test(sqllogictest): add test utils regarding bpm & disk (#764)

Introduce necessary test utils for checking students' external merge sort implementations, including:
1. Maximum number of pages written to disk
2. Minimum number of pages written to disk
3. Minimum number of pages deleted on disk

Author: xx01cyx

src/include/storage/disk/disk_manager.h

@@ -68,6 +68,12 @@ class DiskManager {
    */
   virtual void ReadPage(page_id_t page_id, char *page_data);
 
+  /**
+   * Delete a page from the database file. Reclaim the disk space.
+   * @param page_id id of the page
+   */
+  virtual void DeletePage(page_id_t page_id);
+
   /**
    * Flush the entire log buffer into disk.
    * @param log_data raw log data
@@ -93,6 +99,9 @@ class DiskManager {
   /** @return the number of disk writes */
   auto GetNumWrites() const -> int;
 
+  /** @return the number of deletions */
+  auto GetNumDeletes() const -> int;
+
   /**
    * Sets the future which is used to check for non-blocking flushes.
    * @param f the non-blocking flush check
@@ -115,6 +124,7 @@ class DiskManager {
   std::filesystem::path file_name_;
   int num_flushes_{0};
   int num_writes_{0};
+  int num_deletes_{0};
   bool flush_log_{false};
   std::future<void> *flush_log_f_{nullptr};
   // With multiple buffer pool instances, need to protect file access

src/include/storage/disk/disk_manager_memory.h

@@ -123,6 +123,7 @@ class DiskManagerUnlimitedMemory : public DiskManager {
     l.unlock();
 
     memcpy(ptr->first.data(), page_data, BUSTUB_PAGE_SIZE);
+    num_writes_ += 1;
 
     PostProcessLatency(page_id);
   }
@@ -158,6 +159,14 @@ class DiskManagerUnlimitedMemory : public DiskManager {
     PostProcessLatency(page_id);
   }
 
+  /**
+   * Delete a page from the database file. Reclaim the disk space.
+   * Note: This is a no-op for now without a more complex data structure to
+   * track deallocated pages.
+   * @param page_id id of the page
+   */
+  void DeletePage(page_id_t page_id) override { num_deletes_ += 1; }
+
   void ProcessLatency(page_id_t page_id) {
     uint64_t sleep_micro_sec = 1000;  // for random access, 1ms latency
     if (latency_simulator_enabled_) {

src/include/storage/disk/disk_scheduler.h

@@ -100,7 +100,7 @@ class DiskScheduler {
    *
    * @param page_id The page ID of the page to deallocate from disk.
    */
-  void DeallocatePage(page_id_t page_id) {}
+  void DeallocatePage(page_id_t page_id) { disk_manager_->DeletePage(page_id); }
 
  private:
   /** Pointer to the disk manager. */

src/storage/disk/disk_manager.cpp

@@ -148,6 +148,12 @@ void DiskManager::ReadPage(page_id_t page_id, char *page_data) {
   }
 }
 
+/**
+ * Note: This is a no-op for now without a more complex data structure to
+ * track deallocated pages.
+ */
+void DiskManager::DeletePage(page_id_t page_id) { num_deletes_ += 1; }
+
 /**
  * Write the contents of the log into disk file
  * Only return when sync is done, and only perform sequence write
@@ -215,15 +221,20 @@ auto DiskManager::ReadLog(char *log_data, int size, int offset) -> bool {
  */
 auto DiskManager::GetNumFlushes() const -> int { return num_flushes_; }
 
+/**
+ * Returns true if the log is currently being flushed
+ */
+auto DiskManager::GetFlushState() const -> bool { return flush_log_; }
+
 /**
  * Returns number of Writes made so far
  */
 auto DiskManager::GetNumWrites() const -> int { return num_writes_; }
 
 /**
- * Returns true if the log is currently being flushed
+ * Returns number of deletions made so far
  */
-auto DiskManager::GetFlushState() const -> bool { return flush_log_; }
+auto DiskManager::GetNumDeletes() const -> int { return num_deletes_; }
 
