Lephisto/src/pack.c

#include <stdio.h>
#include <sys/stat.h> /* S_IRUSR */
#include <unistd.h>
#include <errno.h>
#include <string.h>
#include <malloc.h>

#include "pack.h"
#include "log.h"
#include "md5.h"

/**
 * @file pack.c
 *
 * @brief Stores data in funky format.
 *
 * Format:
 *  -- Index.
 *    -- Magic number (16 bytes).
 *    -- Number of files (uint32_t).
 *    -- Files in format name/location.
 *      -- File name (128 bytes max, ending in \0).
 *      -- File location (uint32_t).
 *  -- File data in format Size/Data.
 *    -- File size (uint32_t).
 *    -- File data (char*).
 *    -- File MD5 (16 byte char*).
 *  -- EOF.
 *
 * -- Write magic number and number of files.
 * -- Write the index.
 * -- Pack the files.
 */

#undef DEBUG /* This will be spammy. */
#define DEBUG(str, args...) do{;} while(0)

#define BLOCKSIZE 128*1024 /**< The read/write block size */

/* Max filename length. */
#define MAX_FILENAME 100 /**< Maximum file name length. */

#define PERMS S_IRUSR | S_IWUSR | S_IRGRP | S_IWGRP | S_IROTH /**< Default permissions. */

const uint64_t magic = 0x25524573; /**< File magic number: sER% */

/**
 * @fn static off_t getfilesize(const char* filename)
 *
 * @brief Get the file size.
 *    @param filename File to get the size of.
 *    @return The size of the file.
 */
static off_t getfilesize(const char* filename) {
#ifdef _POSIX_SOURCE
  struct stat file;

  if(!stat(filename, &file))
    return file.st_size;

  ERR("Unable to get filesize of %s", filename);
  return 0;
#else
  long size;
  FILE* fp = fopen(filename, "rb");
  if(fp == NULL) return 0;

  fseek(fp, 0, SEEK_END);
  size = ftell(fp);

  fclose(fp);

  return size;
#endif
}

/**
 * @fn int pack_check(const char* filename)
 *
 * @brief Check to see if a file is a packfile
 *    @param filename Name of the file to check.
 *    @return 1 if it is a packfile, 0 if it isn't and -1 on error.
 */
int pack_check(const char* filename) {
  int ret;
  char* buf;

  buf = malloc(sizeof(magic));

#ifdef _POSIX_SOURCE
  int fd = open(filename, O_RDONLY);
  if(fd == -1) {
    ERR("Error opening %s: %s", filename, strerror(errno));
    return -1;
  }

  if(read(fd, buf, sizeof(magic)) != sizeof(magic)) {
    ERR("Error reading magic number: %s", strerror(errno));
    free(buf);
    return -1;
  }
  ret = (memcmp(buf, &magic, sizeof(magic))==0) ? 0 : 1;
  close(fd);
#else
  FILE* file = fopen(filename, "rb");
  if(file == NULL) {
    ERR("Error opening '%s': %s", filename, strerror(errno));
    return -1;
  }

  buf = malloc(sizeof(magic));
  if(fread(buf, 1, sizeof(magic), file) != sizeof(magic)) {
    ERR("Error reading magic number: %s", strerror(errno));
    free(buf);
    return -1;
  }

  ret = (memcmp(buf, &magic, sizeof(magic))==0) ? 0 : 1;
  fclose(file);
#endif

  free(buf);
  return ret;
}

/**
 * @fn int pack_files(const char* outfile, const char** infiles, const uint32_t nfiles)
 *
 * @brief Packages files into a packfile.
 *    @param outfile Name of the file to output to.
 *    @param infiles Array of filenames to package.
 *    @param nfiles Number of filenames in infiles.
 *    @return 0 on success.
 */
#ifdef _POSIX_SOURCE
#define WRITE(b,n) if(write(outfd, b, n)==-1) { \
  ERR("Error writing to file: %s", strerror(errno)); \
  free(buf); return -1; }
#else
#define WRITE(b,n) if(fwrite(b, 1, n, outf)==0) { \
  ERR("Error writing to file: %s", strerror(errno)); \
  free(buf); return -1; }
#endif
int pack_files(const char* outfile, const char** infiles, const uint32_t nfiles) {
  void* buf;
#ifdef _POSIX_SOURCE
  struct stat file;
  int outfd, infd;
#else
  FILE* outf, *inf;
#endif
  uint32_t i;
  int namesize;
  uint32_t indexsize, pointer;
  int bytes;
  const uint8_t b = '\0';

  for(namesize = 0, i = 0; i < nfiles; i++) {
    /* Make sure file exists before writing. */
#ifdef _POSIX_SOURCE
    if(stat(infiles[i], &file)) {
#else
    if(getfilesize(infiles[i]) == 0) {
#endif
      ERR("File %s does not exist", infiles[i]);
      return -1;
    }
    if(strlen(infiles[i]) > MAX_FILENAME) {
      ERR("filename '%s' is too long, should be only %d characters", infiles[i],
          MAX_FILENAME);
      return -1;
    }
    namesize += strlen(infiles[i]);
  }

  indexsize = (sizeof(magic) + 4 + /* Magic number and number of files. */
               namesize + /* Total length of file names. */
               (1+4)*nfiles);  /* File size and extra end of string char '\0'. */
  DEBUG("Index size is %d", indexsize);

