nure/semester-4/ОС/lb-7/src/editor/analizer.c

#define _XOPEN_SOURCE
#include <time.h>

#include <fcntl.h>
#include <fts.h>
#include <iconv.h>
#include <stdio.h>
#include <stdlib.h>

#include "shared.h"

typedef enum { ENC_UTF8, ENC_UTF16LE, ENC_UTF16BE, ENC_CP1251 } Encoding;

Encoding get_encoding(char *input, size_t length) {
  unsigned char *data = (unsigned char *)input;

  if (length >= 2) {
    if (data[0] == 0xFE && data[1] == 0xFF) { // UTF-16 BE BOM
      return ENC_UTF16BE;
    } else if (data[0] == 0xFF && data[1] == 0xFE) { // UTF-16 LE BOM
      return ENC_UTF16LE;
    }
  }

  size_t i = 0;
  while (i < length) {
    unsigned char c = data[i];

    if (c < 0x80) { // ASCII
      i++;
      continue;
    }

    if ((c & 0xE0) == 0xC0) { // 2-byte UTF8
      if (i + 1 >= length || (data[i + 1] & 0xC0) != 0x80 || c < 0xC2)
        return ENC_CP1251; // overflow
      i += 2;
      continue;
    }

    if ((c & 0xF0) == 0xE0) { // 3-byte UTF8
      if (i + 2 >= length || (data[i + 1] & 0xC0) != 0x80 ||
          (data[i + 2] & 0xC0) != 0x80 || (c == 0xE0 && data[i + 1] < 0xA0))
        return ENC_CP1251; // overflow
      i += 3;
      continue;
    }

    if ((c & 0xF8) == 0xF0) { // 4-byte UTF8
      if (i + 3 >= length || (data[i + 1] & 0xC0) != 0x80 ||
          (data[i + 2] & 0xC0) != 0x80 || (data[i + 3] & 0xC0) != 0x80 ||
          (c == 0xF0 && data[i + 1] < 0x90))
        return ENC_CP1251; // overflow
      i += 4;
      continue;
    }

    return ENC_CP1251; // Invalid utf-8 byte
  }

  return ENC_UTF8;
}

char *to_utf8(char *input, size_t length, Encoding encoding) {
  if (!input || length == 0)
    return NULL;

  char *encstr;
  int skip_bom = 0;
  switch (encoding) {
  case ENC_CP1251:
    encstr = "CP1251";
    break;
  case ENC_UTF8:
    encstr = "UTF-8";
    break;
  case ENC_UTF16LE:
    encstr = "UTF-16LE";
    skip_bom = 2;
    break;
  case ENC_UTF16BE:
    encstr = "UTF-16BE";
    skip_bom = 2;
    break;
  }

  iconv_t cd = iconv_open("UTF-8", encstr);
  if (cd != (iconv_t)-1) {
    size_t out_size = length * 4 + 1;
    char *output = malloc(out_size);
    if (output) {
      char *in_ptr = (char *)input + skip_bom;
      size_t in_left = length - skip_bom;
      char *out_ptr = output;
      size_t out_left = out_size - 1;

      if (iconv(cd, &in_ptr, &in_left, &out_ptr, &out_left) != (size_t)-1) {
        *out_ptr = '\0';
        iconv_close(cd);
        return output;
      }
      free(output);
    }
    iconv_close(cd);
  }

  return NULL;
}

int main(int argc, char *argv[]) {
  if (argc > 2) {
    fprintf(stderr, "Error: too many arguments.\n");
    return 1;
  }

  if (argc < 2) {
    fprintf(stderr, "Error: too few arguments.\n");
    return 1;
  }

  struct tm tm = {0};
  if (strptime(argv[1], "%s", &tm) == NULL) {
    fprintf(stderr, "Error: please provide time in UNIX timestamp format.\n");
    return 1;
  }
  time_t norm_time = mktime(&tm);

  char *paths[] = {CWD, NULL};
  FTS *fts = fts_open(paths, FTS_NOCHDIR, NULL);
  if (fts == NULL) {
    err("Can't initialise FTS", __LINE__ - 1);
    return 2;
  }

  char *buf = (char *)malloc(1024 * sizeof(char));

  for (FTSENT *ent = fts_read(fts); ent != NULL; ent = fts_read(fts)) {
    switch (ent->fts_info) {
    case FTS_F:
      if (ent->fts_statp->st_mtime > norm_time) {
        printf(RED);
        printf("--------------------------------\n");

        printf(GREEN);
        printf("File: %s\n", ent->fts_name);

        char mtime[1024] = {0};
        strftime(mtime, 1024, "%T", localtime(&ent->fts_statp->st_mtime));
        printf("Modified at: %s\n", mtime);

        printf("Size: %ld\n", ent->fts_statp->st_size);

        FILE *file = fopen(ent->fts_path, "r");
        if (file == NULL) {
          err("Can't open the file", __LINE__ - 1);
          break;
        }

        size_t bytes_read = fread(buf, 1, 1023, file);
        if (ferror(file) != 0) {
          err("Can't read the file", __LINE__ - 2);
          fclose(file);
          break;
        }
        buf[bytes_read] = '\0';

        Encoding enc = get_encoding(buf, bytes_read);

        printf("Lines length:");
        size_t lines = 0;
        size_t line_len = 0;
        for (size_t i = (enc == 2 || enc == 1) ? 2 : 0; i < bytes_read; i++) {
          line_len++;

          int newline_detected = 0;
          switch (enc) {
          case ENC_UTF16BE:
            if (i + 1 < bytes_read && buf[i] == '\0' && buf[i + 1] == '\n') {
              newline_detected = 1;
              line_len++;
              i++;
            }
            break;
          case ENC_UTF16LE:
            if (i + 1 < bytes_read && buf[i] == '\n' && buf[i + 1] == '\0') {
              newline_detected = 1;
              line_len++;
              i++;
            }
            break;
          case ENC_CP1251:
          case ENC_UTF8:
          default:
            if (buf[i] == '\n')
              newline_detected = 1;
            break;
          }

          if (newline_detected) {
            printf(" %zu", line_len);
            lines++;
            line_len = 0;
          }
        }
        printf("\nTotal lines: %zu\n", lines);

        if (enc != ENC_UTF8)
          buf = to_utf8(buf, bytes_read, enc);

        printf(NORMAL);
        printf("%s\n", buf);

        fclose(file);
      }
      break;
    default:
      break;
    }
  }

  free(buf);

  return 0;
}