[PATCH v4 0/4] mem: extend with a few additional helpers similar to existing string-only functions

[PATCH v4 0/4] mem: extend with a few additional helpers similar to existing string-only functions

[Cody P Schafer]

since v3:

 - fix memeq() return value
 - add memends_str() helper

since v2:

 - switch all the macros to static inline to avoid multiple evaluations of args
 - add patch adding PURE_FUNCTION attributes

since v1:

 - drop a patch that David pulled in
 - add extra macro parens
 - add blank lines after defintions
 - covert memcchr's while loop to a for.
 - change examples for mempbrkm and mempbrk to use '\0' in places
 - fix spelling
 - use memeq() even where we know the mem lengths are equal in memstarts()
 - add a reviewed-by to the 2nd patch, no other changes to that patch.

--

This is essentially a rebased version of a patch I sent out quite some time ago.

At the time, David indicated that it would probably make sense for bytestring
to call the plain mem-versions of these functions, so I've included a helper to
do that as well (and 1 patch to help ccanlint for bytestring pass a bit better).

The selection of mem* functions is entirely based on which ones I need for some
code I was writing, and they are in no way complete (notably, they include mem
variants of strpbrk instead of strspn variants as used in bytestring). I expect
the missing functions can be added as people find a need for them.


Cody P Schafer (4):
  mem: add mem helper functions
  bytestring: use newly added mem helpers
  mem: mark all functions as PURE
  mem: add memends_str() helper for symmetry

 ccan/bytestring/bytestring.h |   9 +--
 ccan/mem/_info               |   1 +
 ccan/mem/mem.c               |  24 ++++++
 ccan/mem/mem.h               | 185 +++++++++++++++++++++++++++++++++++++++++++
 ccan/mem/test/api.c          |  41 +++++++++-
 5 files changed, 253 insertions(+), 7 deletions(-)

-- 
2.5.1

_______________________________________________
ccan mailing list
ccan@lists.ozlabs.org
https://lists.ozlabs.org/listinfo/ccan

[PATCH v4 1/4] mem: add mem helper functions

[Cody P Schafer]

Signed-off-by: Cody P Schafer <dev@codyps.com>
---
 ccan/mem/mem.c      |  24 ++++++++
 ccan/mem/mem.h      | 157 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 ccan/mem/test/api.c |  37 ++++++++++++-
 3 files changed, 217 insertions(+), 1 deletion(-)

diff --git a/ccan/mem/mem.c b/ccan/mem/mem.c
index 853f975..91c9961 100644
--- a/ccan/mem/mem.c
+++ b/ccan/mem/mem.c
@@ -40,3 +40,27 @@ void *memrchr(const void *s, int c, size_t n)
 	return NULL;
 }
 #endif
+
+void *mempbrkm(const void *data_, size_t len, const void *accept_, size_t accept_len)
+{
+	const char *data = data_, *accept = accept_;
+	size_t i, j;
+
+	for (i = 0; i < len; i++)
+		for (j = 0; j < accept_len; j++)
+			if (accept[j] == data[i])
+				return (void *)&data[i];
+	return NULL;
+}
+
+void *memcchr(void const *data, int c, size_t data_len)
+{
+	char const *p = data;
+	size_t i;
+
+	for (i = 0; i < data_len; i++)
+		if (p[i] != c)
+			return (void *)&p[i];
+
+	return NULL;
+}
diff --git a/ccan/mem/mem.h b/ccan/mem/mem.h
index dcb44b8..89b16d4 100644
--- a/ccan/mem/mem.h
+++ b/ccan/mem/mem.h
@@ -5,6 +5,7 @@
 #include "config.h"
 
 #include <string.h>
+#include <stdbool.h>
 
 #if !HAVE_MEMMEM
 void *memmem(const void *haystack, size_t haystacklen,
@@ -15,4 +16,160 @@ void *memmem(const void *haystack, size_t haystacklen,
 void *memrchr(const void *s, int c, size_t n);
 #endif
 
