Issue 408: Fix escaped newline parsing

diff --git a/libarchive/archive_read_support_format_mtree.c b/libarchive/archive_read_support_format_mtree.c
index b58f592a3f29b962ebe9fa64ee7f6719f093dff2..64d5e6734ac0f05acd0198db491899fd46fd7925 100644 (file)
--- a/libarchive/archive_read_support_format_mtree.c
+++ b/libarchive/archive_read_support_format_mtree.c
@@ -1775,6 +1775,10 @@ parse_escapes(char *src, struct mtree_entry *mentry)
                                c = '\v';
                                ++src;
                                break;
+                       case '\\':
+                               c = '\\';
+                               ++src;
+                               break;
                        }
                }
                *dest++ = c;
@@ -1919,8 +1923,7 @@ readline(struct archive_read *a, struct mtree *mtree, char **start,
        ssize_t total_size = 0;
        ssize_t find_off = 0;
        const void *t;
-       const char *s;
-       void *p;
+       void *nl;
        char *u;
 
        /* Accumulate line in a line buffer. */
@@ -1931,11 +1934,10 @@ readline(struct archive_read *a, struct mtree *mtree, char **start,
                        return (0);
                if (bytes_read < 0)
                        return (ARCHIVE_FATAL);
-               s = t;  /* Start of line? */
-               p = memchr(t, '\n', bytes_read);
-               /* If we found '\n', trim the read. */
-               if (p != NULL) {
-                       bytes_read = 1 + ((const char *)p) - s;
+               nl = memchr(t, '\n', bytes_read);
+               /* If we found '\n', trim the read to end exactly there. */
+               if (nl != NULL) {
+                       bytes_read = ((const char *)nl) - ((const char *)t) + 1;
                }
                if (total_size + bytes_read + 1 > limit) {
                        archive_set_error(&a->archive,
@@ -1949,38 +1951,34 @@ readline(struct archive_read *a, struct mtree *mtree, char **start,
                            "Can't allocate working buffer");
                        return (ARCHIVE_FATAL);
                }
+               /* Append new bytes to string. */
                memcpy(mtree->line.s + total_size, t, bytes_read);
                __archive_read_consume(a, bytes_read);
                total_size += bytes_read;
-               /* Null terminate. */
                mtree->line.s[total_size] = '\0';
-               /* If we found an unescaped '\n', clean up and return. */
+
                for (u = mtree->line.s + find_off; *u; ++u) {
                        if (u[0] == '\n') {
+                               /* Ends with unescaped newline. */
                                *start = mtree->line.s;
                                return total_size;
-                       }
-                       if (u[0] == '#') {
-                               if (p == NULL)
+                       } else if (u[0] == '#') {
+                               /* Ends with comment sequence #...\n */
+                               if (nl == NULL) {
+                                       /* But we've not found the \n yet */
                                        break;
-                               *start = mtree->line.s;
-                               return total_size;
-                       }
-                       if (u[0] != '\\')
-                               continue;
-                       if (u[1] == '\\') {
-                               ++u;
-                               continue;
-                       }
-                       if (u[1] == '\n') {
-                               memmove(u, u + 1,
-                                   total_size - (u - mtree->line.s) + 1);
-                               --total_size;
-                               ++u;
-                               break;
+                               }
+                       } else if (u[0] == '\\') {
+                               if (u[1] == '\n') {
+                                       /* Trim escaped newline. */
+                                       total_size -= 2;
+                                       mtree->line.s[total_size] = '\0';
+                                       break;
+                               } else if (u[1] != '\0') {
+                                       /* Skip the two-char escape sequence */
+                                       ++u;
+                               }
                        }
-                       if (u[1] == '\0')
-                               break;
                }
                find_off = u - mtree->line.s;
        }

diff --git a/libarchive/test/test_read_format_mtree.c b/libarchive/test/test_read_format_mtree.c
index f96529daf7a9c2292e83a5bf6ab051ccfb7bc1fa..a8342f55790e12de08cef03b60c747c457687a5e 100644 (file)
--- a/libarchive/test/test_read_format_mtree.c
+++ b/libarchive/test/test_read_format_mtree.c
@@ -133,6 +133,12 @@ test_read_format_mtree1(void)
        assertEqualInt(archive_entry_is_encrypted(ae), 0);
        assertEqualIntA(a, archive_read_has_encrypted_entries(a), ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED);
 
