1
0
mirror of https://github.com/moparisthebest/pacman synced 2024-11-10 03:25:01 -05:00
pacman/lib/libalpm/dload.c

404 lines
11 KiB
C
Raw Normal View History

/*
* download.c
*
* Copyright (c) 2006-2011 Pacman Development Team <pacman-dev@archlinux.org>
* Copyright (c) 2002-2006 by Judd Vinet <jvinet@zeroflux.org>
*
* This program 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,
* 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, see <http://www.gnu.org/licenses/>.
*/
#include "config.h"
#include <stdlib.h>
#include <stdio.h>
#include <errno.h>
#include <string.h>
#include <unistd.h>
download: major refactor to address lingering issues Sorry for this being such a huge patch, but I believe it is necessary for quite a few reasons which I will attempt to explain herein. I've been mulling this over for a while, but wasn't super happy with making the download interface more complex. Instead, if we carefully order things in the internal download code, we can actually make the interface simpler. 1. FS#15657 - This involves `name.db.tar.gz.part` files being left around the filesystem, and then causing all sorts of issues when someone attempts to rerun the operation they canceled. We need to ensure that if we resume a download, we are resuming it on exactly the same file; if we cannot be almost postive of that then we need to start over. 2. http://www.mail-archive.com/pacman-dev@archlinux.org/msg03536.html - Here we have a lighttpd bug to ruin the day. If we send both a Range: header and If-Modified-Since: header across the wire in a GET request, lighttpd doesn't do what we want in several cases. If the file hadn't been modified, it returns a '304 Not Modified' instead of a '206 Partial Content'. We need to do a stat (e.g. HEAD in HTTP terms) operation here, and the proceed accordingly based off the values we get back from it. 3. The mtime stuff was rather ugly, and relied on the called function to write back to a passed in reference, which isn't the greatest. Instead, use the power of the filesystem to contain this info. Every file downloaded internally is now carefully timestamped with the remote file time. This should allow the resume logic to work. In order to guarantee this, we need to implement a signal handler that catches interrupts, notifies the running code, and causes it to set the mtimes on the file. It then rethrows the signal so the pacman signal handler (or any frontend) works as expected. 4. We did a lot of funky stuff in trying to track the DB last modified time. It is a lot easier to just keep the downloaded DB file around and track the time on that rather than in a funky dot file. It also kills a lot of code. 5. For GPG verification of the databases down the road, we are going to need the DB file around for at least a short bit of time anyway, so this gets us closer to that. Signed-off-by: Dan McGee <dan@archlinux.org> [Xav: fixed printf with off_t] Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
2009-11-12 00:39:26 -05:00
#include <sys/time.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <signal.h>
#ifdef HAVE_LIBCURL
#include <curl/curl.h>
#endif
/* libalpm */
#include "dload.h"
#include "alpm_list.h"
#include "alpm.h"
#include "log.h"
#include "util.h"
#include "handle.h"
static int prevprogress; /* last download amount */
static char *get_filename(const char *url) {
char *filename = strrchr(url, '/');
if(filename != NULL) {
filename++;
}
return filename;
}
#ifdef HAVE_LIBCURL
static char *get_destfile(const char *path, const char *filename)
{
char *destfile;
/* len = localpath len + filename len + null */
size_t len = strlen(path) + strlen(filename) + 1;
CALLOC(destfile, len, sizeof(char), RET_ERR(PM_ERR_MEMORY, NULL));
snprintf(destfile, len, "%s%s", path, filename);
return destfile;
}
static char *get_tempfile(const char *path, const char *filename)
{
char *tempfile;
/* len = localpath len + filename len + '.part' len + null */
size_t len = strlen(path) + strlen(filename) + 6;
