Add support to pcre2grep for binary zeros in -f files.

ChangeLog

@@ -25,6 +25,9 @@ multi-code-unit characters caused bad behaviour and possibly a crash. This
 issue was fixed for other kinds of repeat in release 10.20 by change 19, but
 repeating character classes were overlooked.
 
+5. pcre2grep now supports the inclusion of binary zeros in patterns that are 
+read from files via the -f option.
+
 
 Version 10.31 12-February-2018
 ------------------------------

RunGrepTest

@@ -641,6 +641,12 @@ echo "RC=$?" >>testtrygrep
 $valgrind $vjs $pcre2grep --colour=always '(?=[ac]\K)' testNinputgrep >>testtrygrep
 echo "RC=$?" >>testtrygrep
 
+echo "---------------------------- Test 126 -----------------------------" >>testtrygrep
+printf "Next line pattern has binary zero\nABC\x00XYZ\n" >testtemp1grep
+printf "ABC\x00XYZ\nABCDEF\nDEFABC\n" >testtemp2grep
+$valgrind $vjs $pcre2grep -a -f testtemp1grep testtemp2grep >>testtrygrep
+echo "RC=$?" >>testtrygrep
+
 
 # Now compare the results.
 

doc/pcre2grep.1

@@ -1,4 +1,4 @@
-.TH PCRE2GREP 1 "13 November 2017" "PCRE2 10.31"
+.TH PCRE2GREP 1 "24 February 2018" "PCRE2 10.32"
 .SH NAME
 pcre2grep - a grep with Perl-compatible regular expressions.
 .SH SYNOPSIS
@@ -121,6 +121,14 @@ a binary file is not applied. See the \fB--binary-files\fP option for a means
 of changing the way binary files are handled.
 .
 .
+.SH "BINARY ZEROS IN PATTERNS"
+.rs
+.sp
+Patterns passed from the command line are strings that are terminated by a
+binary zero, so cannot contain internal zeros. However, patterns that are read 
+from a file via the \fB-f\fP option may contain binary zeros.
+.
+.
 .SH OPTIONS
 .rs
 .sp
@@ -304,12 +312,15 @@ files; it does not apply to patterns specified by any of the \fB--include\fP or
 .TP
 \fB-f\fP \fIfilename\fP, \fB--file=\fP\fIfilename\fP
 Read patterns from the file, one per line, and match them against each line of
-input. What constitutes a newline when reading the file is the operating
+input. As is the case with patterns on the command line, no delimiters should
-system's default. The \fB--newline\fP option has no effect on this option.
+be used. What constitutes a newline when reading the file is the operating
-Trailing white space is removed from each line, and blank lines are ignored. An
+system's default interpretation of \en. The \fB--newline\fP option has no
-empty file contains no patterns and therefore matches nothing. See also the
+effect on this option. Trailing white space is removed from each line, and
-comments about multiple patterns versus a single pattern with alternatives in
+blank lines are ignored. An empty file contains no patterns and therefore
-the description of \fB-e\fP above.
+matches nothing. Patterns read from a file in this way may contain binary
+zeros, which are treated as ordinary data characters. See also the comments
+about multiple patterns versus a single pattern with alternatives in the
+description of \fB-e\fP above.
 .sp
 If this option is given more than once, all the specified files are read. A
 data line is output if any of the patterns match it. A file name can be given
@@ -320,14 +331,15 @@ command line; all arguments are treated as the names of paths to be searched.
 .TP
 \fB--file-list\fP=\fIfilename\fP
 Read a list of files and/or directories that are to be scanned from the given
-file, one per line. Trailing white space is removed from each line, and blank
+file, one per line. What constitutes a newline when reading the file is the
-lines are ignored. These paths are processed before any that are listed on the
+operating system's default. Trailing white space is removed from each line, and
-command line. The file name can be given as "-" to refer to the standard input.
+blank lines are ignored. These paths are processed before any that are listed
-If \fB--file\fP and \fB--file-list\fP are both specified as "-", patterns are
+on the command line. The file name can be given as "-" to refer to the standard
-read first. This is useful only when the standard input is a terminal, from
+input. If \fB--file\fP and \fB--file-list\fP are both specified as "-",
-which further lines (the list of files) can be read after an end-of-file
+patterns are read first. This is useful only when the standard input is a
-indication. If this option is given more than once, all the specified files are
+terminal, from which further lines (the list of files) can be read after an
-read.
+end-of-file indication. If this option is given more than once, all the
+specified files are read.
 .TP
 \fB--file-offsets\fP
 Instead of showing lines or parts of lines that match, show each match as an
@@ -679,12 +691,13 @@ The \fB-N\fP (\fB--newline\fP) option allows \fBpcre2grep\fP to scan files with
 different newline conventions from the default. Any parts of the input files
 that are written to the standard output are copied identically, with whatever
 newline sequences they have in the input. However, the setting of this option
-does not affect the interpretation of files specified by the \fB-f\fP,
+affects only the way scanned files are processed. It does not affect the
-\fB--exclude-from\fP, or \fB--include-from\fP options, which are assumed to use
+interpretation of files specified by the \fB-f\fP, \fB--file-list\fP,
-the operating system's standard newline sequence, nor does it affect the way in
+\fB--exclude-from\fP, or \fB--include-from\fP options, nor does it affect the
-which \fBpcre2grep\fP writes informational messages to the standard error and
+way in which \fBpcre2grep\fP writes informational messages to the standard
-output streams. For these it uses the string "\en" to indicate newlines,
+error and output streams. For these it uses the string "\en" to indicate
-relying on the C I/O library to convert this to an appropriate sequence.
+newlines, relying on the C I/O library to convert this to an appropriate
+sequence.
 .
 .
 .SH "OPTIONS COMPATIBILITY"
@@ -862,6 +875,6 @@ Cambridge, England.
 .rs
 .sp
 .nf
-Last updated: 13 November 2017
+Last updated: 24 February 2018
-Copyright (c) 1997-2017 University of Cambridge.
+Copyright (c) 1997-2018 University of Cambridge.
 .fi

