO Attachment (
+ id,
+
+ msgid ,
+ queue_id ,
+ mime_filename ,
+ mime_type,
+
+ stored_filename ,
+ mime_charset ,
+ mime_cdisp ,
+ mime_is_cover ,
+
+ mime_is_multi ,
+ mime_is_mail,
+ mime_size ,
+ filesize,
+
+ checksum,
+ created
+
+ ) VALUES (
+ %d, -- id
+
+ '%s' , -- msgid
+ 0,
+ '%s' , -- filename
+ '%s', -- mimetype
+
+ '%s', -- stored file anme
+ '', -- charset
+ 'attachment',
+ 0,
+ 0,
+ 0,
+ %d, -- size
+ %d, -- size
+
+ '%s', -- checkum
+ '%s' -- created:
+ )
+
+
+ """.printf(
+ id,
+ this.mysql_escape(this.active_message_exim_id),
+ this.mysql_escape(filename),
+ this.mysql_escape(mime_type),
+ this.mysql_escape(file_path),
+ file_size,
+ file_size,
+ this.mysql_escape(chksum),
+ this.created_date
+ ));
+ // this is done to fix the queue_id or maillog_id ??
+ this.query("""
+ SELECT attachment_update(
+ %d, -- in_id INT(11),
+ '', -- mime type
+ '%s', -- in_created DATETIME,
+ '%s' -- in_mailfort_sig varchar(64)
+ )
+ """.printf(
+ id,
+ this.created_date,
+ this.mysql_escape(this.active_message_x_mailfort_sig)
+
+ )
+ );
+ // GLib.error("added attachment?");
+ }
+
+
public void replace_attachment(GMime.Multipart parent, GMime.Part attachment)
{
var sid = attachment.get_header("X-strip-id");
var filename = attachment.get_filename().replace("/", "-").replace("\n", "").replace("\t", " ");
var fn = GLib.Environment.get_tmp_dir() +
- "/"+ this.active_name + "."+ filename;
+ "/"+ this.active_name + "."+ GLib.Uri.escape_string(filename,"", false);
var outfile = new GMime.StreamFile.for_path(fn, "w");
outfile.set_owner(true);
""".printf(
this.mysql_escape(this.active_message_exim_id),
chksum,
- this.mysql_escape( attachment.get_filename() ), // what is thsi is invalid?
+ this.mysql_escape( GLib.Uri.escape_string(attachment.get_filename(),"", false) ), // what is thsi is invalid?
file_size)
);
var target_fn = "";
if (StripApplication.opt_is_extracting) {
- target_fn = StripApplication.opt_target_path + "/" + this.created_dir +"/"+ file_id + "-" + filename;
+ target_fn = StripApplication.opt_target_path + "/" + this.created_dir +"/"+ file_id + "-" + GLib.Uri.escape_string(filename,"", false);
}
- var stored = "/" + this.created_dir +"/"+ file_id + "-" + filename;
+ var stored = "/" + this.created_dir +"/"+ file_id + "-" + GLib.Uri.escape_string(filename,"", false);
this.query("""
SELECT attachment_update_store(
rep.set_filename(filename);
string txt = "<html><body>"+
"<a href=\"" + StripApplication.opt_replace_link + "/" +
- file_id + "/" + this.created_dir + "/"+chksum+"/"+ GLib.Uri.escape_string( filename) +"\">" +
- GLib.Uri.escape_string( filename) + // fixme needs html escaping...
+ file_id + "/" + this.created_dir + "/"+chksum+"/"+ GLib.Uri.escape_string( filename, "", false) +"\">" +
+ GLib.Uri.escape_string( filename, "", false) + // fixme needs html escaping...
"</a>" +
"</body></html>";
rep.get_content_type().set_parameter("charset", "utf-8");
rep.set_header("X-strip-id", file_id);
rep.set_header("X-strip-content-name", filename);
- rep.set_header("X-strip-path", this.created_dir + "/" + file_id + "-" + filename);
+ rep.set_header("X-strip-path", this.created_dir + "/" + file_id + "-" + GLib.Uri.escape_string(filename,"", false));
rep.set_header("X-strip-content-type", mime_type);
var stream = new GMime.StreamMem.with_buffer(txt.data);
var con = new GMime.DataWrapper.with_stream(stream,GMime.ContentEncoding.DEFAULT);
}
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 string query(string str)
{
- return this.real_query(true, str);
+ return this.real_query(1, str);
}
public string execute(string str)
{
- return this.real_query(false, str);
+ return this.real_query(0, str);
}
- public string real_query(bool need_return, string str)
+ /**
+ * need_return
+ 0 = no
+ 1 = yes
+ -1 = don't try.
+ */
+ public string real_query(int need_return, string str)
{
GLib.debug("Before Query : %u : %s\n", this.mysql.errno(), this.mysql.error());
- var rc= this.mysql.query(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);
}
-
-
- var rs = mysql.use_result();
+ var rs = mysql.use_result();
+ if (need_return == -1) {
+ return "";
+ }
+
+
+ //GLib.debug("got %d rows", (int) rs.num_rows());
var got_row = false;
string[] row;
ret = row[0];
}
- if (!need_return) {
- return "";
+ if (need_return == 0) {
+ if (StripApplication.opt_debug_sql) {
+ GLib.debug("got %s", got_row ? "=Nothing=" : ret);
+ }
+ return got_row ? "" : ret;
}
if (!got_row) {
+
GLib.debug("ERROR : no rows returned");
Posix.exit(1);
return "";
}
- GLib.debug("got %s", ret);
+ if (StripApplication.opt_debug_sql) {
+ GLib.debug("got %s", ret);
+ }
return ret;
var oldest = new DateTime.now_local();
oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
- var rtspan = oldest.difference(oldest) / GLib.TimeSpan.DAY;
-
+ 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;
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 basepath, string subpath)
{
+
// 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 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() );
+ //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 year: %d/%d", subpath, oldest.get_year() , 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 year: %d/%d/%d", subpath, oldest.get_year() , oldest.get_month(), 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 min: %d/%d/%d", subpath, newest.get_year() , newest.get_month(), newest.get_day_of_month() );
+ //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 min year: %d", subpath, 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 > newest.get_month()) {
- GLib.debug("Skip directory %s is newer than min year: %d/%d", subpath, newest.get_year() , newest.get_month() );
+ 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 > newest.get_day_of_month()) {
- GLib.debug("Skip directory %s is newer than min year: %d/%d/%d", subpath, newest.get_year() , newest.get_month(), newest.get_day_of_month() );
+ 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;
}
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;
}
// other files to ignore?
-
- this.scan_file(basepath + subpath , next_file.get_display_name());
+ 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(basepath + subpath , next_file.get_display_name()));
+ this.report_state("After scanning %s/%s".printf(basepath + subpath , ds));
}
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;
projects / app.mailtrimmer / blobdiff