From d51f8f7de4ef95e69eb545f46479d96769162359 Mon Sep 17 00:00:00 2001
From: Codetector <codetector@codetector.org>
Date: Sat, 3 Aug 2024 22:54:31 -0700
Subject: [PATCH] parsing more pages.

---
 src/innodb/page/lob/data_page.rs | 69 ++++++++++++++++++++++++++++++++
 src/innodb/page/lob/mod.rs       |  2 +
 src/innodb/table/row.rs          | 20 +++++----
 3 files changed, 84 insertions(+), 7 deletions(-)
 create mode 100644 src/innodb/page/lob/data_page.rs

diff --git a/src/innodb/page/lob/data_page.rs b/src/innodb/page/lob/data_page.rs
new file mode 100644
index 0000000..d5b104d
--- /dev/null
+++ b/src/innodb/page/lob/data_page.rs
@@ -0,0 +1,69 @@
+use anyhow::{anyhow, Result};
+
+use crate::innodb::{page::{Page, PageType}, InnoDBError};
+
+#[derive(Debug, Clone)]
+pub struct LobDataHeader {
+    pub version: u8,
+    pub data_len: u32,
+    pub trx_id: u64, // 6 bytes
+}
+
+impl LobDataHeader {
+    pub fn size() -> usize {
+        1 + 4 + 6
+    }
+
+    pub fn try_from_bytes(buf: &[u8]) -> Result<Self> {
+        if buf.len() < 11 {
+            return Err(anyhow!("Buffer too short for LobDataHeader"));
+        }
+
+        let version = buf[0];
+        let data_len = u32::from_be_bytes(buf[1..5].try_into()?);
+
+        // trx_id is 6 bytes, so we need to pad it with two zero bytes for u64
+        let trx_id = u64::from_be_bytes([0, 0, buf[5], buf[6], buf[7], buf[8], buf[9], buf[10]]);
+
+        Ok(LobDataHeader {
+            version,
+            data_len,
+            trx_id,
+        })
+    }
+}
+
+#[derive(Debug)]
+pub struct LobData<'a> {
+    pub page: &'a Page<'a>,
+    pub header: LobDataHeader,
+}
+
+impl <'a> LobData<'a> {
+    pub fn try_from_page(p: &'a Page<'a>) -> Result<Self> {
+        match p.header.page_type {
+            PageType::LobData => Ok(LobData{
+                header: LobDataHeader::try_from_bytes(p.body())?,
+                page: p,
+            }),
+            _ => Err(anyhow!(InnoDBError::InvalidPageType {
+                expected: PageType::LobData,
+                has: p.header.page_type
+            })),
+        }
+    }
+
+    pub fn read(&self, offset: usize, buf: &mut [u8]) -> usize {
+        let data_len = self.header.data_len as usize;
+        let data = &self.body()[..data_len];
+        assert!(offset < data.len(), "offset too large");
+        let data = &data[offset..];
+        let bytes_to_copy = std::cmp::min(buf.len(), data.len());
+        buf[..bytes_to_copy].copy_from_slice(&data[..bytes_to_copy]);
+        bytes_to_copy
+    }
+
+    pub fn body(&self) -> &[u8] {
+        &self.page.body()[LobDataHeader::size()..]
+    }
+}
diff --git a/src/innodb/page/lob/mod.rs b/src/innodb/page/lob/mod.rs
index 65b09b6..6e41d4b 100644
--- a/src/innodb/page/lob/mod.rs
+++ b/src/innodb/page/lob/mod.rs
@@ -6,6 +6,8 @@ use anyhow::{anyhow, Ok, Result};
 
 use super::{Page, PageType};
 
+pub mod data_page;
+
 /*
  * General Flow for reading extern records
  *
diff --git a/src/innodb/table/row.rs b/src/innodb/table/row.rs
index f55ba1f..c537fe6 100644
--- a/src/innodb/table/row.rs
+++ b/src/innodb/table/row.rs
@@ -6,11 +6,11 @@ use std::{
 };
 
 use crate::innodb::{
-    buffer_manager::BufferManager,
+    buffer_manager::{self, BufferManager},
     file_list::FileListInnerNode,
     page::{
         index::record::{Record, RECORD_HEADER_FIXED_LENGTH},
-        lob::{LobFirst, LobIndexEntry},
+        lob::{data_page::LobData, LobFirst, LobIndexEntry},
     },
     table::blob_header::ExternReference,
     InnoDBError,
@@ -139,8 +139,9 @@ impl<'a> Row<'a> {
         extern_header: &ExternReference,
         buffer_mgr: &mut dyn BufferManager,
     ) -> Result<Box<[u8]>> {
+        let space_id = extern_header.space_id;
         let first_page_number = extern_header.page_number;
-        let lob_first_page = buffer_mgr.open_page(extern_header.space_id, first_page_number)?;
+        let lob_first_page = buffer_mgr.open_page(space_id, first_page_number)?;
         if lob_first_page.header.offset != extern_header.page_number {
             return Err(anyhow!(InnoDBError::InvalidPage));
         }
@@ -165,18 +166,23 @@ impl<'a> Row<'a> {
             let node = LobIndexEntry::try_from_bytes(buf)?;
             trace!("Index Node: {:#?}", node);
 
+            let mut bytes_read = 0usize;
             if node.page_number == first_page_number {
-                let bytes_read = lob_first.read(page_offset, &mut output_buffer[filled..]);
-                filled += bytes_read;
-                page_offset = page_offset.saturating_sub(bytes_read);
+                bytes_read = lob_first.read(page_offset, &mut output_buffer[filled..]);
                 trace!(
                     "Read {} bytes from first page, in total expecting {} bytes",
                     bytes_read,
                     output_buffer.len()
                 );
             } else {
-                // TODO: implemtn LobData
+                let page_guard = buffer_mgr.open_page(space_id, node.page_number)?;
+                let data_page = LobData::try_from_page(&page_guard)?;
+                trace!("Data page: {:#?}", data_page);
+                bytes_read = data_page.read(page_offset, &mut output_buffer[filled..]);
+                trace!("Read {} bytes from data page", bytes_read);
             }
+            filled += bytes_read;
+            page_offset = page_offset.saturating_sub(bytes_read);
 
             node_location = node.file_list_node.next;
         }