+       assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
+       assertEqualString(archive_entry_pathname(ae), "dir2/dir3b/filename\\with_esc\b\t\fapes");
+       assertEqualInt(archive_entry_filetype(ae), AE_IFREG);
+       assertEqualInt(archive_entry_is_encrypted(ae), 0);
+       assertEqualIntA(a, archive_read_has_encrypted_entries(a), ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED);
+
        assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
        assertEqualString(archive_entry_pathname(ae), "notindir");
        assertEqualInt(archive_entry_is_encrypted(ae), 0);
@@ -191,7 +197,7 @@ test_read_format_mtree1(void)
        assertEqualIntA(a, archive_read_has_encrypted_entries(a), ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED);
 
        assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae));
-       assertEqualInt(19, archive_file_count(a));
+       assertEqualInt(20, archive_file_count(a));
        assertEqualInt(ARCHIVE_OK, archive_read_close(a));
        assertEqualInt(ARCHIVE_OK, archive_read_free(a));
 }

diff --git a/libarchive/test/test_read_format_mtree.mtree.uu b/libarchive/test/test_read_format_mtree.mtree.uu
index a0dff18e442eff349653192ce928f7d6143814c0..f1c9d60306986cc7d396566eb52b992192514ab2 100644 (file)
--- a/libarchive/test/test_read_format_mtree.mtree.uu
+++ b/libarchive/test/test_read_format_mtree.mtree.uu
@@ -5,14 +5,16 @@ M92!U:60],3@*("XN"F9I;&5<,#0P=VET:%PP-#!S<&%C92!T>7!E/69I;&4*
 M9&ER,B!T>7!E/61I<@H@9&ER,V$@='EP93UD:7(*("!I;F1I<C-A('1Y<&4]
 M9FEL90ID:7(R+V9U;&QI;F1I<C(@='EP93UF:6QE(&UO9&4],#<W-PH@("XN
 M"B!I;F1I<C(@='EP93UF:6QE"B!D:7(S8B!T>7!E/61I<@H@(&EN9&ER,V(@
-M='EP93UF:6QE"B`@+BX*("XN"FYO=&EN9&ER('1Y<&4]9FEL90ID:7(R+V9U
-M;&QI;F1I<C(@;6]D93TP-C0T"F1I<C(O96UP='EF:6QE('1Y<&4]9FEL92!S
-M:7IE/3!X,`ID:7(R+W-M86QL9FEL92!T>7!E/69I;&4@<VEZ93TP,#`P,#`P
-M,#`P,#`P,#`P,#`P,#`P,#`P,#`P,#$*9&ER,B]T;V]S;6%L;&9I;&4@='EP
-M93UF:6QE('-I>F4]+3$*9&ER,B]B:6=F:6QE('1Y<&4]9FEL92!S:7IE/3DR
-M,C,S-S(P,S8X-30W-S4X,#<*9&ER,B]T;V]B:6=F:6QE('1Y<&4]9FEL92!S
-M:7IE/3DR,C,S-S(P,S8X-30W-S4X,#@*9&ER,B]V97)Y;VQD9FEL92!T>7!E
-M/69I;&4@=&EM93TM.3(R,S,W,C`S-C@U-#<W-3@P.`ID:7(R+W1O;V]L9&9I
-H;&4@='EP93UF:6QE('1I;64]+3DR,C,S-S(P,S8X-30W-S4X,#D*"@``
+M='EP93UF:6QE"B`@9FEL96YA;65<7'=I=&A<"E]E<V-<8EQT7&9A<&5S('1Y
+M<&4]9FEL92!M;V1E/3`T-#0@<VEZ93TP"B`@+BX*("XN"FYO=&EN9&ER('1Y
+M<&4]9FEL90ID:7(R+V9U;&QI;F1I<C(@;6]D93TP-C0T"F1I<C(O96UP='EF
+M:6QE('1Y<&4]9FEL92!S:7IE/3!X,`ID:7(R+W-M86QL9FEL92!T>7!E/69I
+M;&4@<VEZ93TP,#`P,#`P,#`P,#`P,#`P,#`P,#`P,#`P,#`P,#$*9&ER,B]T
+M;V]S;6%L;&9I;&4@='EP93UF:6QE('-I>F4]+3$*9&ER,B]B:6=F:6QE('1Y
+M<&4]9FEL92!S:7IE/3DR,C,S-S(P,S8X-30W-S4X,#<*9&ER,B]T;V]B:6=F
+M:6QE('1Y<&4]9FEL92!S:7IE/3DR,C,S-S(P,S8X-30W-S4X,#@*9&ER,B]V
+M97)Y;VQD9FEL92!T>7!E/69I;&4@=&EM93TM.3(R,S,W,C`S-C@U-#<W-3@P
+M.`ID:7(R+W1O;V]L9&9I;&4@='EP93UF:6QE('1I;64]+3DR,C,S-S(P,S8X
+*-30W-S4X,#D*"@``
 `
 end