X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Fhttp.c;h=e46d2ae3442cec2ac05607290dcb08b283187bd2;hb=1075dbdd4d718fa6e8a2652e9f59d01fc150e0dc;hp=401f5ae4ceb931ed23daf91b663efb0aeba5b25b;hpb=fe3383c0559a453df1e5076fc6faac6d1a11685c;p=pazpar2-moved-to-github.git diff --git a/src/http.c b/src/http.c index 401f5ae..e46d2ae 100644 --- a/src/http.c +++ b/src/http.c @@ -1,5 +1,22 @@ -/* - * $Id: http.c,v 1.1 2006-12-20 20:47:16 quinn Exp $ +/* $Id: http.c,v 1.31 2007-05-15 08:51:49 adam Exp $ + Copyright (c) 2006-2007, Index Data. + +This file is part of Pazpar2. + +Pazpar2 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, or (at your option) any later +version. + +Pazpar2 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 Pazpar2; see the file LICENSE. If not, write to the +Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA +02111-1307, USA. */ #include @@ -14,12 +31,21 @@ #include #include #include +#include + +#if HAVE_CONFIG_H +#include +#endif + +#include +#include +#include #include #include -#include +#include -#include "command.h" +#include "cconfig.h" #include "util.h" #include "eventl.h" #include "pazpar2.h" @@ -27,13 +53,13 @@ #include "http_command.h" static void proxy_io(IOCHAN i, int event); -static struct http_channel *http_create(void); +static struct http_channel *http_create(const char *addr); static void http_destroy(IOCHAN i); -extern IOCHAN channel_list; - -static struct sockaddr_in *proxy_addr = 0; // If this is set, we proxy normal HTTP requests +// If this is set, we proxy normal HTTP requests +static struct sockaddr_in *proxy_addr = 0; static char proxy_url[256] = ""; +static char myurl[256] = ""; static struct http_buf *http_buf_freelist = 0; static struct http_channel *http_channel_freelist = 0; @@ -89,10 +115,10 @@ static struct http_buf *http_buf_bybuf(char *b, int len) while (len) { - *p = http_buf_create(); int tocopy = len; if (tocopy > HTTP_BUF_SIZE) tocopy = HTTP_BUF_SIZE; + *p = http_buf_create(); memcpy((*p)->buf, b, tocopy); (*p)->len = tocopy; len -= tocopy; @@ -162,7 +188,8 @@ static int http_buf_read(struct http_buf **b, char *buf, int len) return rd; } -void static urldecode(char *i, char *o) +// Buffers may overlap. +static void urldecode(char *i, char *o) { while (*i) { @@ -184,6 +211,23 @@ void static urldecode(char *i, char *o) *o = '\0'; } +// Warning: Buffers may not overlap +void urlencode(const char *i, char *o) +{ + while (*i) + { + if (strchr(" /:", *i)) + { + sprintf(o, "%%%.2X", (int) *i); + o += 3; + } + else + *(o++) = *i; + i++; + } + *o = '\0'; +} + void http_addheader(struct http_response *r, const char *name, const char *value) { struct http_channel *c = r->channel; @@ -205,12 +249,11 @@ char *http_argbyname(struct http_request *r, char *name) return 0; } -char *http_headerbyname(struct http_request *r, char *name) +char *http_headerbyname(struct http_header *h, char *name) { - struct http_header *p; - for (p = r->headers; p; p = p->next) - if (!strcmp(p->name, name)) - return p->value; + for (; h; h = h->next) + if (!strcmp(h->name, name)) + return h->value; return 0; } @@ -225,16 +268,10 @@ struct http_response *http_create_response(struct http_channel *c) return r; } -// Check if we have a complete request. Return 0 or length (including trailing newline) -// FIXME: Does not deal gracefully with requests carrying payload -// but this is kind of OK since we will reject anything other than an empty GET -static int request_check(struct http_buf *queue) +// Check if buf contains a package (minus payload) +static int package_check(const char *buf) { - char tmp[4096]; int len = 0; - char *buf = tmp; - - http_buf_peek(queue, tmp, 4096); while (*buf) // Check if we have a sequence of lines terminated by an empty line { char *b = strstr(buf, "\r\n"); @@ -250,6 +287,69 @@ static int request_check(struct http_buf *queue) return 0; } +// Check