sys_prog/splitter/splitter.c

// vim: set ts=2 sw=2 et tw=80:
/*
 * Assignment 1 - File Splitter
 * Claudio Maggioni
 *
 * This code was written by using documentation on man(3) and cplusplus.com (C
 * standard library section).
 *
 * The flag VERBOSE_ERRORS, if defined, activates proper error message handling
 * and printing on stderr. This is disabled to comply with tests.
 */

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <libgen.h>

/*
 * Maximum number of part files expected, as defined by the assignment. By
 * giving a size parameter too small thus having a number of part files greater
 * than this may cause buffer overflows while handling part file names.
 */
const unsigned int MAX_SPLIT = 1024;

/*
 * Size of the byte buffer used in merge to read part file. Increase or decrease
 * at will.
 */
const int BUFFER_SIZE = 1024;

int digits(int);
int usage();
int split(const char* filename, long max_size);
int merge(char* index, const char* merged);

/**
 * Compute an appoximate figure of the number of decimal digits in an unsigned
 * int by using an approximation of log base 8 (faster than log_10). Used only
 * to compute the mamimum length of part filenames.
 */
inline int digits(int x) {
  return (x ? (1 + digits(x << 3)) : 1);
}

/*
 * Print usage message on stderr if VERBOSE_ERRORS is enabled.
 */
inline int usage(const char* s) {
#ifdef VERBOSE_ERRORS
  fprintf(stderr, "Usage: %s split [filename] [max-size]\n"
      "       %s merge [index] [merged]\n", s, s);
#endif
  return 64;
}

int main(int argc, char** const argv) {
  if (argc != 4) return usage(argv[0]);
  if (strcmp(argv[1], "split") == 0) {
    long max = strtol(argv[3], NULL, 10);
    if (max <= 0) {
      return usage(argv[0]);
    }
    return split(argv[2], max);
  } else if (strcmp(argv[1], "merge") == 0) {
    return merge(argv[2], argv[3]);
  } else {
    return usage(argv[0]);
  }
}

/*
 * Compute checksum on part files by using XOR. An initial value must be given
 * in order to support computation of the checksum in chunks.
 */
unsigned char checksum(unsigned char start,
    const unsigned char* buffer, long size) {
  unsigned char sum = start;
  for (long i = 0; i < size; i++) {
    sum ^= buffer[i];
  }
  return sum;
}

/*
 * Entire of split subcommand. Returns the exit status code.
 */
int split(const char* filename, long max_size) {
  FILE* source = fopen(filename, "r");
  if (source == NULL) {
#ifdef VERBOSE_ERRORS
    perror("Cannot open source file");
#endif
    return 1;
  }

  const int fnlen = strlen(filename);

  // Compute the mamimum filename length and allocate a VLA to match that
  char index_name[fnlen + strlen(".part") + digits(MAX_SPLIT) + 1];

  sprintf(index_name, "%s.index", filename);
  FILE* index = fopen(index_name, "wx");

  if (index == NULL) {
#ifdef VERBOSE_ERRORS
    perror("Cannot create index file");
#endif
    fclose(source);
    return 2;
  }

  char part_name[fnlen + 10];
  unsigned char buffer[BUFFER_SIZE];

  for(int part = 1; !feof(source); part++) {
    // assign file name to part
    sprintf(part_name, "%s.part%d", filename, part);

    FILE* part_file = NULL;
    unsigned char sum = 0x00;

    int first = part == 1;
    for (long i = max_size; i > 0 && (!feof(source) || first);) {

      const long to_read = i > BUFFER_SIZE ? BUFFER_SIZE : i;
      const long bytes = fread(buffer, 1, to_read, source);

      if (ferror(source)) {
#ifdef VERBOSE_ERRORS
        perror("Cannot read source file");
#endif
        fclose(source);
        fclose(index);
        return 3;
      } else if (bytes == 0 && part > 1) {
        goto all_read;
      }

      if (part_file == NULL) {
        part_file = fopen(part_name, "wx");

        if (part_file == NULL) {
#ifdef VERBOSE_ERRORS
          perror("Cannot create part file");
#endif
          fclose(source);
          fclose(index);
          return 4;
        }
      }

      sum = checksum(sum, buffer, bytes);

      const long w_bytes = fwrite(buffer, 1, bytes, part_file);

      if (w_bytes == 0 && !first) {
#ifdef VERBOSE_ERRORS
        perror("Cannot write on part file");
#endif
        fclose(source);
        fclose(index);
        fclose(part_file);
        return 5;
      }

      i -= bytes;
      first = 0;
    }

    if (fprintf(index, "%s %02x\n", part_name, sum) < 0) {
#ifdef VERBOSE_ERRORS
      fprintf(stderr, "Cannot update index file");
#endif
      fclose(source);
      fclose(index);
      fclose(part_file);
      return 6;
    }

    fclose(part_file);
  }

all_read:
  fclose(source);
  fclose(index);
  return 0;
}

int merge(char* n_index, const char* n_merged) {
  FILE* index = fopen(n_index, "r");

  if (index == NULL) {
#ifdef VERBOSE_ERRORS
    perror("Cannot open index file");
#endif
    return 7;
  }

  FILE* merged = fopen(n_merged, "wx");

  if (merged == NULL) {
#ifdef VERBOSE_ERRORS
    perror("Cannot create merged file");
#endif
    fclose(index);
    return 8;
  }

  // Allocate enough memory to store the full path of each part
  // We don't care about the .part suffix since the file name length contains
  // .index which is already longer. VLAs again btw
  char part_name[strlen(n_index) + digits(MAX_SPLIT) + 1];

  unsigned int n_part = 1;

  while(!feof(index)) {
    unsigned int sum;

    fscanf(index, "%s %x\n", part_name, &sum);

    FILE* part = fopen(part_name, "r");

    if (part == NULL) {
#ifdef VERBOSE_ERRORS
      perror("Cannot open part");
#endif
      fclose(index);
      fclose(merged);
      return 9;
    }
    unsigned char check = 0x00;
    while (!feof(part)) {
      unsigned char buffer[1000];
      long bytes = fread(buffer, 1, 1000, part);

      if (ferror(part)) {
#ifdef VERBOSE_ERRORS
        perror("Cannot read part");
#endif
        fclose(index);
        fclose(merged);
        fclose(part);
        return 10;
      }

      if (bytes == 0) {
        break;
      }

      check = checksum(check, buffer, bytes);
      fwrite(buffer, 1, bytes, merged);

      if (ferror(part)) {
#ifdef VERBOSE_ERRORS
        perror("Cannot wrtite on merged");
#endif

        fclose(index);
        fclose(merged);
        fclose(part);
        return 11;
      }
    }

    fclose(part);

    if ((unsigned char) sum != check) {
#ifdef VERBOSE_ERRORS
      perror("Checksums differ for part");
#endif

      printf("%u\n", n_part);
      fclose(index);
      fclose(merged);
      return 12;
    }

    n_part++;
  }

  fclose(index);
  fclose(merged);
  return 0;
}