 /**
  * Private helper function to get disk file size

tools/sqllogictest/sqllogictest.cpp

@@ -10,6 +10,7 @@
 
 #include "argparse/argparse.hpp"
 #include "common/bustub_instance.h"
+#include "common/config.h"
 #include "common/exception.h"
 #include "common/util/string_util.h"
 #include "execution/check_options.h"
@@ -137,30 +138,30 @@ auto ProcessExtraOptions(const std::string &sql, bustub::BusTubInstance &instanc
         auto expected_cols_proj = std::stoi(args[2]);
         auto expected_cols_agg = std::stoi(args[3]);
         // find agg & proj plan and test if the output schema has the expected number of columns
-          auto lines = bustub::StringUtil::Split(result.str(), "\n");
-          for (auto &line : lines) {
-            bustub::StringUtil::LTrim(&line);
-            if (bustub::StringUtil::StartsWith(line, "Agg")) {
-              auto cols = bustub::StringUtil::Split(line, "],");
-              if (cols.size() != 3) {
-                fmt::print("Agg plan wrong formatting!\n");
-                return false;
-              }
-              for (int i = 0; i < 2; i++) {
-                if (bustub::StringUtil::Count(cols[i], "\",")+1 > static_cast<size_t>(expected_cols_agg)) {
-                  fmt::print("Agg wrong column pruning count!\n");
-                  return false;
-                }
-              }
-              break;
+        auto lines = bustub::StringUtil::Split(result.str(), "\n");
+        for (auto &line : lines) {
+          bustub::StringUtil::LTrim(&line);
+          if (bustub::StringUtil::StartsWith(line, "Agg")) {
+            auto cols = bustub::StringUtil::Split(line, "],");
+            if (cols.size() != 3) {
+              fmt::print("Agg plan wrong formatting!\n");
+              return false;
             }
-            if (bustub::StringUtil::StartsWith(line, "Projection")) {
-              if (bustub::StringUtil::Count(line, "\",")+1 > static_cast<size_t>(expected_cols_proj)) {
-                fmt::print("Projection wrong column pruning count!\n");
+            for (int i = 0; i < 2; i++) {
+              if (bustub::StringUtil::Count(cols[i], "\",") + 1 > static_cast<size_t>(expected_cols_agg)) {
+                fmt::print("Agg wrong column pruning count!\n");
                 return false;
               }
             }
+            break;
           }
+          if (bustub::StringUtil::StartsWith(line, "Projection")) {
+            if (bustub::StringUtil::Count(line, "\",") + 1 > static_cast<size_t>(expected_cols_proj)) {
+              fmt::print("Projection wrong column pruning count!\n");
+              return false;
+            }
+          }
+        }
       } else {
         throw bustub::NotImplementedException(fmt::format("unsupported extra option: {}", opt));
       }
@@ -224,6 +225,15 @@ auto main(int argc, char **argv) -> int {  // NOLINT
   program.add_argument("--verbose").help("increase output verbosity").default_value(false).implicit_value(true);
   program.add_argument("-d", "--diff").help("write diff file").default_value(false).implicit_value(true);
   program.add_argument("--in-memory").help("use in-memory backend").default_value(false).implicit_value(true);
+  program.add_argument("--bpm-size")
+      .help("size of the buffer pool")
+      .default_value(std::to_string(bustub::BUFFER_POOL_SIZE));
+  program.add_argument("--check-min-disk-write")
+      .help("the minimum disk write threshold to be checked at the end of the program");
+  program.add_argument("--check-max-disk-write")
+      .help("the maximum disk write threshold to be checked at the end of the program");
+  program.add_argument("--check-min-disk-delete")
+      .help("the maximum disk deletion threshold to be checked at the end of the program");
 
   try {
     program.parse_args(argc, argv);
@@ -235,6 +245,8 @@ auto main(int argc, char **argv) -> int {  // NOLINT
 
   bool verbose = program.get<bool>("verbose");
   bool diff = program.get<bool>("diff");
+  auto check_min_disk_write = program.present("check-min-disk-write");
+
   std::string filename = program.get<std::string>("file");
   std::ifstream t(filename);
 
@@ -253,11 +265,12 @@ auto main(int argc, char **argv) -> int {  // NOLINT
   }
 
   std::unique_ptr<bustub::BusTubInstance> bustub;
+  size_t bpm_size = std::stoul(program.get<std::string>("bpm-size"));
 
   if (program.get<bool>("--in-memory")) {
-    bustub = std::make_unique<bustub::BusTubInstance>();
+    bustub = std::make_unique<bustub::BusTubInstance>(bpm_size);
   } else {
-    bustub = std::make_unique<bustub::BusTubInstance>("test.bustub");
+    bustub = std::make_unique<bustub::BusTubInstance>("test.bustub", bpm_size);
   }
 
   bustub->GenerateMockTable();
@@ -365,5 +378,30 @@ auto main(int argc, char **argv) -> int {  // NOLINT
     }
   }
 
+  if (program.is_used("--check-min-disk-write")) {
+    int min_disk_write_num = std::stoi(program.get("--check-min-disk-write"));
+    int actual_disk_write_num = bustub->disk_manager_->GetNumWrites();
+    if (actual_disk_write_num < min_disk_write_num) {
+      fmt::print("test incurred {} times of disk write, which is too low\n", actual_disk_write_num);
+      return 1;
+    }
+  }
+  if (program.is_used("--check-max-disk-write")) {
+    int max_disk_write_num = std::stoi(program.get("--check-max-disk-write"));
+    int actual_disk_write_num = bustub->disk_manager_->GetNumWrites();
+    if (actual_disk_write_num > max_disk_write_num) {
+      fmt::print("test incurred {} times of disk write, which is too high\n", actual_disk_write_num);
+      return 1;
+    }
+  }
+  if (program.is_used("--check-min-disk-delete")) {
+    int min_disk_delete_num = std::stoi(program.get("--check-min-disk-delete"));
+    int actual_disk_delete_num = bustub->disk_manager_->GetNumDeletes();
+    if (actual_disk_delete_num < min_disk_delete_num) {
+      fmt::print("test incurred {} times of disk deletion, which is too low\n", actual_disk_delete_num);
+      return 1;
+    }
+  }
+
   return 0;
 }