CALLOC(tempfile, len, sizeof(char), RET_ERR(PM_ERR_MEMORY, NULL));
snprintf(tempfile, len, "%s%s.part", path, filename);
return tempfile;
}
#define check_stop() if(dload_interrupted) { ret = -1; goto cleanup; }
enum sighandlers { OLD = 0, NEW = 1 };
int dload_interrupted;
static void inthandler(int signum)
{
dload_interrupted = 1;
}
download: major refactor to address lingering issues Sorry for this being such a huge patch, but I believe it is necessary for quite a few reasons which I will attempt to explain herein. I've been mulling this over for a while, but wasn't super happy with making the download interface more complex. Instead, if we carefully order things in the internal download code, we can actually make the interface simpler. 1. FS#15657 - This involves `name.db.tar.gz.part` files being left around the filesystem, and then causing all sorts of issues when someone attempts to rerun the operation they canceled. We need to ensure that if we resume a download, we are resuming it on exactly the same file; if we cannot be almost postive of that then we need to start over. 2. http://www.mail-archive.com/pacman-dev@archlinux.org/msg03536.html - Here we have a lighttpd bug to ruin the day. If we send both a Range: header and If-Modified-Since: header across the wire in a GET request, lighttpd doesn't do what we want in several cases. If the file hadn't been modified, it returns a '304 Not Modified' instead of a '206 Partial Content'. We need to do a stat (e.g. HEAD in HTTP terms) operation here, and the proceed accordingly based off the values we get back from it. 3. The mtime stuff was rather ugly, and relied on the called function to write back to a passed in reference, which isn't the greatest. Instead, use the power of the filesystem to contain this info. Every file downloaded internally is now carefully timestamped with the remote file time. This should allow the resume logic to work. In order to guarantee this, we need to implement a signal handler that catches interrupts, notifies the running code, and causes it to set the mtimes on the file. It then rethrows the signal so the pacman signal handler (or any frontend) works as expected. 4. We did a lot of funky stuff in trying to track the DB last modified time. It is a lot easier to just keep the downloaded DB file around and track the time on that rather than in a funky dot file. It also kills a lot of code. 5. For GPG verification of the databases down the road, we are going to need the DB file around for at least a short bit of time anyway, so this gets us closer to that. Signed-off-by: Dan McGee <dan@archlinux.org> [Xav: fixed printf with off_t] Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
2009-11-12 00:39:26 -05:00
static int curl_progress(void *filename, double dltotal, double dlnow,
double ultotal, double ulnow)
{
/* unused parameters */
(void)ultotal;
(void)ulnow;
if(dltotal == 0 || prevprogress == dltotal) {
return 0;
}
if(dload_interrupted) {
return 1;
}
if(handle->dlcb) {
handle->dlcb((const char*)filename, (long)dlnow, (long)dltotal);
}
prevprogress = dlnow;
return 0;
}
static int curl_gethost(const char *url, char *buffer)
{
int hostlen;
char *p;
if(strncmp(url, "file://", 7) == 0) {
strcpy(buffer, _("disk"));
} else {
p = strstr(url, "//");
if(!p) {
return 1;
}
p += 2; /* jump over the found // */
hostlen = strcspn(p, "/");
if(hostlen > 255) {
/* buffer overflow imminent */
_alpm_log(PM_LOG_ERROR, _("buffer overflow detected"));
return 1;
}
snprintf(buffer, hostlen + 1, "%s", p);
}
return 0;
}
static int curl_download_internal(const char *url, const char *localpath,
int force)
{
int ret = -1;
FILE *localf = NULL;
char *destfile, *filename, *tempfile;
char hostname[256]; /* RFC1123 states applications should support this length */
struct stat st;
long httpresp, timecond, remote_time, local_time;
double remote_size, bytes_dl;
struct sigaction sig_pipe[2], sig_int[2];
filename = get_filename(url);
if(!filename || curl_gethost(url, hostname) != 0) {
