From 7927ac0ee31b60954c9c0ad42236b25498c4f264 Mon Sep 17 00:00:00 2001
From: "Philip.Hazel" <Philip.Hazel@gmail.com>
Date: Sat, 13 May 2017 17:46:27 +0000
Subject: [PATCH] Updates to experimental conversion code.

---
 src/pcre2_convert.c   | 49 ++++++++++++++++++++++++++++++++++---------
 testdata/testinput24  | 17 +++++++++++++++
 testdata/testoutput24 | 29 +++++++++++++++++++++++++
 3 files changed, 85 insertions(+), 10 deletions(-)

diff --git a/src/pcre2_convert.c b/src/pcre2_convert.c
index c391521..d4e44b4 100644
--- a/src/pcre2_convert.c
+++ b/src/pcre2_convert.c
@@ -118,7 +118,9 @@ PCRE2_UCHAR *pp = p;
 PCRE2_UCHAR *endp = p + use_length - 1;  /* Allow for trailing zero */
 PCRE2_SIZE convlength = 0;
 
+uint32_t bracount = 0;
 uint32_t posix_class_state = POSIX_CLASS_NOT_STARTED;
+uint32_t lastspecial = 0;
 BOOL extended = (pattype & PCRE2_CONVERT_POSIX_EXTENDED) != 0;
 BOOL inclass = FALSE;
 BOOL nextisliteral = FALSE;
@@ -130,7 +132,13 @@ BOOL nextisliteral = FALSE;
 
 *bufflenptr = plength;
 
-/* Now scan the input */
+/* Now scan the input. In non-extended patterns, an initial asterisk is treated 
+as literal. Still figuring out what happens in extended patterns... */
+
+if (plength > 0 && *posix == CHAR_ASTERISK)
+  {
+  if (!extended) nextisliteral = TRUE; 
+  } 
 
 while (plength > 0)
   {
@@ -262,35 +270,56 @@ while (plength > 0)
       {
       if (isdigit(*posix)) PUTCHARS(STR_BACKSLASH); 
       if (p + 1 > endp) return PCRE2_ERROR_NOMEMORY;
-      *p++ = *posix++;
+      lastspecial = *p++ = *posix++;
       plength--;  
       }
     else nextisliteral = TRUE;
     break;
     
+    case CHAR_RIGHT_PARENTHESIS:
+    if (!extended || bracount == 0) goto ESCAPE_LITERAL;
+    bracount--;
+    goto COPY_SPECIAL;
+
+    case CHAR_LEFT_PARENTHESIS:
+    bracount++;
+    /* Fall through */  
+
     case CHAR_QUESTION_MARK:
     case CHAR_PLUS:
     case CHAR_LEFT_CURLY_BRACKET:   
     case CHAR_RIGHT_CURLY_BRACKET:   
     case CHAR_VERTICAL_LINE:
-    case CHAR_LEFT_PARENTHESIS:
-    case CHAR_RIGHT_PARENTHESIS:
-    if (!extended) PUTCHARS(STR_BACKSLASH);
+    if (!extended) goto ESCAPE_LITERAL;
     /* Fall through */ 
     
-    case CHAR_ASTERISK:
     case CHAR_DOT:
-    case CHAR_CIRCUMFLEX_ACCENT:
     case CHAR_DOLLAR_SIGN:   
+    COPY_SPECIAL:
+    lastspecial = c; 
     if (p + 1 > endp) return PCRE2_ERROR_NOMEMORY;
-    *p++ = sc;
-    break;  
-    
+    *p++ = c;
+    break; 
+
+    case CHAR_ASTERISK:
+    if (lastspecial != CHAR_ASTERISK) goto COPY_SPECIAL;
+    break;   /* Ignore second and subsequent asterisks */  
+
+    case CHAR_CIRCUMFLEX_ACCENT:
+    if (extended || 
+          lastspecial == 0 || 
+          lastspecial == CHAR_LEFT_PARENTHESIS ||
+          lastspecial == CHAR_VERTICAL_LINE) 
+      goto COPY_SPECIAL;
+    /* Fall through */      
+
     default:
     if (c < 256 && strchr("\\{}?*+[]()|.^$", c) != NULL)
       {
+      ESCAPE_LITERAL: 
       PUTCHARS(STR_BACKSLASH);
       }
+    lastspecial = 0xff;  /* Indicates nothing special */   
     if (p + clength > endp) return PCRE2_ERROR_NOMEMORY;
     memcpy(p, posix - clength, CU2BYTES(clength));
     p += clength;
diff --git a/testdata/testinput24 b/testdata/testinput24
index c98fd25..f18277b 100644
--- a/testdata/testinput24
+++ b/testdata/testinput24
@@ -247,6 +247,11 @@
 \= Expect no match
     aab
 
+/(ab)c)d]/
+    Xabc)d]Y
+
+/a***b/
+
 #pattern convert=unset
 #pattern convert=posix_basic
 
@@ -261,6 +266,18 @@
 
 /^how to \^how to/
 
+/*abc/
+    X*abcY
+
+/**abc/
+    XabcY
+    X*abcY
+    X**abcY  
+
+/^b\(c^d\)\(^e^f\)/
+
+/a***b/
+
 #pattern convert=unset
 
 /abc/
diff --git a/testdata/testoutput24 b/testdata/testoutput24
index bd5a8d2..86ed7a5 100644
--- a/testdata/testoutput24
+++ b/testdata/testoutput24
@@ -396,6 +396,15 @@ No match
     aab
 No match
 
+/(ab)c)d]/
+(ab)c\)d\]
+    Xabc)d]Y
+ 0: abc)d]
+ 1: ab
+
+/a***b/
+a*b
+
 #pattern convert=unset
 #pattern convert=posix_basic
 
@@ -417,6 +426,26 @@ how.to how\.to
 /^how to \^how to/
 ^how to \^how to
 
+/*abc/
+\*abc
+    X*abcY
+ 0: *abc
+
+/**abc/
+\**abc
+    XabcY
+ 0: abc
+    X*abcY
+ 0: *abc
+    X**abcY  
+ 0: **abc
+
+/^b\(c^d\)\(^e^f\)/
+^b(c\^d)(^e\^f)
+
+/a***b/
+a*b
+
 #pattern convert=unset
 
 /abc/