Improve parsing data type

CMakeLists.txt

@@ -11,6 +11,7 @@ file(MAKE_DIRECTORY ${PGNP_INCLUDE_DIR})
 configure_file(src/PGN.hpp ${PGNP_INCLUDE_DIR}/pgnp.hpp COPYONLY)
 configure_file(src/HalfMove.hpp ${PGNP_INCLUDE_DIR} COPYONLY)
 configure_file(src/LargeFileStream.hpp ${PGNP_INCLUDE_DIR} COPYONLY)
+configure_file(src/Types.hpp ${PGNP_INCLUDE_DIR} COPYONLY)
 
 include_directories(${PGNP_INCLUDE_DIR})
 

README.md

@@ -9,8 +9,8 @@ PGN specification can be found [here](https://www.chessclub.com/help/PGN-spec).
 # Features
 - Basic PGN parsing (tags, move, comments, variations, NAG, etc.)
 - Merged PGN files parsing (several games in one file)
-- Handle very large file (severals GB)
-- Very efficient
+- Handle very large file (max is 2^(sizeof(unsigned long long)) bytes)
+- Efficiency
 
 # How to use it ?
 PGNP can be used as a shared library in your project.

src/HalfMove.hpp

@@ -1,3 +1,6 @@
+#pragma once
+
+#include "Types.hpp"
 #include <sstream>
 #include <string>
 #include <vector>
@@ -35,11 +38,13 @@ public:
   std::string Dump();
   /// @brief Perform a deep copy of a HalfMove
   void Copy(HalfMove *copy);
-  /// @brief Get HalfMove located x down the MainLine 
-  HalfMove* GetHalfMoveAt(int);
+  /// @brief Get HalfMove located x down the MainLine
+  HalfMove *GetHalfMoveAt(int);
 };
 
 struct HalfMoveOutOfRange : public std::exception {
-  const char *what() const throw() { return "HalfMove distance is out of range"; }
+  const char *what() const throw() {
+    return "HalfMove distance is out of range";
+  }
 };
 } // namespace pgnp

src/LargeFileStream.cpp

@@ -23,7 +23,7 @@ void LargeFileStream::ReadNextChunk() {
   last_read_size = file.gcount();
 }
 
-char LargeFileStream::operator[](long loc) {
+char LargeFileStream::operator[](ull loc) {
   // Perform various checks
   if (eof) {
     throw ReadToFar();
@@ -42,11 +42,11 @@ char LargeFileStream::operator[](long loc) {
   }
 
   // Goto the right memory chuck
-  long loc_chunk_count = loc / BUFFER_SIZE;
+  ull loc_chunk_count = loc / BUFFER_SIZE;
   while (chuck_count < loc_chunk_count) {
     ReadNextChunk();
   }
-  long offset = loc - (loc_chunk_count * BUFFER_SIZE);
+  ull offset = loc - (loc_chunk_count * BUFFER_SIZE);
   
   // Ensure for EOF
   if (!file && offset >= last_read_size) {

src/LargeFileStream.hpp

@@ -1,9 +1,12 @@
-#define BUFFER_SIZE (1024 * 1024 / 2)
+#pragma once
 
+#include "Types.hpp"
 #include <fstream>
 #include <iostream>
 #include <string>
 
+#define BUFFER_SIZE (1024 * 1024 / 2)
+
 namespace pgnp {
 using namespace std;
 
@@ -13,11 +16,11 @@ class LargeFileStream {
   /// @brief In memory buffer
   char buffer[BUFFER_SIZE];
   /// @brief Number of chuck read minus 1
-  long chuck_count;
+  ull chuck_count;
   /// @brief Number of byte read during the last file access
-  long last_read_size;
+  ull last_read_size;
   /// @brief Keep track of the file offset (to prevent backward read)
-  long last_loc;
+  ull last_loc;
   /// @brief Use a string as file content
   std::string content;
   /// @brief Use to shortcut some methods
@@ -34,7 +37,7 @@ public:
   /// @brief Emulate file access with a string
   void FromString(std::string content);
   /// @brief Allow array like access to the file
-  char operator[](long loc);
+  char operator[](ull loc);
   /// @brief Check if we reach the EOF
   bool IsEOF();
 

src/PGN.cpp

@@ -45,7 +45,7 @@ void PGN::ParseNextGame() {
   if (IS_EOF) {
     throw NoGameFound();
   }
-  long loc = GotoNextToken(LastGameEndLoc);
+  ull loc = GotoNextToken(LastGameEndLoc);
   if (IS_EOF) {
     throw NoGameFound();
   }
@@ -103,7 +103,7 @@ bool PGN::HasTag(std::string key) {
   return (std::find(tags.begin(), tags.end(), key) != tags.end());
 }
 
