repo: update readme

also remove now defunct functions +
rewrite state machine

.gitignore

@@ -1,3 +1,4 @@
 *.o
 proxy
 compile_commands.json
+.cache

README

@@ -4,12 +4,17 @@
 
     Proxy C library for HTTP/s protocol 
 
+    NOTE: I have yet to turn this piece of clutter into a library. 
+
     RFC: datatracker.ietf.org/doc/html/rfc1945
     RES: cs.princeton.edu/courses/archive/spr13/cos461/assignments-proxy.html
     RES: Beej's Guide to Network Programming - Using Internet Sockets
     RES: en.wikipedia.org/wiki/Proxy_server
     RES: TCP/IP sockets in C - Practical guide for programmers 2nd edition
 
+    RES: tinyproxy.github.io/
+    RES: github.com/nginx/nginx
+
 TECHNICALS 
 
     * Parsing is carried by a dedicated sub-library which I wrote named parslib.
@@ -18,6 +23,10 @@ TECHNICALS
     no matter the amount of strings you want to compare it against (I am indeed 
     aware it is nonetheless worse than hashtables - maybe one day I will implement 
     those on my own as well).
+    * After the connection is established with the upstream server, a relay loop 
+    is started in which the client data is relayed to the server's socket and 
+    server data is relayed to client's socket. This is highly inspired from 
+    tinyproxy.
 
 TASKS
 
@@ -26,15 +35,16 @@ TASKS
         [DONE] Implement client message parsing
         [DONE] Implement server message parsing
     [DONE] Add loose string checking for headers
+    [DONE] Relaying mechanism
     [DOING] Cover all possible body segmentation standards
         [DONE] Cover Content-Length
-        [DOING] Cover "chunked transfer encoding"
-        [DOING] Cover "compress transfer encoding"
-        [DOING] Cover "deflate transfer encoding"
-        [DOING] Cover "gzip/x-gzip transfer encoding"
+        [DONE] Cover "chunked transfer encoding"
+        [TODO] Cover "compress transfer encoding"
+        [TODO] Cover "deflate transfer encoding"
+        [TODO] Cover "gzip/x-gzip transfer encoding"
     [ACTIVELY DOING] More testing, debugging, fixing
     [ACTIVELY DOING] Verify and search for memory leaks
-    [TODO] Implement HTTPS with OpenSSL, LibreSSL, or BearSSL
+    [TODO] Implement HTTPS
     [TODO] Caching?
 
 COMMITS

proxlib.c

@@ -7,12 +7,14 @@
 #include <arpa/inet.h>
 #include <netinet/in.h>
 #include <netdb.h>
+#include <poll.h>
 #include "proxlib.h"
 #include "parslib/parslib.h"
 
 int on = 1; 
-int debug = 3;
-int statem; 
+int debug = 1;
+int statem = 0; 
+int err = 0;
 
 #define SEGMENT_LEN 512
 #define MAX_BUFF_LEN 128 * 1024
@@ -92,181 +94,107 @@ int read_line(int fd,
     return 0;
 }
 
