Schema generation

MayBMS-tpch/uncertain-tpch/.gitignore

@@ -0,0 +1,4 @@
+*.o
+qgen
+dbgen
+*.tbl

MayBMS-tpch/uncertain-tpch/Makefile

@@ -91,7 +91,7 @@ WORKLOAD =	TPCH
 #
 # add -EDTERABYTE if orderkey will execeed 32 bits (SF >= 300)
 # and make the appropriate change in gen_schema() of runit.sh
-CFLAGS	= -O -DDBNAME=\"dss\" -D$(MACHINE) -D$(DATABASE) -D$(WORKLOAD)
+CFLAGS	= -g -DDBNAME=\"dss\" -D$(MACHINE) -D$(DATABASE) -D$(WORKLOAD)
 LDFLAGS = -g
 # The OBJ,EXE and LIB macros will need to be changed for compilation under
 #  Windows NT
@@ -174,4 +174,4 @@ rnd$(OBJ): rnd.h
 $(OBJ1): $(HDR1)
 $(OBJ2): dss.h tpcd.h config.h rng64.h release.h
 update_release:
-	update_release.sh ${VERSION} ${RELEASE} ${PATCH}
+	./update_release.sh ${VERSION} ${RELEASE} ${PATCH}

MayBMS-tpch/uncertain-tpch/Rakefile

@@ -0,0 +1,139 @@
+
+$files = {
+  cust:     [ :c_acctbal, :c_address, :c_comment, :c_custkey, :c_mktsegment, :c_name, :c_nationkey, :c_phone ],
+  lineitem: [ :l_comment, :l_commitdate, :l_discount, :l_extendedprice, :l_linenumber, :l_linestatus, :l_orderkey, :l_partkey, :l_quantity, :l_receiptdate, :l_returnflag, :l_shipdate, :l_shipinstruct, :l_shipmode, :l_suppkey, :l_tax ],
+  nation:   [ :n_nationkey, :n_name, :n_regionkey, :n_comment ],
+  orders:   [ :o_clerk, :o_comment, :o_custkey, :o_orderdate, :o_orderkey, :o_orderpriority, :o_orderstatus, :o_shippriority, :o_totalprice ],
+  part:     [ :p_brand, :p_comment, :p_container, :p_mfgr, :p_name, :p_partkey, :p_retailprice, :p_size, :p_type ],
+  psupp:    [ :ps_availqty, :ps_comment, :ps_partkey, :ps_suppkey, :ps_supplycost ],
+  region:   [ :r_comment, :r_name, :r_regionkey ],
+  supp:     [ :s_acctbal, :s_address, :s_comment, :s_name, :s_nationkey, :s_phone, :s_suppkey ]
+}
+$types = {
+  :c_acctbal    => "decimal", 
+  :c_address    => "varchar(40)", 
+  :c_comment    => "varchar(117)", 
+  :c_custkey    => "int", 
+  :c_mktsegment => "char(10)", 
+  :c_name       => "varchar(25)", 
+  :c_nationkey  => "int", 
+  :c_phone      => "char(15)",
+
+  :l_comment       => "char(44)", 
+  :l_commitdate    => "date", 
+  :l_discount      => "decimal", 
+  :l_extendedprice => "decimal", 
+  :l_linenumber    => "int", 
+  :l_linestatus    => "char(1)", 
+  :l_orderkey      => "int", 
+  :l_partkey       => "int", 
+  :l_quantity      => "decimal", 
+  :l_receiptdate   => "date", 
+  :l_returnflag    => "char(1)", 
+  :l_shipdate      => "date", 
+  :l_shipinstruct  => "char(25)", 
+  :l_shipmode      => "char(10)", 
+  :l_suppkey       => "int", 
+  :l_tax           => "decimal",
+
+  :n_nationkey => "int", 
+  :n_name      => "char(25)", 
+  :n_regionkey => "int", 
+  :n_comment   => "char(152)",
+
+  :o_clerk         => "char(15)", 
+  :o_comment       => "varchar(79)", 
+  :o_custkey       => "int", 
+  :o_orderdate     => "date", 
