From 9fe190f5d500e22b0894095e7c917f9c652e0a64 Mon Sep 17 00:00:00 2001
From: Douglas Rumbaugh <dbr4@psu.edu>
Date: Wed, 20 Mar 2024 17:30:14 -0400
Subject: Updates/progress towards succinct trie support

---
 tests/include/shard_standard.h |   6 +-
 tests/include/shard_string.h   | 168 +++++++++++++++++++++++++++++++++++++++++
 tests/include/testing.h        |  59 +++++++++++++--
 3 files changed, 223 insertions(+), 10 deletions(-)
 create mode 100644 tests/include/shard_string.h

(limited to 'tests/include')

diff --git a/tests/include/shard_standard.h b/tests/include/shard_standard.h
index 7d17dcb..55e4c7b 100644
--- a/tests/include/shard_standard.h
+++ b/tests/include/shard_standard.h
@@ -37,17 +37,17 @@ START_TEST(t_mbuffer_init)
     auto buffer = new MutableBuffer<R>(512, 1024);
     for (uint64_t i = 512; i > 0; i--) {
         uint32_t v = i;
-        buffer->append({i, v, 1});
+        buffer->append({i, v});
     }
     
     for (uint64_t i = 1; i <= 256; ++i) {
         uint32_t v = i;
-        buffer->append({i, v, 1}, true);
+        buffer->append({i, v}, true);
     }
 
     for (uint64_t i = 257; i <= 512; ++i) {
         uint32_t v = i + 1;
-        buffer->append({i, v, 1});
+        buffer->append({i, v});
     }
 
     Shard* shard = new Shard(buffer->get_buffer_view());
diff --git a/tests/include/shard_string.h b/tests/include/shard_string.h
new file mode 100644
index 0000000..27ee782
--- /dev/null
+++ b/tests/include/shard_string.h
@@ -0,0 +1,168 @@
+/*
+ * tests/include/shard_string.h
+ *
+ * Standardized unit tests for Shard objects with string keys
+ *
+ * Copyright (C) 2023 Douglas Rumbaugh <drumbaugh@psu.edu> 
+ *
+ * Distributed under the Modified BSD License.
+ *
+ * WARNING: This file must be included in the main unit test set
+ *          after the definition of an appropriate Shard and R
+ *          type. In particular, R needs to implement the key-value
+ *          pair interface. For other types of record, you'll need to
+ *          use a different set of unit tests.
+ */
+#pragma once
+
+/*
+ * Uncomment these lines temporarily to remove errors in this file
+ * temporarily for development purposes. They should be removed prior
+ * to building, to ensure no duplicate definitions. These includes/defines
+ * should be included in the source file that includes this one, above the
+ * include statement.
+ */
+#include "shard/FSTrie.h"
+#include "testing.h"
+#include <check.h>
+using namespace de;
+typedef StringRec R;
+typedef FSTrie<R> Shard;
+
+START_TEST(t_mbuffer_init)
+{
+
+    auto recs = read_string_data(kjv_wordlist, 1024);
+
+    auto buffer = new MutableBuffer<R>(512, 1024);
+
+    for (uint64_t i = 0; i < 512; i++) {
+        buffer->append(recs[i]);
+    }
+    
+    for (uint64_t i = 0; i < 256; ++i) {
+        buffer->delete_record(recs[i]);
+    }
+
+    for (uint64_t i = 512; i < 768; ++i) {
+        buffer->append(recs[i]);
+    }
+
+    Shard* shard = new Shard(buffer->get_buffer_view());
+    ck_assert_uint_eq(shard->get_record_count(), 512);
+
+    delete buffer;
+    delete shard;
+}
+
+
+START_TEST(t_shard_init)
+{
+    size_t n = 512;
+    auto mbuffer1 = create_test_mbuffer<R>(n);
+    auto mbuffer2 = create_test_mbuffer<R>(n);
+    auto mbuffer3 = create_test_mbuffer<R>(n);
+
+    auto shard1 = new Shard(mbuffer1->get_buffer_view());
+    auto shard2 = new Shard(mbuffer2->get_buffer_view());
+    auto shard3 = new Shard(mbuffer3->get_buffer_view());
+
+    std::vector<Shard*> shards = {shard1, shard2, shard3};
+    auto shard4 = new Shard(shards);
+
+    ck_assert_int_eq(shard4->get_record_count(), n * 3);
+    ck_assert_int_eq(shard4->get_tombstone_count(), 0);
+
+    size_t total_cnt = 0;
+    size_t shard1_idx = 0;
+    size_t shard2_idx = 0;
+    size_t shard3_idx = 0;
+
+    for (size_t i = 0; i < shard4->get_record_count(); ++i) {
+        auto rec1 = shard1->get_record_at(shard1_idx);
+        auto rec2 = shard2->get_record_at(shard2_idx);
+        auto rec3 = shard3->get_record_at(shard3_idx);
+
+        auto cur_rec = shard4->get_record_at(i);
+
+        if (shard1_idx < n && cur_rec->rec == rec1->rec) {
+            ++shard1_idx;
+        } else if (shard2_idx < n && cur_rec->rec == rec2->rec) {
+            ++shard2_idx;
+        } else if (shard3_idx < n && cur_rec->rec == rec3->rec) {
+            ++shard3_idx;
+        } else {
+           assert(false);
+        }
+    }
+
+    delete mbuffer1;
+    delete mbuffer2;
+    delete mbuffer3;
+
+    delete shard1;
+    delete shard2;
+    delete shard3;
+    delete shard4;
+}
+
+START_TEST(t_point_lookup) 
+{
+    size_t n = 10000;
+
+    auto buffer = create_test_mbuffer<R>(n);
+    auto shard = Shard(buffer->get_buffer_view());
+
+    {
+        auto view = buffer->get_buffer_view();
+
+        for (size_t i=0; i<n; i++) {
+            R r;
+            auto rec = view.get(i);
+            r.key = rec->rec.key;
+            r.value = rec->rec.value;
+
+            auto result = shard.point_lookup(r);
+            ck_assert_ptr_nonnull(result);
+            ck_assert_str_eq(result->rec.key.c_str(), r.key.c_str());
+            ck_assert_int_eq(result->rec.value, r.value);
+            fprintf(stderr, "%ld\n", i);
+        }
+    }
+
+    delete buffer;
+}
+END_TEST
+
+
+START_TEST(t_point_lookup_miss) 
+{
+    size_t n = 10000;
+
+    auto buffer = create_test_mbuffer<R>(n);
+    auto shard = Shard(buffer->get_buffer_view());
+
+    for (size_t i=n + 100; i<2*n; i++) {
+        R r;
+        r.key = std::string("computer");
+        r.value = 1234;
+
+        auto result = shard.point_lookup(r);
+        ck_assert_ptr_null(result);
+    }
+
+    delete buffer;
+}
+
+static void inject_shard_tests(Suite *suite) {
+    TCase *create = tcase_create("Shard constructor Testing");
+    tcase_add_test(create, t_mbuffer_init);
+    tcase_add_test(create, t_shard_init);
+    tcase_set_timeout(create, 100);
+    suite_add_tcase(suite, create);
+
+    TCase *pointlookup = tcase_create("Shard point lookup Testing"); 
+    tcase_add_test(pointlookup, t_point_lookup);
+    tcase_add_test(pointlookup, t_point_lookup_miss); 
+    suite_add_tcase(suite, pointlookup);
+}
diff --git a/tests/include/testing.h b/tests/include/testing.h
index f935b53..a3c54c0 100644
--- a/tests/include/testing.h
+++ b/tests/include/testing.h
@@ -15,6 +15,8 @@
 
