/* File retrieval.
- Copyright (C) 1995, 1996, 1997, 1998, 2000 Free Software Foundation, Inc.
+ Copyright (C) 1995, 1996, 1997, 1998, 2000, 2001 Free Software Foundation, Inc.
-This file is part of Wget.
+This file is part of GNU Wget.
-This program is free software; you can redistribute it and/or modify
+GNU Wget is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
-This program is distributed in the hope that it will be useful,
+GNU Wget is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
-along with this program; if not, write to the Free Software
+along with Wget; if not, write to the Free Software
Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */
#include <config.h>
#else
# include <strings.h>
#endif /* HAVE_STRING_H */
-#include <ctype.h>
#include <assert.h>
#include "wget.h"
#include "utils.h"
#include "retr.h"
+#include "progress.h"
#include "url.h"
#include "recur.h"
#include "ftp.h"
#include "connect.h"
#include "hash.h"
-#ifdef WINDOWS
-LARGE_INTEGER internal_time;
-#else
-/* Internal variables used by the timer. */
-static long internal_secs, internal_msecs;
+#ifndef errno
+extern int errno;
#endif
-void logflush PARAMS ((void));
+/* See the comment in gethttp() why this is needed. */
+int global_download_count;
-/* From http.c. */
-uerr_t http_loop PARAMS ((struct urlinfo *, char **, int *));
\f
-/* Flags for show_progress(). */
-enum spflags { SP_NONE, SP_INIT, SP_FINISH };
-
-static int show_progress PARAMS ((long, long, enum spflags));
-
#define MIN(i, j) ((i) <= (j) ? (i) : (j))
/* Reads the contents of file descriptor FD, until it is closed, or a
{
int res = 0;
static char c[8192];
+ void *progress = NULL;
*len = restval;
if (opt.verbose)
- show_progress (restval, expected, SP_INIT);
+ progress = progress_create (restval, expected);
+
if (rbuf && RBUF_FD (rbuf) == fd)
{
+ int need_flush = 0;
while ((res = rbuf_flush (rbuf, c, sizeof (c))) != 0)
{
if (fwrite (c, sizeof (char), res, fp) < res)
return -2;
if (opt.verbose)
- {
- if (show_progress (res, expected, SP_NONE))
- fflush (fp);
- }
+ progress_update (progress, res);
*len += res;
+ need_flush = 1;
}
+ if (need_flush)
+ fflush (fp);
+ if (ferror (fp))
+ return -2;
}
/* Read from fd while there is available data.
int amount_to_read = (use_expected
? MIN (expected - *len, sizeof (c))
: sizeof (c));
- res = iread (fd, c, amount_to_read);
+#ifdef HAVE_SSL
+ if (rbuf->ssl!=NULL) {
+ res = ssl_iread (rbuf->ssl, c, amount_to_read);
+ } else {
+#endif /* HAVE_SSL */
+ res = iread (fd, c, amount_to_read);
+#ifdef HAVE_SSL
+ }
+#endif /* HAVE_SSL */
if (res > 0)
{
- if (fwrite (c, sizeof (char), res, fp) < res)
+ fwrite (c, sizeof (char), res, fp);
+ /* Always flush the contents of the network packet. This
+ should not be adverse to performance, as the network
+ packets typically won't be too tiny anyway. */
+ fflush (fp);
+ if (ferror (fp))
return -2;
if (opt.verbose)
- {
- if (show_progress (res, expected, SP_NONE))
- fflush (fp);
- }
+ progress_update (progress, res);
*len += res;
}
else
if (res < -1)
res = -1;
if (opt.verbose)
- show_progress (0, expected, SP_FINISH);
+ progress_finish (progress);
return res;
}
-
-static void
-print_percentage (long bytes, long expected)
+\f
+/* Return a printed representation of the download rate, as
+ appropriate for the speed. If PAD is non-zero, strings will be
+ padded to the width of 7 characters (xxxx.xx). */
+char *
+retr_rate (long bytes, long msecs, int pad)
{
- int percentage = (int)(100.0 * bytes / expected);
- logprintf (LOG_VERBOSE, " [%3d%%]", percentage);
-}
+ static char res[20];
+ static char *rate_names[] = {"B/s", "KB/s", "MB/s", "GB/s" };
+ int units = 0;
-/* Show the dotted progress report of file loading. Called with
- length and a flag to tell it whether to reset or not. It keeps the
- offset information in static local variables.
