Xapian discuss - Jan 2014 - seg fault on search

I have written a very simple function to return the match count based on the
simplesearch.cc code. It fails with a seg fault. The relevant code is:
--------------------
int ftQuery(char* qs, const char* dbname,char* results, int msize) {
        long            docid;
        char*           op; 
        char            fullDB[1024];
        string          queryString;

        sprintf(fullDB,"/var/lib/fulltext/%s",dbname);
        queryString=qs;
    try {
        Database db;
        db.add_database(Database(fullDB));
        Enquire enquire(db);
        try {
                Xapian::QueryParser qp;
                Xapian::Stem stemmer("english");

                qp.set_database(db);
                qp.set_default_op(Query::OP_OR);
                qp.set_stemmer(stemmer);
                qp.set_stemming_strategy(QueryParser::STEM_SOME);
                Xapian::Query q=qp.parse_query(queryString);
...

As soon as it hits the q=qp.parse_query line it gets the seg fault. 

The function is called like so:

int main (int argc, char* const argv[]) {
        int     i;
        char    results[100000];

        if (argc<2) {
                printf("Usage: tft 'query string'\n");
                return 1;
        }
        I=ftQuery(argv[1],"testdb",results,2000);
        printf("%d results returned\n",i);
        return (0);

The query string is "austria" and it works with simplesearch.com. I
have verified that the string is being passed to the function correctly as is
the database name. I am out of ideas and wonder if anyone can point out anything
that I am overlooking.

Thanks,

Michael Lewis
_____________________________________________
Xapian-discuss mailing list
Xapian-discuss at lists.xapian.org
http://lists.xapian.org/mailman/listinfo/xapian-discuss

On Tue, Jan 21, 2014 at 04:53:41PM +0000, Michael Lewis
wrote:
> I have written a very simple function to return the match count based
> on the simplesearch.cc code. It fails with a seg fault. The relevant
> code is:
> --------------------
> int ftQuery(char* qs, const char* dbname,char* results, int msize) {
>         long            docid;
It would be better to use the type Xapian::docid instead of long, though
that won't give you a segfault here.
> The query string is "austria" and it works with simplesearch.com.
I
> have verified that the string is being passed to the function
> correctly as is the database name. I am out of ideas and wonder if
> anyone can point out anything that I am overlooking.
Code looks OK - I suspect the problem is elsewhere - e.g. maybe you've
corrupted the malloc heap or have a stray pointer write.

If you're on a supported platform, I'd run it under valgrind, which can
often spot such problems.

If you aren't, or that doesn't find anything, run it under gdb and get a
backtrace for exactly where the segfault happens - that will likely give
more of a clue.

Cheers,
    Olly

Thanks for the response. I ran it under GDB and single stepped the queryparser
call. I have a realloc() function in a common module. As soon as I stepped into
the query_parser in GDB it called my realloc() instead of the one it was
expecting. I removed my routine and it works fine now.

Thanks,
Michael

-----Original Message-----
From: Olly Betts [mailto:olly at survex.com] 
Sent: Tuesday, January 21, 2014 2:29 PM
To: Michael Lewis
Cc: xapian-discuss at lists.xapian.org
Subject: Re: [Xapian-discuss] seg fault on search

On Tue, Jan 21, 2014 at 04:53:41PM +0000, Michael Lewis
wrote:
> I have written a very simple function to return the match count based 
> on the simplesearch.cc code. It fails with a seg fault. The relevant 
> code is:
> --------------------
> int ftQuery(char* qs, const char* dbname,char* results, int msize) {
>         long            docid;
It would be better to use the type Xapian::docid instead of long, though that
won't give you a segfault here.
> The query string is "austria" and it works with simplesearch.com.
I
> have verified that the string is being passed to the function 
> correctly as is the database name. I am out of ideas and wonder if 
> anyone can point out anything that I am overlooking.
Code looks OK - I suspect the problem is elsewhere - e.g. maybe you've
corrupted the malloc heap or have a stray pointer write.

If you're on a supported platform, I'd run it under valgrind, which can
often spot such problems.

If you aren't, or that doesn't find anything, run it under gdb and get a
backtrace for exactly where the segfault happens - that will likely give more of
a clue.

Cheers,
    Olly