_alpm_log(PM_LOG_ERROR, _("url '%s' is invalid\n"), url);
RET_ERR(PM_ERR_SERVER_BAD_URL, -1);
}
destfile = get_destfile(localpath, filename);
tempfile = get_tempfile(localpath, filename);
/* the curl_easy handle is initialized with the alpm handle, so we only need
* to reset the curl handle set parameters for each time it's used. */
curl_easy_reset(handle->curl);
curl_easy_setopt(handle->curl, CURLOPT_URL, url);
curl_easy_setopt(handle->curl, CURLOPT_FAILONERROR, 1L);
curl_easy_setopt(handle->curl, CURLOPT_ENCODING, "deflate, gzip");
curl_easy_setopt(handle->curl, CURLOPT_CONNECTTIMEOUT, 10L);
curl_easy_setopt(handle->curl, CURLOPT_FILETIME, 1L);
curl_easy_setopt(handle->curl, CURLOPT_NOPROGRESS, 0L);
curl_easy_setopt(handle->curl, CURLOPT_FOLLOWLOCATION, 1L);
curl_easy_setopt(handle->curl, CURLOPT_PROGRESSFUNCTION, curl_progress);
curl_easy_setopt(handle->curl, CURLOPT_PROGRESSDATA, filename);
if(!force && stat(destfile, &st) == 0) {
/* assume its a sync, so we're starting from scratch. but, only download
* our local is out of date. */
local_time = (long)st.st_mtime;
curl_easy_setopt(handle->curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_IFMODSINCE);
curl_easy_setopt(handle->curl, CURLOPT_TIMEVALUE, local_time);
} else if(stat(tempfile, &st) == 0 && st.st_size > 0) {
/* assume its a partial package download. we do not support resuming of
* transfers on partially downloaded sync DBs. */
localf = fopen(tempfile, "ab");
curl_easy_setopt(handle->curl, CURLOPT_RESUME_FROM, (long)st.st_size);
_alpm_log(PM_LOG_DEBUG, "tempfile found, attempting continuation");
}
/* no destfile and no tempfile. start from scratch */
if(localf == NULL) {
localf = fopen(tempfile, "wb");
if(localf == NULL) {
goto cleanup;
}
}
/* this has to be set _after_ figuring out which file we're opening */
curl_easy_setopt(handle->curl, CURLOPT_WRITEDATA, localf);
/* print proxy info for debug purposes */
_alpm_log(PM_LOG_DEBUG, "HTTP_PROXY: %s\n", getenv("HTTP_PROXY"));
_alpm_log(PM_LOG_DEBUG, "http_proxy: %s\n", getenv("http_proxy"));
_alpm_log(PM_LOG_DEBUG, "FTP_PROXY: %s\n", getenv("FTP_PROXY"));
_alpm_log(PM_LOG_DEBUG, "ftp_proxy: %s\n", getenv("ftp_proxy"));
/* ignore any SIGPIPE signals- these may occur if our FTP socket dies or
* something along those lines. Store the old signal handler first. */
sig_pipe[NEW].sa_handler = SIG_IGN;
sigemptyset(&sig_pipe[NEW].sa_mask);
sig_pipe[NEW].sa_flags = 0;
sigaction(SIGPIPE, NULL, &sig_pipe[OLD]);
sigaction(SIGPIPE, &sig_pipe[NEW], NULL);
dload_interrupted = 0;
sig_int[NEW].sa_handler = &inthandler;
sigemptyset(&sig_int[NEW].sa_mask);
sig_int[NEW].sa_flags = 0;
sigaction(SIGINT, NULL, &sig_int[OLD]);
sigaction(SIGINT, &sig_int[NEW], NULL);
/* set initial value of prevprogress to -1 which causes curl_progress() to
* initialize the progress bar with 0% once. */
prevprogress = -1;
/* perform transfer */
handle->curlerr = curl_easy_perform(handle->curl);
/* retrieve info about the state of the transfer */
curl_easy_getinfo(handle->curl, CURLINFO_HTTP_CODE, &httpresp);
curl_easy_getinfo(handle->curl, CURLINFO_FILETIME, &remote_time);
curl_easy_getinfo(handle->curl, CURLINFO_CONTENT_LENGTH_DOWNLOAD, &remote_size);
curl_easy_getinfo(handle->curl, CURLINFO_SIZE_DOWNLOAD, &bytes_dl);
curl_easy_getinfo(handle->curl, CURLINFO_CONDITION_UNMET, &timecond);
/* time condition was met and we didn't download anything. we need to
* clean up the 0 byte .part file that's left behind. */
if(bytes_dl == 0 && timecond == 1) {
ret = 1;
unlink(tempfile);
goto cleanup;
}
if(handle->curlerr == CURLE_ABORTED_BY_CALLBACK) {
goto cleanup;
} else if(handle->curlerr != CURLE_OK) {
pm_errno = PM_ERR_LIBCURL;
_alpm_log(PM_LOG_ERROR, _("failed retrieving file '%s' from %s : %s\n"),
filename, hostname, curl_easy_strerror(handle->curlerr));
unlink(tempfile);
goto cleanup;
}
/* remote_size isn't necessarily the full size of the file, just what the
* server reported as remaining to download. compare it to what curl reported