+ double dlrate = calc_rate (bytes, msecs, &units);
+ sprintf (res, pad ? "%7.2f %s" : "%.2f %s", dlrate, rate_names[units]);
- Return value: 1 or 0, designating whether any dots have been drawn.
+ return res;
+}
- If the init argument is set, the routine will initialize.
+/* Calculate the download rate and trim it as appropriate for the
+ speed. Appropriate means that if rate is greater than 1K/s,
+ kilobytes are used, and if rate is greater than 1MB/s, megabytes
+ are used.
- If the res is non-zero, res/line_bytes lines are skipped
- (meaning the appropriate number ok kilobytes), and the number of
- "dots" fitting on the first line are drawn as ','. */
-static int
-show_progress (long res, long expected, enum spflags flags)
+ UNITS is zero for B/s, one for KB/s, two for MB/s, and three for
+ GB/s. */
+double
+calc_rate (long bytes, long msecs, int *units)
{
- static long line_bytes;
- static long offs;
- static int ndot, nrow;
- int any_output = 0;
+ double dlrate;
- if (flags == SP_FINISH)
- {
- if (expected)
- {
- int dot = ndot;
- char *tmpstr = (char *)alloca (2 * opt.dots_in_line + 1);
- char *tmpp = tmpstr;
- for (; dot < opt.dots_in_line; dot++)
- {
- if (!(dot % opt.dot_spacing))
- *tmpp++ = ' ';
- *tmpp++ = ' ';
- }
- *tmpp = '\0';
- logputs (LOG_VERBOSE, tmpstr);
- print_percentage (nrow * line_bytes + ndot * opt.dot_bytes + offs,
- expected);
- }
- logputs (LOG_VERBOSE, "\n\n");
- return 0;
- }
+ assert (msecs >= 0);
+ assert (bytes >= 0);
- /* Temporarily disable flushing. */
- opt.no_flush = 1;
- /* init set means initialization. If res is set, it also means that
- the retrieval is *not* done from the beginning. The part that
- was already retrieved is not shown again. */
- if (flags == SP_INIT)
- {
- /* Generic initialization of static variables. */
- offs = 0L;
- ndot = nrow = 0;
- line_bytes = (long)opt.dots_in_line * opt.dot_bytes;
- if (res)
- {
- if (res >= line_bytes)
- {
- nrow = res / line_bytes;
- res %= line_bytes;
- logprintf (LOG_VERBOSE,
- _("\n [ skipping %dK ]"),
- (int) ((nrow * line_bytes) / 1024));
- ndot = 0;
- }
- }
- logprintf (LOG_VERBOSE, "\n%5ldK ->", nrow * line_bytes / 1024);
- }
- /* Offset gets incremented by current value. */
- offs += res;
- /* While offset is >= opt.dot_bytes, print dots, taking care to
- precede every 50th dot with a status message. */
- for (; offs >= opt.dot_bytes; offs -= opt.dot_bytes)
- {
- if (!(ndot % opt.dot_spacing))
- logputs (LOG_VERBOSE, " ");
- any_output = 1;
- logputs (LOG_VERBOSE, flags == SP_INIT ? "," : ".");
- ++ndot;
- if (ndot == opt.dots_in_line)
- {
- ndot = 0;
- ++nrow;
- if (expected)
- print_percentage (nrow * line_bytes, expected);
- logprintf (LOG_VERBOSE, "\n%5ldK ->", nrow * line_bytes / 1024);
- }
- }
- /* Reenable flushing. */
- opt.no_flush = 0;
- if (any_output)
- /* Force flush. #### Oh, what a kludge! */
- logflush ();
- return any_output;
+ if (msecs == 0)
+ /* If elapsed time is 0, it means we're under the granularity of
+ the timer. This often happens on systems that use time() for
+ the timer. */
+ msecs = wtimer_granularity ();
+
+ dlrate = (double)1000 * bytes / msecs;
+ if (dlrate < 1024.0)
+ *units = 0;
+ else if (dlrate < 1024.0 * 1024.0)
+ *units = 1, dlrate /= 1024.0;
+ else if (dlrate < 1024.0 * 1024.0 * 1024.0)
+ *units = 2, dlrate /= (1024.0 * 1024.0);
+ else
+ /* Maybe someone will need this one day. More realistically, it
+ will get tickled by buggy timers. */
+ *units = 3, dlrate /= (1024.0 * 1024.0 * 1024.0);
+
+ return dlrate;
}
\f
-/* Reset the internal timer. */
-void
-reset_timer (void)
+static int
+register_redirections_mapper (void *key, void *value, void *arg)
{
-#ifndef WINDOWS
- /* Under Unix, the preferred way to measure the passage of time is
- through gettimeofday() because of its granularity. However, on
- some old or weird systems, gettimeofday() might not be available.
