convert.c

Wget很好的处理了http和ftp的下载,很值得学习的经典代码

    {
      hash_table_remove (dl_url_file_map, mapping_url);
      xfree (mapping_url);
      xfree (mapping_file);
    }

  /* Continue mapping. */
  return 0;
}

/* Remove all associations from various URLs to FILE from dl_url_file_map. */

static void
dissociate_urls_from_file (const char *file)
{
  /* Can't use hash_table_iter_* because the table mutates while mapping.  */
  hash_table_for_each (dl_url_file_map, dissociate_urls_from_file_mapper,
                       (char *) file);
}

/* Register that URL has been successfully downloaded to FILE.  This
   is used by the link conversion code to convert references to URLs
   to references to local files.  It is also being used to check if a
   URL has already been downloaded.  */

void
register_download (const char *url, const char *file)
{
  char *old_file, *old_url;

  ENSURE_TABLES_EXIST;

  /* With some forms of retrieval, it is possible, although not likely
     or particularly desirable.  If both are downloaded, the second
     download will override the first one.  When that happens,
     dissociate the old file name from the URL.  */

  if (hash_table_get_pair (dl_file_url_map, file, &old_file, &old_url))
    {
      if (0 == strcmp (url, old_url))
        /* We have somehow managed to download the same URL twice.
           Nothing to do.  */
        return;

      if (match_except_index (url, old_url)
          && !hash_table_contains (dl_url_file_map, url))
        /* The two URLs differ only in the "index.html" ending.  For
           example, one is "http://www.server.com/", and the other is
           "http://www.server.com/index.html".  Don't remove the old
           one, just add the new one as a non-canonical entry.  */
        goto url_only;

      hash_table_remove (dl_file_url_map, file);
      xfree (old_file);
      xfree (old_url);

      /* Remove all the URLs that point to this file.  Yes, there can
         be more than one such URL, because we store redirections as
         multiple entries in dl_url_file_map.  For example, if URL1
         redirects to URL2 which gets downloaded to FILE, we map both
         URL1 and URL2 to FILE in dl_url_file_map.  (dl_file_url_map
         only points to URL2.)  When another URL gets loaded to FILE,
         we want both URL1 and URL2 dissociated from it.

         This is a relatively expensive operation because it performs
         a linear search of the whole hash table, but it should be
         called very rarely, only when two URLs resolve to the same
         file name, *and* the "<file>.1" extensions are turned off.
         In other words, almost never.  */
      dissociate_urls_from_file (file);
    }

  hash_table_put (dl_file_url_map, xstrdup (file), xstrdup (url));

 url_only:
  /* A URL->FILE mapping is not possible without a FILE->URL mapping.
     If the latter were present, it should have been removed by the
     above `if'.  So we could write:

         assert (!hash_table_contains (dl_url_file_map, url));

     The above is correct when running in recursive mode where the
     same URL always resolves to the same file.  But if you do
     something like:

         wget URL URL

     then the first URL will resolve to "FILE", and the other to
     "FILE.1".  In that case, FILE.1 will not be found in
     dl_file_url_map, but URL will still point to FILE in
     dl_url_file_map.  */
  if (hash_table_get_pair (dl_url_file_map, url, &old_url, &old_file))
    {
      hash_table_remove (dl_url_file_map, url);
      xfree (old_url);
      xfree (old_file);
    }

  hash_table_put (dl_url_file_map, xstrdup (url), xstrdup (file));
}

/* Register that FROM has been redirected to TO.  This assumes that TO
   is successfully downloaded and already registered using
   register_download() above.  */

void
register_redirection (const char *from, const char *to)
{
  char *file;

  ENSURE_TABLES_EXIST;

  file = hash_table_get (dl_url_file_map, to);
  assert (file != NULL);
  if (!hash_table_contains (dl_url_file_map, from))
    hash_table_put (dl_url_file_map, xstrdup (from), xstrdup (file));
}

/* Register that the file has been deleted. */

void
register_delete_file (const char *file)
{
  char *old_url, *old_file;

  ENSURE_TABLES_EXIST;

  if (!hash_table_get_pair (dl_file_url_map, file, &old_file, &old_url))
    return;

  hash_table_remove (dl_file_url_map, file);
  xfree (old_file);
  xfree (old_url);
  dissociate_urls_from_file (file);
}

/* Register that FILE is an HTML file that has been downloaded. */

void
register_html (const char *url, const char *file)
{
  if (!downloaded_html_set)
    downloaded_html_set = make_string_hash_table (0);
  string_set_add (downloaded_html_set, file);
}

static void downloaded_files_free (void);

/* Cleanup the data structures associated with this file.  */

void
convert_cleanup (void)
{
  if (dl_file_url_map)
    {
      free_keys_and_values (dl_file_url_map);
      hash_table_destroy (dl_file_url_map);
      dl_file_url_map = NULL;
    }
  if (dl_url_file_map)
    {
      free_keys_and_values (dl_url_file_map);
      hash_table_destroy (dl_url_file_map);
      dl_url_file_map = NULL;
    }
  if (downloaded_html_set)
    string_set_free (downloaded_html_set);
  downloaded_files_free ();
  if (converted_files)
    string_set_free (converted_files);
}

