zip: check filename crc in Info-ZIP Unicode Path Extra Field

The 0x7075 extension includes a crc of the filename that's in the CDE
to make sure that the UTF8 string is always up to date.  If an older
tool updates the CDE but doesn't update the 0x7075 field, we want to
ignore the UTF8 string since it's stale.

diff --git a/libarchive/archive_read_support_format_zip.c b/libarchive/archive_read_support_format_zip.c
index c6a3cf94329a19606cf93ee7c98242cb7be5e23d..1fa98fb76f1b44a5d787173eec44fd748f569497 100644 (file)
--- a/libarchive/archive_read_support_format_zip.c
+++ b/libarchive/archive_read_support_format_zip.c
@@ -755,6 +755,23 @@ process_extra(struct archive_read *a, struct archive_entry *entry,
                                        break;
                        }
 
+                       /* Make sure the CRC32 of the filename matches. */
+                       if (!zip->ignore_crc32) {
+                               const char *cp = archive_entry_pathname(entry);
+                               if (cp) {
+                                       unsigned long file_crc = zip->crc32func(0, cp, strlen(cp));
+                                       unsigned long utf_crc = archive_le32dec(p + offset - 4);
+                                       if (file_crc != utf_crc) {
+#ifdef DEBUG
+                                               fprintf(stderr, "CRC filename mismatch; CDE is %lx, "
+                                                   "but UTF8 is outdated with %lx\n",
+                                                   file_crc, utf_crc);
+#endif
+                                               break;
+                                       }
+                               }
+                       }
+
                        if (archive_entry_copy_pathname_l(entry,
                            p + offset, datasize, zip->sconv_utf8) != 0) {
                                /* Ignore the error, and fallback to the path

diff --git a/libarchive/test/test_read_format_zip_7075_utf8_paths.c b/libarchive/test/test_read_format_zip_7075_utf8_paths.c
index 85c44c0edfc97af27f8c0ceaec9e9d7aaa6287e3..7b78770aae4377d07754e8437e5e85e2946b2dfc 100644 (file)
--- a/libarchive/test/test_read_format_zip_7075_utf8_paths.c
+++ b/libarchive/test/test_read_format_zip_7075_utf8_paths.c
@@ -58,15 +58,10 @@ verify(struct archive *a) {
        assertEqualUTF8String(p, "File 3 - \xC3\xA4.txt");
 #endif
 
+       /* The CRC of the filename fails, so fall back to CDE. */
        assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
        assert((p = archive_entry_pathname_utf8(ae)) != NULL);
-#if defined(__APPLE__)
-       /* Compare NFD string. */
-       assertEqualUTF8String(p, "File 4 - a\xCC\x8A.txt");
-#else
-       /* Compare NFC string. */
-       assertEqualUTF8String(p, "File 4 - \xC3\xA5.txt");
-#endif
+       assertEqualUTF8String(p, "File 4 - xx.txt");
 
        assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae));
 }

diff --git a/libarchive/test/test_read_format_zip_7075_utf8_paths.zip.uu b/libarchive/test/test_read_format_zip_7075_utf8_paths.zip.uu
index 65c34f517a037628218726a745e2b8525014efb8..332f996898e07f35ab7c76d463e78c744fafc568 100644 (file)
--- a/libarchive/test/test_read_format_zip_7075_utf8_paths.zip.uu
+++ b/libarchive/test/test_read_format_zip_7075_utf8_paths.zip.uu
@@ -1,18 +1,18 @@
 begin 644 test_read_format_zip_7075_utf8_paths.zip
 M4$L#!!0````(`,$^9D5BZ95P"P````D````*````1FEL92`Q+G1X=`M)+2Y1
 M2,O,204`4$L#!!0````(`,$^9D5BZ95P"P````D````/`!@`1FEL92`R("T@
-M>'@N='AT=7`4``'W!4IU1FEL92`R("T@P[8N='AT"TDM+E%(R\Q)!0!02P,$
+M>'@N='AT=7`4``$NSPQQ1FEL92`R("T@P[8N='AT"TDM+E%(R\Q)!0!02P,$
 M%`````@`P3YF16+IE7`+````"0````\`&`!&:6QE(#,@+2!X>"YT>'1U<!0`
-M`3:&T0)&:6QE(#,@+2##I"YT>'0+22TN44C+S$D%`%!+`P04````"`#!/F9%
+M`1"DSIY&:6QE(#,@+2##I"YT>'0+22TN44C+S$D%`%!+`P04````"`#!/F9%
 M8NF5<`L````)````#P`8`$9I;&4@-"`M('AX+G1X='5P%``!G[AP'$9I;&4@
 M-"`M(,.E+G1X=`M)+2Y12,O,204`4$L!`A\`%`````@`P3YF16+IE7`+````
 M"0````H`)``````````@`````````$9I;&4@,2YT>'0*`"````````$`&``Q
 M6UASCOG/`5^OQVV.^<\!7Z_';8[YSP%02P$"'P`4````"`#!/F9%8NF5<`L`
 M```)````#@`\`````````"`````S````1FEL92`R("T@E"YT>'0*`"``````
-M``$`&``Q6UASCOG/`2M.B72.^<\!*TZ)=([YSP%U<!0``?<%2G5&:6QE(#(@
+M``$`&``Q6UASCOG/`2M.B72.^<\!*TZ)=([YSP%U<!0``2[/#'%&:6QE(#(@
 M+2##MBYT>'102P$"'P`4````"`#!/F9%8NF5<`L````)````#@`\````````
 M`"````"#````1FEL92`S("T@A"YT>'0*`"````````$`&``Q6UASCOG/`5<$
-M&W>.^<\!5P0;=X[YSP%U<!0``3:&T0)&:6QE(#,@+2##I"YT>'102P$"'P`4
+M&W>.^<\!5P0;=X[YSP%U<!0``1"DSIY&:6QE(#,@+2##I"YT>'102P$"'P`4
 M````"`#!/F9%8NF5<`L````)````#@`\`````````"````#3````1FEL92`T
 M("T@ABYT>'0*`"````````$`&``Q6UASCOG/`6#)ZG:.^<\!8,GJ=H[YSP%U
 M<!0``9^X<!Q&:6QE(#0@+2##I2YT>'102P4&``````0`!`#$`0``(P$`````