* as actually being transferred during curl_easy_perform() */
if((remote_size != -1 && bytes_dl != -1) && bytes_dl != remote_size) {
pm_errno = PM_ERR_RETRIEVE;
_alpm_log(PM_LOG_ERROR, _("%s appears to be truncated: %jd/%jd bytes\n"),
filename, (intmax_t)bytes_dl, (intmax_t)remote_size);
goto cleanup;
}
fclose(localf);
localf = NULL;
/* set the times on the file to the same as that of the remote file */
if(remote_time != -1) {
struct timeval tv[2];
memset(&tv, 0, sizeof(tv));
tv[0].tv_sec = tv[1].tv_sec = remote_time;
utimes(tempfile, tv);
}
rename(tempfile, destfile);
ret = 0;
cleanup:
FREE(tempfile);
FREE(destfile);
if(localf != NULL) {
/* if we still had a local file open, we got interrupted. set the mtimes on
* the file accordingly. */
fflush(localf);
if(remote_time != -1) {
struct timeval tv[2];
memset(&tv, 0, sizeof(tv));
tv[0].tv_sec = tv[1].tv_sec = remote_time;
futimes(fileno(localf), tv);
}
fclose(localf);
}
/* restore the old signal handlers */
sigaction(SIGINT, &sig_int[OLD], NULL);
sigaction(SIGPIPE, &sig_pipe[OLD], NULL);
/* if we were interrupted, trip the old handler */
if(dload_interrupted) {
raise(SIGINT);
}
return ret;
}
#endif
static int download(const char *url, const char *localpath,
int force)
{
if(handle->fetchcb == NULL) {
#ifdef HAVE_LIBCURL
return curl_download_internal(url, localpath, force);
#else
RET_ERR(PM_ERR_EXTERNAL_DOWNLOAD, -1);
#endif
} else {
download: major refactor to address lingering issues Sorry for this being such a huge patch, but I believe it is necessary for quite a few reasons which I will attempt to explain herein. I've been mulling this over for a while, but wasn't super happy with making the download interface more complex. Instead, if we carefully order things in the internal download code, we can actually make the interface simpler. 1. FS#15657 - This involves `name.db.tar.gz.part` files being left around the filesystem, and then causing all sorts of issues when someone attempts to rerun the operation they canceled. We need to ensure that if we resume a download, we are resuming it on exactly the same file; if we cannot be almost postive of that then we need to start over. 2. http://www.mail-archive.com/pacman-dev@archlinux.org/msg03536.html - Here we have a lighttpd bug to ruin the day. If we send both a Range: header and If-Modified-Since: header across the wire in a GET request, lighttpd doesn't do what we want in several cases. If the file hadn't been modified, it returns a '304 Not Modified' instead of a '206 Partial Content'. We need to do a stat (e.g. HEAD in HTTP terms) operation here, and the proceed accordingly based off the values we get back from it. 3. The mtime stuff was rather ugly, and relied on the called function to write back to a passed in reference, which isn't the greatest. Instead, use the power of the filesystem to contain this info. Every file downloaded internally is now carefully timestamped with the remote file time. This should allow the resume logic to work. In order to guarantee this, we need to implement a signal handler that catches interrupts, notifies the running code, and causes it to set the mtimes on the file. It then rethrows the signal so the pacman signal handler (or any frontend) works as expected. 4. We did a lot of funky stuff in trying to track the DB last modified time. It is a lot easier to just keep the downloaded DB file around and track the time on that rather than in a funky dot file. It also kills a lot of code. 5. For GPG verification of the databases down the road, we are going to need the DB file around for at least a short bit of time anyway, so this gets us closer to that. Signed-off-by: Dan McGee <dan@archlinux.org> [Xav: fixed printf with off_t] Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
2009-11-12 00:39:26 -05:00
int ret = handle->fetchcb(url, localpath, force);
if(ret == -1) {
RET_ERR(PM_ERR_EXTERNAL_DOWNLOAD, -1);
}
return ret;
}
}
/*
* Download a single file
* - servers must be a list of urls WITHOUT trailing slashes.
