MGL upload eating rows

[GoogleCodeExporter]

What steps will reproduce the problem?
1. log in as admin
2. Upload a file (samples sent to bajaj)

What is the expected output? What do you see instead?
The number of entries created (either in stats->branch, or through SQL)
should match the lines in the txt file. But it is not so. Not sure what is
happenning. Also, the 1998.txt file is causing a timeout in the script.

+++dump of initial email with this issue +++

I uploaded this file (1992.txt), which has 260 lines, each with
graduation year as 1992

However, the query

 SELECT *
FROM `user`
WHERE `graduationyear` =1992
LIMIT 0 , 30

results in

"Showing rows 0 - 29 (258 total, Query took 0.0009 sec)"

It shows only 258 entries total. Not sure what happened to the 2
missing entries. (each of them has an enrolment number. None are
blank. However, there might have been collisions due to typing error
at the institute's part)

On a different note, the upload of the 1998.txt file failed (got blank
screen after a while). So I queried and realized that it had been
partially uploaded. I therefore put all the IDs below "Ajay Kumar
Jain"(it had the highest user ID in the 197 names returned by the SQL
query, but interestingly, it is line #198 in the text file. Somehow it
is still eating lines), and created a new file with the balance names.
That file (1998-part2.txt) refuses to upload even one user. The script
simply crashes.

+++

Original issue reported on code.google.com by animesh on 7 Aug 2009 at 4:23

[GoogleCodeExporter]

Detailed log mechanism added. Working properly on local. please verify on server

Original comment by gs.bajaj on 17 Aug 2009 at 3:15

• Changed state: Fixed

[GoogleCodeExporter]

I don't see a commit log for Issue #79 (up to r157 ). Please mention the 
revision
that fixes this.

Original comment by animesh on 17 Aug 2009 at 3:21

[GoogleCodeExporter]

The detailed log mechanism works, but that is actually issue #70.

The current issue still holds, since when uploading the files from a clean 
install,
this is what happened:

2003.txt: all OK
1992.txt: 2 rows already in DB [as reported in this issue]
1993.txt: all OK
1996.txt: 3 rows already in DB [as reported in this issue]
1998.txt: script crashes [as reported in this issue]

Original comment by animesh on 19 Aug 2009 at 11:46

• Changed state: Accepted

[GoogleCodeExporter]

I checked, and indeed, there were entries with duplicate enrolment numbers in
1992.txt and 1996.txt.

Consider that part of the bug solved.

Still, not sure why 1998.txt causes the script to crash.

Original comment by animesh on 19 Aug 2009 at 12:04

[GoogleCodeExporter]

Ok fixed this bug. Problem was mainly with 3 entries for "amit kumar 
gupta","amit
gupta", and "amit kumar gupta" respectively in 1998.txt.
Have also cleaned the format in which such cases are handled. So previously the
second amit gupta had username "amit.gupta@xyz981" this has been changed to the
cleaner "amit.gupta1@xyz98"

Original comment by R.DUSHYA...@gmail.com on 17 Mar 2010 at 3:41

Attachments:

• issue79.patch