+  :o_orderkey      => "int", 
+  :o_orderpriority => "char(15)", 
+  :o_orderstatus   => "char(1)", 
+  :o_shippriority  => "int", 
+  :o_totalprice    => "decimal",
+
+  :p_brand        => "char(10)", 
+  :p_comment      => "varchar(23)", 
+  :p_container    => "char(10)", 
+  :p_mfgr         => "char(25)", 
+  :p_name         => "varchar(55)", 
+  :p_partkey      => "int", 
+  :p_retailprice  => "decimal", 
+  :p_size         => "int", 
+  :p_type         => "varchar(25)",
+
+  :ps_availqty   => "int", 
+  :ps_comment    => "varchar(199)", 
+  :ps_partkey    => "int", 
+  :ps_suppkey    => "int", 
+  :ps_supplycost => "decimal",
+
+  :r_comment   => "char(152)", 
+  :r_name      => "varchar(25)", 
+  :r_regionkey => "int",
+
+  :s_acctbal    => "decimal", 
+  :s_address    => "varchar(40)", 
+  :s_comment    => "varchar(101)", 
+  :s_name       => "char(25)", 
+  :s_nationkey  => "int", 
+  :s_phone      => "char(15)", 
+  :s_suppkey    => "int",
+
+  :tid  => "int"
+}
+
+def init_data
+  # TUPLE_ID      ->     CONFLICT_GROUP  ->     WORLD        ->        FIELD    -> VALUE
+  Hash.new {|h,k| h[k] = Hash.new {|h,k| h[k] = Hash.new {|h,k| h[k] = Hash.new } } }
+end
+
+def load(data, table, field)
+  puts "Loading #{table}_#{field}.tbl"
+  File.open("input/#{table}_#{field}.tbl") do |f|
+    f.readlines.each do |l|
+      conflict_group, world, tuple_id, value = l.chomp.split(/\|/)
+      data[tuple_id.to_i][conflict_group.to_i][world.to_i][field] = value
+    end
+  end
+end
+
+def render(data)
+  puts "Rendering data..."
+  data.map do |tuple_id, conflict_groups| 
+    conflict_groups = conflict_groups.values.to_a
+    world_ids = conflict_groups.map { |worlds| worlds.keys }
+
+    # compute the cartesian product of the resulting key sets
+    world_groups = [ [nil] ]
+    world_ids.each { |id_set| world_groups = world_groups.product(id_set) }
+    world_groups = world_groups.map { |x| x.flatten[1..-1] }
+
+    # render the corresponding tuple
+    world_groups.map do |per_group_world_ids| 
+      conflict_groups.zip(per_group_world_ids).map { |worlds, world_id| worlds[world_id].to_a }.flatten(1)
+    end.map { |tuple| [tuple_id, tuple.to_h] }
+  end.flatten(1)
+end
+
+$files.each do |table, fields|
+  file "output/#{table}.tbl" do
+    data = init_data
+    fields.each { |field| load(data, table, field) }
+    File.open("output/#{table}.tbl", "w+") do |f|
+      f.puts "tid|#{fields.join("|")}"
+      render(data).each do |tid, tuple| 
+        f.puts "#{tid}|#{fields.map {|f| tuple[f]}.join("|")}"
+      end
+    end
+  end
+
+  task table => "output/#{table}.tbl"
+end
+
+file "schema.sql" do
+  File.open("schema.sql", "w+") do |f|
+    $files.each do |table, fields|
+      f.puts("CREATE TABLE #{table}(#{([:tid]+fields).map {|field| "#{field} #{$types[field]}"}.join(", ")});")
+    end
+  end
+end