-int parse_line(char *line, int line_count) {
+int pull_content_length(int fd, int len, int *msgbuff_len, char **msgbuff) {
     int ret = 0; 
-
-    return ret;
-}
-
-void do_err(void) {
-    int statem_code = statem & (~STATEM_ERR);
-    fprintf(stderr, "[%d,%d,%d] Errored out!\n", statem, statem_code,
-            STATEM_ERR);
-}
-
-int do_fwd_clt(struct conn *conn) {
-    int bytes = 0;
-    int ret = 0;
-    while (bytes < conn->srvbuff_len) {
-        ret = write(conn->cltfd, conn->srvbuff+bytes, conn->srvbuff_len-bytes);
-        if (ret < 0)
-            return -1;
-
-        bytes += ret;
-    }
-
-    return 0;
-}
-
-int do_rcv_srv(struct conn *conn) {
-    int ret = 0;
-    char *line = NULL; 
-    char *msgbuff = NULL;
-    int line_len = 0; 
-    int msgbuff_len = 0;
-
-    // response line 
-    ret = read_line(conn->srvfd, &line_len, &line, &msgbuff_len, &msgbuff);
-    if (ret < 0) {
-        fprintf(stderr, "Failed receiving response line from upstream\n");
-        return -1;
-    }
-
-    if (debug == 1) {
-        fprintf(stdout, "debug - [upstream] received line: %s\n", line);
-    }
-
-    ret = parestitl(line, line_len, &(conn->srvres.titl));    
-    if (ret < 0) {
-        fprintf(stderr, "Failed parsing response line\n");
-        return -1;
-    }
-
-    if (debug == 1) {
-        fprintf(stdout, "debug - [upstream] parsed response line\n");
-    }
-
-    free(line);
-
-    // headers
-    int next_header = 1; 
-    while (next_header) {
-        ret = read_line(conn->srvfd, &line_len, &line, &msgbuff_len, &msgbuff);
-        if (ret < 0) {
-            fprintf(stderr, "Failed receiving header line\n");
-            return -1;
-        }
-
-        if (line_len == 0) {
-            if (debug == 1) {
-                fprintf(stdout, "debug - [upstream] reached end of headers\n");
-            }
-            next_header = 0;
-            continue;
-        }
-
-        if (debug == 1) {
-            fprintf(stdout, "debug - [upstream] received line: %s\n", line);
-        }
-
-        ret = parshfield(line, line_len, conn->srvres.hentries);
-        if (ret < 0) {
-            fprintf(stderr, "Failed parsing header field\n");
-            return -1;
-        }
-
-        if (debug == 1) {
-            fprintf(stdout, "debug - parsed header field\n");
-        }
-
-        free(line);
-    }
-
-    // body
-body:
-    struct httpares *res = &conn->srvres;
-    struct point *content_length_entry = &res->hentries[header_content_length];
-    if (content_length_entry->er == NULL) {
-        fprintf(stderr, "[upstream] no content length header\n");
-        return -1;
-    }
-
-    int content_length = 0;
-    ret = stoin(content_length_entry->er, content_length_entry->len, &content_length);
-    if (ret < 0) {
-        fprintf(stderr, "[upstream] failed parsing content length header\n");
-        return -1;
-    }
-
-    line_len = content_length;
-    line = (char *) calloc(1, line_len);
+    int line_len = len;
+    char *line = (char *) calloc(1, line_len);
     if (!line) {
-        fprintf(stderr, "[upstream] not enough dynamic memory\n");
-        return -1;
+        return ERR_MEM;
     }
 
     int bytes = 0;
     do {
-        ret = recv(conn->srvfd, line+bytes, line_len-bytes, MSG_WAITALL);
+        ret = recv(fd, line+bytes, line_len-bytes, MSG_WAITALL);
         if (ret < 0) {
-            fprintf(stderr, "[upstream] failed reading body from response\n");
-            return -1;
+            return ERR_MEM;
         }
         bytes += ret;
     } while (bytes < line_len);
 
-    msgbuff = (char *) realloc(msgbuff, msgbuff_len+line_len);
-    if (!msgbuff) {
-        fprintf(stderr, "[upstream] not enough dynamic memory\n");
-        return -1;
+    *msgbuff = (char *) realloc(*msgbuff, *msgbuff_len+line_len);
+    if (!*msgbuff) {
+        return ERR_MEM;
     }
 
-    memcpy(msgbuff+msgbuff_len, line, line_len);
-    msgbuff_len += line_len;
+    memcpy(*msgbuff+*msgbuff_len, line, line_len);
+    *msgbuff_len += line_len;
 
-    if (debug <= 2) {
-        fprintf(stdout, "------------------------------\n");
-        fprintf(stdout, "debug - [upstream] received body %d: %.*s\n", line_len, line_len, line);
-        fprintf(stdout, "------------------------------\n");
-    }
+    return 0;
+}
 
