<rdar://problem/11131245> PDF detection in cups needs to be tightened up

Add regex() rule support.


git-svn-id: svn+ssh://src.apple.com/svn/cups/cups.org/trunk@11272 a1ca3aef-8c08-0410-bb20-df032aa958be

CHANGES.txt

@@ -1,5 +1,7 @@
-CHANGES.txt - 2.0b1 - 2013-08-06
+CHANGES.txt - 2.0b1 - 2013-09-10
 --------------------------------
 
 CHANGES IN CUPS V2.0b1
 
+	- Added support for regular expression matching in the MIME type rules
+	  (<rdar://problem/11131245>)

conf/mime.types

@@ -7,7 +7,7 @@
 #   VERSIONS OF CUPS.  Instead, create a "local.types" file that
 #   reflects your local configuration changes.
 #
-#   Copyright 2007-2011 by Apple Inc.
+#   Copyright 2007-2013 by Apple Inc.
 #   Copyright 1997-2007 by Easy Software Products.
 #
 #   These coded instructions, statements, and computer programs are the
@@ -37,6 +37,7 @@
 #                                       100=default, 200=highest)
 #   printable(offset,length)		True if bytes are printable 8-bit chars
 #					(CR, NL, TAB, BS, 32-126, 128-254)
+#   regex(offset,"regex")		True if bytes match regular expression
 #   string(offset,"string")		True if bytes are identical to string
 #   istring(offset,"string")		True if bytes are identical to
 #                                       case-insensitive string
@@ -70,7 +71,7 @@
 #
 
 #application/msword		doc string(0,<D0CF11E0A1B11AE1>)
-application/pdf			pdf string(0,%PDF)
+application/pdf			pdf regex(0,^[\\n\\r]*%PDF)
 application/postscript		ai eps ps string(0,%!) string(0,<04>%!) \
 				contains(0,128,<1B>%-12345X) + \
 				(contains(0,4096,"LANGUAGE=POSTSCRIPT") \

scheduler/mime.c

@@ -3,7 +3,7 @@
  *
  *   MIME database file routines for CUPS.
  *
- *   Copyright 2007-2012 by Apple Inc.
+ *   Copyright 2007-2013 by Apple Inc.
  *   Copyright 1997-2006 by Easy Software Products, all rights reserved.
  *
  *   These coded instructions, statements, and computer programs are the

scheduler/mime.h

@@ -3,7 +3,7 @@
  *
  *   MIME type/conversion database definitions for CUPS.
  *
- *   Copyright 2007-2011 by Apple Inc.
+ *   Copyright 2007-2013 by Apple Inc.
  *   Copyright 1997-2007 by Easy Software Products, all rights reserved.
  *
  *   These coded instructions, statements, and computer programs are the
@@ -19,6 +19,7 @@
 #  include <cups/array.h>
 #  include <cups/ipp.h>
 #  include <cups/file.h>
+#  include <regex.h>
 
 
 /*
@@ -58,7 +59,8 @@ typedef enum
   MIME_MAGIC_INT,			/* Integer/32-bit word matches */
   MIME_MAGIC_LOCALE,			/* Current locale matches string */
   MIME_MAGIC_CONTAINS,			/* File contains a string */
-  MIME_MAGIC_ISTRING			/* Case-insensitive string matches */
+  MIME_MAGIC_ISTRING,			/* Case-insensitive string matches */
+  MIME_MAGIC_REGEX			/* Regular expression matches */
 } mime_op_t;
 
 typedef struct _mime_magic_s		/**** MIME Magic Data ****/
@@ -80,6 +82,7 @@ typedef struct _mime_magic_s		/**** MIME Magic Data ****/
     unsigned char charv;		/* Byte value */
     unsigned short shortv;		/* Short value */
     unsigned	intv;			/* Integer value */
+    regex_t	rev;			/* Regular expression value */
   }		value;
 } mime_magic_t;
 

scheduler/type.c

@@ -3,7 +3,7 @@
  *
  *   MIME typing routines for CUPS.
  *
- *   Copyright 2007-2012 by Apple Inc.
+ *   Copyright 2007-2013 by Apple Inc.
  *   Copyright 1997-2006 by Easy Software Products, all rights reserved.
  *
  *   These coded instructions, statements, and computer programs are the
@@ -425,6 +425,8 @@ mimeAddTypeRule(mime_type_t *mt,	/* I - Type to add to */
 	  op = MIME_MAGIC_ASCII;
 	else if (!strcmp(name, "printable"))
 	  op = MIME_MAGIC_PRINTABLE;
+	else if (!strcmp(name, "regex"))
+	  op = MIME_MAGIC_REGEX;
 	else if (!strcmp(name, "string"))
 	  op = MIME_MAGIC_STRING;
 	else if (!strcmp(name, "istring"))
@@ -524,6 +526,12 @@ mimeAddTypeRule(mime_type_t *mt,	/* I - Type to add to */
 	    if (temp->length > MIME_MAX_BUFFER)
 	      temp->length = MIME_MAX_BUFFER;
 	    break;
+	case MIME_MAGIC_REGEX :
+	    temp->offset = strtol(value[0], NULL, 0);
+	    temp->length = MIME_MAX_BUFFER;
+	    if (regcomp(&(temp->value.rev), value[1], REG_NOSUB | REG_EXTENDED))
+	      return (-1);
+	    break;
 	case MIME_MAGIC_STRING :
 	case MIME_MAGIC_ISTRING :
 	    temp->offset = strtol(value[0], NULL, 0);
@@ -852,6 +860,49 @@ mime_check_rules(
 	  result = (n == 0);
 	  break;
 
+      case MIME_MAGIC_REGEX :
+          DEBUG_printf(("5mime_check_rules: regex(%d, \"%s\")", rules->offset,
+	                rules->value.stringv));
+
+         /*
+	  * Load the buffer if necessary...
+	  */
+
+          if (fb->offset < 0 || rules->offset < fb->offset ||
+	      (rules->offset + rules->length) > (fb->offset + fb->length))
+	  {
+	   /*
+	    * Reload file buffer...
+	    */
+
+            cupsFileSeek(fb->fp, rules->offset);
+	    fb->length = cupsFileRead(fb->fp, (char *)fb->buffer,
+	                              sizeof(fb->buffer));
+	    fb->offset = rules->offset;
+
+            DEBUG_printf(("5mime_check_rules: loaded %d byte fb->buffer at %d, starts "
+	                  "with \"%c%c%c%c\".",
+	                  fb->length, fb->offset, fb->buffer[0], fb->buffer[1],
+			  fb->buffer[2], fb->buffer[3]));
+	  }
+
+         /*
+	  * Compare the buffer against the string.  If the file is too
+	  * short then don't compare - it can't match...
+	  */
+
+          {
+            char temp[MIME_MAX_BUFFER + 1];
+					/* Temporary buffer */
+
+            memcpy(temp, fb->buffer, fb->length);
+            temp[fb->length] = '\0';
+            result = !regexec(&(rules->value.rev), temp, 0, NULL, 0);
+          }
+
+          DEBUG_printf(("5mime_check_rules: result=%d", result));
+	  break;
+
       case MIME_MAGIC_STRING :
           DEBUG_printf(("5mime_check_rules: string(%d, \"%s\")", rules->offset,
 	                rules->value.stringv));