if we have a request. Return 0 or length +// (including trailing CRNL) FIXME: Does not deal gracefully with requests +// carrying payload but this is kind of OK since we will reject anything +// other than an empty GET +static int request_check(struct http_buf *queue) +{ + char tmp[4096]; + + http_buf_peek(queue, tmp, 4096); + return package_check(tmp); +} + +struct http_response *http_parse_response_buf(struct http_channel *c, const char *buf, int len) +{ + char tmp[4096]; + struct http_response *r = http_create_response(c); + char *p, *p2; + struct http_header **hp = &r->headers; + + if (len >= 4096) + return 0; + memcpy(tmp, buf, len); + for (p = tmp; *p && *p != ' '; p++) // Skip HTTP version + ; + p++; + // Response code + for (p2 = p; *p2 && *p2 != ' ' && p2 - p < 3; p2++) + r->code[p2 - p] = *p2; + if (!(p = strstr(tmp, "\r\n"))) + return 0; + p += 2; + while (*p) + { + if (!(p2 = strstr(p, "\r\n"))) + return 0; + if (p == p2) // End of headers + break; + else + { + struct http_header *h = *hp = nmem_malloc(c->nmem, sizeof(*h)); + char *value = strchr(p, ':'); + if (!value) + return 0; + *(value++) = '\0'; + h->name = nmem_strdup(c->nmem, p); + while (isspace(*value)) + value++; + if (value >= p2) // Empty header; + { + h->value = ""; + p = p2 + 2; + continue; + } + *p2 = '\0'; + h->value = nmem_strdup(c->nmem, value); + h->next = 0; + hp = &h->next; + p = p2 + 2; + } + } + return r; +} + struct http_request *http_parse_request(struct http_channel *c, struct http_buf **queue, int len) { @@ -263,6 +363,7 @@ struct http_request *http_parse_request(struct http_channel *c, struct http_buf if (http_buf_read(queue, buf, len) < len) return 0; + r->search = ""; r->channel = c; r->arguments = 0; r->headers = 0; @@ -293,6 +394,7 @@ struct http_request *http_parse_request(struct http_channel *c, struct http_buf r->path = nmem_strdup(c->nmem, buf); if (p2) { + r->search = nmem_strdup(c->nmem, p2); // Parse Arguments while (*p2) { @@ -366,18 +468,21 @@ struct http_request *http_parse_request(struct http_channel *c, struct http_buf return r; } - static struct http_buf *http_serialize_response(struct http_channel *c, struct http_response *r) { - wrbuf_rewind(c->wrbuf); struct http_header *h; + wrbuf_rewind(c->wrbuf); wrbuf_printf(c->wrbuf, "HTTP/1.1 %s %s\r\n", r->code, r->msg); for (h = r->headers; h; h = h->next) wrbuf_printf(c->wrbuf, "%s: %s\r\n", h->name, h->value); - wrbuf_printf(c->wrbuf, "Content-length: %d\r\n", r->payload ? (int) strlen(r->payload) : 0); - wrbuf_printf(c->wrbuf, "Content-type: text/xml\r\n"); + if (r->payload) + { + wrbuf_printf(c->wrbuf, "Content-length: %d\r\n", r->payload ? + (int) strlen(r->payload) : 0); + wrbuf_printf(c->wrbuf, "Content-type: text/xml\r\n"); + } wrbuf_puts(c->wrbuf, "\r\n"); if (r->payload) @@ -390,10 +495,10 @@ static struct http_buf *http_serialize_response(struct http_channel *c, static struct http_buf *http_serialize_request(struct http_request *r) { struct http_channel *c = r->channel; - wrbuf_rewind(c->wrbuf); struct http_header *h; struct http_argument *a; + wrbuf_rewind(c->wrbuf); wrbuf_printf(c->wrbuf, "%s %s", r->method, r->path); if (r->arguments) @@ -424,12 +529,46 @@ static int http_weshouldproxy(struct http_request *rq) return 0; } + +struct http_header * http_header_append(struct http_channel *ch, + struct http_header * hp, + const char *name, + const char *value) +{ + struct http_header *hpnew = 0; + + if (!hp | !ch) + return 0; + + while (hp && hp->next) + hp = hp->next; + + if(name && strlen(name)&& value && strlen(value)){ + hpnew = nmem_malloc(ch->nmem, sizeof *hpnew); + hpnew->name = nmem_strdup(ch->nmem, name); + hpnew->value = nmem_strdup(ch->nmem, value); + + hpnew->next = 0; + hp->next = hpnew; + hp = hp->next; + + return hpnew; + } + + return hp; +} + + + static int http_proxy(struct http_request *rq) { struct http_channel *c = rq->channel; struct http_proxy *p = c->proxy; struct http_header *hp; struct http_buf *requestbuf; + char server_via[128] = ""; + char server_port[16] = ""; + struct conf_server *ser = global_parameters.server; if (!p) // This is a new connection. Create a proxy channel { @@ -453,7 +592,8 @@ static int http_proxy(struct http_request *rq) yaz_log(YLOG_FATAL|YLOG_ERRNO, "fcntl"); if (fcntl(sock, F_SETFL, flags | O_NONBLOCK) < 0) yaz_log(YLOG_FATAL|YLOG_ERRNO, "fcntl2"); - if (connect(sock, (struct sockaddr *) proxy_addr, sizeof(*proxy_addr)) < 0) + if (connect(sock, (struct sockaddr *) proxy_addr, + sizeof(*proxy_addr)) < 0) if (errno != EINPROGRESS) { yaz_log(YLOG_WARN|YLOG_ERRNO, "Proxy connect"); @@ -463,15 +603,15 @@ static int http_proxy(struct http_request *rq) p = xmalloc(sizeof(struct http_proxy)); p->oqueue = 0; p->channel = c; + p->first_response = 1; c->proxy = p; // We will add EVENT_OUTPUT below p->iochan = iochan_create(sock, proxy_io, EVENT_INPUT); iochan_setdata(p->iochan, p); - p->iochan->next = channel_list; - channel_list = p->iochan; + pazpar2_add_channel(p->iochan); } - // Modify Host: header + // Do _not_ modify Host: header, just checking it's existence for (hp = rq->headers; hp; hp = hp->next) if (!strcmp(hp->name, "Host")) break; @@ -480,7 +620,23 @@ static int http_proxy(struct http_request *rq) yaz_log(YLOG_WARN, "Failed to find Host header in proxy"); return -1; } - hp->value = nmem_strdup(c->nmem, proxy_url); + + // Add new header about paraz2 version, host, remote client address, etc. + { + hp = rq->headers; + hp = http_header_append(c, hp, + "X-Pazpar2-Version", PACKAGE_VERSION); + hp = http_header_append(c, hp, + "X-Pazpar2-Server-Host", ser->host); + sprintf(server_port, "%d", ser->port); + hp = http_header_append(c, hp, + "X-Pazpar2-Server-Port", server_port); + sprintf(server_via, "1.1 %s:%s (%s/%s)", + ser->host, server_port, PACKAGE_NAME, PACKAGE_VERSION); + hp = http_header_append(c, hp, "Via" , server_via); + hp = http_header_append(c, hp, "X-Forwarded-For", c->addr); + } + requestbuf = http_serialize_request(rq); http_buf_enqueue(&p->oqueue, requestbuf); iochan_setflag(p->iochan, EVENT_OUTPUT); @@ -490,8 +646,10 @@ static int http_proxy(struct http_request *rq) void http_send_response(struct http_channel *ch) { struct http_response *rs = ch->response; + struct http_buf *hb; + assert(rs); - struct http_buf *hb = http_serialize_response(ch, rs); + hb = http_serialize_response(ch, rs); if (!hb) { yaz_log(YLOG_WARN, "Failed to serialize HTTP response"); @@ -517,7 +675,12 @@ static void http_io(IOCHAN i, int event) case EVENT_INPUT: htbuf = http_buf_create(); res = read(iochan_getfd(i), htbuf->buf, HTTP_BUF_SIZE -1); - if (res <= 0 && errno != EAGAIN) + if (res == -1 && errno == EAGAIN) + { + http_buf_destroy(htbuf); + return; + } + if (res <= 0) { http_buf_destroy(htbuf); http_destroy(i); @@ -532,7 +695,6 @@ static void http_io(IOCHAN i, int event) if (hc->state == Http_Busy) return; - if ((reqlen = request_check(hc->iqueue)) <= 2) return; @@ -544,8 +706,10 @@ static void http_io(IOCHAN i, int event) return; } hc->response = 0; - yaz_log(YLOG_LOG, "Request: %s %s v %s", hc->request->method, - hc->request->path, hc->request->http_version); + yaz_log(YLOG_LOG, "Request: %s %s%s%s", hc->request->method, + hc->request->path, + *hc->request->search ? "?" : "", + hc->request->search); if (http_weshouldproxy(hc->request)) http_proxy(hc->request); else @@ -607,6 +771,25 @@ static void http_io(IOCHAN i, int event) } } +#ifdef GAGA +// If this hostname contains our proxy host as a prefix, replace with myurl +static char *sub_hostname(struct http_channel *c, char *buf) +{ + char tmp[1024]; + if (strlen(buf) > 1023) + return buf; + if (strncmp(buf, "http://", 7)) + return buf; + if (!strncmp(buf + 7, proxy_url, strlen(proxy_url))) + { + strcpy(tmp, myurl); + strcat(tmp, buf + strlen(proxy_url) + 7); + return nmem_strdup(c->nmem, tmp); + } + return buf; +} +#endif + // Handles I/O on a client connection to a backend web server (proxy mode) static void proxy_io(IOCHAN pi, int event) { @@ -641,8 +824,38 @@ static void proxy_io(IOCHAN pi, int event) else { htbuf->buf[res] = '\0'; + htbuf->offset = 0; htbuf->len = res; - http_buf_enqueue(&hc->oqueue, htbuf); +#ifdef GAGA + if (pc->first_response) // Check if this is a redirect + { + int len; + if ((len = package_check(htbuf->buf))) + { + struct http_response *res = http_parse_response_buf(hc, htbuf->buf, len); + if (res) + { + struct http_header *h; + for (h = res->headers; h; h = h->next) + if (!strcmp(h->name, "Location")) + { + // We found a location header. Rewrite it. + struct http_buf *buf; + h->value = sub_hostname(hc, h->value); + buf = http_serialize_response(hc, res); + yaz_log(YLOG_LOG, "Proxy rewrite"); + http_buf_enqueue(&hc->oqueue, buf); + htbuf->offset = len; + break; + } + } + } + pc->first_response = 0; + } +#endif + // Write any remaining payload + if (htbuf->len - htbuf->offset > 0) + http_buf_enqueue(&hc->oqueue, htbuf); } iochan_setflag(hc->iochan, EVENT_OUTPUT); break; @@ -702,7 +915,7 @@ static void http_destroy(IOCHAN i) iochan_destroy(i); } -static struct http_channel *http_create(void) +static struct http_channel *http_create(const char *addr) { struct http_channel *r = http_channel_freelist; @@ -724,6 +937,12 @@ static struct http_channel *http_create(void) r->state = Http_Idle; r->request = 0; r->response = 0; + if (!addr) + { + yaz_log(YLOG_WARN, "Invalid HTTP forward address"); + exit(1); + } + strcpy(r->addr, addr); return r; } @@ -750,15 +969,14 @@ static void http_accept(IOCHAN i, int event) if (fcntl(s, F_SETFL, flags | O_NONBLOCK) < 0) yaz_log(YLOG_FATAL|YLOG_ERRNO, "fcntl2"); - yaz_log(YLOG_LOG, "New command connection"); + yaz_log(YLOG_DEBUG, "New command connection"); c = iochan_create(s, http_io, EVENT_INPUT | EVENT_EXCEPT); - - ch = http_create(); + + ch = http_create(inet_ntoa(addr.sin_addr)); ch->iochan = c; iochan_setdata(c, ch); - c->next = channel_list; - channel_list = c; + pazpar2_add_channel(c); } /* Create a http-channel listener, syntax [host:]port */ @@ -772,9 +990,9 @@ void http_init(const char *addr) const char *pp; int port; - yaz_log(YLOG_LOG, "HTTP listener is %s", addr); + yaz_log(YLOG_LOG, "HTTP listener %s", addr); - bzero(&myaddr, sizeof myaddr); + memset(&myaddr, 0, sizeof myaddr); myaddr.sin_family = AF_INET; pp = strchr(addr, ':'); if (pp) @@ -785,19 +1003,25 @@ void http_init(const char *addr) strncpy(hostname, addr, len); hostname[len] = '\0'; - if (!(he = gethostbyname(hostname))) - { + if (!(he = gethostbyname(hostname))){ yaz_log(YLOG_FATAL, "Unable to resolve '%s'", hostname); exit(1); } + memcpy(&myaddr.sin_addr.s_addr, he->h_addr_list[0], he->h_length); port = atoi(pp + 1); + + yaz_log(YLOG_LOG, "HTTP address %s:%d", + "" == he->h_addr_list[0] ? he->h_addr_list[0] : "127.0.0.1" , + port); + } else { port = atoi(addr); myaddr.sin_addr.s_addr = INADDR_ANY; } + myaddr.sin_port = htons(port); if (!(p = getprotobyname("tcp"))) { @@ -810,24 +1034,31 @@ void http_init(const char *addr) abort(); if (bind(l, (struct sockaddr *) &myaddr, sizeof myaddr) < 0) + { yaz_log(YLOG_FATAL|YLOG_ERRNO, "bind"); + exit(1); + } if (listen(l, SOMAXCONN) < 0) + { yaz_log(YLOG_FATAL|YLOG_ERRNO, "listen"); + exit(1); + } c = iochan_create(l, http_accept, EVENT_INPUT | EVENT_EXCEPT); - c->next = channel_list; - channel_list = c; + pazpar2_add_channel(c); } -void http_set_proxyaddr(char *host) +void http_set_proxyaddr(char *host, char *base_url) { char *p; int port; struct hostent *he; + strcpy(myurl, base_url); strcpy(proxy_url, host); p = strchr(host, ':'); yaz_log(YLOG_DEBUG, "Proxying for %s", host); + yaz_log(YLOG_LOG, "HTTP backend %s", proxy_url); if (p) { port = atoi(p + 1); *p = '\0';