rpm  5.2.1
rpmio/url.c
Go to the documentation of this file.
00001 
00005 #include "system.h"
00006 
00007 #include <netinet/in.h>
00008 
00009 #include <rpmmacro.h>
00010 #include <rpmcb.h>
00011 #include <rpmio_internal.h>
00012 #ifdef WITH_NEON
00013 #include <rpmdav.h>
00014 #endif
00015 
00016 #include "debug.h"
00017 
00018 /*@access FD_t@*/               /* XXX compared with NULL */
00019 /*@access urlinfo@*/
00020 
00021 #ifndef IPPORT_FTP
00022 #define IPPORT_FTP      21
00023 #endif
00024 #ifndef IPPORT_HTTP
00025 #define IPPORT_HTTP     80
00026 #endif
00027 #ifndef IPPORT_HTTPS
00028 #define IPPORT_HTTPS    443
00029 #endif
00030 #ifndef IPPORT_PGPKEYSERVER
00031 #define IPPORT_PGPKEYSERVER     11371
00032 #endif
00033 
00036 /*@-redecl@*/
00037 int (*urlNotify) (const urlinfo u, unsigned status)
00038         /*@*/;
00039 /*@=redecl@*/
00040 
00043 /*@unchecked@*/ /*@null@*/
00044 void * urlNotifyArg;
00045 
00048 /*@unchecked@*/
00049 int _url_iobuf_size = RPMURL_IOBUF_SIZE;
00050 
00053 /*@unchecked@*/
00054 int _url_debug = 0;
00055 
00056 #define URLDBG(_f, _m, _x)      if ((_url_debug | (_f)) & (_m)) fprintf _x
00057 
00058 #define URLDBGIO(_f, _x)        URLDBG((_f), RPMURL_DEBUG_IO, _x)
00059 #define URLDBGREFS(_f, _x)      URLDBG((_f), RPMURL_DEBUG_REFS, _x)
00060 
00063 /*@unchecked@*/
00064 /*@only@*/ /*@null@*/
00065 urlinfo *_url_cache = NULL;
00066 
00067 static void urlFini(void * _u)
00068         /*@globals fileSystem, internalState @*/
00069         /*@modifies _u, fileSystem, internalState @*/
00070 {
00071     urlinfo u =_u;
00072     int xx;
00073 
00074     if (u->ctrl) {
00075 #ifndef NOTYET
00076         void * fp = fdGetFp(u->ctrl);
00077         if (fp) {
00078             fdPush(u->ctrl, fpio, fp, -1);   /* Push fpio onto stack */
00079             xx = Fclose(u->ctrl);
00080         } else if (fdFileno(u->ctrl) >= 0)
00081             xx = fdio->close(u->ctrl);
00082 #else
00083         xx = Fclose(u->ctrl);
00084 #endif
00085 
00086 /*@-usereleased@*/
00087         u->ctrl = (FD_t)rpmioFreePoolItem((rpmioItem)u->ctrl, "persist ctrl (urlFree)", __FILE__, __LINE__);
00088         if (u->ctrl)
00089             fprintf(stderr, _("warning: u %p ctrl %p nrefs != 0 (%s %s)\n"),
00090                         u, u->ctrl, (u->host ? u->host : ""),
00091                         (u->scheme ? u->scheme : ""));
00092 /*@=usereleased@*/
00093     }
00094     if (u->data) {
00095 #ifndef NOTYET
00096         void * fp = fdGetFp(u->data);
00097         if (fp) {
00098             fdPush(u->data, fpio, fp, -1);   /* Push fpio onto stack */
00099             xx = Fclose(u->data);
00100         } else if (fdFileno(u->data) >= 0)
00101             xx = fdio->close(u->data);
00102 #else
00103         xx = Fclose(u->ctrl);
00104 #endif
00105 
00106 /*@-usereleased@*/
00107         u->data = (FD_t)rpmioFreePoolItem((rpmioItem)u->data, "persist data (urlFree)", __FILE__, __LINE__);
00108         if (u->data)
00109             fprintf(stderr, _("warning: u %p data %p nrefs != 0 (%s %s)\n"),
00110                         u, u->data, (u->host ? u->host : ""),
00111                         (u->scheme ? u->scheme : ""));
00112 /*@=usereleased@*/
00113     }
00114 #ifdef WITH_NEON
00115     xx = davFree(u);
00116 #endif
00117     u->etag = _free(u->etag);
00118     u->location = _free(u->location);
00119     u->rop = _free(u->rop);
00120     u->sop = _free(u->sop);
00121     u->top = _free(u->top);
00122     u->buf = _free(u->buf);
00123     u->url = _free(u->url);
00124     u->scheme = _free((void *)u->scheme);
00125     u->user = _free((void *)u->user);
00126     u->password = _free((void *)u->password);
00127     u->host = _free((void *)u->host);
