Merge pull request #861 from aryanA101a/main

src/fileimpl.cpp

@@ -19,6 +19,7 @@
  *
  */
 
+#define CHUNK_SIZE 1024
 #include "fileimpl.h"
 #include <zim/error.h>
 #include <zim/tools.h>
@@ -547,28 +548,45 @@ class Grouping
 
  struct zim_MD5_CTX md5ctx;
  zim_MD5Init(&md5ctx);
-
+
+ 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);
 
- char ch;
- for(/*NOTHING*/ ; currentPos < checksumPos && stream.get(ch).good(); currentPos++) {
- zim_MD5Update(&md5ctx, reinterpret_cast<const uint8_t*>(&ch), 1);
+ while(toRead>=CHUNK_SIZE && stream.read(reinterpret_cast<char*>(ch),CHUNK_SIZE).good()) {
+ zim_MD5Update(&md5ctx, ch, CHUNK_SIZE);
+ toRead-=CHUNK_SIZE;
+ }
+
+ // Previous read was good, so we have exited the previous `while` because
+ // `toRead<CHUNK_SIZE`. Let's try to read `toRead` chars and process them later.
+ // Else, the previous `while` exited because we didn't succeed to read
+ // `CHUNK_SIZE`, and we still have some data to process before changing part.
+ // It reads the remaining amount of part when we reach the end of the file
+ if(stream.good()){
+ 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;
  }
 
@@ -580,7 +598,6 @@ class Grouping
  {
  return false;
  }
-
  return true;
  }