*
* RETURN: 0 for successful download
download: major refactor to address lingering issues Sorry for this being such a huge patch, but I believe it is necessary for quite a few reasons which I will attempt to explain herein. I've been mulling this over for a while, but wasn't super happy with making the download interface more complex. Instead, if we carefully order things in the internal download code, we can actually make the interface simpler. 1. FS#15657 - This involves `name.db.tar.gz.part` files being left around the filesystem, and then causing all sorts of issues when someone attempts to rerun the operation they canceled. We need to ensure that if we resume a download, we are resuming it on exactly the same file; if we cannot be almost postive of that then we need to start over. 2. http://www.mail-archive.com/pacman-dev@archlinux.org/msg03536.html - Here we have a lighttpd bug to ruin the day. If we send both a Range: header and If-Modified-Since: header across the wire in a GET request, lighttpd doesn't do what we want in several cases. If the file hadn't been modified, it returns a '304 Not Modified' instead of a '206 Partial Content'. We need to do a stat (e.g. HEAD in HTTP terms) operation here, and the proceed accordingly based off the values we get back from it. 3. The mtime stuff was rather ugly, and relied on the called function to write back to a passed in reference, which isn't the greatest. Instead, use the power of the filesystem to contain this info. Every file downloaded internally is now carefully timestamped with the remote file time. This should allow the resume logic to work. In order to guarantee this, we need to implement a signal handler that catches interrupts, notifies the running code, and causes it to set the mtimes on the file. It then rethrows the signal so the pacman signal handler (or any frontend) works as expected. 4. We did a lot of funky stuff in trying to track the DB last modified time. It is a lot easier to just keep the downloaded DB file around and track the time on that rather than in a funky dot file. It also kills a lot of code. 5. For GPG verification of the databases down the road, we are going to need the DB file around for at least a short bit of time anyway, so this gets us closer to that. Signed-off-by: Dan McGee <dan@archlinux.org> [Xav: fixed printf with off_t] Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
2009-11-12 00:39:26 -05:00
* 1 if the files are identical
* -1 on error
*/
int _alpm_download_single_file(const char *filename,
alpm_list_t *servers, const char *localpath,
download: major refactor to address lingering issues Sorry for this being such a huge patch, but I believe it is necessary for quite a few reasons which I will attempt to explain herein. I've been mulling this over for a while, but wasn't super happy with making the download interface more complex. Instead, if we carefully order things in the internal download code, we can actually make the interface simpler. 1. FS#15657 - This involves `name.db.tar.gz.part` files being left around the filesystem, and then causing all sorts of issues when someone attempts to rerun the operation they canceled. We need to ensure that if we resume a download, we are resuming it on exactly the same file; if we cannot be almost postive of that then we need to start over. 2. http://www.mail-archive.com/pacman-dev@archlinux.org/msg03536.html - Here we have a lighttpd bug to ruin the day. If we send both a Range: header and If-Modified-Since: header across the wire in a GET request, lighttpd doesn't do what we want in several cases. If the file hadn't been modified, it returns a '304 Not Modified' instead of a '206 Partial Content'. We need to do a stat (e.g. HEAD in HTTP terms) operation here, and the proceed accordingly based off the values we get back from it. 3. The mtime stuff was rather ugly, and relied on the called function to write back to a passed in reference, which isn't the greatest. Instead, use the power of the filesystem to contain this info. Every file downloaded internally is now carefully timestamped with the remote file time. This should allow the resume logic to work. In order to guarantee this, we need to implement a signal handler that catches interrupts, notifies the running code, and causes it to set the mtimes on the file. It then rethrows the signal so the pacman signal handler (or any frontend) works as expected. 4. We did a lot of funky stuff in trying to track the DB last modified time. It is a lot easier to just keep the downloaded DB file around and track the time on that rather than in a funky dot file. It also kills a lot of code. 5. For GPG verification of the databases down the road, we are going to need the DB file around for at least a short bit of time anyway, so this gets us closer to that. Signed-off-by: Dan McGee <dan@archlinux.org> [Xav: fixed printf with off_t] Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