00128     u->portstr = _free((void *)u->portstr);
00129     u->query = _free(u->query);
00130     u->fragment = _free(u->fragment);
00131     u->proxyu = _free((void *)u->proxyu);
00132     u->proxyh = _free((void *)u->proxyh);
00133 }
00134 
00137 /*@unchecked@*/
00138 int _url_count = 0;
00139 
00140 /*@unchecked@*/ /*@only@*/ /*@null@*/
00141 rpmioPool _urlPool;
00142 
00143 static urlinfo urlGetPool(/*@null@*/ rpmioPool pool)
00144         /*@globals _urlPool, fileSystem @*/
00145         /*@modifies pool, _urlPool, fileSystem @*/
00146 {
00147     urlinfo u;
00148 
00149     if (_urlPool == NULL) {
00150         _urlPool = rpmioNewPool("u", sizeof(*u), -1, _url_debug,
00151                         NULL, NULL, urlFini);
00152         pool = _urlPool;
00153     }
00154     return (urlinfo) rpmioGetPool(pool, sizeof(*u));
00155 }
00156 
00157 urlinfo XurlNew(const char *msg, const char *fn, unsigned ln)
00158 {
00159     urlinfo u = urlGetPool(_urlPool);
00160 
00161     u->proxyp = -1;
00162     u->port = -1;
00163     u->urltype = URL_IS_UNKNOWN;
00164     u->ctrl = NULL;
00165     u->data = NULL;
00166     u->location = NULL;
00167     u->etag = NULL;
00168     u->notify = urlNotify;
00169 /*@-assignexpose@*/
00170     u->arg = urlNotifyArg;
00171 /*@=assignexpose@*/
00172     u->rop = xcalloc(1, sizeof(*u->rop));
00173     u->sop = xcalloc(1, sizeof(*u->sop));
00174     u->top = xcalloc(1, sizeof(*u->top));
00175     u->bufAlloced = 0;
00176     u->buf = NULL;
00177     u->allow = RPMURL_SERVER_HASRANGE;
00178     u->httpVersion = 0;
00179     u->magic = URLMAGIC;
00180     return (urlinfo) rpmioLinkPoolItem((rpmioItem)u, msg, fn, ln);
00181 }
00182 
00183 void urlFreeCache(void)
00184 {
00185     if (_url_cache) {
00186         int i;
00187         for (i = 0; i < _url_count; i++) {
00188             if (_url_cache[i] == NULL) continue;
00189             _url_cache[i] = urlFree(_url_cache[i], "_url_cache");
00190             if (_url_cache[i] == NULL)
00191                 continue;
00192             yarnPossess(_url_cache[i]->_item.use);
00193             fprintf(stderr,
00194                 _("warning: _url_cache[%d] %p nrefs(%ld) != 1 (%s %s)\n"),
00195                 i, _url_cache[i], yarnPeekLock(_url_cache[i]->_item.use),
00196                 (_url_cache[i]->host ? _url_cache[i]->host : ""),
00197                 (_url_cache[i]->scheme ? _url_cache[i]->scheme : ""));
00198             yarnRelease(_url_cache[i]->_item.use);
00199         }
00200     }
00201     _url_cache = _free(_url_cache);
00202     _url_count = 0;
00203 }
00204 
00205 static int urlStrcmp(/*@null@*/ const char * str1, /*@null@*/ const char * str2)
00206         /*@*/
00207 {
00208     if (str1)
00209         if (str2)
00210             return strcmp(str1, str2);
00211     if (str1 != str2)
00212         return -1;
00213     return 0;
00214 }
00215 
00216 /*@-mods@*/
00217 static void urlFind(/*@null@*/ /*@in@*/ /*@out@*/ urlinfo * uret, int mustAsk)
00218         /*@globals rpmGlobalMacroContext, h_errno, fileSystem, internalState @*/
00219         /*@modifies *uret, rpmGlobalMacroContext, fileSystem, internalState @*/
00220 {
00221     urlinfo u;
00222     int ucx;
00223     int i = 0;
00224 
00225     if (uret == NULL)
00226         return;
00227 
00228     u = *uret;
00229     URLSANE(u);
00230 
00231     ucx = -1;
00232     for (i = 0; i < _url_count; i++) {
00233         urlinfo ou = NULL;
00234         if (_url_cache == NULL || (ou = _url_cache[i]) == NULL) {
00235             if (ucx < 0)
00236                 ucx = i;
00237             continue;
00238         }
00239 
00240         /* Check for cache-miss condition. A cache miss is
00241          *    a) both items are not NULL and don't compare.
