libmp/src/url.c

#include "../include/url.h"
#include "../include/error.h"
#include "../include/util.h"
#include <assert.h>
#include <stdlib.h>
#include <string.h>

char valid_path[] = "-._~:/?#[]@!$&'()*+,;%=";

typedef enum lm_state {
  URL_PROTOCOL_0 = 0,
  URL_SPLIT_1    = 1,
  URL_HOST_2     = 2,
  URL_PATH_3     = 3,
} lm_state_t;

uint16_t lm_url_default(char *protocol) {
  if (eq(protocol, "ftp"))
    return 21;
  else if (eq(protocol, "ftps"))
    return 990;
  else if (eq(protocol, "http"))
    return 80;
  else if (eq(protocol, "https"))
    return 443;
  else if (eq(protocol, "mptp"))
    return 5858;
  return 0;
}

bool lm_url_parse(lm_url_t *url, char *str) {
  // clear out every variable
  bzero(url->protocol, sizeof(url->protocol));
  url->host = NULL;
  url->path = NULL;
  url->port = 0;

  // stores the string size
  size_t strl = 0, index = 0, pos = 0;

  // make sure the URL string size is not too large
  // extra 4 for "://" and ":"
  if ((strl = strlen(str)) > URL_PROTOCOL_MAX + URL_PATH_MAX + URL_HOST_MAX + 4) {
    lm_error_set(LM_ERR_URLTooLarge);
    return false;
  }

  lm_state_t state = URL_PROTOCOL_0;
  char       buffer[strl + 1], *save; // temporary buffer, strok_r save pointer
  bool       ret = false;             // return value

  // clear out the temporary buffer
  bzero(buffer, strl + 1);

  while ((buffer[index] = *(str + pos)) != 0) {
    switch (state) {
    case URL_PROTOCOL_0:
      if (index > URL_PROTOCOL_MAX) {
        lm_error_set(LM_ERR_URLBadProtocol);
        goto end;
      }

      if (!is_letter(buffer[index]) && !is_digit(buffer[index]) && buffer[index] != ':') {
        lm_error_set(LM_ERR_URLBadChar);
        goto end;
      }

      if (buffer[index] != ':')
        break;

      if (0 == index) {
        lm_error_set(LM_ERR_URLBadProtocol);
        goto end;
      }

      buffer[index] = 0;
      memcpy(url->protocol, buffer, index + 1);
      goto next;

    case URL_SPLIT_1:
      if (index > 1) {
        lm_error_set(LM_ERR_URLBadProtocol);
        goto end;
      }

      if (buffer[index] != '/') {
        lm_error_set(LM_ERR_URLBadChar);
        goto end;
      }

      if (index != 1)
        break;

      if (buffer[index - 1] != '/' || buffer[index] != '/') {
        lm_error_set(LM_ERR_URLBadProtocol);
        goto end;
      }

      goto next;

    case URL_HOST_2:
      if (index > URL_HOST_MAX) {
        lm_error_set(LM_ERR_URLHostLarge);
        goto end;
      }

      if (!is_letter(buffer[index]) && !is_digit(buffer[index]) && buffer[index] != '.' && buffer[index] != ':' &&
          buffer[index] != '/') {
        lm_error_set(LM_ERR_URLBadChar);
        goto end;
      }

      if (buffer[index] != '/')
        break;

      if (index == 0) {
        lm_error_set(LM_ERR_URLBadHost);
        goto end;
      }

      buffer[index] = 0;
      url->host     = malloc((index + 1) * sizeof(char));
      memcpy(url->host, buffer, index + 1);
      goto next;

    case URL_PATH_3:
      if (index > URL_PATH_MAX) {
        lm_error_set(LM_ERR_URLPathLarge);
        goto end;
      }

      if (!is_letter(buffer[index]) && !is_digit(buffer[index]) && !contains(valid_path, buffer[index])) {
        lm_error_set(LM_ERR_URLBadChar);
        goto end;
      }

      break;

    default:
      assert(false);
    }

    index++;
    pos++;
    continue;

  next:
    bzero(buffer, strl + 1);
    state++;
    index = 0;
    pos++;
  }

  switch (state) {
  case URL_HOST_2:
    if (index == 0) {
      lm_error_set(LM_ERR_URLBadHost);
      goto end;
    }

    if (index > URL_HOST_MAX) {
      lm_error_set(LM_ERR_URLHostLarge);
      goto end;
    }

    url->host = malloc((index + 1) * sizeof(char));
    memcpy(url->host, buffer, index + 1);

    url->path    = malloc(2 * sizeof(char));
    url->path[0] = '/';
    url->path[1] = 0;
    break;

  case URL_PATH_3:
    url->path    = malloc((index + 2) * sizeof(char));
    url->path[0] = '/';
    memcpy(url->path + 1, buffer, index + 1);
    break;

  default:
    lm_error_set(LM_ERR_URLEnd);
    goto end;
  }

  if (strtok_r(url->host, ":", &save) == NULL) {
    lm_error_set(LM_ERR_URLBadHost);
    goto end;
  }

  char *portc = strtok_r(NULL, ":", &save);
  if (NULL == portc) {
    url->port = lm_url_default(url->protocol);
    if (url->port == 0) {
      lm_error_set(LM_ERR_URLPortUnknown);
      goto end;
    }
    ret = true;
    goto end;
  }

  int port = atoi(portc);
  if (port <= 0 || port > UINT16_MAX) {
    lm_error_set(LM_ERR_URLBadPort);
    goto end;
  }

  url->port = port;
  ret       = true;

end:
  if (!ret && NULL != url->host)
    free(url->host);

  if (!ret && NULL != url->path)
    free(url->path);

  return ret;
}

void lm_url_free(lm_url_t *url) {
  if (NULL != url->host)
    free(url->host);

  if (NULL != url->path)
    free(url->path);
}