src/pcre2grep.c

@@ -13,7 +13,7 @@ distribution because other apparatus is needed to compile pcre2grep for z/OS.
 The header can be found in the special z/OS distribution, which is available
 from www.zaconsultants.net or from www.cbttape.org.
 
-           Copyright (c) 1997-2017 University of Cambridge
+           Copyright (c) 1997-2018 University of Cambridge
 
 -----------------------------------------------------------------------------
 Redistribution and use in source and binary forms, with or without
@@ -303,6 +303,7 @@ also for include/exclude patterns. */
 typedef struct patstr {
   struct patstr *next;
   char *string;
+  PCRE2_SIZE length; 
   pcre2_code *compiled;
 } patstr;
 
@@ -557,13 +558,14 @@ exit(rc);
 
 Arguments:
   s          pattern string to add
+  patlen     length of pattern 
   after      if not NULL points to item to insert after
 
 Returns:     new pattern block or NULL on error
 */
 
 static patstr *
-add_pattern(char *s, patstr *after)
+add_pattern(char *s, PCRE2_SIZE patlen, patstr *after)
 {
 patstr *p = (patstr *)malloc(sizeof(patstr));
 if (p == NULL)
@@ -571,7 +573,7 @@ if (p == NULL)
   fprintf(stderr, "pcre2grep: malloc failed\n");
   pcre2grep_exit(2);
   }
-if (strlen(s) > MAXPATLEN)
+if (patlen > MAXPATLEN)
   {
   fprintf(stderr, "pcre2grep: pattern is too long (limit is %d bytes)\n",
     MAXPATLEN);
@@ -580,6 +582,7 @@ if (strlen(s) > MAXPATLEN)
   }
 p->next = NULL;
 p->string = s;
+p->length = patlen;
 p->compiled = NULL;
 
 if (after != NULL)
@@ -1276,12 +1279,14 @@ return om;
 *            Read one line of input              *
 *************************************************/
 
-/* Normally, input is read using fread() (or gzread, or BZ2_read) into a large
+/* Normally, input that is to be scanned is read using fread() (or gzread, or
-buffer, so many lines may be read at once. However, doing this for tty input
+BZ2_read) into a large buffer, so many lines may be read at once. However,
-means that no output appears until a lot of input has been typed. Instead, tty
+doing this for tty input means that no output appears until a lot of input has
-input is handled line by line. We cannot use fgets() for this, because it does
+been typed. Instead, tty input is handled line by line. We cannot use fgets()
-not stop at a binary zero, and therefore there is no way of telling how many
+for this, because it does not stop at a binary zero, and therefore there is no
-characters it has read, because there may be binary zeros embedded in the data.
+way of telling how many characters it has read, because there may be binary
+zeros embedded in the data. This function is also used for reading patterns 
+from files (the -f option).
 