00242          *    b) either of the items is not NULL.
00243          */
00244         if (urlStrcmp(u->scheme, ou->scheme))
00245             continue;
00246         if (urlStrcmp(u->host, ou->host))
00247             continue;
00248         if (urlStrcmp(u->user, ou->user))
00249             continue;
00250         if (urlStrcmp(u->portstr, ou->portstr))
00251             continue;
00252         break;  /* Found item in cache */
00253     }
00254 
00255     if (i == _url_count) {
00256         if (ucx < 0) {
00257             ucx = _url_count++;
00258             _url_cache = xrealloc(_url_cache, sizeof(*_url_cache) * _url_count);
00259         }
00260         if (_url_cache)         /* XXX always true */
00261             _url_cache[ucx] = urlLink(u, "_url_cache (miss)");
00262         u = urlFree(u, "urlSplit (urlFind miss)");
00263     } else {
00264         ucx = i;
00265         u = urlFree(u, "urlSplit (urlFind hit)");
00266     }
00267 
00268     /* This URL is now cached. */
00269 
00270     if (_url_cache)             /* XXX always true */
00271         u = urlLink(_url_cache[ucx], "_url_cache");
00272     *uret = u;
00273     /*@-usereleased@*/
00274     u = urlFree(u, "_url_cache (urlFind)");
00275     /*@=usereleased@*/
00276 assert(u != NULL);
00277 
00278     /* Zap proxy host and port in case they have been reset */
00279     u->proxyp = -1;
00280     u->proxyh = _free(u->proxyh);
00281 
00282     /* Perform one-time FTP initialization */
00283     if (u->urltype == URL_IS_FTP) {
00284 
00285         if (mustAsk || (u->user != NULL && u->password == NULL)) {
00286             const char * host = (u->host ? u->host : "");
00287             const char * user = (u->user ? u->user : "");
00288             char * prompt;
00289             prompt = alloca(strlen(host) + strlen(user) + 256);
00290             sprintf(prompt, _("Password for %s@%s: "), user, host);
00291             u->password = _free(u->password);
00292 /*@-dependenttrans -moduncon @*/
00293             u->password = Getpass(prompt);
00294 /*@=dependenttrans =moduncon @*/
00295             if (u->password)
00296                 u->password = xstrdup(u->password);
00297         }
00298 
00299         if (u->proxyh == NULL) {
00300             const char *proxy = rpmExpand("%{_ftpproxy}", NULL);
00301             if (proxy && *proxy != '%') {
00302 /*@observer@*/
00303                 const char * host = (u->host ? u->host : "");
00304                 const char *uu = (u->user ? u->user : "anonymous");
00305                 char *nu = xmalloc(strlen(uu) + sizeof("@") + strlen(host));
00306                 (void) stpcpy( stpcpy( stpcpy(nu, uu), "@"), host);
00307                 u->proxyu = nu;
00308                 u->proxyh = xstrdup(proxy);
00309             }
00310             proxy = _free(proxy);
00311         }
00312 
00313         if (u->proxyp < 0) {
00314             const char *proxy = rpmExpand("%{_ftpport}", NULL);
00315             if (proxy && *proxy != '%') {
00316                 char *end = NULL;
00317                 int port = strtol(proxy, &end, 0);
00318                 if (!(end && *end == '\0')) {
00319                     fprintf(stderr, _("error: %sport must be a number\n"),
00320                         (u->scheme ? u->scheme : ""));
00321                     return;
00322                 }
00323                 u->proxyp = port;
00324             }
00325             proxy = _free(proxy);
00326         }
00327     }
00328 
00329     /* Perform one-time HTTP initialization */
