From 8863e95fb4ba39c43188bb0a2cb2cc7cf110a67c Mon Sep 17 00:00:00 2001
From: Aryan Arora <aryanarora.w1@gmail.com>
Date: Tue, 5 Mar 2024 00:19:54 +0530
Subject: [PATCH] refactor FileImpl::verify()

---
 src/fileimpl.cpp | 37 +++++++++++++++----------------------
 src/fileimpl.h   |  2 ++
 2 files changed, 17 insertions(+), 22 deletions(-)

diff --git a/src/fileimpl.cpp b/src/fileimpl.cpp
index 746298769..a19f6d0d3 100644
--- a/src/fileimpl.cpp
+++ b/src/fileimpl.cpp
@@ -551,47 +551,40 @@ class Grouping
     struct zim_MD5_CTX md5ctx;
     zim_MD5Init(&md5ctx);
 
-    const int piece_size=1024;
-    unsigned char ch[piece_size];
+    unsigned char ch[CHUNK_SIZE];
     offset_type checksumPos = header.getChecksumPos();
-    offset_type currentPos = 0;
+    offset_type toRead = checksumPos;
     
     for(auto part = zimFile->begin();
         part != zimFile->end();
         part++) {
       std::ifstream stream(part->second->filename(), std::ios_base::in|std::ios_base::binary);
 
-
-      while(currentPos < checksumPos-piece_size && stream.read(reinterpret_cast<char*>(ch),piece_size).good()) {
-        zim_MD5Update(&md5ctx, ch, piece_size);
-        currentPos+=stream.gcount();
+      while(toRead>=CHUNK_SIZE && stream.read(reinterpret_cast<char*>(ch),CHUNK_SIZE).good()) {
+        zim_MD5Update(&md5ctx, ch, CHUNK_SIZE);
+        toRead-=stream.gcount();
       }
 
-      // It reads and updates the checksum with the remaining amount of file
-      // when we reach the end of the file
+      // It reads the remaining amount of file when we reach the end of the file
       if(stream.good()){
-        int remaining=checksumPos-currentPos;
-        stream.read(reinterpret_cast<char*>(ch),remaining);
-        zim_MD5Update(&md5ctx, ch, remaining);
-        currentPos+=stream.gcount();
-      }
-      // It updates the checksum with the remaining amount of part when we
-      // reach the end of the part or filesize is less than piece_size
-      else if(stream.gcount()>0){
-        int remaining=checksumPos<piece_size?checksumPos:stream.gcount();
-        zim_MD5Update(&md5ctx, ch, remaining);
-        currentPos+=remaining;
+        stream.read(reinterpret_cast<char*>(ch),toRead);
       }
+      
+      // It updates the checksum with the remaining amount of data when we
+      // reach the end of the file or part
+        zim_MD5Update(&md5ctx, ch, stream.gcount());
+        toRead-=stream.gcount();
+    
       if (stream.bad()) {
         perror("error while reading file");
         return false;
       }
-      if (currentPos == checksumPos) {
+      if (!toRead) {
         break;
       }
     }
 
-    if (currentPos != checksumPos) {
+    if (toRead) {
       return false;
     }
 
diff --git a/src/fileimpl.h b/src/fileimpl.h
index b07d2c7f1..25ea19006 100644
--- a/src/fileimpl.h
+++ b/src/fileimpl.h
@@ -145,6 +145,8 @@ namespace zim
       const std::string& getMimeType(uint16_t idx) const;
 
       std::string getChecksum();
+
+      #define CHUNK_SIZE 1024
       bool verify();
       bool is_multiPart() const;