TRE binding: add wide-character functions.

11 files changed, 86 insertions, 56 deletions

diff --git a/src/common.c b/src/common.c
index 09d595c..0b444c0 100644
--- a/src/common.c
+++ b/src/common.c
@@ -166,7 +166,7 @@ void buffer_addvalue (TBuffer *buf, int stackpos) {
   buffer_addlstring (buf, p, len);
 }
 
-static void bufferZ_addlstring (TBuffer *buf, const void *src, size_t len) {
+void bufferZ_addlstring (TBuffer *buf, const void *src, size_t len) {
   int n;
   size_t header[2] = { ID_STRING };
   header[1] = len;
@@ -176,7 +176,7 @@ static void bufferZ_addlstring (TBuffer *buf, const void *src, size_t len) {
   if (n) buffer_addlstring (buf, NULL, N_ALIGN - n);
 }
 
-static void bufferZ_addnum (TBuffer *buf, size_t num) {
+void bufferZ_addnum (TBuffer *buf, size_t num) {
   size_t header[2] = { ID_NUMBER };
   header[1] = num;
   buffer_addlstring (buf, header, sizeof (header));
@@ -220,47 +220,6 @@ void bufferZ_putrepstring (TBuffer *BufRep, int reppos, int nsub) {
   }
 }
 
-#ifdef WIDE_CHAR_GSUB
-/* 1. When called repeatedly on the same TBuffer, its existing data
-      is discarded and overwritten by the new data.
-   2. The TBuffer's array is never shrunk by this function.
-*/
-void bufferZ_putrepstringW (TBuffer *BufRep, int reppos, int nsub) {
-  wchar_t dbuf[] = { 0, 0 };
-  size_t replen;
-  const wchar_t *p = (const wchar_t*) lua_tolstring (BufRep->L, reppos, &replen);
-  replen /= sizeof(wchar_t);
-  const wchar_t *end = p + replen;
-  BufRep->top = 0;
-  while (p < end) {
-    const wchar_t *q;
-    for (q = p; q < end && *q != L'%'; ++q)
-      {}
-    if (q != p)
-      bufferZ_addlstring (BufRep, p, (q - p) * sizeof(wchar_t));
-    if (q < end) {
-      if (++q < end) {  /* skip % */
-        if (iswdigit (*q)) {
-          int num;
-          *dbuf = *q;
-          num = wcstol (dbuf, NULL, 10);
-          if (num == 1 && nsub == 0)
-            num = 0;
-          else if (num > nsub) {
-            freelist_free (BufRep->freelist);
-            luaL_error (BufRep->L, "invalid capture index");
-          }
-          bufferZ_addnum (BufRep, num);
-        }
-        else bufferZ_addlstring (BufRep, q, 1 * sizeof(wchar_t));
-      }
-      p = q + 1;
-    }
-    else break;
-  }
-}
-#endif
-
 /******************************************************************************
   The intended use of this function is as follows:
         size_t iter = 0;
@@ -287,4 +246,3 @@ int bufferZ_next (TBuffer *buf, size_t *iter, size_t *num, const char **str) {
   }
   return 0;
 }
-
diff --git a/src/common.h b/src/common.h
index 04ef7cb..95aa79d 100644
--- a/src/common.h
+++ b/src/common.h
@@ -83,8 +83,9 @@ void buffer_addvalue (TBuffer *buf, int stackpos);
 void buffer_pushresult (TBuffer *buf);
 
 void bufferZ_putrepstring (TBuffer *buf, int reppos, int nsub);
-void bufferZ_putrepstringW (TBuffer *buf, int reppos, int nsub);
 int  bufferZ_next (TBuffer *buf, size_t *iter, size_t *len, const char **str);
+void bufferZ_addlstring (TBuffer *buf, const void *src, size_t len);
+void bufferZ_addnum (TBuffer *buf, size_t num);
 
 int  get_int_field (lua_State *L, const char* field);
 void set_int_field (lua_State *L, const char* field, int val);
diff --git a/src/tre/Makefile b/src/tre/Makefile
index 2350108..2d3964e 100644
--- a/src/tre/Makefile
+++ b/src/tre/Makefile
@@ -26,7 +26,6 @@ REGNAME = tre
 # === END OF USER SETTINGS ===
 
 OBJ    = ltre.o ltre_w.o ../common.o
-MYCFLAGS = -DWIDE_CHAR_GSUB
 
 include ../common.mak
 
diff --git a/src/tre/ltre_w.c b/src/tre/ltre_w.c
index 7f8f87b..d3bceb7 100644
--- a/src/tre/ltre_w.c
+++ b/src/tre/ltre_w.c
@@ -10,6 +10,8 @@
 
 #include <tre/tre.h>
 
+void bufferZ_putrepstringW (TBuffer *BufRep, int reppos, int nsub);
+
 /* These 2 settings may be redefined from the command-line or the makefile.
  * They should be kept in sync between themselves and with the target name.
  */
@@ -132,8 +134,8 @@ static int generic_atfind (lua_State *L, int tfind) {
   res_match.pmatch = ud->match;
 