- There we use the simple time(). */
-# ifdef HAVE_GETTIMEOFDAY
- struct timeval t;
- gettimeofday (&t, NULL);
- internal_secs = t.tv_sec;
- internal_msecs = t.tv_usec / 1000;
-# else /* not HAVE_GETTIMEOFDAY */
- internal_secs = time (NULL);
- internal_msecs = 0;
-# endif /* not HAVE_GETTIMEOFDAY */
-#else /* WINDOWS */
- /* Under Windows, use Windows-specific APIs. */
- FILETIME ft;
- SYSTEMTIME st;
- GetSystemTime(&st);
- SystemTimeToFileTime(&st,&ft);
- internal_time.HighPart = ft.dwHighDateTime;
- internal_time.LowPart = ft.dwLowDateTime;
-#endif /* WINDOWS */
+ const char *redirected_from = (const char *)key;
+ const char *redirected_to = (const char *)arg;
+ if (0 != strcmp (redirected_from, redirected_to))
+ register_redirection (redirected_from, redirected_to);
+ return 0;
}
-/* Return the time elapsed from the last call to reset_timer(), in
- milliseconds. */
-long
-elapsed_time (void)
-{
-#ifndef WINDOWS
-# ifdef HAVE_GETTIMEOFDAY
- struct timeval t;
- gettimeofday (&t, NULL);
- return ((t.tv_sec - internal_secs) * 1000
- + (t.tv_usec / 1000 - internal_msecs));
-# else /* not HAVE_GETTIMEOFDAY */
- return 1000 * ((long)time (NULL) - internal_secs);
-# endif /* not HAVE_GETTIMEOFDAY */
-#else /* WINDOWS */
- FILETIME ft;
- SYSTEMTIME st;
- LARGE_INTEGER li;
- GetSystemTime(&st);
- SystemTimeToFileTime(&st,&ft);
- li.HighPart = ft.dwHighDateTime;
- li.LowPart = ft.dwLowDateTime;
- return (long) ((li.QuadPart - internal_time.QuadPart) / 1e4);
-#endif /* WINDOWS */
-}
+/* Register the redirections that lead to the successful download of
+ this URL. This is necessary so that the link converter can convert
+ redirected URLs to the local file. */
-/* Print out the appropriate download rate. Appropriate means that if
- rate is > 1024 bytes per second, kilobytes are used, and if rate >
- 1024 * 1024 bps, megabytes are used. */
-char *
-rate (long bytes, long msecs)
+static void
+register_all_redirections (struct hash_table *redirections, const char *final)
{
- static char res[15];
- double dlrate;
-
- if (!msecs)
- ++msecs;
- dlrate = (double)1000 * bytes / msecs;
- /* #### Should these strings be translatable? */
- if (dlrate < 1024.0)
- sprintf (res, "%.2f B/s", dlrate);
- else if (dlrate < 1024.0 * 1024.0)
- sprintf (res, "%.2f KB/s", dlrate / 1024.0);
- else
- sprintf (res, "%.2f MB/s", dlrate / (1024.0 * 1024.0));
- return res;
+ hash_table_map (redirections, register_redirections_mapper, (void *)final);
}
-\f
-#define USE_PROXY_P(u) (opt.use_proxy && getproxy((u)->proto) \
+
+#define USE_PROXY_P(u) (opt.use_proxy && getproxy((u)->scheme) \
&& no_proxy_match((u)->host, \
(const char **)opt.no_proxy))
-/* Retrieve the given URL. Decides which loop to call -- HTTP, FTP,
+/* Retrieve the given URL. Decides which loop to call -- HTTP(S), FTP,
or simply copy it with file:// (#### the latter not yet
implemented!). */
uerr_t
uerr_t result;
char *url;
int location_changed, dummy;
- int local_use_proxy;
+ int use_proxy;
char *mynewloc, *proxy;
- struct urlinfo *u;
