Add memelemscan

Signed-off-by: Mattias Andrée <maandree@kth.se>

10 files changed, 216 insertions, 5 deletions

diff --git a/Makefile b/Makefile
index 3428753..5944817 100644
--- a/Makefile
+++ b/Makefile
@@ -67,6 +67,7 @@ OBJ =\
 	memcasestarts.o\
 	memdup.o\
 	memelem.o\
+	memelemscan.o\
 	memends.o\
 	memeqlen.o\
 	memisutf8.o\
diff --git a/libsimple/mem.h b/libsimple/mem.h
index dacc78c..67834bc 100644
--- a/libsimple/mem.h
+++ b/libsimple/mem.h
@@ -269,6 +269,26 @@ void *libsimple_memelem(const void *, size_t, const void *, size_t);
 
 
 /**
+ * Finds the first element in an array, the comparison is case-sensitive
+ * 
+ * @param   haystack   The array of bytes to search
+ * @param   nhaystack  The length of `haystack`, divided by `needle`
+ * @param   needle     The substring to search for
+ * @param   nneedle    The length of `needle`
+ * @return             `haystack` with a minimal offset such that,
+ *                     `!memcmp(r, needle, nneedle)` where `r` is the
+ *                     returned pointer and such that `(r - haystack) % nneedle == 0`,
+ *                     `(void *)&((char *)haystack)[nhaystack * nneedle]`
+ *                     if no such offset exists
+ */
+_LIBSIMPLE_GCC_ONLY(__attribute__((__pure__, __warn_unused_result__)))
+void *libsimple_memelemscan(const void *, size_t, const void *, size_t);
+#ifndef memelemscan
+# define memelemscan libsimple_memelemscan
+#endif
+
+
+/**
  * Finds the last element in an array, the comparison is case-sensitive
  * 
  * @param   haystack   The array of bytes to search
diff --git a/man0/libsimple.h.0 b/man0/libsimple.h.0
index 7cb93dd..d3c018d 100644
--- a/man0/libsimple.h.0
+++ b/man0/libsimple.h.0
@@ -542,6 +542,13 @@ that operate on multibyte units
 rather than simple bytes.
 