   /* execute the search */
-  res = tre_reganexec (&ud->r, argE.text, argE.textlen - argE.startoffset,
-                   &res_match, argP, argE.eflags);
+  res = tre_regawnexec (&ud->r, (const wchar_t*)argE.text,
+    (argE.textlen - argE.startoffset)/ALG_CHARSIZE, &res_match, argP, argE.eflags);
   if (ALG_ISMATCH (res)) {
     ALG_PUSHOFFSETS (L, ud, argE.startoffset, 0);
     if (tfind)
@@ -218,3 +220,42 @@ void add_wide_lib (lua_State *L, int methods)
 {
   luaL_register(L, NULL, methods ? posixmeta : rexlib);
 }
+
+/* 1. When called repeatedly on the same TBuffer, its existing data
+      is discarded and overwritten by the new data.
+   2. The TBuffer's array is never shrunk by this function.
+*/
+void bufferZ_putrepstringW (TBuffer *BufRep, int reppos, int nsub) {
+  wchar_t dbuf[] = { 0, 0 };
+  size_t replen;
+  const wchar_t *p = (const wchar_t*) lua_tolstring (BufRep->L, reppos, &replen);
+  replen /= sizeof(wchar_t);
+  const wchar_t *end = p + replen;
+  BufRep->top = 0;
+  while (p < end) {
+    const wchar_t *q;
+    for (q = p; q < end && *q != L'%'; ++q)
+      {}
+    if (q != p)
+      bufferZ_addlstring (BufRep, p, (q - p) * sizeof(wchar_t));
+    if (q < end) {
+      if (++q < end) {  /* skip % */
+        if (iswdigit (*q)) {
+          int num;
+          *dbuf = *q;
+          num = wcstol (dbuf, NULL, 10);
+          if (num == 1 && nsub == 0)
+            num = 0;
+          else if (num > nsub) {
+            freelist_free (BufRep->freelist);
+            luaL_error (BufRep->L, "invalid capture index");
+          }
+          bufferZ_addnum (BufRep, num);
+        }
+        else bufferZ_addlstring (BufRep, q, 1 * sizeof(wchar_t));
+      }
+      p = q + 1;
+    }
+    else break;
+  }
+}
diff --git a/test/tre_sets.lua b/test/tre_sets.lua
index 89f00b6..e755e03 100644
--- a/test/tre_sets.lua
+++ b/test/tre_sets.lua
@@ -115,6 +115,21 @@ local function set_m_wexec (lib, flg)
   }
 end
 
+local function set_m_waexec (lib, flg)
+  return {
+    Name = "Method waexec",
+    Method = "waexec",
+  --{patt},                 {subj, st}           { results }
+    { {L".+"},              {L"abcd",{}},        {1,4,{}}  }, -- [none]
+    { {L".+"},              {L"abcd",{},2},      {2,4,{}}  }, -- positive st
+    { {L".+"},              {L"abcd",{},-2},     {3,4,{}}  }, -- negative st
+    { {L".*"},              {L"abcd",{}},        {1,4,{}}  }, -- [none]
+    { {L"bc"},              {L"abc", {}},        {2,3,{}}  }, -- [none]
+    { {L "(.)b.(d)"},       {L"abcd",{}},        {1,4,{1,1,4,4}}},--[captures]
+    { {L"(a+)6+(b+)"},      {L"Taa66bbT",{},2},  {2,7,{2,3,6,7}}},--[st+captures]
+  }
+end
+
 local function set_m_wtfind (lib, flg)
   return {
     Name = "Method wtfind",
@@ -129,6 +144,20 @@ local function set_m_wtfind (lib, flg)
   }
 end
 
+local function set_m_watfind (lib, flg)
+  return {
+    Name = "Method watfind",
+    Method = "watfind",
+  --{patt},                 {subj, st}           { results }
+    { {L".+"},              {L"abcd",{}},        {1,4,{}}  }, -- [none]
+    { {L".+"},              {L"abcd",{},2},      {2,4,{}}  }, -- positive st
+    { {L".+"},              {L"abcd",{},-2},     {3,4,{}}  }, -- negative st
+    { {L".*"},              {L"abcd",{}},        {1,4,{}}  }, -- [none]
+    { {L"bc"},              {L"abc", {}},        {2,3,{}}  }, -- [none]
+    { {L"(.)b.(d)"},        {L"abcd",{}},        {1,4,{L"a",L"d"}}},--[captures]
+  }
+end
+
 local function set_m_wfind (lib, flg)
   return {
     Name = "Method wfind",
@@ -308,7 +337,9 @@ return function (libname)
     set_f_wfind      (lib),
     set_f_wmatch     (lib),
     set_m_wexec      (lib),
+    set_m_waexec     (lib),
     set_m_wtfind     (lib),
+    set_m_watfind    (lib),
     set_m_wfind      (lib),
     set_m_wmatch     (lib),
     set_f_wgsub1     (lib),
diff --git a/windows/mingw/_mingw.mak b/windows/mingw/_mingw.mak
index c7d9ba3..169c962 100644
--- a/windows/mingw/_mingw.mak
+++ b/windows/mingw/_mingw.mak
@@ -42,8 +42,8 @@ $(BIN): $(OBJ) $(DEFFILE)
 	$(CC) $(DEFFILE) $(OBJ) $(LIBS) -o $@ -shared
 