MayBMS-tpch/uncertain-tpch/bm_utils.c

@@ -74,7 +74,9 @@
 #include <ctype.h>
 #include <math.h>
 #ifndef _POSIX_SOURCE
+#ifndef __APPLE__
 #include <malloc.h>
+#endif
 #endif /* POSIX_SOURCE */
 #include <fcntl.h>
 #include <sys/types.h>
@@ -390,6 +392,7 @@ tbl_open(int tbl, char *mode)
     struct stat fstats;
     int      retcode;
 
+    bzero(&fstats, sizeof(fstats));
 
     if (*tdefs[tbl].name == PATH_SEP)
         strcpy(fullpath, tdefs[tbl].name);

MayBMS-tpch/uncertain-tpch/print.c

@@ -719,7 +719,7 @@ pr_cust(customer_t *c, int mode)
    if (TUPLE_LEVEL) {
      if ((long)i <  threshold+1) {
        add_placeholder(CUST, (long)c->custkey, 0);
-       return;
+       return 1;
      }
      threshold = -1;
    }
@@ -864,7 +864,7 @@ pr_order(order_t *o, int mode)
    if (TUPLE_LEVEL) {
      if ((long)i <  threshold+1) {
        add_placeholder(ORDER,(long)o->okey,0);
-       return;
+       return 1;
      }
      threshold = -1;
    }
@@ -1301,7 +1301,7 @@ pr_part(part_t *part, int mode)
    if (TUPLE_LEVEL) {
      if ((long)j <threshold+1) {
        add_placeholder(PART,(long)part->partkey,0);
-       return;
+       return 1;
      }
      threshold = -1;
    }
@@ -1573,7 +1573,7 @@ pr_supp(supplier_t *supp, int mode)
   if (TUPLE_LEVEL) {
     if ((long)j <threshold+1) {
       add_placeholder(SUPP,(long)supp->suppkey,0);
-      return;
+      return 1;
     }
     threshold = -1;
   }
@@ -1581,6 +1581,13 @@ pr_supp(supplier_t *supp, int mode)
 
   if ((long)j > threshold) {
     if (!COUNT_ONLY) {
+      printf("%ld\n",crt_cid);
+      printf("%c1\n",SEPARATOR);
+      printf("%c\n",SEPARATOR);
+      printf("%lld\n",supp->suppkey);
+      printf("%c\n",SEPARATOR);
+      printf("%lld\n",supp->suppkey);
+      printf("%d\n",supp_s_suppkey);
       PR_STRT(supp_s_suppkey);
       fprintf(supp_s_suppkey, "%ld%c1%c%lld%c%lld",crt_cid,SEPARATOR,SEPARATOR,supp->suppkey,SEPARATOR,
 	      supp->suppkey);
@@ -1699,7 +1706,7 @@ pr_nation(code_t *c, int mode)
   if (TUPLE_LEVEL) {
     if ((long)j <threshold+1) {
       add_placeholder(NATION,(long)c->code,0);
-      return;
+      return 1;
     }
     threshold = -1;
   }
@@ -1779,7 +1786,7 @@ pr_region(code_t *c, int mode)
   if (TUPLE_LEVEL) {
     if ((long)j <threshold+1) {
       add_placeholder(REGION,(long)c->code,0);
-      return;
+      return 1;
     }
     threshold = -1;
   }

MayBMS-tpch/uncertain-tpch/release.h

@@ -1,4 +1,4 @@
-#ORIGINAL TPC-H distribution
+/* #ORIGINAL TPC-H distribution */
 #define VERSION 2
 #define RELEASE 6
 #define PATCH 0

MayBMS-tpch/uncertain-tpch/varsub.c

@@ -41,7 +41,9 @@
 */
 #include <stdio.h>
 #ifndef _POSIX_SOURCE
+#ifndef __APPLE__
 #include <malloc.h>
+#endif
 #endif /* POSIX_SOURCE */
 #if (defined(_POSIX_)||!defined(WIN32))
 #include <unistd.h>