-    if (debug <= 2) {
-        fprintf(stdout, "printing parsed response\n");
-        printfpares(&conn->srvres);
-    }
+int pull_chunked_encoding(int fd, int *msgbuff_len, char **msgbuff) {
+	int ret = 0; 
+	char *line = NULL;
+	int line_len = 0; 
 
-    conn->srvbuff = msgbuff;
-    conn->srvbuff_len = msgbuff_len;
+	while (1) {
+		ret = read_line(fd, &line_len, &line, msgbuff_len, msgbuff);
+		if (ret < 0) {
+			fprintf(stderr, "Failed receiving chunked body from upstream\n");
+			return -1;
+		}
 
-    return 0; 
+		line_len = strtol(line, (char **) 0, 16); 
+		if (!line_len) {
+			break;
+		}
+
+		line_len += 2;
+
+		free(line);
+
+		line = (char *) calloc(1, line_len);
+		if (!line) {
+			fprintf(stderr, "Not enough dynamic memory\n");
+			return -1;
+		}
+
+		int bytes = 0; 
+		do {
+			ret = recv(fd, line+bytes, line_len-bytes, MSG_WAITALL); 
+			if (ret < 0) {
+				fprintf(stderr, "Failed reading respones body from server\n");
+				return -1;
+			}
+			bytes += ret;
+		} while (bytes < line_len); 
+
+		*msgbuff = (char *) realloc(*msgbuff, *msgbuff_len+line_len);
+		if (!msgbuff) {
+			fprintf(stderr, "Not enough dynamic memory\n");
+			return -1;
+		}
+
+		memcpy(*msgbuff+*msgbuff_len, line, line_len);
+		*msgbuff_len += line_len;
+		free(line);
+	}
+    return 0;
+
+}
+
+void do_err(void) {
+    fprintf(stderr, "failed with error code %d\n", err);
 }
 
 int do_con_srv(struct conn *conn) {
+    statem = state_con_srv;
+
     int ret = 0;
     struct httpareq *req = &conn->cltreq;
     struct point *host = &req->hentries[header_host];
     if (host->er == NULL) {
-        if (debug <= 2) {
-            fprintf(stderr, "debug - request does not have HOST header\n");
-        }
-        goto _exit;
+        return ERR_PARS;
     }
 
     struct hostinfo *info = (struct hostinfo *) calloc(1, sizeof(struct hostinfo));
     if (!info) {
-        goto _exit;
+        return ERR_MEM;
     }
 
     ret = pahostinfo(host->er, host->len, info);
     if (ret < 0) {
-        if (debug <= 2) {
-            fprintf(stderr, "Failed parsing upstream host header\n");
-        }
-        goto _exit_hostinfo;
-    }
-
-    if (debug <= 2) {
-        fprintf(stdout, "Establishing connection with upstream: %.*s : %.*s\n", info->hostname_len, info->hostname, info->service_len, info->service);
+        return ERR_PARS;
     }
 
     struct addrinfo hints; 
@@ -278,45 +206,37 @@ int do_con_srv(struct conn *conn) {
 
     ret = getaddrinfo(info->hostname, info->service, &hints, &res);
     if (ret < 0) {
-        goto _exit_hostinfo;
+        free(info->hostname);
+        free(info->service);
+        free(info);
+        return ERR_PARS;
     }
 
     ret = conn->srvfd = socket(res->ai_family, res->ai_socktype,
             res->ai_protocol);
     if (ret < 0) {
-        goto _exit_getaddrinfo;
+        freeaddrinfo(res);
+        free(info->hostname);
+        free(info->service);
+        free(info);
+        return ERR_PARS;
     }
 
     ret = connect(conn->srvfd, res->ai_addr, res->ai_addrlen);
     if (ret < 0) {
-        goto _exit_getaddrinfo;
+        freeaddrinfo(res);
+        free(info->hostname);
+        free(info->service);
+        free(info);
+        return ERR_PARS;
     }
 
-_exit_getaddrinfo:
-    freeaddrinfo(res);
-_exit_hostinfo:
-    free(info->hostname);
-    free(info->service);
-    free(info);
-_exit:
     return ret;
 }
 