  /* Create the output file. */
#ifdef _POSIX_SOURCE
  outfd = creat(outfile, PERMS);
  if(outfd == -1) {
#else
  outf = fopen(outfile, "wb");
  if(outf == NULL) {
#endif
    ERR("Unable to open %s for writing", outfile);
    return -1;
  }

  /* Index! */
  buf = malloc(BLOCKSIZE);

  /* Magic number. */
  WRITE(&magic, sizeof(magic));
  DEBUG("Wrote magic number");
  /* Number of files. */
  WRITE(&nfiles, sizeof(nfiles));
  DEBUG("Wrote number of files: %d", nfiles);
  /* Create file dependent index part. */
  pointer = indexsize;
  for(i = 0; i < nfiles; i++) {
    WRITE(infiles[i], strlen(infiles[i]));
    DEBUG("File: '%s' at %d", infiles[i], pointer);
    WRITE(&b, 1);
    WRITE(&pointer, 4);
    pointer += 4 + getfilesize(infiles[i]) + 16; /* Set pointer to be next file pos. */
  }

  /* Data! */
  md5_state_t md5;
  md5_byte_t* md5val = malloc(16);
  for(i = 0; i < nfiles; i++) {
    bytes = (uint32_t)getfilesize(infiles[i]);
    WRITE(&bytes, 4); /* filesize. */
    DEBUG("About to write '%s' of %d bytes", infiles[i], bytes);
    md5_init(&md5);
#ifdef _POSIX_SOURCE
    infd = open(infiles[i], O_RDONLY);
    while((bytes = read(infd, buf, BLOCKSIZE))) {
#else
    inf = fopen(infiles[i], "rb");
    while((bytes = fread(buf, 1, BLOCKSIZE, inf))) {
#endif
      WRITE(buf, bytes); /* Data. */
      md5_append(&md5, buf, bytes);
    }
    md5_finish(&md5, md5val);
    WRITE(md5val, 16);
#ifdef _POSIX_SOURCE
    close(infd);
#else
    fclose(inf);
#endif
    DEBUG("Wrote file '%s'", infiles[i]);
  }
  free(md5val);

#ifdef _POSIX_SOURCE
  close(outfd);
#else
  fclose(outf);
#endif
  free(buf);

  DEBUG("Packfile success\n\t%d files\n\t%d bytes",
        nfiles, (int)getfilesize(outfile));
  return 0;
}
#undef WRITE

/**
 * @fn int pack_open(Packfile* file, const char* packfile, const char* filename)
 *
 * @brief Open a file in the packfile for reading.
 *    @param file Packfile to store data into.
 *    @param packfile Path to real packfile.
 *    @param filename Name of the file within th. packfile.
 *    @return 0 on success.
 */
#ifdef _POSIX_SOURCE
#define READ(b,n) if(read(file->fd, (b), (n))!=(n)) { \
  ERR("Too few bytes read. Expected more."); \
  free(buf); return -1; }
#else
#define READ(b,n) if(fread((b), 1, (n), file->fp)!=(n)) { \
  ERR("Fewer bytes read then expected"); \
  free(buf); return -1; }
#endif
int pack_open(Packfile* file, const char* packfile, const char* filename) {
  int j;
  uint32_t nfiles, i;
  char* buf = malloc(MAX_FILENAME);

  file->start = file->end = 0;

#ifdef _POSIX_SOURCE
  file->fd = open(packfile, O_RDONLY);
  if(file->fd == -1) {
#else
  file->fp = fopen(packfile, "rb");
  if(file->fp == NULL) {
#endif
    ERR("Error opening %s: %s", filename, strerror(errno));
    return -1;
  }

  READ(buf, sizeof(magic)); /* Make sure it's a packfile. */
  if(memcmp(buf, &magic, sizeof(magic))) {
    ERR("File %s is not a valid packfile", filename);
    return -1;
  }

