Add the ability to use raw memory blocks as subjects.

The implementation is in algo.h, in the new check_subject function.

Usage is documented in manual.txt.

Optional tests have been added, using alien buffers.

6 files changed, 87 insertions, 15 deletions

diff --git a/doc/manual.txt b/doc/manual.txt
index 6af4d5f..72790ba 100644
--- a/doc/manual.txt
+++ b/doc/manual.txt
@@ -38,9 +38,19 @@ Notes
    too. In this case, the cf_ and larg_ arguments are ignored (should
    be either supplied as nils or omitted).
 
+6. All functions that take a string-type subject accept a table (in Lua >= 5.2)
+   or userdata that has a ``topointer`` method and ``__len`` metamethod, and
+   take the subject to be a block of memory starting at the address returned by
+   ``subject:topointer()`` and of length ``#subject``. This works with buffers
+   objects from the alien library (https://github.com/mascarenhas/alien). Note
+   that special attention is needed with POSIX regex libraries that do not
+   support ``REG_STARTEND``, and hence need NUL-terminated subjects: the NUL is
+   not included in the string length, so alien buffers must be wrapped to
+   report a length that excludes the NUL.
+
 .. _cf:
 
-6. The default value for *compilation flags* (*cf*) that Lrexlib uses when
+7. The default value for *compilation flags* (*cf*) that Lrexlib uses when
    the parameter is not supplied or ``nil`` is:
 
      * REG_EXTENDED for POSIX and TRE
@@ -65,7 +75,7 @@ Notes
 
 .. _ef:
 
-7. The default value for *execution flags* (*ef*) that Lrexlib uses when
+8. The default value for *execution flags* (*ef*) that Lrexlib uses when
    the parameter is not supplied or ``nil``, is:
 
      * 0 for standard POSIX regex library
@@ -75,7 +85,7 @@ Notes
 
 .. _larg:
 
-8. The notation *larg...* is used to indicate optional library-specific
+9. The notation *larg...* is used to indicate optional library-specific
    arguments, which are documented in the ``new`` method of each library.
 
 ------------------------------------------------------------
diff --git a/src/algo.h b/src/algo.h
index a1ea61c..e268a02 100644
--- a/src/algo.h
+++ b/src/algo.h
@@ -115,6 +115,36 @@ static TUserdata* check_ud (lua_State *L)
 }
 
 
+static void check_subject (lua_State *L, int pos, TArgExec *argE)
+{
+  int stype;
+  argE->text = lua_tolstring (L, pos, &argE->textlen);
+  stype = lua_type (L, pos);
+  if (stype != LUA_TSTRING && stype != LUA_TTABLE && stype != LUA_TUSERDATA) {
+    luaL_typerror (L, pos, "string, table or userdata");
+  } else if (argE->text == NULL) {
+    int type;
+    lua_getfield (L, pos, "topointer");
+    if (lua_type (L, -1) != LUA_TFUNCTION)
+      luaL_error (L, "subject has no topointer method");
+    lua_pushvalue (L, pos);
+    lua_call (L, 1, 1);
+    type = lua_type (L, -1);
+    if (type != LUA_TLIGHTUSERDATA)
+      luaL_error (L, "subject's topointer method returned %s (expected lightuserdata)",
+                  lua_typename (L, type));
+    argE->text = lua_touserdata (L, -1);
+    lua_pop (L, 1);
+    lua_len (L, pos);
+    type = lua_type (L, -1);
+    if (type != LUA_TNUMBER)
+      luaL_error (L, "subject's length is %s (expected number)",
+                  lua_typename (L, type));
+    argE->textlen = lua_tointeger (L, -1);
+    lua_pop (L, 1);
+  }
+}
+
 static void check_pattern (lua_State *L, int pos, TArgComp *argC)
 {
   if (lua_isstring (L, pos)) {
@@ -134,7 +164,7 @@ static void checkarg_new (lua_State *L, TArgComp *argC) {
 