2009-11-12 00:39:26 -05:00
int force)
{
alpm_list_t *i;
int ret = -1;
ASSERT(servers != NULL, RET_ERR(PM_ERR_SERVER_NONE, -1));
for(i = servers; i; i = i->next) {
const char *server = i->data;
char *fileurl = NULL;
size_t len;
/* print server + filename into a buffer */
len = strlen(server) + strlen(filename) + 2;
CALLOC(fileurl, len, sizeof(char), RET_ERR(PM_ERR_MEMORY, -1));
snprintf(fileurl, len, "%s/%s", server, filename);
download: major refactor to address lingering issues Sorry for this being such a huge patch, but I believe it is necessary for quite a few reasons which I will attempt to explain herein. I've been mulling this over for a while, but wasn't super happy with making the download interface more complex. Instead, if we carefully order things in the internal download code, we can actually make the interface simpler. 1. FS#15657 - This involves `name.db.tar.gz.part` files being left around the filesystem, and then causing all sorts of issues when someone attempts to rerun the operation they canceled. We need to ensure that if we resume a download, we are resuming it on exactly the same file; if we cannot be almost postive of that then we need to start over. 2. http://www.mail-archive.com/pacman-dev@archlinux.org/msg03536.html - Here we have a lighttpd bug to ruin the day. If we send both a Range: header and If-Modified-Since: header across the wire in a GET request, lighttpd doesn't do what we want in several cases. If the file hadn't been modified, it returns a '304 Not Modified' instead of a '206 Partial Content'. We need to do a stat (e.g. HEAD in HTTP terms) operation here, and the proceed accordingly based off the values we get back from it. 3. The mtime stuff was rather ugly, and relied on the called function to write back to a passed in reference, which isn't the greatest. Instead, use the power of the filesystem to contain this info. Every file downloaded internally is now carefully timestamped with the remote file time. This should allow the resume logic to work. In order to guarantee this, we need to implement a signal handler that catches interrupts, notifies the running code, and causes it to set the mtimes on the file. It then rethrows the signal so the pacman signal handler (or any frontend) works as expected. 4. We did a lot of funky stuff in trying to track the DB last modified time. It is a lot easier to just keep the downloaded DB file around and track the time on that rather than in a funky dot file. It also kills a lot of code. 5. For GPG verification of the databases down the road, we are going to need the DB file around for at least a short bit of time anyway, so this gets us closer to that. Signed-off-by: Dan McGee <dan@archlinux.org> [Xav: fixed printf with off_t] Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
2009-11-12 00:39:26 -05:00
ret = download(fileurl, localpath, force);
FREE(fileurl);
if(ret != -1) {
break;
}
}
return ret;
}
int _alpm_download_files(alpm_list_t *files,
alpm_list_t *servers, const char *localpath)
{
int ret = 0;
alpm_list_t *lp;
for(lp = files; lp; lp = lp->next) {
char *filename = lp->data;
if(_alpm_download_single_file(filename, servers,
download: major refactor to address lingering issues Sorry for this being such a huge patch, but I believe it is necessary for quite a few reasons which I will attempt to explain herein. I've been mulling this over for a while, but wasn't super happy with making the download interface more complex. Instead, if we carefully order things in the internal download code, we can actually make the interface simpler. 