[commit] master: add support for incomplete headers to copy_msg_convert()

[Oswald B. <os...@us...>]

commit 5b9256f5dcd32e217637938d904d582d5f81c0fd
Author: Oswald Buddenhagen <os...@us...>
Date:   Tue May 3 20:01:29 2022 +0200

    add support for incomplete headers to copy_msg_convert()
    
    in the spirit of being liberal about what we accept.
    
    this completely re-structures the header processing loop, with the nice
    side effect of eliminating the gotos.
    
    REFMAIL: 87b...@cu...

 src/sync_msg_cvt.c | 118 ++++++++++++++++++++++++++++++---------------
 src/tst_msg_cvt.c  |  83 +++++++++++++++++++++++++++++--
 2 files changed, 159 insertions(+), 42 deletions(-)

diff --git a/src/sync_msg_cvt.c b/src/sync_msg_cvt.c
index 472fcf6b..1e49cf7b 100644
--- a/src/sync_msg_cvt.c
+++ b/src/sync_msg_cvt.c
@@ -44,52 +44,82 @@ copy_msg_convert( int in_cr, int out_cr, copy_vars_t *vars )
 	uint in_len = vars->data.len;
 	uint idx = 0, sbreak = 0, ebreak = 0, break2 = UINT_MAX;
 	uint lines = 0, hdr_crs = 0, bdy_crs = 0, app_cr = 0, extra = 0;
-	uint add_subj = 0;
+	uint add_subj = 0, fix_tuid = 0, fix_subj = 0, fix_hdr = 0, end_hdr = 0;
 
 	if (vars->srec) {
-	  nloop: ;
-		uint start = idx;
-		uint line_cr = 0;
-		char pc = 0;
-		while (idx < in_len) {
-			char c = in_buf[idx++];
-			if (c == '\n') {
+		for (;;) {
+			uint start = idx;
+			uint line_cr = 0;
+			uint got_line = 0;
+			char pc = 0;
+			while (idx < in_len) {
+				char c = in_buf[idx++];
+				if (c == '\n') {
+					if (pc == '\r')
+						line_cr = 1;
+					got_line = 1;
+					break;
+				}
+				pc = c;
+			}
+			if (!ebreak && starts_with_upper( in_buf + start, (int)(in_len - start), "X-TUID: ", 8 )) {
+				extra = (sbreak = start) - (ebreak = idx);
+				if (!vars->minimal)
+					break;
+				continue;
+			}
+			if (break2 == UINT_MAX && vars->minimal &&
+			    starts_with_upper( in_buf + start, (int)(in_len - start), "SUBJECT:", 8 )) {
+				break2 = start + 8;
+				if (break2 < in_len && in_buf[break2] == ' ')
+					break2++;
+			}
+			hdr_crs += line_cr;
+			if (got_line) {
+				lines++;
+				if (idx - line_cr - 1 != start)
+					continue;
+				// Empty line => end of headers
+			} else {
+				// The line is incomplete.
 				if (pc == '\r')
-					line_cr = 1;
-				if (!ebreak && starts_with_upper( in_buf + start, (int)(in_len - start), "X-TUID: ", 8 )) {
-					extra = (sbreak = start) - (ebreak = idx);
-					if (!vars->minimal)
-						goto oke;
-				} else {
-					if (break2 == UINT_MAX && vars->minimal &&
-					    starts_with_upper( in_buf + start, (int)(in_len - start), "SUBJECT:", 8 )) {
-						break2 = start + 8;
-						if (break2 < in_len && in_buf[break2] == ' ')
-							break2++;
-					}
-					lines++;
-					hdr_crs += line_cr;
+					idx--;  // For simplicity, move back before trailing CR
+				if (idx != start) {
+					// The line is non-empty, so schedule completing it
+					fix_hdr = 1;
+					// ... and put our headers after it. (It would seem easier
+					// to prepend them, as then we could avoid the fixing - but
+					// the line might be a continuation. We could also prepend
+					// it to _all_ pre-exiting headers, but then we would risk
+					// masking an (incorrectly present) leading 'From ' header.)
+					start = idx;
 				}
-				if (idx - line_cr - 1 == start) {
-					if (!ebreak)
-						sbreak = ebreak = start;
-					if (vars->minimal) {
-						in_len = idx;
-						if (break2 == UINT_MAX) {
-							break2 = start;
-							add_subj = 1;
-						}
-					}
-					goto oke;
+				end_hdr = 1;
+			}
+			if (!ebreak) {
+				sbreak = ebreak = start;
+				fix_tuid = fix_hdr;
+				fix_hdr = 0;
+			}
+			if (vars->minimal) {
+				in_len = idx;
+				if (break2 == UINT_MAX) {
+					break2 = start;
+					add_subj = 1;
+					fix_subj = fix_hdr;
+					fix_hdr = 0;
 				}
-				goto nloop;
+			} else {
+				fix_hdr = 0;
+				end_hdr = 0;
 			}
-			pc = c;
+			break;
 		}
-		free( in_buf );
-		return "has incomplete header";
-	  oke:
 		app_cr = out_cr && (!in_cr || hdr_crs);
+		if (fix_tuid || fix_subj || fix_hdr)
+			extra += app_cr + 1;
+		if (end_hdr)
+			extra += app_cr + 1;
 		extra += 8 + TUIDL + app_cr + 1;
 	}
 	if (out_cr != in_cr) {
@@ -157,6 +187,8 @@ copy_msg_convert( int in_cr, int out_cr, copy_vars_t *vars )
 		}
 		copy_msg_bytes( &out_buf, in_buf, &idx, sbreak, in_cr, out_cr );
 
