From 675cf7f7558ebaef15f398d90cc3d1d91457b219 Mon Sep 17 00:00:00 2001
From: Douglas Rumbaugh <dbr4@psu.edu>
Date: Fri, 3 May 2024 11:01:47 -0400
Subject: FST benchmarks

---
 benchmarks/include/file_util.h | 45 +++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 44 insertions(+), 1 deletion(-)

(limited to 'benchmarks/include')

diff --git a/benchmarks/include/file_util.h b/benchmarks/include/file_util.h
index ebcf17e..586b44f 100644
--- a/benchmarks/include/file_util.h
+++ b/benchmarks/include/file_util.h
@@ -1,3 +1,5 @@
+#pragma once
+
 #include <cstdlib>
 #include <cstdio>
 #include <iostream>
@@ -7,8 +9,10 @@
 #include <gsl/gsl_rng.h>
 #include <cstring>
 #include <vector>
+#include <memory>
+
+#include "psu-util/progress.h"
 
-#pragma once
 
 template <typename QP>
 static std::vector<QP> read_lookup_queries(std::string fname, double selectivity) {
@@ -35,6 +39,20 @@ static std::vector<QP> read_lookup_queries(std::string fname, double selectivity
     return queries;
 }
 
+template <typename QP>
+static std::vector<QP> generate_string_lookup_queries(std::vector<std::unique_ptr<char[]>> &strings, size_t cnt, gsl_rng *rng) {
+    std::vector<QP> queries;
+
+    for (size_t i=0; i<cnt; i++) {
+        auto idx = gsl_rng_uniform_int(rng, strings.size());
+        QP q;
+        q.search_key = strings[idx].get();
+        queries.push_back(q);
+    }
+
+    return queries;
+}
+
 template <typename QP>
 static std::vector<QP> read_range_queries(std::string &fname, double selectivity) {
     std::vector<QP> queries;
@@ -173,3 +191,28 @@ static std::vector<R> read_vector_file(std::string &fname, size_t n) {
 
     return records;
 }
+
+
+static std::vector<std::unique_ptr<char[]>>read_string_file(std::string fname, size_t n=10000000) {
+
+    std::fstream file;
+    file.open(fname, std::ios::in);
+
+    if (!file.is_open()) {
+        fprintf(stderr, "ERROR: Failed to open file %s\n", fname.c_str());
+        exit(EXIT_FAILURE);
+    }
+
+    std::vector<std::unique_ptr<char[]>> strings;
+    strings.reserve(n);
+
+    size_t i=0;
+    std::string line;
+    while (i < n && std::getline(file, line, '\n')) {
+        strings.emplace_back(std::unique_ptr<char[]>(strdup(line.c_str())));
+        i++;
+        psudb::progress_update((double) i / (double) n, "Reading file:");
+    }
+
+    return strings;
+}
-- 
cgit v1.2.3