-long PGN::ParseComment(long loc, HalfMove *hm) {
+ull PGN::ParseComment(ull loc, HalfMove *hm) {
   // Goto next char
   loc = GotoNextToken(loc);
   EOF_CHECK(loc);
@@ -131,7 +131,7 @@ long PGN::ParseComment(long loc, HalfMove *hm) {
   return (loc);
 }
 
-long PGN::ParseHalfMove(long loc, HalfMove *hm) {
+ull PGN::ParseHalfMove(ull loc, HalfMove *hm) {
   // Goto next char
   loc = GotoNextToken(loc);
   EOF_CHECK(loc);
@@ -249,10 +249,10 @@ long PGN::ParseHalfMove(long loc, HalfMove *hm) {
   return (loc);
 }
 
-long PGN::ParseNextTag(long start_loc) {
+ull PGN::ParseNextTag(ull start_loc) {
   // Parse key
   std::string key;
-  long keyloc = start_loc + 1;
+  ull keyloc = start_loc + 1;
   EOF_CHECK(keyloc);
   char c = pgn_content[keyloc];
   while (!IS_BLANK(c)) {
@@ -264,7 +264,7 @@ long PGN::ParseNextTag(long start_loc) {
 
   // Parse value
   std::string value;
-  long valueloc = GotoNextToken(keyloc) + 1;
+  ull valueloc = GotoNextToken(keyloc) + 1;
   EOF_CHECK(keyloc);
   c = pgn_content[valueloc];
   while (c != '"' or IS_EOF) {
@@ -312,7 +312,7 @@ std::string PGN::Dump() {
   return (ss.str());
 }
 
-long PGN::GotoNextToken(long loc) {
+ull PGN::GotoNextToken(ull loc) {
   char c = pgn_content[loc];
   while (IS_BLANK(c)) {
     loc++;
@@ -331,7 +331,7 @@ long PGN::GotoNextToken(long loc) {
   return (loc);
 }
 
-long PGN::GotoEOL(long loc) {
+ull PGN::GotoEOL(ull loc) {
   char c = pgn_content[loc];
   while (true) {
     loc++;

src/PGN.hpp

@@ -1,5 +1,8 @@
+#pragma once
+
 #include "HalfMove.hpp"
 #include "LargeFileStream.hpp"
+#include "Types.hpp"
 #include <algorithm>
 #include <exception>
 #include <fstream>
@@ -21,7 +24,7 @@ private:
   LargeFileStream pgn_content;
   /// @brief Contains the location of the end of the last parsed game (1 PGN
   /// file may have multiple games)
-  long LastGameEndLoc;
+  ull LastGameEndLoc;
 
 public:
   PGN();
@@ -52,16 +55,16 @@ public:
 private:
   /// @brief Populate @a tags with by parsing the one starting at location in
   /// argument
-  long ParseNextTag(long);
+  ull ParseNextTag(ull);
   /// @brief Parse a HalfMove at a specific location into @a pgn_content
-  long ParseHalfMove(long, HalfMove *);
+  ull ParseHalfMove(ull, HalfMove *);
   /// @brief Parse a consecutive sequence of comment
-  long ParseComment(long, HalfMove *);
+  ull ParseComment(ull, HalfMove *);
   /// @brief Get the next non-blank char location ignoring line comments ('%'
   /// and ';')
-  long GotoNextToken(long);
+  ull GotoNextToken(ull);
   /// @brief Goto the end of the current line
-  long GotoEOL(long);
+  ull GotoEOL(ull);
 };
 
 struct UnexpectedEOF : public std::exception {
@@ -82,7 +85,7 @@ struct NoGameFound : public std::exception {
 
 struct UnexpectedCharacter : public std::exception {
   std::string msg;
-  UnexpectedCharacter(char actual, char required, long loc) {
+  UnexpectedCharacter(char actual, char required, ull loc) {
     std::stringstream ss;
     ss << "Expected \'" << required << "\' at location " << loc
        << " but read \'" << actual << "\'";

src/Types.hpp

@@ -0,0 +1,5 @@
+#pragma once
+
+namespace pgnp {
+typedef unsigned long long ull;
+}