+		if (fix_tuid)
+			ADD_NL();
 		memcpy( out_buf, "X-TUID: ", 8 );
 		out_buf += 8;
 		memcpy( out_buf, vars->srec->tuid, TUIDL );
@@ -170,6 +202,8 @@ copy_msg_convert( int in_cr, int out_cr, copy_vars_t *vars )
 				memcpy( out_buf, dummy_pfx, strlen(dummy_pfx) );
 				out_buf += strlen(dummy_pfx);
 			} else {
+				if (fix_subj)
+					ADD_NL();
 				memcpy( out_buf, dummy_subj, strlen(dummy_subj) );
 				out_buf += strlen(dummy_subj);
 				ADD_NL();
@@ -178,8 +212,14 @@ copy_msg_convert( int in_cr, int out_cr, copy_vars_t *vars )
 	}
 	copy_msg_bytes( &out_buf, in_buf, &idx, in_len, in_cr, out_cr );
 
-	if (vars->minimal)
+	if (vars->minimal) {
+		if (end_hdr) {
+			if (fix_hdr)
+				ADD_NL();
+			ADD_NL();
+		}
 		memcpy( out_buf, dummy_msg_buf, dummy_msg_len );
+	}
 
 	free( in_buf );
 	return NULL;
diff --git a/src/tst_msg_cvt.c b/src/tst_msg_cvt.c
index 399c5c2b..399c5653 100644
--- a/src/tst_msg_cvt.c
+++ b/src/tst_msg_cvt.c
@@ -131,10 +131,13 @@ mintests( const char *name, const char *in, const char *out, int flagged )
 }
 
 #define FROM "From: de\rvil\r\n"
-#define TO "To: me\r\n"
-#define IN_TUID "X-TUID: garbage\r\n"
+#define R_TO "To: me"
+#define TO R_TO "\r\n"
+#define R_IN_TUID "X-TUID: garbage"
+#define IN_TUID R_IN_TUID "\r\n"
 #define OUT_TUID "X-TUID: " TUID "\r\n"
-#define SUBJECT "Subject: hell\r\n"
+#define R_SUBJECT "Subject: hell"
+#define SUBJECT R_SUBJECT "\r\n"
 #define PH_SUBJECT "Subject: [placeholder] hell\r\n"
 #define NO_SUBJECT "Subject: [placeholder] (No Subject)\r\n"
 #define BODY "\r\nHi,\r\n\r\n...\r\n"
@@ -177,5 +180,79 @@ main( void )
 	scc out_tuid_from_subj_to[] = OUT_TUID FROM PH_SUBJECT TO PH_BODY;
 	mintests( "tuid / from / subject / to", in_tuid_from_subj_to, out_tuid_from_subj_to, REGULAR );
 
