url.c

Wget很好的处理了http和ftp的下载,很值得学习的经典代码

      /* We're looking for the first slash, but want to ignore
         double slash. */
    again:
      slash = memchr (pos, '/', end - pos);
      if (slash && !seen_slash_slash)
        if (*(slash + 1) == '/')
          {
            pos = slash + 2;
            seen_slash_slash = true;
            goto again;
          }

      /* At this point, SLASH is the location of the first / after
         "//", or the first slash altogether.  START_INSERT is the
         pointer to the location where LINK will be inserted.  When
         examining the last two examples, keep in mind that LINK
         begins with '/'. */

      if (!slash && !seen_slash_slash)
        /* example: "foo" */
        /*           ^    */
        start_insert = base;
      else if (!slash && seen_slash_slash)
        /* example: "http://foo" */
        /*                     ^ */
        start_insert = end;
      else if (slash && !seen_slash_slash)
        /* example: "foo/bar" */
        /*           ^        */
        start_insert = base;
      else if (slash && seen_slash_slash)
        /* example: "http://something/" */
        /*                           ^  */
        start_insert = slash;

      span = start_insert - base;
      merge = xmalloc (span + linklength + 1);
      if (span)
        memcpy (merge, base, span);
      memcpy (merge + span, link, linklength);
      merge[span + linklength] = '\0';
    }
  else
    {
      /* LINK is a relative URL: we need to replace everything
         after last slash (possibly empty) with LINK.

         So, if BASE is "whatever/foo/bar", and LINK is "qux/xyzzy",
         our result should be "whatever/foo/qux/xyzzy".  */
      bool need_explicit_slash = false;
      int span;
      const char *start_insert;
      const char *last_slash = find_last_char (base, end, '/');
      if (!last_slash)
        {
          /* No slash found at all.  Replace what we have with LINK. */
          start_insert = base;
        }
      else if (last_slash && last_slash >= base + 2
               && last_slash[-2] == ':' && last_slash[-1] == '/')
        {
          /* example: http://host"  */
          /*                      ^ */
          start_insert = end + 1;
          need_explicit_slash = true;
        }
      else
        {
          /* example: "whatever/foo/bar" */
          /*                        ^    */
          start_insert = last_slash + 1;
        }

      span = start_insert - base;
      merge = xmalloc (span + linklength + 1);
      if (span)
        memcpy (merge, base, span);
      if (need_explicit_slash)
        merge[span - 1] = '/';
      memcpy (merge + span, link, linklength);
      merge[span + linklength] = '\0';
    }

  return merge;
}

#define APPEND(p, s) do {                       \
  int len = strlen (s);                         \
  memcpy (p, s, len);                           \
  p += len;                                     \
} while (0)

/* Use this instead of password when the actual password is supposed
   to be hidden.  We intentionally use a generic string without giving
   away the number of characters in the password, like previous
   versions did.  */
#define HIDDEN_PASSWORD "*password*"

/* Recreate the URL string from the data in URL.

   If HIDE is true (as it is when we're calling this on a URL we plan
   to print, but not when calling it to canonicalize a URL for use
   within the program), password will be hidden.  Unsafe characters in
   the URL will be quoted.  */

char *
url_string (const struct url *url, enum url_auth_mode auth_mode)
{
  int size;
  char *result, *p;
  char *quoted_host, *quoted_user = NULL, *quoted_passwd = NULL;

  int scheme_port = supported_schemes[url->scheme].default_port;
  const char *scheme_str = supported_schemes[url->scheme].leading_string;
  int fplen = full_path_length (url);

  bool brackets_around_host;

  assert (scheme_str != NULL);

  /* Make sure the user name and password are quoted. */
  if (url->user)
    {
      if (auth_mode != URL_AUTH_HIDE)
        {
          quoted_user = url_escape_allow_passthrough (url->user);
          if (url->passwd)
            {
              if (auth_mode == URL_AUTH_HIDE_PASSWD)
                quoted_passwd = HIDDEN_PASSWORD;
              else
                quoted_passwd = url_escape_allow_passthrough (url->passwd);
            }
        }
    }