 .TP
+.BR libsimple_memelemscan (3)
+Version of
+.BR memscan (3)
+that operate on multibyte units
+rather than simple bytes.
+
+.TP
 .BR libsimple_memends (3),
 .RS 0
 .BR libsimple_memcaseends (3)
diff --git a/man3/libsimple_memelem.3 b/man3/libsimple_memelem.3
index 833e246..6eb3c0b 100644
--- a/man3/libsimple_memelem.3
+++ b/man3/libsimple_memelem.3
@@ -79,6 +79,7 @@ None.
 .SH BUGS
 None.
 .SH SEE ALSO
+.BR libsimple_memelemscan (3),
 .BR libsimple_memrelem (3),
 .BR libsimple_memmem (3),
 .BR memchr (3)
diff --git a/man3/libsimple_memelemscan.3 b/man3/libsimple_memelemscan.3
new file mode 100644
index 0000000..de70105
--- /dev/null
+++ b/man3/libsimple_memelemscan.3
@@ -0,0 +1,87 @@
+.TH LIBSIMPLE_MEMELEMSCAN 3 2018-11-17 libsimple
+.SH NAME
+libsimple_memelemscan \- find aligned byte string in memory
+.SH SYNOPSIS
+.nf
+#include <libsimple.h>
+
+void *libsimple_memelemscan(const void *\fIhaystack\fP, size_t \fInhaystack\fP, const void *\fIneedle\fP, size_t \fInneedle\fP);
+
+#ifndef memelemscan
+# define memelemscan libsimple_memelemscan
+#endif
+.fi
+.PP
+Link with
+.IR \-lsimple .
+.SH DESCRIPTION
+The
+.BR libsimple_memelemscan ()
+function scans the memory segment
+.IR haystack ,
+with the size
+.IR nhaystack*nneedle ,
+for the first occurence of the byte string
+.I needle
+with the size
+.IR nneedle ,
+and with an offset equivalent to zero modulo
+.IR nneedle .
+If no such character exist in the memory
+segment, the memory segment's end is returned.
+.PP
+The comparison is case-sensitive.
+.SH RETURN VALUE
+The
+.BR libsimple_memelemscan ()
+function returns the pointer
+.I s
+with a minimal offset such that
+.IR !memcmp(r,needle,nneedle) ,
+where
+.I r
+is the returned pointer, and such that
+the offset is equivalent to zero modulo
+.IR nneedle .
+If no such offset exists,
+.I (void *)&((char *)haystack)[nhaystack*nneedle]
+is returned.
+.SH ERRORS
+The
+.BR libsimple_memelemscan ()
+function cannot fail.
+.SH ATTRIBUTES
+For an explanation of the terms used in this section, see
+.BR attributes (7).
+.TS
+allbox;
+lb lb lb
+l l l.
+Interface	Attribute	Value
+T{
+.BR libsimple_memelemscan ()
+T}	Thread safety	MT-Safe
+T{
+.BR libsimple_memelemscan ()
+T}	Async-signal safety	AS-Safe
+T{
+.BR libsimple_memelemscan ()
+T}	Async-cancel safety	AC-Safe
+.TE
+.SH EXAMPLES
+None.
+.SH APPLICATION USAGE
+None.
+.SH RATIONALE
+None.
+.SH FUTURE DIRECTIONS
+None.
+.SH NOTES
+None.
+.SH BUGS
+None.
+.SH SEE ALSO
+.BR libsimple_memelem (3),
+.BR libsimple_memrelem (3),
+.BR libsimple_memmem (3),
+.BR libsimple_memchrscan (3)
diff --git a/man3/libsimple_memscan.3 b/man3/libsimple_memscan.3
index c9db7da..0c5d14a 100644
--- a/man3/libsimple_memscan.3
+++ b/man3/libsimple_memscan.3
@@ -25,6 +25,8 @@ for the first occurence of the byte
 .I c
 (it is converted to a
 .BR char ).
+If no such character exist in the memory
+segment, the memory segment's end is returned.
 .PP
 The comparison is case-sensitive.
 .SH RETURN VALUE
@@ -38,7 +40,7 @@ where
 .I r
 is the returned pointer.
 If no such offset exists,
-.B NULL
+.I &r[n]
 is returned.
 .SH ERRORS
 The
diff --git a/man3/memelemscan.3libsimple b/man3/memelemscan.3libsimple
new file mode 120000
index 0000000..3c612db
--- /dev/null
+++ b/man3/memelemscan.3libsimple
@@ -0,0 +1 @@
+libsimple_memelemscan.3
\ No newline at end of file
diff --git a/memelem.c b/memelem.c
index 43cb298..3881832 100644
--- a/memelem.c
+++ b/memelem.c
@@ -52,7 +52,6 @@ libsimple_memelem(const void *hay_, size_t hayn, const void *sub_, size_t subn)
 			next:;
 			}
 			break;
-
 		}
 	}
 
@@ -70,7 +69,7 @@ main(void)
 	assert(!strcmpnul(libsimple_memelem("12345634", 0, "", 0), "12345634"));
 
 	assert(!strcmpnul(libsimple_memelem("12345634", 8, "3", 1), "345634"));
-	assert(!libsimple_memelem("12345634", 8, "x", 1));;
+	assert(!libsimple_memelem("12345634", 8, "x", 1));
 	assert(!strcmpnul(libsimple_memelem("13456342", 8, "3", 1), "3456342"));
 	assert(!libsimple_memelem("12345634", 0, "3", 1));
 
