refactor xxhash

randy3k · randy3k · commit 71f4cf58edf3 · 2020-06-15T20:49:27.000-07:00
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -23,9 +23,7 @@ Description: Provides high performance container data types such
     those offered by other packages.
 License: MIT + file LICENSE
 URL: https://github.com/randy3k/collections
-Depends:
-    R (>= 3.4.0)
-Suggests: 
+Suggests:
     covr (>= 3.4.0),
     testthat (>= 2.3.1)
 ByteCompile: yes
diff --git a/src/Makevars b/src/Makevars
@@ -1,2 +1,2 @@
-SOURCES = collections.c deque.c dict.c priority_queue.c queue.c stack.c utils.c
-OBJECTS = tommyds/tommy.o xxhash/xxh.o $(SOURCES:.c=.o)
+SOURCES = collections.c deque.c dict.c priority_queue.c queue.c stack.c xxh.c utils.c
+OBJECTS = tommyds/tommy.o $(SOURCES:.c=.o)
diff --git a/src/dict.c b/src/dict.c
@@ -1,7 +1,7 @@
 #include "tommyds/tommyhashlin.h"
 #include "dict.h"
+#include "xxh.h"
 #include "utils.h"
-#include "xxhash/xxh.h"
 
 #define INITIAL_SIZE 16
 #define GROW_FACTOR 1.5
@@ -127,16 +127,20 @@ tommy_hash_t digest(SEXP key) {
         SEXP c = Rf_asChar(key);
         s = Rf_translateCharUTF8(c);
         return XXH3_64bits(s, strlen(s));
+    }
 
-    } else if (is_hashable(key)) {
+    if (is_hashable(key)) {
         return xxh_digest(key);
+    }
 
-    } else if (Rf_isEnvironment(key)) {
+    if (Rf_isEnvironment(key)) {
         s = R_alloc(sizeof(char), 30);
         sprintf((char*) s, "env<%p>", key);
         return XXH3_64bits(s, strlen(s));
 
-    } else if (Rf_isFunction(key)) {
+    }
+
+    if (Rf_isFunction(key)) {
         SEXP key2 = PROTECT(Rf_shallow_duplicate(key));
         // the digest function will also hash the closure environment and attributes
         SET_CLOENV(key2, R_NilValue);
@@ -146,6 +150,7 @@ tommy_hash_t digest(SEXP key) {
         return h;
 
     }
+
     Rf_error("key is not hashable");
 }
 
diff --git a/src/xxh.c b/src/xxh.c
@@ -17,6 +17,27 @@ static void OutBytes(R_outpstream_t stream, void *buf, int length)
 
 
 XXH64_hash_t xxh_digest(SEXP x) {
+
+    if (Rf_isVectorAtomic(x) && !ALTREP(x)) {
+        char *p;
+        if (TYPEOF(x) == INTSXP) {
+            p = (char*) INTEGER(x);
+            return XXH3_64bits(p, Rf_length(x) * sizeof(int));
+        }
+        if (TYPEOF(x) == REALSXP) {
+            p = (char*) REAL(x);
+            return XXH3_64bits(p, Rf_length(x) * sizeof(double));
+        }
+        if (TYPEOF(x) == LGLSXP) {
+            p = (char*) LOGICAL(x);
+            return XXH3_64bits(p, Rf_length(x) * sizeof(int));
+        }
+        if (TYPEOF(x) == RAWSXP) {
+            p = (char*) RAW(x);
+            return XXH3_64bits(p, Rf_length(x));
+        }
+    }
+
     XXH3_state_t* const xxh_state = XXH3_createState();
     XXH3_64bits_reset_withSeed(xxh_state, 0);
     struct R_outpstream_st stream;
diff --git a/src/xxh.h b/src/xxh.h
@@ -5,13 +5,21 @@
 #include <config.h>
 #endif
 
+#include "Rversion.h"
+
+#if (R_VERSION < R_Version(3, 5, 0))
+
+# define ALTREP(x) false
+
+#endif
+
 #define USE_RINTERNALS
 #include <R.h>
 #include <Rinternals.h>
 #include <R_ext/Rdynload.h>
 #undef USE_RINTERNALS
 
-#include "xxh3.h"
+#include "xxhash/xxh3.h"
 
 XXH64_hash_t xxh_digest(SEXP x);
 