  READ(&nfiles, 4);
  for(i = 0; i < nfiles; i++) {
    /* Start to search files. */
    j = 0;
    READ(&buf[j], 1); /* Get the name. */
    while(buf[j++] != '\0')
      READ(&buf[j], 1);

    if(strcmp(filename, buf) == 0) {
      /* We found the file. */
      READ(&file->start, 4);
      DEBUG("'%s' found at %d", filename, file->start);
      break;
    }
#ifdef _POSIX_SOURCE
    lseek(file->fd, 4, SEEK_CUR); /* Ignore the file location. */
#else
    fseek(file->fp, 4, SEEK_CUR);
#endif
  }
  free(buf);

  if(file->start) {
    /* Go to the beginning of the file. */
#ifdef _POSIX_SOURCE
    if((uint32_t)lseek(file->fd, file->start, SEEK_SET) != file->start) {
#else
    fseek(file->fp, file->start, SEEK_SET);
    if(errno) {
#endif
      ERR("Failure to seek to file start: %s", strerror(errno));
      return -1;
    }
    READ(&file->end, 4);
    DEBUG("\t%d bytes", file->end);
    file->pos = file->start;
    file->end += file->start;
  } else {
    ERR("File '%s' not found in packfile '%s'", filename, packfile);
    return -1;
  }
  return 0;
}
#undef READ

/**
 * @fn ssize_t pack_read(Packfile* file, void* buf, size_t count)
 *
 * @brief Reads data from a packfile.
 *
 * Behaves like POSIX read.
 *    @param file Opened packfile to read data from.
 *    @param buf Allocated buffer to read into.
 *    @param count Bytes to read.
 *    @return Bytes to read or -1 on error.
 */
ssize_t pack_read(Packfile* file, void* buf, size_t count) {
  if(file->pos + count > file->end)
    count = file->end - file->pos; /* Can't go past the end! */
  if(count == 0) return 0;

  int bytes;

#ifdef _POSIX_SOURCE
  if((bytes = read(file->fd, buf, count)) == -1) {
#else
  if((bytes = fread(buf, 1, count, file->fp)) == -1) {
#endif
    ERR("Error while reading file: %s", strerror(errno));
    return -1;
  }
  file->pos += bytes;

  return bytes;
}

/**
 * @fn off_t pack_seek(Packfile* file, off_t offset, int whence)
 *
 * @brief Seeks within a file inside a packfile.
 *
 * Behaves like lseek/fseek.
 *
 * @todo It's broken, needs fixing.
 *    @param file File to seek.
 *    @param offset Position to seek to.
 *    @param whence Either SEEK_SET, SEEK_CUR or SEEK_END.
 *    @return The position moved to.
 */
off_t pack_seek(Packfile* file, off_t offset, int whence) {
  DEBUG("Attempting to seek offset: %d, whence: %d", offset, whence);
  off_t ret;
  switch(whence) {
#ifdef _POSIX_SOURCE
  case SEEK_SET:
    if((file->start + offset) > file->end) return -1;
    ret = lseek(file->fd, file->start + offset, SEEK_SET);
    if(ret != ((off_t)file->start + offset)) return -1;
    break;
  case SEEK_CUR:
    if((file->start + offset) > file->end) return -1;
    ret = lseek(file->fd, file->pos + offset, SEEK_SET);
    if(ret != ((off_t)file->pos + offset)) return -1;
    break;
  case SEEK_END:
    if((file->end - offset) < file->start) return -1;
    ret = lseek(file->fd, file->end - offset - 1, SEEK_SET);
    if(ret != ((off_t)file->end - offset)) return -1;
    break;
#else
  case SEEK_SET:
    if((file->start + offset) > file->end) return -1;
    ret = fseek(file->fd, file->start + offset, SEEK_SET);
    if(ret != (file->start + offset)) return -1;
    break;
  case SEEK_CUR:
    if((file->start + offset) > file->end) return -1;
    ret = flseek(file->fd, file->pos + offset, SEEK_SET);
    if(ret != (file->pos + offset)) return -1;
    break;
  case SEEK_END:
    if((file->end - offset) < file->start) return -1;
    ret = flseek(file->fd, file->end - offset - 1, SEEK_SET);
    if(ret != (file->end - offset)) return -1;
    break;
#endif
  default:
    ERR("Whence is not one of SEEK_SET, SEEK_CUR or SEEK_END");
    return -1;
  }
  return ret - file->start;
}

/**
 * @fn long pack_tell(Packfile* file)
 *
 * @brief Get the current position in the file.
 *    @param file Packfile to get the position from.
 *    @return The current position in the file.
 */
long pack_tell(Packfile* file) {
  return file->pos - file->start;
}