 Arguments:
   buffer     the buffer to read into
@@ -1291,7 +1296,7 @@ Arguments:
 Returns:     the number of characters read, zero at end of file
 */
 
-static unsigned int
+static PCRE2_SIZE
 read_one_line(char *buffer, int length, FILE *f)
 {
 int c;
@@ -1651,11 +1656,11 @@ Returns:      TRUE if there was a match
 */
 
 static BOOL
-match_patterns(char *matchptr, size_t length, unsigned int options,
+match_patterns(char *matchptr, PCRE2_SIZE length, unsigned int options,
-  size_t startoffset, int *mrc)
+  PCRE2_SIZE startoffset, int *mrc)
 {
 int i;
-size_t slen = length;
+PCRE2_SIZE slen = length;
 patstr *p = patterns;
 const char *msg = "this text:\n\n";
 
@@ -2317,7 +2322,7 @@ unsigned long int count = 0;
 char *lastmatchrestart = NULL;
 char *ptr = main_buffer;
 char *endptr;
-size_t bufflength;
+PCRE2_SIZE bufflength;
 BOOL binary = FALSE;
 BOOL endhyphenpending = FALSE;
 BOOL input_line_buffered = line_buffered;
@@ -2339,7 +2344,7 @@ bufflength = fill_buffer(handle, frtype, main_buffer, bufsize,
   input_line_buffered);
 
 #ifdef SUPPORT_LIBBZ2
-if (frtype == FR_LIBBZ2 && (int)bufflength < 0) return 2;   /* Gotcha: bufflength is size_t; */
+if (frtype == FR_LIBBZ2 && (int)bufflength < 0) return 2;   /* Gotcha: bufflength is PCRE2_SIZE; */
 #endif
 
 endptr = main_buffer + bufflength;
@@ -2368,8 +2373,8 @@ while (ptr < endptr)
   unsigned int options = 0;
   BOOL match;
   char *t = ptr;
-  size_t length, linelength;
+  PCRE2_SIZE length, linelength;
-  size_t startoffset = 0;
+  PCRE2_SIZE startoffset = 0;
 
   /* At this point, ptr is at the start of a line. We need to find the length
   of the subject string to pass to pcre2_match(). In multiline mode, it is the
@@ -2381,7 +2386,7 @@ while (ptr < endptr)
 
   t = end_of_line(t, endptr, &endlinelength);
   linelength = t - ptr - endlinelength;
-  length = multiline? (size_t)(endptr - ptr) : linelength;
+  length = multiline? (PCRE2_SIZE)(endptr - ptr) : linelength;
 
   /* Check to see if the line we are looking at extends right to the very end
   of the buffer without a line terminator. This means the line is too long to
@@ -2560,7 +2565,7 @@ while (ptr < endptr)
       {
       if (!invert)
         {
-        size_t oldstartoffset;
+        PCRE2_SIZE oldstartoffset;
 
         if (printname != NULL) fprintf(stdout, "%s:", printname);
         if (number) fprintf(stdout, "%lu:", linenumber);
@@ -2647,7 +2652,7 @@ while (ptr < endptr)
           startoffset -= (int)(linelength + endlinelength);
           t = end_of_line(ptr, endptr, &endlinelength);
           linelength = t - ptr - endlinelength;
-          length = (size_t)(endptr - ptr);
+          length = (PCRE2_SIZE)(endptr - ptr);
           }
 
         goto ONLY_MATCHING_RESTART;
@@ -2812,7 +2817,7 @@ while (ptr < endptr)
             endprevious -= (int)(linelength + endlinelength);
             t = end_of_line(ptr, endptr, &endlinelength);
             linelength = t - ptr - endlinelength;
-            length = (size_t)(endptr - ptr);
+            length = (PCRE2_SIZE)(endptr - ptr);
             }
 
           /* If startoffset is at the exact end of the line it means this
@@ -2895,7 +2900,7 @@ while (ptr < endptr)
   /* If input is line buffered, and the buffer is not yet full, read another
   line and add it into the buffer. */
 
-  if (input_line_buffered && bufflength < (size_t)bufsize)
+  if (input_line_buffered && bufflength < (PCRE2_SIZE)bufsize)
     {
     int add = read_one_line(ptr, bufsize - (int)(ptr - main_buffer), in);
     bufflength += add;
@@ -2907,7 +2912,7 @@ while (ptr < endptr)
   1/3 and refill it. Before we do this, if some unprinted "after" lines are
   about to be lost, print them. */
 