+
+	scc in_from_to_b1[] = FROM TO;
+	fulltests( "from / to w/o end", in_from_to_b1, in_from_to_b1, AS_IS );
+	scc out_from_to_b1[] = FROM TO OUT_TUID;
+	fulltests( "from / to w/o end", in_from_to_b1, out_from_to_b1, ADD_TUID );
+	scc in_from_tuid_to_b1[] = FROM IN_TUID TO;
+	scc out_from_tuid_to_b1[] = FROM OUT_TUID TO;
+	fulltests( "from / tuid / to w/o end", in_from_tuid_to_b1, out_from_tuid_to_b1, ADD_TUID );
+	scc in_from_to_tuid_b1[] = FROM TO IN_TUID;
+	scc out_from_to_tuid_b1[] = FROM TO OUT_TUID;
+	fulltests( "from / to / tuid w/o end", in_from_to_tuid_b1, out_from_to_tuid_b1, ADD_TUID );
+
+	mintests( "from / to w/o end", in_from_to_b1, out_from_to_ph, REGULAR );
+	mintests( "from / tuid / to w/o end", in_from_tuid_to_b1, out_from_tuid_to_ph, REGULAR );
+	scc in_from_subj_to_b1[] = FROM SUBJECT TO;
+	mintests( "from / subject / to w/o end", in_from_subj_to_b1, out_from_subj_to, REGULAR );
+	scc in_from_subj_tuid_to_b1[] = FROM SUBJECT IN_TUID TO;
+	mintests( "from / subject / tuid / to w/o end", in_from_subj_tuid_to_b1, out_from_subj_tuid_to, REGULAR );
+	scc in_from_subj_to_tuid_b1[] = FROM SUBJECT TO IN_TUID;
+	scc out_from_subj_to_tuid_b1[] = FROM PH_SUBJECT TO OUT_TUID PH_BODY;
+	mintests( "from / subject / to / tuid w/o end", in_from_subj_to_tuid_b1, out_from_subj_to_tuid_b1, REGULAR );
+	scc in_from_tuid_subj_to_b1[] = FROM IN_TUID SUBJECT TO;
+	mintests( "from / tuid / subject / to w/o end", in_from_tuid_subj_to_b1, out_from_tuid_subj_to, REGULAR );
+	scc in_from_tuid_to_subj_b1[] = FROM IN_TUID TO SUBJECT;
+	scc out_from_tuid_to_subj_b1[] = FROM OUT_TUID TO PH_SUBJECT PH_BODY;
+	mintests( "from / tuid / to / subject w/o end", in_from_tuid_to_subj_b1, out_from_tuid_to_subj_b1, REGULAR );
+
+
+	scc in_from_to_b2[] = FROM R_TO "\r";
+	fulltests( "from / to w/o lf", in_from_to_b2, in_from_to_b2, AS_IS );
+	scc out_from_to_b2[] = FROM TO OUT_TUID "\r";
+	fulltests( "from / to w/o lf", in_from_to_b2, out_from_to_b2, ADD_TUID );
+	scc in_from_tuid_to_b2[] = FROM IN_TUID R_TO "\r";
+	scc out_from_tuid_to_b2[] = FROM OUT_TUID R_TO "\r";
+	fulltests( "from / tuid / to w/o lf", in_from_tuid_to_b2, out_from_tuid_to_b2, ADD_TUID );
+	scc in_from_to_tuid_b2[] = FROM TO R_IN_TUID "\r";
+	fulltests( "from / to / tuid w/o lf", in_from_to_tuid_b2, out_from_to_tuid_b1, ADD_TUID );
+
+	mintests( "from / to w/o lf", in_from_to_b2, out_from_to_ph, REGULAR );