00330     if (u->urltype == URL_IS_HTTP || u->urltype == URL_IS_HTTPS || u->urltype == URL_IS_HKP) {
00331 
00332         if (u->proxyh == NULL) {
00333             const char *proxy = rpmExpand("%{_httpproxy}", NULL);
00334             if (proxy && *proxy != '%')
00335                 u->proxyh = xstrdup(proxy);
00336             proxy = _free(proxy);
00337         }
00338 
00339         if (u->proxyp < 0) {
00340             const char *proxy = rpmExpand("%{_httpport}", NULL);
00341             if (proxy && *proxy != '%') {
00342                 char *end;
00343                 int port = strtol(proxy, &end, 0);
00344                 if (!(end && *end == '\0')) {
00345                     fprintf(stderr, _("error: %sport must be a number\n"),
00346                         (u->scheme ? u->scheme : ""));
00347                     return;
00348                 }
00349                 u->proxyp = port;
00350             }
00351             proxy = _free(proxy);
00352         }
00353 
00354     }
00355 
00356     return;
00357 }
00358 /*@=mods@*/
00359 
00362 /*@observer@*/ /*@unchecked@*/
00363 static struct urlstring {
00364 /*@observer@*/ /*@null@*/
00365     const char * leadin;
00366     urltype     ret;
00367 } urlstrings[] = {
00368     { "file://",        URL_IS_PATH },
00369     { "ftp://",         URL_IS_FTP },
00370     { "hkp://",         URL_IS_HKP },
00371     { "http://",        URL_IS_HTTP },
00372     { "https://",       URL_IS_HTTPS },
00373     { "-",              URL_IS_DASH },
00374     { NULL,             URL_IS_UNKNOWN }
00375 };
00376 
00377 urltype urlIsURL(const char * url)
00378 {
00379     struct urlstring *us;
00380 
00381     if (url && *url) {
00382         for (us = urlstrings; us->leadin != NULL; us++) {
00383             if (strncmp(url, us->leadin, strlen(us->leadin)))
00384                 continue;
00385             return us->ret;
00386         }
00387     }
00388 
00389     return URL_IS_UNKNOWN;
00390 }
00391 
00392 /* Return path portion of url (or pointer to NUL if url == NULL) */
00393 urltype urlPath(const char * url, const char ** pathp)
00394 {
00395     const char *path;
00396     int urltype;
00397 
00398     path = url;
00399     urltype = urlIsURL(url);
00400     switch (urltype) {
00401     case URL_IS_FTP:
00402         url += sizeof("ftp://") - 1;
00403         path = strchr(url, '/');
00404         if (path == NULL) path = url + strlen(url);
00405         break;
00406     case URL_IS_PATH:
00407         url += sizeof("file://") - 1;
00408         path = strchr(url, '/');
00409         if (path == NULL) path = url + strlen(url);
00410         break;
00411     case URL_IS_HKP:
00412         url += sizeof("hkp://") - 1;
00413         path = strchr(url, '/');
00414         if (path == NULL) path = url + strlen(url);
00415         break;
00416     case URL_IS_HTTP:
00417         url += sizeof("http://") - 1;
00418         path = strchr(url, '/');
00419         if (path == NULL) path = url + strlen(url);
00420         break;
00421     case URL_IS_HTTPS:
00422         url += sizeof("https://") - 1;
00423         path = strchr(url, '/');
00424         if (path == NULL) path = url + strlen(url);
00425         break;
00426     case URL_IS_UNKNOWN:
00427         if (path == NULL) path = "";
00428         break;
00429     case URL_IS_DASH:
00430         path = "";
00431         break;
00432     }
00433     if (pathp)
00434         /*@-observertrans@*/
00435         *pathp = path;
00436         /*@=observertrans@*/
00437     return urltype;
