rexim / freq-parser Goto Github PK

I didn't find any support for h2 yet. As a backup plan we are going to generate an SQL script from the logs. For this we need some sort of variable binding to avoid accidental SQL injections.

Currently we retrieve this information from the log file path.

When I try to insert data into log table I see the following error message:

DBD::Pg::st execute failed: ERROR:  General error: "java.lang.ArrayIndexOutOfBoundsException: 0" [50000-173]
DETAIL:  org.h2.jdbc.JdbcSQLException: General error: "java.lang.ArrayIndexOutOfBoundsException: 0" [50000-173]
ERROR:  Parameter "#1" is not set; SQL statement:
INSERT INTO log (time, room, sender, type, message) VALUES ($1, $2, $3, $4, $5) [90012-173]
DETAIL:  org.h2.jdbc.JdbcSQLException: Parameter "#1" is not set; SQL statement:
INSERT INTO log (time, room, sender, type, message) VALUES ($1, $2, $3, $4, $5) [90012-173] at ../freq-parser/parse.pl line 88, <$fh> line 50.

The problem is reproducible on h2 1.3.173. When I use 1.3.170 everything works as expected. This is probably a bug of h2. We have to resolve the problem somehow.

Currently the log parsing process consists of two phases:

1. Extract so called raw messages (a tuple of four strings: time, sender, and two chunks of text) from the HTML
2. Convert raw message into particular type of a message (regular message, leave message, kick message and so on)

Currently all unrecognized raw messages are ignored. We should print a warning message for each unrecognized raw message instead.

Nicknames or messages may contain HTML escape characters. We should handle them properly.

In case it is not possible we have a backup plan #2