  /* In the unlikely event that the host name contains non-printable
     characters, quote it for displaying to the user.  */
  quoted_host = url_escape_allow_passthrough (url->host);

  /* Undo the quoting of colons that URL escaping performs.  IPv6
     addresses may legally contain colons, and in that case must be
     placed in square brackets.  */
  if (quoted_host != url->host)
    unescape_single_char (quoted_host, ':');
  brackets_around_host = strchr (quoted_host, ':') != NULL;

  size = (strlen (scheme_str)
          + strlen (quoted_host)
          + (brackets_around_host ? 2 : 0)
          + fplen
          + 1);
  if (url->port != scheme_port)
    size += 1 + numdigit (url->port);
  if (quoted_user)
    {
      size += 1 + strlen (quoted_user);
      if (quoted_passwd)
        size += 1 + strlen (quoted_passwd);
    }

  p = result = xmalloc (size);

  APPEND (p, scheme_str);
  if (quoted_user)
    {
      APPEND (p, quoted_user);
      if (quoted_passwd)
        {
          *p++ = ':';
          APPEND (p, quoted_passwd);
        }
      *p++ = '@';
    }

  if (brackets_around_host)
    *p++ = '[';
  APPEND (p, quoted_host);
  if (brackets_around_host)
    *p++ = ']';
  if (url->port != scheme_port)
    {
      *p++ = ':';
      p = number_to_string (p, url->port);
    }

  full_path_write (url, p);
  p += fplen;
  *p++ = '\0';

  assert (p - result == size);

  if (quoted_user && quoted_user != url->user)
    xfree (quoted_user);
  if (quoted_passwd && auth_mode == URL_AUTH_SHOW
      && quoted_passwd != url->passwd)
    xfree (quoted_passwd);
  if (quoted_host != url->host)
    xfree (quoted_host);

  return result;
}

/* Return true if scheme a is similar to scheme b.
 
   Schemes are similar if they are equal.  If SSL is supported, schemes
   are also similar if one is http (SCHEME_HTTP) and the other is https
   (SCHEME_HTTPS).  */
bool
schemes_are_similar_p (enum url_scheme a, enum url_scheme b)
{
  if (a == b)
    return true;
#ifdef HAVE_SSL
  if ((a == SCHEME_HTTP && b == SCHEME_HTTPS)
      || (a == SCHEME_HTTPS && b == SCHEME_HTTP))
    return true;
#endif
  return false;
}

static int
getchar_from_escaped_string (const char *str, char *c)
{  
  const char *p = str;

  assert (str && *str);
  assert (c);
  
  if (p[0] == '%')
    {
      if (!ISXDIGIT(p[1]) || !ISXDIGIT(p[2]))
        {
          *c = '%';
          return 1;
        }
      else
        {
          if (p[2] == 0)
            return 0; /* error: invalid string */

          *c = X2DIGITS_TO_NUM (p[1], p[2]);
          if (URL_RESERVED_CHAR(*c))
            {
              *c = '%';
              return 1;
            }
          else
            return 3;
        }
    }
  else
    {
      *c = p[0];
    }

  return 1;
}

bool
are_urls_equal (const char *u1, const char *u2)
{
  const char *p, *q;
  int pp, qq;
  char ch1, ch2;
  assert(u1 && u2);

  p = u1;
  q = u2;

  while (*p && *q
         && (pp = getchar_from_escaped_string (p, &ch1))
         && (qq = getchar_from_escaped_string (q, &ch2))
         && (TOLOWER(ch1) == TOLOWER(ch2)))
    {
      p += pp;
      q += qq;
    }
  
  return (*p == 0 && *q == 0 ? true : false);
}

#if 0
/* Debugging and testing support for path_simplify. */

/* Debug: run path_simplify on PATH and return the result in a new
   string.  Useful for calling from the debugger.  */
static char *
ps (char *path)
{
  char *copy = xstrdup (path);
  path_simplify (copy);
  return copy;
}