+/**
+ * mempbrkm - locates the first occurrence in @data of any bytes in @accept
+ * @data: where we search
+ * @len: length of data in bytes
+ * @accept: array of bytes we search for
+ * @accept_len: # of bytes in accept
+ *
+ * Returns a pointer to the byte in @data that matches one of the bytes in
+ * @accept, or NULL if no such byte is found.
+ *
+ * Example:
+ *	char otherbytes[] = "Hello \0world";
+ *	size_t otherbytes_len = sizeof(otherbytes) - 1;
+ *	char *r = mempbrkm(otherbytes, otherbytes_len, "\0b", 2);
+ *	if (r) {
+ *		printf("Found %c\n", *r);
+ *	} else {
+ *		printf("Nada\n");
+ *	}
+ *
+ */
+void *mempbrkm(const void *data, size_t len, const void *accept, size_t accept_len);
+
+/**
+ * mempbrk - locates the first occurrence in @data of any bytes in @accept
+ * @data: where we search
+ * @len: length of data in bytes
+ * @accept: NUL terminated string containing the bytes we search for
+ *
+ * Returns a pointer to the byte in @data that matches one of the bytes in
+ * @accept, or NULL if no such byte is found.
+ *
+ * Example:
+ *
+ *	r = mempbrk(otherbytes, otherbytes_len, "abcde");
+ *	if (r) {
+ *		printf("Found %c\n", *r);
+ *	} else {
+ *		printf("Nada\n");
+ *	}
+ */
+static inline char *mempbrk(const void *data, size_t len, const char *accept)
+{
+	return mempbrkm(data, len, accept, strlen(accept));
+}
+
+/**
+ * memcchr - scan memory until a character does _not_ match @c
+ * @data: pointer to memory to scan
+ * @data_len: length of data
+ * @c: character to scan for
+ *
+ * The complement of memchr().
+ *
+ * Returns a pointer to the first character which is _not_ @c. If all memory in
+ * @data is @c, returns NULL.
+ *
+ * Example:
+ *	char somebytes[] = "HI By\0e";
+ *	size_t bytes_len = sizeof(somebytes) - 1;
+ *	r = memcchr(somebytes, ' ', bytes_len);
+ *	if (r) {
+ *		printf("Found %c after trimming spaces\n", *r);
+ *	}
+ */
+void *memcchr(void const *data, int c, size_t data_len);
+
+/**
+ * memeq - Are two byte arrays equal?
+ * @a: first array
+ * @al: bytes in first array
+ * @b: second array
+ * @bl: bytes in second array
+ *
+ * Example:
+ *	if (memeq(somebytes, bytes_len, otherbytes, otherbytes_len)) {
+ *		printf("memory blocks are the same!\n");
+ *	}
+ */
+static inline bool memeq(const void *a, size_t al, const void *b, size_t bl)
+{
+	return al == bl && !memcmp(a, b, bl);
+}
+
+/**
+ * memstarts - determine if @data starts with @prefix
+ * @data: does this begin with @prefix?
+ * @data_len: bytes in @data
+ * @prefix: does @data begin with these bytes?
+ * @prefix_len: bytes in @prefix
+ *
+ * Returns true if @data starts with @prefix, otherwise return false.
+ *
+ * Example:
+ *	if (memstarts(somebytes, bytes_len, otherbytes, otherbytes_len)) {
+ *		printf("somebytes starts with otherbytes!\n");
+ *	}
+ */
+static inline bool memstarts(void const *data, size_t data_len,
+		void const *prefix, size_t prefix_len)
+{
+	if (prefix_len > data_len)
+		return false;
+	return memeq(data, prefix_len, prefix, prefix_len);
+}
+
+/**
+ * memeqstr - Is a byte array equal to a NUL terminated string?
+ * @data: byte array
+ * @length: length of @data in bytes
+ * @string: NUL terminated string
+ *
+ * The '\0' byte is ignored when checking if @bytes == @string.
+ *
+ * Example:
+ *	if (memeqstr(somebytes, bytes_len, "foo")) {
+ *		printf("somebytes == 'foo'!\n");
+ *	}
+ */
+static inline bool memeqstr(const void *data, size_t length, const char *string)
+{
+	return memeq(data, length, string, strlen(string));
+}
+
+/**
+ * memstarts_str - Does this byte array start with a string prefix?
+ * @a: byte array
+ * @al: length in bytes
+ * @s: string prefix
+ *
+ * Example:
+ *	if (memstarts_str(somebytes, bytes_len, "It")) {
+ *		printf("somebytes starts with 'It'\n");
+ *	}
+ */
+static inline bool memstarts_str(const void *a, size_t al, const char *s)
+{
+	return memstarts(a, al, s, strlen(s));
+}
+
+/**
+ * memends - Does this byte array end with a given byte-array suffix?
+ * @s: byte array
+ * @s_len: length in bytes
+ * @suffix: byte array suffix
+ * @suffix_len: length of suffix in bytes
+ *
+ * Returns true if @suffix appears as a substring at the end of @s,
+ * false otherwise.
+ */
+static inline bool memends(const void *s, size_t s_len, const void *suffix, size_t suffix_len)
+{
+	return (s_len >= suffix_len) && (memcmp((const char *)s + s_len - suffix_len,
+						suffix, suffix_len) == 0);
+}
+
 #endif /* CCAN_MEM_H */
diff --git a/ccan/mem/test/api.c b/ccan/mem/test/api.c
index 2ee15c5..89b6acb 100644
--- a/ccan/mem/test/api.c
+++ b/ccan/mem/test/api.c
@@ -8,9 +8,11 @@ int main(void)
 	char haystack2[] = "ab\0ab\0ab\0ab";
 	char needle1[] = "ab";
 	char needle2[] = "d\0e";
+	char scan1[] = "aaaab";
+	char scan2[] = "\0\0\0b";
 