-int do_fwd_srv(struct conn *conn) {
-    int bytes = 0;
-    int ret = 0;
-    while (bytes < conn->cltbuff_len) {
-        ret = write(conn->srvfd, conn->cltbuff+bytes, conn->cltbuff_len-bytes);
-        if (ret < 0)
-            return -1;
-
-        bytes += ret;
-    }
-
-    return 0;
-}
-
 int do_rcv_clt(struct conn *conn) {
+    statem = state_rcv_clt;
+
     int ret = 0;
     char *line = NULL; 
     char *msgbuff = NULL;
@@ -326,22 +246,12 @@ int do_rcv_clt(struct conn *conn) {
     // request line 
     ret = read_line(conn->cltfd, &line_len, &line, &msgbuff_len, &msgbuff);
     if (ret < 0) {
-        fprintf(stderr, "Failed receiving request line\n");
-        return -1;
-    }
-
-    if (debug == 1) {
-        fprintf(stdout, "debug - received line: %s\n", line);
+        return ERR_RECV;
     }
 
     ret = pareqtitl(line, line_len, &(conn->cltreq.titl));    
     if (ret < 0) {
-        fprintf(stderr, "Failed parsing request line\n");
-        return -1;
-    }
-
-    if (debug == 1) {
-        fprintf(stdout, "debug - parsed request line\n");
+        return ERR_PARSTITLE;
     }
 
     free(line);
@@ -351,94 +261,170 @@ int do_rcv_clt(struct conn *conn) {
     while (next_header) {
         ret = read_line(conn->cltfd, &line_len, &line, &msgbuff_len, &msgbuff);
         if (ret < 0) {
-            fprintf(stderr, "Failed receiving header line\n");
-            return -1;
+            return ERR_RECV;
         }
 
         if (line_len == 0) {
-            if (debug == 1) {
-                fprintf(stdout, "debug - reached end of headers\n");
-            }
             next_header = 0;
             continue;
         }
 
-        if (debug == 1) {
-            fprintf(stdout, "debug - received line: %s\n", line);
-        }
-
         ret = parshfield(line, line_len, conn->cltreq.hentries);
         if (ret < 0) {
-            fprintf(stderr, "Failed parsing header field\n");
-            return -1;
-        }
-
-        if (debug == 1) {
-            fprintf(stdout, "debug - parsed header field\n");
+            return ERR_PARSHEADER;
         }
 
         free(line);
     }
 
     // body
-    // TODO
+    struct httpareq *req = &conn->cltreq;
+    struct point *content_length_entry = &req->hentries[header_content_length];
+    struct point *transfer_encoding_entry = &req->hentries[header_transfer_encoding];
+    if (content_length_entry->er) {
+        int content_length = 0;
 
-    if (debug <= 2) {
-        fprintf(stdout, "printing parsed request\n");
-        printfpareq(&conn->cltreq);
-    }
+        ret = stoin(content_length_entry->er, content_length_entry->len, &content_length);
+        if (ret < 0) {
+            return ERR_PARS;
+        }
 
+        ret = pull_content_length(conn->srvfd, content_length, &msgbuff_len, &msgbuff);
+        if (ret < 0) {
+            return ERR_RECV;
+        }
+    } else if (transfer_encoding_entry->er && strcmp(transfer_encoding_entry->er, "chunked") == 0) {
+	    ret = pull_chunked_encoding(conn->srvfd, &msgbuff_len, &msgbuff);
+	    if (ret < 0) {
+            return ERR_RECV;
+	    }
+    } 
+    
     conn->cltbuff = msgbuff;
     conn->cltbuff_len = msgbuff_len;
 