00438 }
00439 
00443 static const char * urlStrdup(const char * url)
00444         /*@*/
00445 {
00446     size_t nb = strlen(url);
00447     char * t = xmalloc(nb + 1 + 1);
00448     const char * nurl = t;
00449     while (*url != '\0')
00450         *t++ = *url++;
00451     *t = '\0';
00452     return nurl;
00453 }
00454 
00455 /*
00456  * Split URL into components. The URL can look like
00457  *      scheme://user:password@host:port/path
00458   * or as in RFC2732 for IPv6 address
00459   *    service://user:password@[ip:v6:ad:dr:es:s]:port/path?query#fragment
00460  */
00461 /*@-modfilesys@*/
00462 int urlSplit(const char * url, urlinfo *uret)
00463 {
00464     urlinfo u;
00465     char *myurl;
00466     char *s, *se, *f, *fe;
00467 
00468     if (uret == NULL)
00469         return -1;
00470     if ((u = urlNew("urlSplit")) == NULL)
00471         return -1;
00472 
00473     myurl = xstrdup(url);
00474     if ((se = strrchr(myurl, '#')) != NULL) {
00475         *se++ = '\0';
00476         u->fragment = xstrdup(se);
00477     }
00478     if ((se = strrchr(myurl, '?')) != NULL) {
00479         *se++ = '\0';
00480         u->query = xstrdup(se);
00481     }
00482 
00483     u->url = urlStrdup(myurl);          /* XXX +1 byte for pesky trailing '/' */
00484     u->urltype = urlIsURL(myurl);
00485 
00486     se = s = myurl;
00487     while (1) {
00488         /* Point to end of next item */
00489         while (*se && *se != '/') se++;
00490         /* Item was scheme. Save scheme and go for the rest ...*/
00491         if (*se && (se != s) && se[-1] == ':' && se[0] == '/' && se[1] == '/') {
00492                 se[-1] = '\0';
00493             u->scheme = xstrdup(s);
00494             se += 2;    /* skip over "//" */
00495             s = se++;
00496             continue;
00497         }
00498         
00499         /* Item was everything-but-path. Continue parse on rest */
00500         *se = '\0';
00501         break;
00502     }
00503 
00504     /* Look for ...@host... */
00505     fe = f = s;
00506     while (*fe && *fe != '@') fe++;
00507     if (*fe == '@') {
00508         s = fe + 1;
00509         *fe = '\0';
00510         /* Look for user:password@host... */
00511         while (fe > f && *fe != ':') fe--;
00512         if (*fe == ':') {
00513             *fe++ = '\0';
00514             u->password = xstrdup(fe);
00515         }
00516         u->user = xstrdup(f);
00517     }
00518 
00519     /* Look for ...host:port or [v6addr]:port*/
00520     fe = f = s;
00521     if (strchr(fe, '[') && strchr(fe, ']')) {
00522         fe = strchr(f, ']');
00523         *f++ = '\0';
00524         *fe++ = '\0';
00525     }
00526 assert(fe != NULL);     /* XXX can't happen */
00527     while (*fe && *fe != ':') fe++;
00528     if (*fe == ':') {
00529         *fe++ = '\0';
00530         u->portstr = xstrdup(fe);
00531         if (u->portstr != NULL && u->portstr[0] != '\0') {
00532             char *end;
00533             u->port = strtol(u->portstr, &end, 0);
00534             if (!(end && *end == '\0')) {
00535                 rpmlog(RPMLOG_ERR, _("url port must be a number\n"));
00536                 myurl = _free(myurl);
00537                 u = urlFree(u, "urlSplit (error #3)");
00538                 return -1;
00539             }
00540         }
00541     }
00542     u->host = xstrdup(f);
00543 
00544     if (u->port < 0 && u->scheme != NULL) {
00545         struct servent *serv;
00546 /*@-multithreaded -moduncon @*/
00547         /* HACK hkp:// might lookup "pgpkeyserver" */
