lcs_upper_bound's second sum is required all the time and try to handle endians

1 files changed, 27 insertions, 37 deletions

diff --git a/lcs.c b/lcs.c
index f3a6203..2ea6f87 100644
--- a/lcs.c
+++ b/lcs.c
@@ -158,11 +158,7 @@ int lcs_32_rough(unsigned char *str1, unsigned char *str2) {
 #endif
 
 #ifdef USE_PRE_LCS
-#define DO_SECOND_SUM
-
-#ifdef DO_SECOND_SUM
 #include "arpa/inet.h"
-#endif
 
 int lcs_upper_bound(const uint32_t * const str1_int, const uint32_t * const str2_int) {
     static char contains[256] = {0};
@@ -170,7 +166,7 @@ int lcs_upper_bound(const uint32_t * const str1_int, const uint32_t * const str2
     const unsigned char * const str2 = (const unsigned char * const)str2_int;
     int first_sum;
     int ret;
-    
+
     contains[str1[0]] = contains[str1[1]] = contains[str1[2]]
         = contains[str1[3]] = contains[str1[4]] = contains[str1[5]]
         = contains[str1[6]] = contains[str1[7]] = contains[str1[8]]
@@ -185,38 +181,32 @@ int lcs_upper_bound(const uint32_t * const str1_int, const uint32_t * const str2
         + contains[str2[12]] + contains[str2[13]] + contains[str2[14]]
         + contains[str2[15]];
 
-    // can be off up to two nibbles (e.g., "0110" and "1122")
-    ret = first_sum * 2 + 2;
-#ifdef DO_SECOND_SUM
-    // We can only increase our accuracy by 1, so do second_sum only if we can
-    // turn a false-positive into a negative
-    if (ret == MSF) {
-        uint32_t bo_str2_int[4];
-        uint32_t shifted_str2_int[4];
-        const unsigned char * const str2_r = (const unsigned char * const)shifted_str2_int;
-        int second_sum;
-
-        bo_str2_int[0] = ntohl(str2_int[0]);
-        bo_str2_int[1] = ntohl(str2_int[1]);
-        bo_str2_int[2] = ntohl(str2_int[2]);
-        bo_str2_int[3] = ntohl(str2_int[3]);
-
-        shifted_str2_int[0] = (bo_str2_int[0] << 4) | (bo_str2_int[1] >> 28);
-        shifted_str2_int[1] = (bo_str2_int[1] << 4) | (bo_str2_int[2] >> 28);
-        shifted_str2_int[2] = (bo_str2_int[2] << 4) | (bo_str2_int[3] >> 28);
-        shifted_str2_int[3] = (bo_str2_int[3] << 4) | (bo_str2_int[0] >> 28);
-
-        second_sum = contains[str2_r[0]] + contains[str2_r[1]] + contains[str2_r[2]]
-            + contains[str2_r[3]] + contains[str2_r[4]] + contains[str2_r[5]]
-            + contains[str2_r[6]] + contains[str2_r[7]] + contains[str2_r[8]]
-            + contains[str2_r[9]] + contains[str2_r[10]] + contains[str2_r[11]]
-            + contains[str2_r[12]] + contains[str2_r[13]] + contains[str2_r[14]]
-            + contains[str2_r[15]];
-
-        // can be off up to one nibble (e.g., "1110" and "2111")
-        ret = MAX(first_sum, second_sum) * 2 + 1;
-    }
-#endif
+    uint32_t bo_str2_int[4];
+    uint32_t shifted_str2_int[4];
+    const unsigned char * const str2_r = (const unsigned char * const)shifted_str2_int;
+    int second_sum;
+
+    bo_str2_int[0] = ntohl(str2_int[0]);
+    bo_str2_int[1] = ntohl(str2_int[1]);
+    bo_str2_int[2] = ntohl(str2_int[2]);
+    bo_str2_int[3] = ntohl(str2_int[3]);
+
+    shifted_str2_int[0] = (bo_str2_int[0] << 4) | (bo_str2_int[1] >> 28);
+    shifted_str2_int[1] = (bo_str2_int[1] << 4) | (bo_str2_int[2] >> 28);
+    shifted_str2_int[2] = (bo_str2_int[2] << 4) | (bo_str2_int[3] >> 28);
+    shifted_str2_int[3] = (bo_str2_int[3] << 4);
+
+    // don't add str2_r[LE_BYTE_INDEX(15)] because it is known-bad due to zero filling
+    second_sum = contains[str2_r[0]] + contains[str2_r[1]] + contains[str2_r[2]]
+        + contains[str2_r[3]] + contains[str2_r[4]] + contains[str2_r[5]]
+        + contains[str2_r[6]] + contains[str2_r[7]] + contains[str2_r[8]]
+        + contains[str2_r[9]] + contains[str2_r[10]] + contains[str2_r[11]]
+        + contains[str2_r[BE_BYTE_INDEX(12)]]
+        + contains[str2_r[BE_BYTE_INDEX(13)]]
+        + contains[str2_r[BE_BYTE_INDEX(14)]];
+
+    // can be off up to two nibbles
+    ret = MAX(first_sum, second_sum) * 2 + 2;
 
     contains[str1[0]] = contains[str1[1]] = contains[str1[2]]
         = contains[str1[3]] = contains[str1[4]] = contains[str1[5]]