+ struct url *u;
+ int up_error_code; /* url parse error code */
+ char *local_file;
struct hash_table *redirections = NULL;
/* If dt is NULL, just ignore it. */
if (file)
*file = NULL;
- u = newurl ();
- /* Parse the URL. */
- result = parseurl (url, u, 0);
- if (result != URLOK)
+ u = url_parse (url, &up_error_code);
+ if (!u)
{
- logprintf (LOG_NOTQUIET, "%s: %s.\n", url, uerrmsg (result));
- freeurl (u, 1);
+ logprintf (LOG_NOTQUIET, "%s: %s.\n", url, url_error (up_error_code));
if (redirections)
string_set_free (redirections);
- free (url);
- return result;
+ xfree (url);
+ return URLERROR;
}
+ if (!refurl)
+ refurl = opt.referer;
+
redirected:
- /* Set the referer. */
- if (refurl)
- u->referer = xstrdup (refurl);
- else
- {
- if (opt.referer)
- u->referer = xstrdup (opt.referer);
- else
- u->referer = NULL;
- }
+ result = NOCONERROR;
+ mynewloc = NULL;
+ local_file = NULL;
- local_use_proxy = USE_PROXY_P (u);
- if (local_use_proxy)
+ use_proxy = USE_PROXY_P (u);
+ if (use_proxy)
{
- struct urlinfo *pu = newurl ();
-
- /* Copy the original URL to new location. */
- memcpy (pu, u, sizeof (*u));
- pu->proxy = NULL; /* A minor correction :) */
- /* Initialize u to nil. */
- memset (u, 0, sizeof (*u));
- u->proxy = pu;
- /* Get the appropriate proxy server, appropriate for the
- current protocol. */
- proxy = getproxy (pu->proto);
+ struct url *proxy_url;
+
+ /* Get the proxy server for the current scheme. */
+ proxy = getproxy (u->scheme);
if (!proxy)
{
logputs (LOG_NOTQUIET, _("Could not find proxy host.\n"));
- freeurl (u, 1);
+ url_free (u);
if (redirections)
string_set_free (redirections);
- free (url);
+ xfree (url);
return PROXERR;
}
+
/* Parse the proxy URL. */
- result = parseurl (proxy, u, 0);
- if (result != URLOK || u->proto != URLHTTP)
+ proxy_url = url_parse (proxy, &up_error_code);
+ if (!proxy_url)
{
- if (u->proto == URLHTTP)
- logprintf (LOG_NOTQUIET, "Proxy %s: %s.\n", proxy, uerrmsg(result));
- else
- logprintf (LOG_NOTQUIET, _("Proxy %s: Must be HTTP.\n"), proxy);
- freeurl (u, 1);
+ logprintf (LOG_NOTQUIET, _("Error parsing proxy URL %s: %s.\n"),
+ proxy, url_error (up_error_code));
if (redirections)
string_set_free (redirections);
- free (url);
+ xfree (url);
+ return PROXERR;
+ }
+ if (proxy_url->scheme != SCHEME_HTTP)
+ {
+ logprintf (LOG_NOTQUIET, _("Error in proxy URL %s: Must be HTTP.\n"), proxy);
+ url_free (proxy_url);
+ if (redirections)
+ string_set_free (redirections);
+ xfree (url);
return PROXERR;
}
- u->proto = URLHTTP;
- }
-
- assert (u->proto != URLFILE); /* #### Implement me! */
- mynewloc = NULL;
- if (u->proto == URLHTTP)
- result = http_loop (u, &mynewloc, dt);
- else if (u->proto == URLFTP)
+ result = http_loop (u, &mynewloc, &local_file, refurl, dt, proxy_url);
+ url_free (proxy_url);
+ }
+ else if (u->scheme == SCHEME_HTTP
+#ifdef HAVE_SSL
+ || u->scheme == SCHEME_HTTPS
+#endif
+ )
+ {
+ result = http_loop (u, &mynewloc, &local_file, refurl, dt, NULL);
+ }
+ else if (u->scheme == SCHEME_FTP)
{