     return 0; 
 }
 
-void do_clear(struct conn *conn) {
-    statem = STATEM_RCV_CLT;
-    frepareq(&conn->cltreq);
-    frepares(&conn->srvres);
-    free(conn->cltbuff);
-    free(conn->srvbuff);
-} 
+int read_buffer(int fd, char **buff, int *len) {
+    char *tmp = (char *) malloc(RELAY_BUFFER_SIZE);
+    if (!tmp) {
+        return ERR_MEM;
+    }
+
+    memset(tmp, 0, RELAY_BUFFER_SIZE);
+    int bytes = recv(fd, tmp, RELAY_BUFFER_SIZE, 0);
+    if (bytes <= 0) {
+        free(tmp);
+        return ERR_RECV;
+    }
+
+    *buff = realloc(*buff, *len+bytes);
+    if (!*buff) {
+        free(tmp);
+        return ERR_MEM;
+    }
+
+    memcpy(*buff+*len, tmp, bytes);
+    *len += bytes;
+
+    return 0;
+}
+
+int write_buffer(int fd, char **buff, int *len) {
+    if (*len <= 0) {
+        *len = 0;
+        return 0;
+    }
+
+    int writen = send(fd, *buff, *len, 0); 
+    if (writen < 0) {
+        return ERR_SEND;
+    }
+
+    char *trunc = (char *) malloc(*len-writen);
+    if (!trunc) {
+        return ERR_MEM;
+    }
+
+    memcpy(trunc, *buff+writen, *len-writen);
+    
+    char *tofree = *buff; // FIXME: any better solution? 
+    *buff = trunc;
+    *len -= writen;
+    free(tofree);
+
+    return 0;
+}
 
 void do_statem(struct conn *conn) {
-    int ret = 0;
+    int ret = 0; 
 
-    for (int counter = 0; counter < MAX_BOUND; counter++) {
-        switch (statem & (~STATEM_ERR)) {
-        case STATEM_RCV_CLT:
-            ret = do_rcv_clt(conn);
-            break;
-        case STATEM_CON_SRV:
-            ret = do_con_srv(conn);
-            break;
-        case STATEM_FWD_SRV: 
-            ret = do_fwd_srv(conn);
-            break;
-        case STATEM_RCV_SRV: 
-            ret = do_rcv_srv(conn);
-            break;
-        case STATEM_FWD_CLT: 
-            ret = do_fwd_clt(conn);
-            break;
-        }
-
-        if (ret < 0) 
-            statem |= STATEM_ERR;
-
-        if (statem & STATEM_ERR) {
-            do_err();
-            break;
-        }
-
-        if (statem & STATEM_FWD_CLT) {
-            do_clear(conn);
-            continue;
-        }
-
-        statem <<= 1;
+    ret = do_rcv_clt(conn); 
+    if (ret < 0) {
+        err = ret;
+        do_err();
     }
+
+    ret = do_con_srv(conn);
+    if (ret < 0) {
+        err = ret;
+        do_err();
+    }
+
+    // relay the data between the two sockets until the end of time
+    ssize_t bytes_received;
+	struct pollfd fds[2];
+    for (;;) {
+	    memset(fds, 0, 2*sizeof(struct pollfd));
+        fds[0].fd = conn->cltfd;
+        fds[1].fd = conn->srvfd;
+
+        if (conn->srvbuff_len > 0) {
+            fds[0].events |= POLLOUT;
+        }
+        if (conn->cltbuff_len > 0) {
+            fds[1].events |= POLLOUT;
+        }
+        if (!conn->srvbuff_len) {
+            fds[1].events |= POLLIN;
+        }
+        if (!conn->cltbuff_len) {
+            fds[0].events |= POLLIN;
+        }
+
+        ret = poll(fds, 2, 1000);
+
+        if (fds[1].revents & POLLOUT) {
+            ret = write_buffer(conn->srvfd, &conn->cltbuff, &conn->cltbuff_len);
+        }
+        if (fds[1].revents & POLLIN) {
+            ret = read_buffer(conn->srvfd, &conn->srvbuff, &conn->srvbuff_len);
+        }
+        if (fds[0].revents & POLLIN) {
+            ret = read_buffer(conn->cltfd, &conn->cltbuff, &conn->cltbuff_len);
+        }
+        if (fds[0].revents & POLLOUT) {
+            ret = write_buffer(conn->cltfd, &conn->srvbuff, &conn->srvbuff_len);
+        }
+        if (fds[0].revents & POLLHUP) {
+            break;
+        }
+        if (ret < 0) {
+            break;
+        }
+    }
+
+    if (conn->cltbuff_len > 0) {
+        write_buffer(conn->srvfd, &conn->cltbuff, &conn->cltbuff_len);
+    }
+    if (conn->srvbuff_len > 0) {
+        write_buffer(conn->cltfd, &conn->srvbuff, &conn->srvbuff_len);
+    }
+
+    close(conn->cltfd);
+    close(conn->srvfd);
+    exit(0); // child die
 }
 