diff --git a/memelemscan.c b/memelemscan.c
new file mode 100644
index 0000000..744989b
--- /dev/null
+++ b/memelemscan.c
@@ -0,0 +1,94 @@
+/* See LICENSE file for copyright and license details. */
+#include "libsimple.h"
+#ifndef TEST
+
+
+void *
+libsimple_memelemscan(const void *hay_, size_t hayn, const void *sub_, size_t subn)
+{
+	if (!subn)
+		return (void *)hay_;
+
+	switch (subn) {
+	case 1:
+		return libsimple_memscan(hay_, *(char *)sub_, hayn);
+	case 2:
+		{
+			uint16_t *hay = (void *)hay_;
+			uint16_t sub = *(uint16_t *)sub_;
+			for (; hayn-- && *hay != sub; hay++);
+			return hay;
+		}
+	case 4:
+		{
+			uint32_t *hay = (void *)hay_;
+			uint32_t sub = *(uint32_t *)sub_;
+			for (; hayn-- && *hay != sub; hay++);
+			return hay;
+		}
+	case 8:
+		{
+			uint64_t *hay = (void *)hay_;
+			uint64_t sub = *(uint64_t *)sub_;
+			for (; hayn-- && *hay != sub; hay++);
+			return hay;
+		}
+	default:
+		{
+			char *hay = (void *)hay_;
+			const char *sub = sub_;
+			size_t i;
+			for (; hayn--; hay += subn) {
+				for (i = 0; i < subn; i++)
+					if (hay[i] != sub[i])
+						goto next;
+				return hay;
+			next:;
+			}
+			return hay;
+		}
+	}
+
+	return NULL;
+}
+
+
+#else
+#include "test.h"
+
+int
+main(void)
+{
+	assert(!strcmpnul(libsimple_memelemscan("12345634", 8, "", 0), "12345634"));
+	assert(!strcmpnul(libsimple_memelemscan("12345634", 0, "", 0), "12345634"));
+
+	assert(!strcmpnul(libsimple_memelemscan("12345634", 8, "3", 1), "345634"));
+	assert(!strcmpnul(libsimple_memelemscan("12345634", 8, "x", 1), ""));
+	assert(!strcmpnul(libsimple_memelemscan("13456342", 8, "3", 1), "3456342"));
+	assert(!strcmpnul(libsimple_memelemscan("12345634", 0, "3", 1), "12345634"));
+
+	assert(!strcmpnul(libsimple_memelemscan("12345634", 4, "34", 2), "345634"));
+	assert(!strcmpnul(libsimple_memelemscan("12345634z", 4, "xx", 2), "z"));
+	assert(!strcmpnul(libsimple_memelemscan("13456342z", 4, "34", 2), "z"));
+	assert(!strcmpnul(libsimple_memelemscan("12345634z", 0, "34", 2), "12345634z"));
+
+	assert(!strcmpnul(libsimple_memelemscan("abcd1234abcd1234", 4, "1234", 4), "1234abcd1234"));
+	assert(!strcmpnul(libsimple_memelemscan("abcd1234abcd1234z", 4, "zzzz", 4), "z"));
+	assert(!strcmpnul(libsimple_memelemscan("cd1234abcd1234abz", 4, "1234", 4), "z"));
+	assert(!strcmpnul(libsimple_memelemscan("abcd1234abcd1234z", 0, "1234", 4), "abcd1234abcd1234z"));
+
+	assert(!strcmpnul(libsimple_memelemscan("abcdefgh12345678abcdefgh12345678", 4, "12345678", 8), "12345678abcdefgh12345678"));
+	assert(!strcmpnul(libsimple_memelemscan("abcdefgh12345678abcdefgh12345678z", 4, "zzzzzzzz", 8), "z"));
+	assert(!strcmpnul(libsimple_memelemscan("efgh12345678abcdefgh12345678abcdz", 4, "12345678", 8), "z"));
+	assert(!strcmpnul(libsimple_memelemscan("abcdefgh12345678abcdefgh12345678z", 0, "12345678", 8),
+	                  "abcdefgh12345678abcdefgh12345678z"));
+
+	assert(!strcmpnul(libsimple_memelemscan("abc123abc123", 4, "123", 3), "123abc123"));
+	assert(!strcmpnul(libsimple_memelemscan("abc123abc123z", 4, "zzz", 3), "z"));
+	assert(!strcmpnul(libsimple_memelemscan("bc123abc123az", 4, "123", 3), "z"));
+	assert(!strcmpnul(libsimple_memelemscan("abc123abc123z", 0, "123", 3), "abc123abc123z"));
+
+	return 0;
+}
+
+#endif
diff --git a/memrelem.c b/memrelem.c
index b1e69ba..8317e8e 100644
--- a/memrelem.c
+++ b/memrelem.c
@@ -60,7 +60,6 @@ libsimple_memrelem(const void *hay_, size_t hayn, const void *sub_, size_t subn)
 			next:;
 			}
 			break;
-
 		}
 	}
 
@@ -78,7 +77,7 @@ main(void)
 	assert(!strcmpnul(libsimple_memrelem("12345634", 0, "", 0), "12345634"));
 
 	assert(!strcmpnul(libsimple_memrelem("12345634", 8, "3", 1), "34"));
-	assert(!libsimple_memrelem("12345634", 8, "x", 1));;
+	assert(!libsimple_memrelem("12345634", 8, "x", 1));
 	assert(!strcmpnul(libsimple_memrelem("13456342", 8, "3", 1), "342"));
 	assert(!libsimple_memrelem("12345634", 0, "3", 1));