From 26364a229e4ec6f0b04e70fc8696570478970844 Mon Sep 17 00:00:00 2001
From: Sebastian Hiebl <bastidest@mailbox.org>
Date: Sun, 5 Nov 2023 20:41:44 +0100
Subject: [PATCH] add lazy loading of subsequent chunks

---
 src/ChunkedJournal.hpp | 69 +++++++++++++++++++++++++++++++-----------
 1 file changed, 51 insertions(+), 18 deletions(-)

diff --git a/src/ChunkedJournal.hpp b/src/ChunkedJournal.hpp
index 72ec62c..c1cbaed 100644
--- a/src/ChunkedJournal.hpp
+++ b/src/ChunkedJournal.hpp
@@ -4,6 +4,7 @@
 #include "SdJournal.hpp"
 #include "SdLine.hpp"
 
+#include <cassert>
 #include <list>
 #include <span>
 #include <vector>
@@ -13,26 +14,35 @@ namespace jess {
 static constexpr auto uChunkSize = 1024;
 
 class ChunkedJournal {
+  enum class Contiguity {
+    CONTIGUOUS,
+    NON_CONTIGUOUS,
+    OVERLAPPING,
+  };
+
+  enum class InsertPosition {
+    BEFORE,
+    AFTER,
+  };
+
   struct Chunk {
     // we might need multiple cursors depending on the sequence id
     SdCursor beginning;
     SdCursor end;
     std::vector<SdLine> lines;
+    Contiguity contiguityBeginning{Contiguity::NON_CONTIGUOUS};
+    Contiguity contiguityEnd{Contiguity::NON_CONTIGUOUS};
 
     std::partial_ordering operator<=>(const Chunk &other) const { return beginning <=> other.beginning; }
   };
 
   SdJournal m_journal{};
   std::list<Chunk> m_chunks{};
-  Chunk *m_pCurrentChunk{nullptr};
+  decltype(m_chunks.begin()) m_pCurrentChunk{m_chunks.begin()};
   size_t m_uLineOffsetInChunk{0};
 
-  auto addChunk(Chunk chunk) -> decltype(m_chunks.begin()) {
-    auto it = std::upper_bound(m_chunks.begin(), m_chunks.end(), chunk);
-    return m_chunks.insert(it, std::move(chunk));
-  }
-
-  auto loadChunk(const SdCursor &cursor) -> decltype(m_chunks.begin()) {
+  auto createChunkFromCurrentPosition(const SdCursor &cursor, InsertPosition insertPos, Contiguity contiguity)
+      -> decltype(m_chunks.begin()) {
     Chunk newChunk{cursor, cursor, {}};
 
     newChunk.lines.reserve(uChunkSize);
@@ -46,35 +56,58 @@ class ChunkedJournal {
 
     newChunk.end = m_journal.getCursor();
 
-    return addChunk(std::move(newChunk));
+    auto insertIt = m_pCurrentChunk;
+    if (insertPos == InsertPosition::AFTER) {
+      std::advance(insertIt, 1);
+    }
+    auto newChunkIt = m_chunks.insert(insertIt, std::move(newChunk));
+
+    if (insertPos == InsertPosition::BEFORE) {
+      newChunkIt->contiguityEnd = contiguity;
+      if(m_pCurrentChunk != m_chunks.end()) {
+        m_pCurrentChunk->contiguityBeginning = contiguity;
+      }
+    } else if (insertPos == InsertPosition::AFTER) {
+      if(m_pCurrentChunk != m_chunks.end()) {
+        m_pCurrentChunk->contiguityEnd = contiguity;
+      }
+      newChunkIt->contiguityBeginning = contiguity;
+    }
+
+    return newChunkIt;
   }
 
 public:
-
   void seekToBof() {
     m_journal.seekToBof();
     m_journal.next();
     auto cursorBof = m_journal.getCursor();
 
     if (m_chunks.empty() || m_chunks.front().beginning == cursorBof) {
-      m_pCurrentChunk = loadChunk(cursorBof).operator->();
+      m_pCurrentChunk = createChunkFromCurrentPosition(cursorBof, InsertPosition::BEFORE, Contiguity::NON_CONTIGUOUS);
     }
 
-    m_pCurrentChunk = &m_chunks.front();
     m_uLineOffsetInChunk = 0;
   }
 
   void seekLines(int64_t uNumLines) {
-    if(uNumLines < 0 && static_cast<size_t>(std::abs(uNumLines)) > m_uLineOffsetInChunk) {
+    assert(m_pCurrentChunk != m_chunks.end());
+
+    if (uNumLines < 0 && static_cast<size_t>(std::abs(uNumLines)) > m_uLineOffsetInChunk) {
       m_uLineOffsetInChunk = 0;
-      // todo: modulo logic?
       // todo: load previous chunk
       return;
     }
-    if(m_uLineOffsetInChunk + uNumLines > uChunkSize) {
-      m_uLineOffsetInChunk = 0;
-      // todo: modulo logic?
-      // todo: load next chunk
+    if (m_uLineOffsetInChunk + uNumLines > m_pCurrentChunk->lines.size()) {
+      int64_t uLinesAfterEndOfChunk =
+          uNumLines - static_cast<int64_t>(m_pCurrentChunk->lines.size() - m_uLineOffsetInChunk);
+      size_t uNumChunksToSkip = uLinesAfterEndOfChunk / uChunkSize;
+      m_uLineOffsetInChunk = uLinesAfterEndOfChunk - uNumChunksToSkip * uChunkSize;
+      m_journal.seekForward(uNumChunksToSkip * uChunkSize);
+      m_journal.next();
+      auto cursorBeginningOfChunk = m_journal.getCursor();
+      Contiguity contiguity = uNumChunksToSkip == 0 ? Contiguity::CONTIGUOUS : Contiguity::NON_CONTIGUOUS;
+      m_pCurrentChunk = createChunkFromCurrentPosition(cursorBeginningOfChunk, InsertPosition::AFTER, contiguity);
       return;
     }
 
@@ -82,7 +115,7 @@ class ChunkedJournal {
   }
 
   std::span<SdLine> getLines(size_t uNumLines) {
-    if (m_pCurrentChunk == nullptr) {
+    if (m_pCurrentChunk == m_chunks.end()) {
       return {};
     }
     std::span ret{m_pCurrentChunk->lines};