[Assorted-commits] SF.net SVN: assorted: [327] hash-join/trunk/src/hashjoin.cc

[<yan...@us...>]

Revision: 327
          http://assorted.svn.sourceforge.net/assorted/?rev=327&view=rev
Author:   yangzhang
Date:     2008-02-06 08:19:42 -0800 (Wed, 06 Feb 2008)

Log Message:
-----------
fixed the obscene replication

Modified Paths:
--------------
    hash-join/trunk/src/hashjoin.cc

Modified: hash-join/trunk/src/hashjoin.cc
===================================================================
--- hash-join/trunk/src/hashjoin.cc	2008-02-06 16:01:30 UTC (rev 326)
+++ hash-join/trunk/src/hashjoin.cc	2008-02-06 16:19:42 UTC (rev 327)
@@ -204,22 +204,23 @@
 int
 main(int argc, char *argv[])
 {
-  if (argc != 3) {
-    fprintf(stderr, "hashjoin <ncpus> <actresses>\n");
+  if (argc != 4) {
+    fprintf(stderr, "hashjoin <ncpus> <movies> <actresses>\n");
     exit(1);
   }
 
   ncpus = atoi(argv[1]);
-  const char *actresses = argv[2];
+  const char *movies = argv[2];
+  const char *actresses = argv[3];
 
   timer t("main time: ");
 
   cout << "loading movies" << endl;
-  movdb mdb("../movie-data/movies.dat");
+  movdb mdb(movies); // "../movie-data/movies.dat"
   t.print();
 
   cout << "loading actresses" << endl;
-  actdb adb(actresses);//"../movie-data/mdactresses.dat");
+  actdb adb(actresses); // "../movie-data/mdactresses.dat"
   t.print();
 
   cout << "hash-partitioning movies into per-core buckets" << endl;
@@ -409,6 +410,7 @@
   for (unsigned int i = 0; i < ncpus; i++) {
     heads[i] = bs[i].bufs[0];
   }
+  char tmp[1024];
 
   // Statistics (TODO dynamic allocation)
   int counter = 0, mincount = INT_MAX;
@@ -416,15 +418,20 @@
   while (p < end - 999) {
     char *name = p;
     p = strchr(p, '\0') + 1;
+    strcpy(tmp, name);
+    char *subtmp = tmp + strlen(name) + 1;
     char *tuple_end = unsafe_strstr(p, "\0\0", end) + 2;
     while (true) {
       char *title = p;
       p = strchr(p, '\0') + 1;
 
+      strcpy(subtmp, title);
+      size_t strl = strlen(subtmp);
+
       // Copy this line into the correct local bucket.
       //cout << "hashing " << title << endl;
       unsigned int bbb;
-      if (-1 != (bbb = push_bucket(heads, bs, title, name, tuple_end - name))) {
+      if (-1 != (bbb = push_bucket(heads, bs, title, tmp, subtmp + strl + 1 - tmp))) {
         //size_t bucket_size = max(1000000,buflen / ncpus * 2); //2 * buflen / ncpus;
         //int bucket_size = max(1000000UL,buflen / ncpus * 3);
         //cout << "FUCK " << heads[0] - bs[0].buf << " " << bucket_size << " " << heads[1] - bs[1].buf << " " << p - title << endl;


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.