 int do_srv(void) {
@@ -476,9 +462,8 @@ int do_srv(void) {
 		return -1;
 	}
 
-	fprintf(stdout, "Listening on port %d\n", PROXY_PORT);
-
 	for (;;) {
+		fprintf(stdout, "listening for sockets\n");
 		struct sockaddr_in new_clt_addr; 
         socklen_t new_clt_addr_len= sizeof(new_clt_addr);
 		int new_clt_sock; 
@@ -490,6 +475,7 @@ int do_srv(void) {
                             "client\n");	
 			return -1;
 		}
+		fprintf(stdout, "accepted new client socket\n");
 
         ret = fork();
         if (ret < 0) {
@@ -499,12 +485,11 @@ int do_srv(void) {
         } 
 
         if (ret > 0) {
-            fprintf(stdout, "[PROGRAM] Successfully forked a new child process"
-                            " with PID %d\n", ret);
+            fprintf(stdout, "+new request process:%d(pid)\n", ret);
             continue;
         }
 
-        // child 
+        // request process
         struct conn *conn = (struct conn *) calloc(1, sizeof(struct conn));
         if (!conn) {
             fprintf(stderr, "Not enough dynamic memory to establish connection\n");
@@ -512,14 +497,9 @@ int do_srv(void) {
         }
 
         conn->cltfd = new_clt_sock;
-        statem = STATEM_RCV_CLT;
+        statem = state_rcv_clt;
         do_statem(conn);
         free(conn);
-
-        if (debug == 1) {
-            fprintf(stdout, "Finished proxying client\n");
-        }
-
         return 0;
 	}
 
@@ -533,7 +513,10 @@ int main(int argc, char *argv[]) {
         return -1;
     } 
 
-    return do_srv();
+    ret = do_srv();
+    if (ret < 0) {
+        return -1;
+    }
 
     fretres();
 }

proxlib.h

@@ -4,18 +4,41 @@
 #include "parslib/parslib.h"
 
 #ifndef MAX_BOUND
-#define MAX_BOUND 10000
+#define MAX_BOUND 100000000000
 #endif
 
 #define PROXY_PORT 2020
 #define PROXY_CONN 20
+#define RELAY_BUFFER_SIZE 1024*2
+#define RELAY_POLL_TIMEOUT 1000
 
-#define STATEM_RCV_CLT 0b00000001
-#define STATEM_CON_SRV 0b00000010
-#define STATEM_FWD_SRV 0b00000100
-#define STATEM_RCV_SRV 0b00001000
-#define STATEM_FWD_CLT 0b00010000
-#define STATEM_ERR     0b00100000
+enum states {
+    state_rcv_clt = 0,
+    state_con_srv,
+    state_fwd_srv,
+    state_rcv_srv,
+    state_fwd_clt,
+    state_ok
+};
+
+#define ERR_GENERIC     -1 
+#define ERR_MEM         -2
+#define ERR_RECV        -3
+#define ERR_SEND        -4
+#define ERR_PARS        -5
+#define ERR_PARSTITLE   -6
+#define ERR_PARSHEADER  -7
+#define ERR_SUPPORT     -8
+#define ERR_TIMEOUT     -9
+
+char *states_str[] = {
+    "state_rcv_clt",
+    "state_con_srv",
+    "state_fwd_srv",
+    "state_rcv_srv",
+    "state_fwd_clt",
+    "state_ok"
+};
 
 struct conn {
     int cltfd;