Description: Try to reduce confusion around docx files
Author: Olly Betts <olly@survex.com>
Bug-Debian: https://bugs.debian.org/758959
Forwarded: no
Last-Update: 2014-10-17

--- antiword-0.37.orig/Docs/antiword.1
+++ antiword-0.37/Docs/antiword.1
@@ -14,7 +14,10 @@ documents.
 .br
 A wordfile named - stands for a Word document read from the standard input.
 .br
-Only documents made by MS Word version 2 and version 6 or later are supported.
+Only documents made by MS Word version 2, 6, 7, 97, 2000 and 2003 are
+supported.  Newer Word versions default to using a completely different format
+consisting of XML files in a ZIP container (usually with a .docx file
+extension) which antiword doesn't support.
 .SH OPTIONS
 .TP
 .BI "\-a " papersize
--- antiword-0.37.orig/antiword.h
+++ antiword-0.37/antiword.h
@@ -695,6 +695,7 @@ extern int	iInitDocumentDOS(FILE *, long
 extern BOOL	bIsWordForDosFile(FILE *, long);
 extern BOOL	bIsRtfFile(FILE *);
 extern BOOL	bIsWordPerfectFile(FILE *);
+extern BOOL	bIsZipFile(FILE *);
 extern BOOL	bIsWinWord12File(FILE *, long);
 extern BOOL	bIsMacWord45File(FILE *);
 extern int	iGuessVersionNumber(FILE *, long);
--- antiword-0.37.orig/main_u.c
+++ antiword-0.37/main_u.c
@@ -187,10 +187,17 @@ bProcessFile(const char *szFilename)
 			werr(0, "%s is not a Word Document."
 				" It is probably a Rich Text Format file",
 				szFilename);
-		} if (bIsWordPerfectFile(pFile)) {
+		} else if (bIsWordPerfectFile(pFile)) {
 			werr(0, "%s is not a Word Document."
 				" It is probably a Word Perfect file",
 				szFilename);
+		} else if (bIsZipFile(pFile)) {
+			werr(0, "%s is not a Word Document."
+				" It is probably an OpenDocument file or"
+				" a new-style Microsoft Office XML file"
+				" (antiword only handles documents from"
+				" MS Word 2003 and earlier)",
+				szFilename);
 		} else {
 #if defined(__dos)
 			werr(0, "%s is not a Word Document or the filename"
--- antiword-0.37.orig/wordlib.c
+++ antiword-0.37/wordlib.c
@@ -131,6 +131,20 @@ bIsWordPerfectFile(FILE *pFile)
 } /* end of bIsWordPerfectFile */
 
 /*
+ * This function checks whether the given file is or is not a ZIP file
+ */
+BOOL
+bIsZipFile(FILE *pFile)
+{
+	static UCHAR	aucBytes[] =
+		{ 'P', 'K', 0x03, 0x04 };
+
+	DBG_MSG("bIsZipFile");
+
+	return bCheckBytes(pFile, aucBytes, elementsof(aucBytes));
+} /* end of bIsZipFile */
+
+/*
  * This function checks whether the given file is or is not a "Win Word 1 or 2"
  * document
  */