 $(DEFFILE):
-	lua -e"print('EXPORTS') for k,v in ipairs{$(EXPORTED)} do \
-      print('\t'..v) end" > $@
+	echo EXPORTS > $@
+	for %%d in ($(EXPORTED)) do echo   %%d>> $@
 
 $(BININSTALL): $(BIN)
 	copy /Y $< $@
diff --git a/windows/mingw/rex_gnu.mak b/windows/mingw/rex_gnu.mak
index 0675658..965095e 100644
--- a/windows/mingw/rex_gnu.mak
+++ b/windows/mingw/rex_gnu.mak
@@ -10,7 +10,7 @@ MYINCS   = -I$(REGEXINC) -I$(LUAINC)
 MYLIBS   = -L$(DLLPATH) -lregex2 -l$(LUADLL)
 OBJ      = lgnu.o common.o
 MYCFLAGS = -W -Wall -O2
-EXPORTED = 'luaopen_$(PROJECT)'
+EXPORTED = luaopen_$(PROJECT)
 SRCPATH  = ..\..\src;..\..\src\gnu
 TESTPATH = ..\..\test
 TESTNAME = gnu
diff --git a/windows/mingw/rex_onig.mak b/windows/mingw/rex_onig.mak
index b69d3a9..a179362 100644
--- a/windows/mingw/rex_onig.mak
+++ b/windows/mingw/rex_onig.mak
@@ -10,7 +10,7 @@ MYINCS   = -I$(REGEXINC) -I$(LUAINC)
 MYLIBS   = -L$(DLLPATH) -lonig -l$(LUADLL) -Wl,--enable-auto-import
 OBJ      = lonig.o lonig_f.o common.o
 MYCFLAGS = -W -Wall -O2
-EXPORTED = 'luaopen_$(PROJECT)'
+EXPORTED = luaopen_$(PROJECT)
 SRCPATH  = ..\..\src;..\..\src\oniguruma
 TESTPATH = ..\..\test
 TESTNAME = onig
diff --git a/windows/mingw/rex_pcre.mak b/windows/mingw/rex_pcre.mak
index d134b05..ef68352 100644
--- a/windows/mingw/rex_pcre.mak
+++ b/windows/mingw/rex_pcre.mak
@@ -10,7 +10,7 @@ MYINCS   = -I$(REGEXINC) -I$(LUAINC)
 MYLIBS   = -L$(DLLPATH) -lpcre -l$(LUADLL)
 OBJ      = lpcre.o lpcre_f.o common.o
 MYCFLAGS = -W -Wall -O2
-EXPORTED = 'luaopen_$(PROJECT)'
+EXPORTED = luaopen_$(PROJECT)
 SRCPATH  = ..\..\src;..\..\src\pcre
 TESTPATH = ..\..\test
 TESTNAME = pcre
diff --git a/windows/mingw/rex_spencer.mak b/windows/mingw/rex_spencer.mak
index 23ae67c..bc00d8c 100644
--- a/windows/mingw/rex_spencer.mak
+++ b/windows/mingw/rex_spencer.mak
@@ -10,7 +10,7 @@ MYINCS   = -I$(REGEXINC) -I$(LUAINC)
 MYLIBS   = -L$(DLLPATH) -lrxspencer -l$(LUADLL)
 OBJ      = lposix.o common.o
 MYCFLAGS = -W -Wall -O2
-EXPORTED = 'luaopen_$(PROJECT)'
+EXPORTED = luaopen_$(PROJECT)
 SRCPATH  = ..\..\src;..\..\src\posix
 TESTPATH = ..\..\test
 TESTNAME = spencer
diff --git a/windows/mingw/rex_tre.mak b/windows/mingw/rex_tre.mak
index 14507a6..8c167b3 100644
--- a/windows/mingw/rex_tre.mak
+++ b/windows/mingw/rex_tre.mak
@@ -9,8 +9,8 @@ PROJECT  = rex_tre
 MYINCS   = -I$(REGEXINC) -I$(LUAINC) 
 MYLIBS   = -L$(DLLPATH) -ltre -l$(LUADLL)
 OBJ      = ltre.o ltre_w.o common.o
-MYCFLAGS = -W -Wall -O2 -DWIDE_CHAR_GSUB
-EXPORTED = 'luaopen_$(PROJECT)'
+MYCFLAGS = -W -Wall -O2
+EXPORTED = luaopen_$(PROJECT)
 SRCPATH  = ..\..\src;..\..\src\tre
 TESTPATH = ..\..\test
 TESTNAME = tre