00548         serv = getservbyname(u->scheme, "tcp");
00549 /*@=multithreaded =moduncon @*/
00550         if (serv != NULL)
00551             u->port = (int) ntohs(serv->s_port);
00552         else if (u->urltype == URL_IS_FTP)
00553             u->port = IPPORT_FTP;
00554         else if (u->urltype == URL_IS_HKP)
00555             u->port = IPPORT_PGPKEYSERVER;
00556         else if (u->urltype == URL_IS_HTTP)
00557             u->port = IPPORT_HTTP;
00558         else if (u->urltype == URL_IS_HTTPS)
00559             u->port = IPPORT_HTTPS;
00560     }
00561 
00562     myurl = _free(myurl);
00563     if (uret) {
00564         *uret = u;
00565 /*@-globs -mods @*/ /* FIX: rpmGlobalMacroContext not in <rpmlib.h> */
00566         urlFind(uret, 0);
00567 /*@=globs =mods @*/
00568     }
00569     return 0;
00570 }
00571 /*@=modfilesys@*/
00572 
00573 int urlGetFile(const char * url, const char * dest)
00574 {
00575     int rc;
00576     FD_t sfd = NULL;
00577     FD_t tfd = NULL;
00578     const char * sfuPath = NULL;
00579     int urlType = urlPath(url, &sfuPath);
00580     char *result;
00581 
00582     if (*sfuPath == '\0')
00583         return FTPERR_UNKNOWN;
00584 
00585     if (dest == NULL) {
00586         if ((dest = strrchr(sfuPath, '/')) != NULL)
00587             dest++;
00588         else
00589             dest = sfuPath;
00590     }
00591     if (dest == NULL)
00592         return FTPERR_UNKNOWN;
00593 
00594 /*@-globs -mods@*/      /* Avoid including <rpmmacro.h> everywhere for now */
00595     if (rpmExpandNumeric("%{?__urlgetfile:1}%{!?__urlgetfile:0}")) {
00596         result = rpmExpand("%{__urlgetfile ", url, " ", dest, "}", NULL);
00597         if (result != NULL && strcmp(result, "OK") == 0)
00598             rc = 0;
00599         else {
00600             rpmlog(RPMLOG_DEBUG, D_("failed to fetch URL %s via external command\n"), url);
00601             rc = FTPERR_UNKNOWN;
00602         }
00603         result = _free(result);
00604         goto exit;
00605     }
00606 /*@=globs =mods@*/
00607 
00608     sfd = Fopen(url, "r.ufdio");
00609     if (sfd == NULL || Ferror(sfd)) {
00610         rpmlog(RPMLOG_DEBUG, D_("failed to open %s: %s\n"), url, Fstrerror(sfd));
00611         rc = FTPERR_UNKNOWN;
00612         goto exit;
00613     }
00614 
00615     /* XXX this can fail if directory in path does not exist. */
00616     tfd = Fopen(dest, "w");
00617 if (_url_debug)
00618 fprintf(stderr, "*** urlGetFile sfd %p %s tfd %p %s\n", sfd, url, (tfd ? tfd : NULL), dest);
00619     if (tfd == NULL || Ferror(tfd)) {
00620         rpmlog(RPMLOG_DEBUG, D_("failed to create %s: %s\n"), dest, Fstrerror(tfd));
00621         rc = FTPERR_UNKNOWN;
00622         goto exit;
00623     }
00624 
00625     switch (urlType) {
00626     case URL_IS_HTTPS:
00627     case URL_IS_HTTP:
00628     case URL_IS_HKP:
00629     case URL_IS_FTP:
00630     case URL_IS_PATH:
00631     case URL_IS_DASH:
00632     case URL_IS_UNKNOWN:
00633         if ((rc = ufdGetFile(sfd, tfd))) {
00634             (void) Unlink(dest);
00635             /* XXX FIXME: sfd possibly closed by copyData */
00636             /*@-usereleased@*/ (void) Fclose(sfd) /*@=usereleased@*/ ;
00637         }
00638         sfd = NULL;     /* XXX Fclose(sfd) done by ufdGetFile */
00639         break;
00640     default:
00641         rc = FTPERR_UNKNOWN;
00642         break;
00643     }
00644 
00645 exit:
00646     if (tfd)
00647         (void) Fclose(tfd);
00648     if (sfd)
00649         (void) Fclose(sfd);
00650 
00651     return rc;
00652 }