/* If this is a redirection, we must not allow recursive FTP
retrieval, so we save recursion to oldrec, and restore it
opt.recursive = 0;
result = ftp_loop (u, dt);
opt.recursive = oldrec;
+#if 0
/* There is a possibility of having HTTP being redirected to
FTP. In these cases we must decide whether the text is HTML
according to the suffix. The HTML suffixes are `.html' and
- `.htm', case-insensitive.
-
- #### All of this is, of course, crap. These types should be
- determined through mailcap. */
- if (redirections && u->local && (u->proto == URLFTP ))
+ `.htm', case-insensitive. */
+ if (redirections && u->local && (u->scheme == SCHEME_FTP))
{
char *suf = suffix (u->local);
if (suf && (!strcasecmp (suf, "html") || !strcasecmp (suf, "htm")))
*dt |= TEXTHTML;
- FREE_MAYBE (suf);
}
+#endif
}
location_changed = (result == NEWLOCATION);
if (location_changed)
{
char *construced_newloc;
- uerr_t newloc_result;
- struct urlinfo *newloc_struct;
+ struct url *newloc_parsed;
assert (mynewloc != NULL);
+ if (local_file)
+ xfree (local_file);
+
/* The HTTP specs only allow absolute URLs to appear in
redirects, but a ton of boneheaded webservers and CGIs out
there break the rules and use relative URLs, and popular
browsers are lenient about this, so wget should be too. */
- construced_newloc = url_concat (url, mynewloc);
- free (mynewloc);
+ construced_newloc = uri_merge (url, mynewloc);
+ xfree (mynewloc);
mynewloc = construced_newloc;
/* Now, see if this new location makes sense. */
- newloc_struct = newurl ();
- newloc_result = parseurl (mynewloc, newloc_struct, 1);
- if (newloc_result != URLOK)
+ newloc_parsed = url_parse (mynewloc, &up_error_code);
+ if (!newloc_parsed)
{
- logprintf (LOG_NOTQUIET, "%s: %s.\n", mynewloc, uerrmsg (newloc_result));
- freeurl (newloc_struct, 1);
- freeurl (u, 1);
+ logprintf (LOG_NOTQUIET, "%s: %s.\n", mynewloc,
+ url_error (up_error_code));
+ url_free (u);
if (redirections)
string_set_free (redirections);
- free (url);
- free (mynewloc);
+ xfree (url);
+ xfree (mynewloc);
return result;
}
- /* Now mynewloc will become newloc_struct->url, because if the
+ /* Now mynewloc will become newloc_parsed->url, because if the
Location contained relative paths like .././something, we
don't want that propagating as url. */
- free (mynewloc);
- mynewloc = xstrdup (newloc_struct->url);
+ xfree (mynewloc);
+ mynewloc = xstrdup (newloc_parsed->url);
if (!redirections)
{
string_set_add (redirections, u->url);
}
- /* The new location is OK. Let's check for redirection cycle by
+ /* The new location is OK. Check for redirection cycle by
peeking through the history of redirections. */
- if (string_set_exists (redirections, newloc_struct->url))
+ if (string_set_contains (redirections, newloc_parsed->url))
{
logprintf (LOG_NOTQUIET, _("%s: Redirection cycle detected.\n"),
mynewloc);
- freeurl (newloc_struct, 1);
- freeurl (u, 1);
+ url_free (newloc_parsed);
+ url_free (u);
if (redirections)
string_set_free (redirections);
- free (url);
- free (mynewloc);
+ xfree (url);
+ xfree (mynewloc);