static void
run_test (char *test, char *expected_result, bool expected_change)
{
  char *test_copy = xstrdup (test);
  bool modified = path_simplify (test_copy);

  if (0 != strcmp (test_copy, expected_result))
    {
      printf ("Failed path_simplify(\"%s\"): expected \"%s\", got \"%s\".\n",
              test, expected_result, test_copy);
    }
  if (modified != expected_change)
    {
      if (expected_change)
        printf ("Expected modification with path_simplify(\"%s\").\n",
                test);
      else
        printf ("Expected no modification with path_simplify(\"%s\").\n",
                test);
    }
  xfree (test_copy);
}

static void
test_path_simplify (void)
{
  static struct {
    char *test, *result;
    bool should_modify;
  } tests[] = {
    { "",                       "",             false },
    { ".",                      "",             true },
    { "./",                     "",             true },
    { "..",                     "",             true },
    { "../",                    "",             true },
    { "foo",                    "foo",          false },
    { "foo/bar",                "foo/bar",      false },
    { "foo///bar",              "foo///bar",    false },
    { "foo/.",                  "foo/",         true },
    { "foo/./",                 "foo/",         true },
    { "foo./",                  "foo./",        false },
    { "foo/../bar",             "bar",          true },
    { "foo/../bar/",            "bar/",         true },
    { "foo/bar/..",             "foo/",         true },
    { "foo/bar/../x",           "foo/x",        true },
    { "foo/bar/../x/",          "foo/x/",       true },
    { "foo/..",                 "",             true },
    { "foo/../..",              "",             true },
    { "foo/../../..",           "",             true },
    { "foo/../../bar/../../baz", "baz",         true },
    { "a/b/../../c",            "c",            true },
    { "./a/../b",               "b",            true }
  };
  int i;

  for (i = 0; i < countof (tests); i++)
    {
      char *test = tests[i].test;
      char *expected_result = tests[i].result;
      bool  expected_change = tests[i].should_modify;
      run_test (test, expected_result, expected_change);
    }
}
#endif

#ifdef TESTING

const char *
test_append_uri_pathel()
{
  int i;
  struct {
    char *original_url;
    char *input;
    bool escaped;
    char *expected_result;
  } test_array[] = {
    { "http://www.yoyodyne.com/path/", "somepage.html", false, "http://www.yoyodyne.com/path/somepage.html" },
  };
  
  for (i = 0; i < sizeof(test_array)/sizeof(test_array[0]); ++i) 
    {
      struct growable dest;
      const char *p = test_array[i].input;
      
      memset (&dest, 0, sizeof (dest));
      
      append_string (test_array[i].original_url, &dest);
      append_uri_pathel (p, p + strlen(p), test_array[i].escaped, &dest);
      append_char ('\0', &dest);

      mu_assert ("test_append_uri_pathel: wrong result", 
                 strcmp (dest.base, test_array[i].expected_result) == 0);
    }

  return NULL;
}

const char*
test_are_urls_equal()
{
  int i;
  struct {
    char *url1;
    char *url2;
    bool expected_result;
  } test_array[] = {
    { "http://www.adomain.com/apath/", "http://www.adomain.com/apath/",       true },
    { "http://www.adomain.com/apath/", "http://www.adomain.com/anotherpath/", false },
    { "http://www.adomain.com/apath/", "http://www.anotherdomain.com/path/",  false },
    { "http://www.adomain.com/~path/", "http://www.adomain.com/%7epath/",     true },
    { "http://www.adomain.com/longer-path/", "http://www.adomain.com/path/",  false },
    { "http://www.adomain.com/path%2f", "http://www.adomain.com/path/",       false },
  };
  
  for (i = 0; i < sizeof(test_array)/sizeof(test_array[0]); ++i) 
    {
      mu_assert ("test_are_urls_equal: wrong result", 
                 are_urls_equal (test_array[i].url1, test_array[i].url2) == test_array[i].expected_result);
    }

  return NULL;
}

#endif /* TESTING */

/*
 * vim: et ts=2 sw=2
 */