1. FS#15657 - This involves `name.db.tar.gz.part` files being left around the filesystem, and then causing all sorts of issues when someone attempts to rerun the operation they canceled. We need to ensure that if we resume a download, we are resuming it on exactly the same file; if we cannot be almost postive of that then we need to start over. 2. http://www.mail-archive.com/pacman-dev@archlinux.org/msg03536.html - Here we have a lighttpd bug to ruin the day. If we send both a Range: header and If-Modified-Since: header across the wire in a GET request, lighttpd doesn't do what we want in several cases. If the file hadn't been modified, it returns a '304 Not Modified' instead of a '206 Partial Content'. We need to do a stat (e.g. HEAD in HTTP terms) operation here, and the proceed accordingly based off the values we get back from it. 3. The mtime stuff was rather ugly, and relied on the called function to write back to a passed in reference, which isn't the greatest. Instead, use the power of the filesystem to contain this info. Every file downloaded internally is now carefully timestamped with the remote file time. This should allow the resume logic to work. In order to guarantee this, we need to implement a signal handler that catches interrupts, notifies the running code, and causes it to set the mtimes on the file. It then rethrows the signal so the pacman signal handler (or any frontend) works as expected. 4. We did a lot of funky stuff in trying to track the DB last modified time. It is a lot easier to just keep the downloaded DB file around and track the time on that rather than in a funky dot file. It also kills a lot of code. 5. For GPG verification of the databases down the road, we are going to need the DB file around for at least a short bit of time anyway, so this gets us closer to that. Signed-off-by: Dan McGee <dan@archlinux.org> [Xav: fixed printf with off_t] Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
2009-11-12 00:39:26 -05:00
localpath, 0) == -1) {
ret++;
}
}
return ret;
}
/** Fetch a remote pkg.
* @param url URL of the package to download
* @return the downloaded filepath on success, NULL on error
* @addtogroup alpm_misc
*/
char SYMEXPORT *alpm_fetch_pkgurl(const char *url)
{
char *filename, *filepath;
const char *cachedir;
int ret;
ALPM_LOG_FUNC;
filename = get_filename(url);
/* find a valid cache dir to download to */
cachedir = _alpm_filecache_setup();
/* download the file */
download: major refactor to address lingering issues Sorry for this being such a huge patch, but I believe it is necessary for quite a few reasons which I will attempt to explain herein. I've been mulling this over for a while, but wasn't super happy with making the download interface more complex. Instead, if we carefully order things in the internal download code, we can actually make the interface simpler. 1. FS#15657 - This involves `name.db.tar.gz.part` files being left around the filesystem, and then causing all sorts of issues when someone attempts to rerun the operation they canceled. We need to ensure that if we resume a download, we are resuming it on exactly the same file; if we cannot be almost postive of that then we need to start over. 2. http://www.mail-archive.com/pacman-dev@archlinux.org/msg03536.html - Here we have a lighttpd bug to ruin the day. If we send both a Range: header and If-Modified-Since: header across the wire in a GET request, lighttpd doesn't do what we want in several cases. If the file hadn't been modified, it returns a '304 Not Modified' instead of a '206 Partial Content'. We need to do a stat (e.g. HEAD in HTTP terms) operation here, and the proceed accordingly based off the values we get back from it. 3. The mtime stuff was rather ugly, and relied on the called function to write back to a passed in reference, which isn't the greatest. Instead, use the power of the filesystem to contain this info. Every file downloaded internally is now carefully timestamped with the remote file time. This should allow the resume logic to work. In order to guarantee this, we need to implement a signal handler that catches interrupts, notifies the running code, and causes it to set the mtimes on the file. It then rethrows the signal so the pacman signal handler (or any frontend) works as expected. 4. We did a lot of funky stuff in trying to track the DB last modified time. It is a lot easier to just keep the downloaded DB file around and track the time on that rather than in a funky dot file. It also kills a lot of code. 5. For GPG verification of the databases down the road, we are going to need the DB file around for at least a short bit of time anyway, so this gets us closer to that. Signed-off-by: Dan McGee <dan@archlinux.org> [Xav: fixed printf with off_t] Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
2009-11-12 00:39:26 -05:00
ret = download(url, cachedir, 0);
if(ret == -1) {
_alpm_log(PM_LOG_WARNING, _("failed to download %s\n"), url);
return NULL;
}
_alpm_log(PM_LOG_DEBUG, "successfully downloaded %s\n", url);
/* we should be able to find the file the second time around */
filepath = _alpm_filecache_find(filename);
return filepath;
}
/* vim: set ts=2 sw=2 noet: */