/**
 * @fn void* pack_readfile(const char* packfile, const char* filename, uint32_t* filesize)
 *
 * @brief Read an entire file into memory.
 *    @param packfile Name of the packfile to read from.
 *    @param filename Name of the packed file to read.
 *    @param filesize Is set to the size of the file.
 *    @return A pointer to the sata in the file or NULL if an error accured.
 */
void* pack_readfile(const char* packfile, const char* filename, uint32_t* filesize) {
  Packfile* file = (Packfile*)malloc(sizeof(Packfile));
  void* buf;
  char* str;
  int size, bytes;

  if(filesize)
    *filesize = 0;

  if(pack_open(file, packfile, filename)) {
    ERR("Opening packfile");
    return NULL;
  }
  DEBUG("Opened file '%s' from '%s'", filename, packfile);

  /* Read the entire file. */
  size = file->end - file->start;
  buf = malloc(size+1);
  if(buf == NULL) {
    ERR("Unable to allocate %d bytes of memory!", size+1);
    free(file);
    return NULL;
  }
  if((bytes = pack_read(file, buf, size)) != size) {
    ERR("Reading '%s' from packfile '%s'. Expected %d bytes got %d bytes",
        filename, packfile, size, bytes);
    free(buf);
    free(file);
    return NULL;
  }
  DEBUG("Read %d bytes from '%s'", bytes, filename);
  str = buf;
  str[size] = '\0'; /* Append size '\0' for it to validate as a string. */

  /* Check the md5. */
  md5_state_t md5;
  md5_byte_t* md5val = malloc(16);
  md5_byte_t* md5fd = malloc(16);
  md5_init(&md5);
  md5_append(&md5, buf, bytes);
  md5_finish(&md5, md5val);
#ifdef _POSIX_SOURCE
  if((bytes = read(file->fd, md5fd, 16)) == -1)
#else
  if((bytes = fread(md5fd, 1, 16, file->fp)) == -1)
#endif
    WARN("Failure to read MD5, continuing anyway..");
  else if(memcmp(md5val, md5fd, 16))
    WARN("MD5 gives different value, possible memory corruption, continuing..");
  free(md5val);
  free(md5fd);

  /* Cleanup. */
  if(pack_close(file) == -1) {
    ERR("Closing packfile");
    free(file);
    return NULL;
  }
  DEBUG("Closed '%s' in '%s'", filename, packfile);
  free(file);

  if(filesize)
    *filesize = size;
  return buf;
}

/**
 * @fn char** pack_listfiles(const char* packfile, uint32_t* nfiles)
 *
 * @brief Get what files are in the packfile.
 *
 * Each name must be freed individually afterwords and the array of names too.
 *    @param packfile Packfile to query it's internal files.
 *    @param nfiles Stores the amount of files in packfile.
 *    @return An array of filenames in packfile.
 */
#ifdef _POSIX_SOURCE
#define READ(b,n) if(read(fd, (b), (n))!=(n)) { \
  ERR("Too few bytes read"); \
  return NULL; }
#else
#define READ(b,n) if(fread((b), 1, (n), fp) != (n)) { \
  ERR("Too few bytes read"); \
  return NULL; }
#endif
char** pack_listfiles(const char* packfile, uint32_t* nfiles) {
#ifdef _POSIX_SOURCE
  int fd;
#else
  FILE* fp;
#endif
  int j;
  uint32_t i;
  char** filenames;
  char* buf = malloc(sizeof(magic));

  *nfiles = 0;

#ifdef _POSIX_SOURCE
  fd = open(packfile, O_RDONLY);
  if(fd == -1) {
#else
  fp = fopen(packfile, "rb");
  if(fp == NULL) {
#endif
    ERR("opening %s: %s", packfile, strerror(errno));
    return NULL;
  }
  READ(buf, sizeof(magic)); /* Make sure it is a packfile. */
  if(memcmp(buf, &magic, sizeof(magic))) {
    ERR("File %s is not a valid packfile", packfile);
    return NULL;
  }

  READ(nfiles, 4);
  filenames = malloc(((*nfiles)+1)*sizeof(char*));
  for(i = 0; i < *nfiles; i++) {
    /* Start searching files. */
    j = 0;
    filenames[i] = malloc(MAX_FILENAME * sizeof(char));
    READ(&filenames[i][j], 1); /* Get the name. */
    while(filenames[i][j++] != '\0')
      READ(&filenames[i][j], 1);
    READ(buf, 4); /* skip the location. */
  }
  free(buf);

#ifdef _POSIX_SOURCE
  close(fd);
#else
  fclose(fp);
#endif

  return filenames;
}
#undef READ

/**
 * @fn int pack_close(Packfile* file)
 *
 * @brief Closes a packfile.
 *    @param file Packfile to close.
 *    @return 0 on success.
 */
int pack_close(Packfile* file) {
  int i;
#ifdef _POSIX_SOURCE
  i = close(file->fd);
#else
  i = fclose(file->fp);
#endif
  return (i) ? -1 : 0;
}