 	/* This is how many tests you plan to run */
-	plan_tests(19);
+	plan_tests(42);
 
 	ok1(memmem(haystack1, sizeof(haystack1), needle1, 2) == haystack1);
 	ok1(memmem(haystack1, sizeof(haystack1), needle1, 3) == NULL);
@@ -37,6 +39,39 @@ int main(void)
 
 	ok1(memrchr(needle1, '\0', 2) == NULL);
 
+#define S(x) (x), sizeof(x) - 1
+	ok1(mempbrkm(S(haystack1), S("\0efgh")) == haystack1 + 4);
+	ok1(mempbrkm(S(haystack1), S("jklmn")) == NULL);
+	ok1(mempbrkm(S(haystack1), S("sd\0a")) == haystack1 + 0);
+
+	ok1(mempbrk(haystack1, sizeof(haystack1), "bcd\0a") == haystack1 + 1);
+	ok1(mempbrk(haystack1, sizeof(haystack1), "\0") == NULL);
+
+	ok1(memcchr(scan1, 'a', sizeof(scan1)) == scan1 + 4);
+	ok1(memcchr(scan1, 'b', sizeof(scan1)) == scan1);
+	ok1(memcchr(scan2, '\0', sizeof(scan2)) == scan2 + 3);
+	ok1(memcchr(scan2, '\0', sizeof(scan2) - 2) == NULL);
+
+	ok1(memeq(haystack1, sizeof(haystack1), haystack1, sizeof(haystack1)));
+	ok1(!memeq(haystack1, sizeof(haystack1), haystack2, sizeof(haystack2)));
+
+	ok1(memeqstr(scan1, sizeof(scan1) - 1, scan1));
+	ok1(!memeqstr(scan1, sizeof(scan1), scan1));
+	ok1(!memeqstr(scan1, sizeof(scan1), "aaaa"));
+
+	ok1(memstarts(S("a\0bcdef"), S("a\0bc")));
+	ok1(!memstarts(S("a\0bcdef"), S("a\0bcG")));
+	ok1(!memstarts(S("a\0bcdef"), S("a\0bcdefg")));
+
+	ok1(memstarts_str(scan1, sizeof(scan1), scan1));
+	ok1(!memstarts_str(scan1, sizeof(scan1), "ab"));
+
+	ok1(memends(S("abcdef"), S("abcdef")));
+	ok1(!memends(S("abcdef"), S("abcdefg")));
+	ok1(!memends(S("a\0bcdef"), S("a\0b")));
+	ok1(memends(S("a\0bcdef"), S("ef")));
+
+
 	/* This exits depending on whether all tests passed */
 	return exit_status();
 }
-- 
2.5.1

_______________________________________________
ccan mailing list
ccan@lists.ozlabs.org
https://lists.ozlabs.org/listinfo/ccan

[PATCH v4 2/4] bytestring: use newly added mem helpers

[Cody P Schafer]

Reviwed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Cody P Schafer <dev@codyps.com>
---
 ccan/bytestring/bytestring.h | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)

diff --git a/ccan/bytestring/bytestring.h b/ccan/bytestring/bytestring.h
index dd1f913..bc99e79 100644
--- a/ccan/bytestring/bytestring.h
+++ b/ccan/bytestring/bytestring.h
@@ -94,8 +94,7 @@ static inline struct bytestring bytestring_from_string(const char *s)
  */
 static inline bool bytestring_eq(struct bytestring a, struct bytestring b)
 {
-	return (a.len == b.len)
-		&& (memcmp(a.ptr, b.ptr, a.len) == 0);
+	return memeq(a.ptr, a.len, b.ptr, b.len);
 }
 
 /**
@@ -149,8 +148,7 @@ static inline struct bytestring bytestring_slice(struct bytestring s,
 static inline bool bytestring_starts(struct bytestring s,
 				     struct bytestring prefix)
 {
-	return (s.len >= prefix.len) && (memcmp(s.ptr,
-						prefix.ptr, prefix.len) == 0);
+	return memstarts(s.ptr, s.len, prefix.ptr, prefix.len);
 }
 