return WRONGCODE;
}
- string_set_add (redirections, newloc_struct->url);
+ string_set_add (redirections, newloc_parsed->url);
- free (url);
+ xfree (url);
url = mynewloc;
- freeurl (u, 1);
- u = newloc_struct;
+ url_free (u);
+ u = newloc_parsed;
goto redirected;
}
- if (file)
+ if (local_file)
{
- if (u->local)
- *file = xstrdup (u->local);
- else
- *file = NULL;
+ if (*dt & RETROKF)
+ {
+ register_download (url, local_file);
+ if (redirections)
+ register_all_redirections (redirections, url);
+ if (*dt & TEXTHTML)
+ register_html (url, local_file);
+ }
}
- freeurl (u, 1);
+
+ if (file)
+ *file = local_file ? local_file : NULL;
+ else
+ FREE_MAYBE (local_file);
+
+ url_free (u);
if (redirections)
string_set_free (redirections);
if (newloc)
*newloc = url;
else
- free (url);
+ xfree (url);
+
+ ++global_download_count;
return result;
}
retrieve_from_file (const char *file, int html, int *count)
{
uerr_t status;
- urlpos *url_list, *cur_url;
+ struct urlpos *url_list, *cur_url;
url_list = (html ? get_urls_html (file, NULL, FALSE, NULL)
: get_urls_file (file));
status = RETROK; /* Suppose everything is OK. */
*count = 0; /* Reset the URL count. */
- recursive_reset ();
+
for (cur_url = url_list; cur_url; cur_url = cur_url->next, ++*count)
{
- char *filename, *new_file;
+ char *filename = NULL, *new_file;
int dt;
+ if (cur_url->ignore_when_downloading)
+ continue;
+
if (downloaded_exceeds_quota ())
{
status = QUOTEXC;
break;
}
- status = retrieve_url (cur_url->url, &filename, &new_file, NULL, &dt);
- if (opt.recursive && status == RETROK && (dt & TEXTHTML))
- status = recursive_retrieve (filename, new_file ? new_file
- : cur_url->url);
+ if (opt.recursive && cur_url->url->scheme != SCHEME_FTP)
+ status = retrieve_tree (cur_url->url->url);
+ else
+ status = retrieve_url (cur_url->url->url, &filename, &new_file, NULL, &dt);
if (filename && opt.delete_after && file_exists_p (filename))
{
return opt.downloaded > opt.quota;
}
+
+/* If opt.wait or opt.waitretry are specified, and if certain
+ conditions are met, sleep the appropriate number of seconds. See
+ the documentation of --wait and --waitretry for more information.
+
+ COUNT is the count of current retrieval, beginning with 1. */
+
+void
+sleep_between_retrievals (int count)
+{
+ static int first_retrieval = 1;
+
+ if (!first_retrieval && (opt.wait || opt.waitretry))
+ {
+ if (opt.waitretry && count > 1)
+ {
+ /* If opt.waitretry is specified and this is a retry, wait
+ for COUNT-1 number of seconds, or for opt.waitretry
+ seconds. */
+ if (count <= opt.waitretry)
+ sleep (count - 1);
+ else
+ sleep (opt.waitretry);
+ }
+ else if (opt.wait)
+ {
+ /* Otherwise, check if opt.wait is specified. If so, sleep. */
+ if (count > 1 || !opt.random_wait)
+ sleep (opt.wait);
+ else
+ {
+ int waitsecs = random() % (opt.wait * 2 + 1);
+ DEBUGP(("sleep_between_retrievals: norm=%ld,random=%ld,sleep=%d\n",
+ opt.wait, waitsecs - opt.wait, waitsecs));
+ sleep(waitsecs);
+ }
+ }
+ }
+ if (first_retrieval)
+ first_retrieval = 0;
+}