/* Book-keeping code for downloaded files that enables extension
   hacks.  */

/* This table should really be merged with dl_file_url_map and
   downloaded_html_files.  This was originally a list, but I changed
   it to a hash table beause it was actually taking a lot of time to
   find things in it.  */

static struct hash_table *downloaded_files_hash;

/* We're storing "modes" of type downloaded_file_t in the hash table.
   However, our hash tables only accept pointers for keys and values.
   So when we need a pointer, we use the address of a
   downloaded_file_t variable of static storage.  */
   
static downloaded_file_t *
downloaded_mode_to_ptr (downloaded_file_t mode)
{
  static downloaded_file_t
    v1 = FILE_NOT_ALREADY_DOWNLOADED,
    v2 = FILE_DOWNLOADED_NORMALLY,
    v3 = FILE_DOWNLOADED_AND_HTML_EXTENSION_ADDED,
    v4 = CHECK_FOR_FILE;

  switch (mode)
    {
    case FILE_NOT_ALREADY_DOWNLOADED:
      return &v1;
    case FILE_DOWNLOADED_NORMALLY:
      return &v2;
    case FILE_DOWNLOADED_AND_HTML_EXTENSION_ADDED:
      return &v3;
    case CHECK_FOR_FILE:
      return &v4;
    }
  return NULL;
}

/* Remembers which files have been downloaded.  In the standard case,
   should be called with mode == FILE_DOWNLOADED_NORMALLY for each
   file we actually download successfully (i.e. not for ones we have
   failures on or that we skip due to -N).

   When we've downloaded a file and tacked on a ".html" extension due
   to -E, call this function with
   FILE_DOWNLOADED_AND_HTML_EXTENSION_ADDED rather than
   FILE_DOWNLOADED_NORMALLY.

   If you just want to check if a file has been previously added
   without adding it, call with mode == CHECK_FOR_FILE.  Please be
   sure to call this function with local filenames, not remote
   URLs.  */

downloaded_file_t
downloaded_file (downloaded_file_t mode, const char *file)
{
  downloaded_file_t *ptr;

  if (mode == CHECK_FOR_FILE)
    {
      if (!downloaded_files_hash)
        return FILE_NOT_ALREADY_DOWNLOADED;
      ptr = hash_table_get (downloaded_files_hash, file);
      if (!ptr)
        return FILE_NOT_ALREADY_DOWNLOADED;
      return *ptr;
    }

  if (!downloaded_files_hash)
    downloaded_files_hash = make_string_hash_table (0);

  ptr = hash_table_get (downloaded_files_hash, file);
  if (ptr)
    return *ptr;

  ptr = downloaded_mode_to_ptr (mode);
  hash_table_put (downloaded_files_hash, xstrdup (file), ptr);

  return FILE_NOT_ALREADY_DOWNLOADED;
}

static void
downloaded_files_free (void)
{
  if (downloaded_files_hash)
    {
      hash_table_iterator iter;
      for (hash_table_iterate (downloaded_files_hash, &iter);
           hash_table_iter_next (&iter);
           )
        xfree (iter.key);
      hash_table_destroy (downloaded_files_hash);
      downloaded_files_hash = NULL;
    }
}

/* The function returns the pointer to the malloc-ed quoted version of
   string s.  It will recognize and quote numeric and special graphic
   entities, as per RFC1866:

   `&' -> `&amp;'
   `<' -> `&lt;'
   `>' -> `&gt;'
   `"' -> `&quot;'
   SP  -> `&#32;'

   No other entities are recognized or replaced.  */
char *
html_quote_string (const char *s)
{
  const char *b = s;
  char *p, *res;
  int i;

  /* Pass through the string, and count the new size.  */
  for (i = 0; *s; s++, i++)
    {
      if (*s == '&')
        i += 4;                 /* `amp;' */
      else if (*s == '<' || *s == '>')
        i += 3;                 /* `lt;' and `gt;' */
      else if (*s == '\"')
        i += 5;                 /* `quot;' */
      else if (*s == ' ')
        i += 4;                 /* #32; */
    }
  res = xmalloc (i + 1);
  s = b;
  for (p = res; *s; s++)
    {
      switch (*s)
        {
        case '&':
          *p++ = '&';
          *p++ = 'a';
          *p++ = 'm';
          *p++ = 'p';
          *p++ = ';';
          break;
        case '<': case '>':
          *p++ = '&';
          *p++ = (*s == '<' ? 'l' : 'g');
          *p++ = 't';
          *p++ = ';';
          break;
        case '\"':
          *p++ = '&';
          *p++ = 'q';
          *p++ = 'u';
          *p++ = 'o';
          *p++ = 't';
          *p++ = ';';
          break;
        case ' ':
          *p++ = '&';
          *p++ = '#';
          *p++ = '3';
          *p++ = '2';
          *p++ = ';';
          break;
        default:
          *p++ = *s;
        }
    }
  *p = '\0';
  return res;
}

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