 /**
@@ -163,8 +161,7 @@ static inline bool bytestring_starts(struct bytestring s,
 static inline bool bytestring_ends(struct bytestring s,
 				   struct bytestring suffix)
 {
-	return (s.len >= suffix.len) && (memcmp(s.ptr + s.len - suffix.len,
-						suffix.ptr, suffix.len) == 0);
+	return memends(s.ptr, s.len, suffix.ptr, suffix.len);
 }
 
 /**
-- 
2.5.1

_______________________________________________
ccan mailing list
ccan@lists.ozlabs.org
https://lists.ozlabs.org/listinfo/ccan

[PATCH v4 3/4] mem: mark all functions as PURE

[Cody P Schafer]

This might allow compilers that support the anotation to make better
choices when optimizing, and all these functions meet the requirements
for being marked pure.

Signed-off-by: Cody P Schafer <dev@codyps.com>
---
 ccan/mem/_info |  1 +
 ccan/mem/mem.h | 11 +++++++++++
 2 files changed, 12 insertions(+)

diff --git a/ccan/mem/_info b/ccan/mem/_info
index 19b2228..6389e35 100644
--- a/ccan/mem/_info
+++ b/ccan/mem/_info
@@ -18,6 +18,7 @@ int main(int argc, char *argv[])
 		return 1;
 
 	if (strcmp(argv[1], "depends") == 0) {
+		printf("ccan/compiler");
 		return 0;
 	}
 
diff --git a/ccan/mem/mem.h b/ccan/mem/mem.h
index 89b16d4..8d6bba9 100644
--- a/ccan/mem/mem.h
+++ b/ccan/mem/mem.h
@@ -3,16 +3,19 @@
 #define CCAN_MEM_H
 
 #include "config.h"
+#include <ccan/compiler/compiler.h>
 
 #include <string.h>
 #include <stdbool.h>
 
 #if !HAVE_MEMMEM
+PURE_FUNCTION
 void *memmem(const void *haystack, size_t haystacklen,
 	     const void *needle, size_t needlelen);
 #endif
 
 #if !HAVE_MEMRCHR
+PURE_FUNCTION
 void *memrchr(const void *s, int c, size_t n);
 #endif
 
@@ -37,6 +40,7 @@ void *memrchr(const void *s, int c, size_t n);
  *	}
  *
  */
+PURE_FUNCTION
 void *mempbrkm(const void *data, size_t len, const void *accept, size_t accept_len);
 
 /**
@@ -57,6 +61,7 @@ void *mempbrkm(const void *data, size_t len, const void *accept, size_t accept_l
  *		printf("Nada\n");
  *	}
  */
+PURE_FUNCTION
 static inline char *mempbrk(const void *data, size_t len, const char *accept)
 {
 	return mempbrkm(data, len, accept, strlen(accept));
@@ -81,6 +86,7 @@ static inline char *mempbrk(const void *data, size_t len, const char *accept)
  *		printf("Found %c after trimming spaces\n", *r);
  *	}
  */
+PURE_FUNCTION
 void *memcchr(void const *data, int c, size_t data_len);
 
 /**
@@ -95,6 +101,7 @@ void *memcchr(void const *data, int c, size_t data_len);
  *		printf("memory blocks are the same!\n");
  *	}
  */
+PURE_FUNCTION
 static inline bool memeq(const void *a, size_t al, const void *b, size_t bl)
 {
 	return al == bl && !memcmp(a, b, bl);
@@ -114,6 +121,7 @@ static inline bool memeq(const void *a, size_t al, const void *b, size_t bl)
  *		printf("somebytes starts with otherbytes!\n");
  *	}
  */
+PURE_FUNCTION
 static inline bool memstarts(void const *data, size_t data_len,
 		void const *prefix, size_t prefix_len)
 {
@@ -135,6 +143,7 @@ static inline bool memstarts(void const *data, size_t data_len,
  *		printf("somebytes == 'foo'!\n");
  *	}
  */
+PURE_FUNCTION
 static inline bool memeqstr(const void *data, size_t length, const char *string)
 {
 	return memeq(data, length, string, strlen(string));
@@ -151,6 +160,7 @@ static inline bool memeqstr(const void *data, size_t length, const char *string)
  *		printf("somebytes starts with 'It'\n");
  *	}
  */
+PURE_FUNCTION
 static inline bool memstarts_str(const void *a, size_t al, const char *s)
 {
 	return memstarts(a, al, s, strlen(s));
@@ -166,6 +176,7 @@ static inline bool memstarts_str(const void *a, size_t al, const char *s)
  * Returns true if @suffix appears as a substring at the end of @s,
  * false otherwise.
  */
+PURE_FUNCTION
 static inline bool memends(const void *s, size_t s_len, const void *suffix, size_t suffix_len)
 {
 	return (s_len >= suffix_len) && (memcmp((const char *)s + s_len - suffix_len,
-- 
2.5.1