-  if (bufflength >= (size_t)bufsize && ptr > main_buffer + 2*bufthird)
+  if (bufflength >= (PCRE2_SIZE)bufsize && ptr > main_buffer + 2*bufthird)
     {
     if (after_context > 0 &&
         lastmatchnumber > 0 &&
@@ -3395,9 +3400,8 @@ PCRE2_SIZE patlen, erroffset;
 PCRE2_UCHAR errmessbuffer[ERRBUFSIZ];
 
 if (p->compiled != NULL) return TRUE;
-
 ps = p->string;
-patlen = strlen(ps);
+patlen = p->length;
 
 if ((options & PCRE2_LITERAL) != 0)
   {
@@ -3407,8 +3411,8 @@ if ((options & PCRE2_LITERAL) != 0)
 
   if (ellength != 0)
     {
-    if (add_pattern(pe, p) == NULL) return FALSE;
+    patlen = pe - ps - ellength;
-    patlen = (int)(pe - ps - ellength);
+    if (add_pattern(pe, p->length-patlen-ellength, p) == NULL) return FALSE;
     }
   }
 
@@ -3470,6 +3474,7 @@ static BOOL
 read_pattern_file(char *name, patstr **patptr, patstr **patlastptr)
 {
 int linenumber = 0;
+PCRE2_SIZE patlen;
 FILE *f;
 const char *filename;
 char buffer[MAXPATLEN+20];
@@ -3490,20 +3495,18 @@ else
   filename = name;
   }
 
-while (fgets(buffer, sizeof(buffer), f) != NULL)
+while ((patlen = read_one_line(buffer, sizeof(buffer), f)) > 0)
   {
-  char *s = buffer + (int)strlen(buffer);
+  while (patlen > 0 && isspace((unsigned char)(buffer[patlen-1]))) patlen--; 
-  while (s > buffer && isspace((unsigned char)(s[-1]))) s--;
-  *s = 0;
   linenumber++;
-  if (buffer[0] == 0) continue;   /* Skip blank lines */
+  if (patlen == 0) continue;   /* Skip blank lines */
 
   /* Note: this call to add_pattern() puts a pointer to the local variable
   "buffer" into the pattern chain. However, that pointer is used only when
   compiling the pattern, which happens immediately below, so we flatten it
   afterwards, as a precaution against any later code trying to use it. */
 
-  *patlastptr = add_pattern(buffer, *patlastptr);
+  *patlastptr = add_pattern(buffer, patlen, *patlastptr);
   if (*patlastptr == NULL)
     {
     if (f != stdin) fclose(f);
@@ -3513,8 +3516,9 @@ while (fgets(buffer, sizeof(buffer), f) != NULL)
 
   /* This loop is needed because compiling a "pattern" when -F is set may add
   on additional literal patterns if the original contains a newline. In the
-  common case, it never will, because fgets() stops at a newline. However,
+  common case, it never will, because read_one_line() stops at a newline.
-  the -N option can be used to give pcre2grep a different newline setting. */
+  However, the -N option can be used to give pcre2grep a different newline
+  setting. */
 
   for(;;)
     {
@@ -3833,7 +3837,8 @@ for (i = 1; i < argc; i++)
   else if (op->type == OP_PATLIST)
     {
     patdatastr *pd = (patdatastr *)op->dataptr;
-    *(pd->lastptr) = add_pattern(option_data, *(pd->lastptr));
+    *(pd->lastptr) = add_pattern(option_data, (PCRE2_SIZE)strlen(option_data), 
+      *(pd->lastptr));
     if (*(pd->lastptr) == NULL) goto EXIT2;
     if (*(pd->anchor) == NULL) *(pd->anchor) = *(pd->lastptr);
     }
@@ -4095,7 +4100,9 @@ the first argument is the one and only pattern, and it must exist. */
 if (patterns == NULL && pattern_files == NULL)
   {
   if (i >= argc) return usage(2);
-  patterns = patterns_last = add_pattern(argv[i++], NULL);
+  patterns = patterns_last = add_pattern(argv[i], (PCRE2_SIZE)strlen(argv[i]),
+    NULL);
+  i++;   
   if (patterns == NULL) goto EXIT2;
   }
 

testdata/grepoutput

@@ -945,3 +945,6 @@ RC=0
 RC=0
 abcd
 RC=0
+---------------------------- Test 126 -----------------------------
+ABCXYZ
+RC=0