refactor: also return the size prefix size in read_header()

`read_header()` also returns the number of bytes read. Make that number
contain the total amount of bytes read by this function, which includes
the size prefix as well as the byte size of the header itself.

examples/indexinfo.rs

@@ -3,7 +3,7 @@ use std::env;
 use std::fs::File;
 use std::io::BufReader;
 
-use storethehash::index::{self, IndexIter, SIZE_PREFIX_SIZE};
+use storethehash::index::{self, IndexIter};
 use storethehash::recordlist::{RecordList, BUCKET_PREFIX_SIZE};
 
 fn index_info(index_path: &str) {
@@ -13,7 +13,7 @@ fn index_info(index_path: &str) {
     let (_header, bytes_read) = index::read_header(&mut index_file).unwrap();
 
     let mut buffered = BufReader::new(index_file);
-    for entry in IndexIter::new(&mut buffered, SIZE_PREFIX_SIZE + bytes_read) {
+    for entry in IndexIter::new(&mut buffered, bytes_read) {
         match entry {
             Ok((data, _pos)) => {
                 let bucket = u32::from_le_bytes(data[..BUCKET_PREFIX_SIZE].try_into().unwrap());

examples/indexstats.rs

@@ -4,7 +4,7 @@ use std::env;
 use std::fs::File;
 use std::io::BufReader;
 
-use storethehash::index::{self, IndexIter, SIZE_PREFIX_SIZE};
+use storethehash::index::{self, IndexIter};
 use storethehash::recordlist::{RecordList, BUCKET_PREFIX_SIZE};
 
 fn index_stats(index_path: &str) -> BTreeMap<u32, Vec<usize>> {
@@ -16,7 +16,7 @@ fn index_stats(index_path: &str) -> BTreeMap<u32, Vec<usize>> {
     let (_header, bytes_read) = index::read_header(&mut index_file).unwrap();
 
     let mut buffered = BufReader::new(index_file);
-    for entry in IndexIter::new(&mut buffered, SIZE_PREFIX_SIZE + bytes_read) {
+    for entry in IndexIter::new(&mut buffered, bytes_read) {
         match entry {
             Ok((data, _pos)) => {
                 let bucket = u32::from_le_bytes(data[..BUCKET_PREFIX_SIZE].try_into().unwrap());

src/index.rs

@@ -109,7 +109,7 @@ impl<P: PrimaryStorage, const N: u8> Index<P, N> {
                 // TODO vmx 2020-11-30: Find if there's a better way than cloning the file. Perhaps
                 // a BufReader should be used instead of File for this whole module?
                 let mut buffered = BufReader::new(file.try_clone()?);
-                for entry in IndexIter::new(&mut buffered, SIZE_PREFIX_SIZE + bytes_read) {
+                for entry in IndexIter::new(&mut buffered, bytes_read) {
                     match entry {
                         Ok((data, pos)) => {
                             let bucket_prefix = u32::from_le_bytes(
@@ -398,14 +398,20 @@ pub fn read_size_prefix<R: Read>(reader: &mut R) -> Result<usize, io::Error> {
 }
 
 /// Returns the headet together with the bytes read.
+///
+/// The bytes read include all the bytes that were read by this function. Hence it also includes
+/// the 4-byte size prefix of the header besides the size of the header data itself.
 pub fn read_header(file: &mut File) -> Result<(Header, usize), io::Error> {
     let mut header_size_buffer = [0; SIZE_PREFIX_SIZE];
     file.read_exact(&mut header_size_buffer)?;
     let header_size =
         usize::try_from(u32::from_le_bytes(header_size_buffer)).expect(">=32-bit platform needed");
     let mut header_bytes = vec![0u8; header_size];
     file.read_exact(&mut header_bytes)?;
-    Ok((Header::from(&header_bytes[..]), header_size))
+    Ok((
+        Header::from(&header_bytes[..]),
+        SIZE_PREFIX_SIZE + header_size,
+    ))
 }
 
 /// Returns the position of the first character that both given slices have not in common.

tests/lib.rs

@@ -2,7 +2,7 @@ use std::convert::TryInto;
 use std::fs::{self, File};
 use std::path::Path;
 
-use storethehash::index::{self, Header, Index, IndexIter, INDEX_VERSION, SIZE_PREFIX_SIZE};
+use storethehash::index::{self, Header, Index, IndexIter, INDEX_VERSION};
 use storethehash::recordlist::RecordList;
 use storethehash_primary_inmemory::InMemory;
 
@@ -35,7 +35,7 @@ fn assert_common_prefix_trimmed(key1: Vec<u8>, key2: Vec<u8>, expected_key_lengt
 
     // The record list is append only, hence the first record list only contains the first insert
     {
-        let (data, _pos) = IndexIter::new(&mut file, SIZE_PREFIX_SIZE + bytes_read)
+        let (data, _pos) = IndexIter::new(&mut file, bytes_read)
             .next()
             .unwrap()
             .unwrap();
@@ -49,7 +49,7 @@ fn assert_common_prefix_trimmed(key1: Vec<u8>, key2: Vec<u8>, expected_key_lengt
 
     // The second block contains both keys
     {
-        let (data, _pos) = IndexIter::new(&mut file, SIZE_PREFIX_SIZE + bytes_read)
+        let (data, _pos) = IndexIter::new(&mut file, bytes_read)
             .next()
             .unwrap()
             .unwrap();
@@ -81,7 +81,7 @@ fn index_put_single_key() {
     let mut file = File::open(index_path).unwrap();
     let (_header, bytes_read) = index::read_header(&mut file).unwrap();
 
-    let (data, _pos) = IndexIter::new(&mut file, SIZE_PREFIX_SIZE + bytes_read)
+    let (data, _pos) = IndexIter::new(&mut file, bytes_read)
         .next()
         .unwrap()
         .unwrap();
@@ -110,7 +110,7 @@ fn index_put_distinct_key() {
     let mut file = File::open(index_path).unwrap();
     let (_header, bytes_read) = index::read_header(&mut file).unwrap();
 
-    let (data, _pos) = IndexIter::new(&mut file, SIZE_PREFIX_SIZE + bytes_read)
+    let (data, _pos) = IndexIter::new(&mut file, bytes_read)
         .last()
         .unwrap()
         .unwrap();
@@ -165,7 +165,7 @@ fn index_put_prev_and_next_key_common_prefix() {
     let mut file = File::open(index_path).unwrap();
     let (_header, bytes_read) = index::read_header(&mut file).unwrap();
 
-    let (data, _pos) = IndexIter::new(&mut file, SIZE_PREFIX_SIZE + bytes_read)
+    let (data, _pos) = IndexIter::new(&mut file, bytes_read)
         .last()
         .unwrap()
         .unwrap();