_______________________________________________
ccan mailing list
ccan@lists.ozlabs.org
https://lists.ozlabs.org/listinfo/ccan

[PATCH v4 4/4] mem: add memends_str() helper for symmetry

[Cody P Schafer]

Signed-off-by: Cody P Schafer <dev@codyps.com>
---
 ccan/mem/mem.h      | 17 +++++++++++++++++
 ccan/mem/test/api.c |  6 +++++-
 2 files changed, 22 insertions(+), 1 deletion(-)

diff --git a/ccan/mem/mem.h b/ccan/mem/mem.h
index 8d6bba9..1afe508 100644
--- a/ccan/mem/mem.h
+++ b/ccan/mem/mem.h
@@ -183,4 +183,21 @@ static inline bool memends(const void *s, size_t s_len, const void *suffix, size
 						suffix, suffix_len) == 0);
 }
 
+/**
+ * memends_str - Does this byte array end with a string suffix?
+ * @a: byte array
+ * @al: length in bytes
+ * @s: string suffix
+ *
+ * Example:
+ *	if (memends_str(somebytes, bytes_len, "It")) {
+ *		printf("somebytes ends with with 'It'\n");
+ *	}
+ */
+PURE_FUNCTION
+static inline bool memends_str(const void *a, size_t al, const char *s)
+{
+	return memends(a, al, s, strlen(s));
+}
+
 #endif /* CCAN_MEM_H */
diff --git a/ccan/mem/test/api.c b/ccan/mem/test/api.c
index 89b6acb..9ec226d 100644
--- a/ccan/mem/test/api.c
+++ b/ccan/mem/test/api.c
@@ -12,7 +12,7 @@ int main(void)
 	char scan2[] = "\0\0\0b";
 
 	/* This is how many tests you plan to run */
-	plan_tests(42);
+	plan_tests(46);
 
 	ok1(memmem(haystack1, sizeof(haystack1), needle1, 2) == haystack1);
 	ok1(memmem(haystack1, sizeof(haystack1), needle1, 3) == NULL);
@@ -71,6 +71,10 @@ int main(void)
 	ok1(!memends(S("a\0bcdef"), S("a\0b")));
 	ok1(memends(S("a\0bcdef"), S("ef")));
 
+	ok1(memends_str(S("abcdef"), "abcdef"));
+	ok1(!memends_str(S("abcde\0f"), "d\0f"));
+	ok1(!memends_str(S("a\0bcdef"), "a"));
+	ok1(memends_str(S("a\0bcdef"), "ef"));
 
 	/* This exits depending on whether all tests passed */
 	return exit_status();
-- 
2.5.1

_______________________________________________
ccan mailing list
ccan@lists.ozlabs.org
https://lists.ozlabs.org/listinfo/ccan

Re: [PATCH v4 0/4] mem: extend with a few additional helpers similar to existing string-only functions

[David Gibson]

[-- Attachment #1.1: Type: text/plain, Size: 959 bytes --]

On Sat, Sep 05, 2015 at 09:21:02PM -0400, Cody P Schafer wrote:
> since v3:
> 
>  - fix memeq() return value
>  - add memends_str() helper
> 
> since v2:
> 
>  - switch all the macros to static inline to avoid multiple evaluations of args
>  - add patch adding PURE_FUNCTION attributes
> 
> since v1:
> 
>  - drop a patch that David pulled in
>  - add extra macro parens
>  - add blank lines after defintions
>  - covert memcchr's while loop to a for.
>  - change examples for mempbrkm and mempbrk to use '\0' in places
>  - fix spelling
>  - use memeq() even where we know the mem lengths are equal in memstarts()
>  - add a reviewed-by to the 2nd patch, no other changes to that patch.

Looks good, I've applied this to the repository.


-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

[-- Attachment #1.2: Type: application/pgp-signature, Size: 819 bytes --]

[-- Attachment #2: Type: text/plain, Size: 127 bytes --]

_______________________________________________
ccan mailing list
ccan@lists.ozlabs.org
https://lists.ozlabs.org/listinfo/ccan