New upstream version 6.2.0upstream/6.2.0

1 files changed, 30 insertions, 3 deletions

diff --git a/src/regcomp.c b/src/regcomp.c
index 11ba1e7..5c924b5 100644
--- a/src/regcomp.c
+++ b/src/regcomp.c
@@ -1230,6 +1230,11 @@ compile_length_enclose_node(EncloseNode* node, regex_t* reg)
         len += (IS_ENCLOSE_RECURSION(node)
                 ? SIZE_OP_MEMORY_END_REC : SIZE_OP_MEMORY_END);
     }
+    else if (IS_ENCLOSE_RECURSION(node)) {
+      len = SIZE_OP_MEMORY_START_PUSH;
+      len += tlen + (BIT_STATUS_AT(reg->bt_mem_end, node->regnum)
+                     ? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_REC);
+    }
     else
 #endif
     {
@@ -1321,6 +1326,14 @@ compile_enclose_node(EncloseNode* node, regex_t* reg)
       if (r) return r;
       r = add_opcode(reg, OP_RETURN);
     }
+    else if (IS_ENCLOSE_RECURSION(node)) {
+      if (BIT_STATUS_AT(reg->bt_mem_end, node->regnum))
+        r = add_opcode(reg, OP_MEMORY_END_PUSH_REC);
+      else
+        r = add_opcode(reg, OP_MEMORY_END_REC);
+      if (r) return r;
+      r = add_mem_num(reg, node->regnum);
+    }
     else
 #endif
     {
@@ -2231,6 +2244,7 @@ is_not_included(Node* x, Node* y, regex_t* reg)
             return 0;
           }
           else {
+            if (IS_NOT_NULL(xc->mbuf)) return 0;
             for (i = 0; i < SINGLE_BYTE_SIZE; i++) {
               if (! IS_CODE_SB_WORD(reg->enc, i)) {
                 if (!IS_NCCLASS_NOT(xc)) {
@@ -3673,6 +3687,8 @@ setup_comb_exp_check(Node* node, int state, ScanEnv* env)
 #define IN_NOT        (1<<1)
 #define IN_REPEAT     (1<<2)
 #define IN_VAR_REPEAT (1<<3)
+#define IN_CALL       (1<<4)
+#define IN_RECCALL    (1<<5)
 
 /* setup_tree does the following work.
  1. check empty loop. (set qn->target_empty_info)
@@ -3843,10 +3859,16 @@ setup_tree(Node* node, regex_t* reg, int state, ScanEnv* env)
         break;
 
       case ENCLOSE_MEMORY:
-        if ((state & (IN_ALT | IN_NOT | IN_VAR_REPEAT)) != 0) {
+        if ((state & (IN_ALT | IN_NOT | IN_VAR_REPEAT | IN_CALL)) != 0) {
           BIT_STATUS_ON_AT(env->bt_mem_start, en->regnum);
           /* SET_ENCLOSE_STATUS(node, NST_MEM_IN_ALT_NOT); */
         }
+        if (IS_ENCLOSE_CALLED(en))
+          state |= IN_CALL;
+        if (IS_ENCLOSE_RECURSION(en))
+          state |= IN_RECCALL;
+        else if ((state & IN_RECCALL) != 0)
+          SET_CALL_RECURSION(node);
         r = setup_tree(en->target, reg, state, env);
         break;
 
@@ -4160,6 +4182,9 @@ concat_opt_anc_info(OptAncInfo* to, OptAncInfo* left, OptAncInfo* right,
   if (right_len == 0) {
     to->right_anchor |= left->right_anchor;
   }
+  else {
+    to->right_anchor |= (left->right_anchor & ANCHOR_PREC_READ_NOT);
+  }
 }
 
 static int
@@ -5003,12 +5028,14 @@ set_optimize_info_from_tree(Node* node, regex_t* reg, ScanEnv* scan_env)
   if (r) return r;
 
   reg->anchor = opt.anc.left_anchor & (ANCHOR_BEGIN_BUF |
-        ANCHOR_BEGIN_POSITION | ANCHOR_ANYCHAR_STAR | ANCHOR_ANYCHAR_STAR_ML);
+        ANCHOR_BEGIN_POSITION | ANCHOR_ANYCHAR_STAR | ANCHOR_ANYCHAR_STAR_ML |
+        ANCHOR_LOOK_BEHIND);
 
   if ((opt.anc.left_anchor & (ANCHOR_LOOK_BEHIND | ANCHOR_PREC_READ_NOT)) != 0)
     reg->anchor &= ~ANCHOR_ANYCHAR_STAR_ML;
 
-  reg->anchor |= opt.anc.right_anchor & (ANCHOR_END_BUF | ANCHOR_SEMI_END_BUF);
+  reg->anchor |= opt.anc.right_anchor & (ANCHOR_END_BUF | ANCHOR_SEMI_END_BUF |
+       ANCHOR_PREC_READ_NOT);
 
   if (reg->anchor & (ANCHOR_END_BUF | ANCHOR_SEMI_END_BUF)) {
     reg->anchor_dmin = opt.len.min;