+	mintests( "from / tuid / to w/o lf", in_from_tuid_to_b2, out_from_tuid_to_ph, REGULAR );
+	scc in_from_subj_to_b2[] = FROM SUBJECT R_TO "\r";
+	mintests( "from / subject / to w/o lf", in_from_subj_to_b2, out_from_subj_to, REGULAR );
+	scc in_from_subj_tuid_to_b2[] = FROM SUBJECT IN_TUID R_TO "\r";
+	mintests( "from / subject / tuid / to w/o lf", in_from_subj_tuid_to_b2, out_from_subj_tuid_to, REGULAR );
+	scc in_from_subj_to_tuid_b2[] = FROM SUBJECT TO R_IN_TUID "\r";
+	mintests( "from / subject / to / tuid w/o lf", in_from_subj_to_tuid_b2, out_from_subj_to_tuid_b1, REGULAR );
+	scc in_from_tuid_subj_to_b2[] = FROM IN_TUID SUBJECT R_TO "\r";
+	mintests( "from / tuid / subject / to w/o lf", in_from_tuid_subj_to_b2, out_from_tuid_subj_to, REGULAR );
+	scc in_from_tuid_to_subj_b2[] = FROM IN_TUID TO R_SUBJECT "\r";
+	mintests( "from / tuid / to / subject w/o lf", in_from_tuid_to_subj_b2, out_from_tuid_to_subj_b1, REGULAR );
+
+
+	scc in_from_to_b3[] = FROM R_TO;
+	fulltests( "from / to w/o crlf", in_from_to_b3, in_from_to_b3, AS_IS );
+	fulltests( "from / to w/o crlf", in_from_to_b3, out_from_to_b1, ADD_TUID );
+	scc in_from_tuid_to_b3[] = FROM IN_TUID R_TO;
+	scc out_from_tuid_to_b3[] = FROM OUT_TUID R_TO;
+	fulltests( "from / tuid / to w/o crlf", in_from_tuid_to_b3, out_from_tuid_to_b3, ADD_TUID );
+	scc in_from_to_tuid_b3[] = FROM TO R_IN_TUID;
+	fulltests( "from / to / tuid w/o crlf", in_from_to_tuid_b3, out_from_to_tuid_b1, ADD_TUID );
+
+	mintests( "from / to w/o crlf", in_from_to_b3, out_from_to_ph, REGULAR );
+	mintests( "from / tuid / to w/o crlf", in_from_tuid_to_b3, out_from_tuid_to_ph, REGULAR );
+	scc in_from_subj_to_b3[] = FROM SUBJECT R_TO;
+	mintests( "from / subject / to w/o crlf", in_from_subj_to_b3, out_from_subj_to, REGULAR );
+	scc in_from_subj_tuid_to_b3[] = FROM SUBJECT IN_TUID R_TO;
+	mintests( "from / subject / tuid / to w/o crlf", in_from_subj_tuid_to_b3, out_from_subj_tuid_to, REGULAR );
+	scc in_from_subj_to_tuid_b3[] = FROM SUBJECT TO R_IN_TUID;
+	mintests( "from / subject / to / tuid w/o crlf", in_from_subj_to_tuid_b3, out_from_subj_to_tuid_b1, REGULAR );
+	scc in_from_tuid_subj_to_b3[] = FROM IN_TUID SUBJECT R_TO;
+	mintests( "from / tuid / subject / to w/o crlf", in_from_tuid_subj_to_b3, out_from_tuid_subj_to, REGULAR );
+	scc in_from_tuid_to_subj_b3[] = FROM IN_TUID TO R_SUBJECT;
+	mintests( "from / tuid / to / subject w/o crlf", in_from_tuid_to_subj_b3, out_from_tuid_to_subj_b1, REGULAR );
+
 	return 0;
 }