 #include <unistd.h>
 #include <fcntl.h>
+#include <fstream>
+#include <sstream>
 
 #include "util/types.h"
 #include "psu-util/alignment.h"
@@ -25,6 +27,38 @@ typedef de::WeightedRecord<uint64_t, uint32_t, uint64_t> WRec;
 typedef de::Record<uint64_t, uint32_t> Rec;
 typedef de::EuclidPoint<uint64_t> PRec;
 
+typedef de::Record<std::string, uint64_t> StringRec;
+
+std::string kjv_wordlist = "tests/data/kjv-wordlist.txt";
+std::string summa_wordlist = "tests/data/summa-wordlist.txt";
+
+static std::vector<StringRec> read_string_data(std::string fname, size_t n) {
+    std::vector<StringRec> vec;
+    vec.reserve(n);
+
+    std::fstream file;
+    file.open(fname, std::ios::in);
+
+    for (size_t i=0; i<n; i++) {
+        std::string line;
+        if (!std::getline(file, line, '\n')) break;
+
+        std::stringstream ls(line);
+        StringRec r;
+        std::string field;
+
+        std::getline(ls, field, '\t');
+        r.value = atol(field.c_str());
+        std::getline(ls, field, '\n');
+        r.key = std::string(field);
+
+        vec.push_back(r);
+    }
+
+    return vec;
+}
+
+
 template <de::RecordInterface R> 
 std::vector<R> strip_wrapping(std::vector<de::Wrapped<R>> vec) {
     std::vector<R> out(vec.size());
@@ -83,15 +117,26 @@ static de::MutableBuffer<R> *create_test_mbuffer(size_t cnt)
 
     R rec;
     if constexpr (de::KVPInterface<R>) {
-        for (size_t i = 0; i < cnt; i++) {
-            rec.key = rand();
-            rec.value = rand();
-
-            if constexpr (de::WeightedRecordInterface<R>) {
-                rec.weight = 1;
+        if constexpr (std::is_same_v<R, StringRec>) {
+            auto records = read_string_data(kjv_wordlist, cnt);
+            for (size_t i=0; i<cnt; i++) {
+                if constexpr (de::WeightedRecordInterface<R>) {
+                    rec.weight = 1;
+                }
+
+                buffer->append(records[i]);
             }
+        } else {
+            for (size_t i = 0; i < cnt; i++) {
+                rec.key = rand();
+                rec.value = rand();
+
+                if constexpr (de::WeightedRecordInterface<R>) {
+                    rec.weight = 1;
+                }
 
-            buffer->append(rec);
+                buffer->append(rec);
+            }
         }
     } else if constexpr (de::NDRecordInterface<R>) {
         for (size_t i=0; i<cnt; i++) {
-- 
cgit v1.2.3