 /* function gsub (s, patt, f, [n], [cf], [ef], [larg...]) */
 static void checkarg_gsub (lua_State *L, TArgComp *argC, TArgExec *argE) {
-  argE->text = luaL_checklstring (L, 1, &argE->textlen);
+  check_subject (L, 1, argE);
   check_pattern (L, 2, argC);
   lua_tostring (L, 3);    /* converts number (if any) to string */
   argE->reptype = lua_type (L, 3);
@@ -154,7 +184,7 @@ static void checkarg_gsub (lua_State *L, TArgComp *argC, TArgExec *argE) {
 /* function find  (s, patt, [st], [cf], [ef], [larg...]) */
 /* function match (s, patt, [st], [cf], [ef], [larg...]) */
 static void checkarg_find_func (lua_State *L, TArgComp *argC, TArgExec *argE) {
-  argE->text = luaL_checklstring (L, 1, &argE->textlen);
+  check_subject (L, 1, argE);
   check_pattern (L, 2, argC);
   argE->startoffset = get_startoffset (L, 3, argE->textlen);
   argC->cflags = ALG_GETCFLAGS (L, 4);
@@ -166,7 +196,7 @@ static void checkarg_find_func (lua_State *L, TArgComp *argC, TArgExec *argE) {
 /* function gmatch (s, patt, [cf], [ef], [larg...]) */
 /* function split  (s, patt, [cf], [ef], [larg...]) */
 static void checkarg_gmatch_split (lua_State *L, TArgComp *argC, TArgExec *argE) {
-  argE->text = luaL_checklstring (L, 1, &argE->textlen);
+  check_subject (L, 1, argE);
   check_pattern (L, 2, argC);
   argC->cflags = ALG_GETCFLAGS (L, 3);
   argE->eflags = luaL_optint (L, 4, ALG_EFLAGS_DFLT);
@@ -180,7 +210,7 @@ static void checkarg_gmatch_split (lua_State *L, TArgComp *argC, TArgExec *argE)
 /* method r:match (s, [st], [ef]) */
 static void checkarg_find_method (lua_State *L, TArgExec *argE, TUserdata **ud) {
   *ud = check_ud (L);
-  argE->text = luaL_checklstring (L, 2, &argE->textlen);
+  check_subject (L, 2, argE);
   argE->startoffset = get_startoffset (L, 3, argE->textlen);
   argE->eflags = luaL_optint (L, 4, ALG_EFLAGS_DFLT);
 }
diff --git a/test/luatest.lua b/test/luatest.lua
index 5ea7ba8..3949d90 100644
--- a/test/luatest.lua
+++ b/test/luatest.lua
@@ -73,6 +73,22 @@ local function test_function (test, func)
   if t[1] then
     table.remove (t, 1)
     res = t
+    if alien then
+      local subject = test[1][1]
+      local buf = alien.buffer (#subject)
+      if #subject > 0 then
+        alien.memmove (buf:topointer (), subject, #subject)
+      end
+      test[1][1] = buf
+      local t = packNT (pcall (func, unpackNT (test[1])))
+      if t[1] then
+        table.remove (t, 1)
+        res = t
+      else
+        print "alien test failed"
+        res = t[2] --> error_message
+      end
+    end
   else
     res = t[2] --> error_message
   end
@@ -87,6 +103,7 @@ end
 --  3) test results table or error_message
 local function test_method (test, constructor, name)
   local res1, res2
+  local subject = test[2][1]
   local ok, r = pcall (constructor, unpackNT (test[1]))
   if ok then
     local t = packNT (pcall (r[name], r, unpackNT (test[2])))
diff --git a/test/onig_sets.lua b/test/onig_sets.lua
index 83555a6..dd226ec 100644
--- a/test/onig_sets.lua
+++ b/test/onig_sets.lua
@@ -14,14 +14,14 @@ end
 local function set_named_subpatterns (lib, flg)
   return {
     Name = "Named Subpatterns",
-    Func = function (methodname, subj, patt, name1, name2)
+    Func = function (subj, methodname, patt, name1, name2)
       local r = lib.new (patt)
       local _,_,caps = r[methodname] (r, subj)
       return norm(caps[name1]), norm(caps[name2])
     end,
-    --{}
-    { {"tfind", "abcd", "(?<dog>.)b.(?<cat>d)", "dog", "cat"},  {"a","d"} },
-    { {"exec",  "abcd", "(?<dog>.)b.(?<cat>d)", "dog", "cat"},  {"a","d"} },
+    --{} N.B. subject is always first element
+    { {"abcd", "tfind", "(?<dog>.)b.(?<cat>d)", "dog", "cat"},  {"a","d"} },
+    { {"abcd", "exec",  "(?<dog>.)b.(?<cat>d)", "dog", "cat"},  {"a","d"} },
   }
 end
 
diff --git a/test/pcre_sets.lua b/test/pcre_sets.lua
index 9c19bbe..e2db7a9 100644
--- a/test/pcre_sets.lua
+++ b/test/pcre_sets.lua
@@ -14,14 +14,14 @@ end
 local function set_named_subpatterns (lib, flg)
   return {
     Name = "Named Subpatterns",
-    Func = function (methodname, subj, patt, name1, name2)
+    Func = function (subj, methodname, patt, name1, name2)
       local r = lib.new (patt)
       local _,_,caps = r[methodname] (r, subj)
       return norm(caps[name1]), norm(caps[name2])
     end,
-    --{}
-    { {"tfind", "abcd", "(?P<dog>.)b.(?P<cat>d)", "dog", "cat"},  {"a","d"} },
-    { {"exec",  "abcd", "(?P<dog>.)b.(?P<cat>d)", "dog", "cat"},  {"a","d"} },
+    --{} N.B. subject is always first element
+    { {"abcd", "tfind", "(?P<dog>.)b.(?P<cat>d)", "dog", "cat"},  {"a","d"} },
+    { {"abcd", "exec",  "(?P<dog>.)b.(?P<cat>d)", "dog", "cat"},  {"a","d"} },
   }
 end
 
diff --git a/test/runtest.lua b/test/runtest.lua
index 3c79970..d0d1435 100644
--- a/test/runtest.lua
+++ b/test/runtest.lua
@@ -1,5 +1,12 @@
 -- See Copyright Notice in the file LICENSE
 
+-- See if we have alien, so we can do tests with buffer subjects
+local ok
+ok, alien = pcall (require, "alien")
+if not ok then
+  io.stderr:write ("Warning: alien not found, so cannot run tests with buffer subjects\n")
+end
+
 do
   local path = "./?.lua;"
   if package.path:sub(1, #path) ~= path then
@@ -16,6 +23,13 @@ local function test_library (libname, setfile, verbose)
   local lib = require (libname)
   local f = require (setfile)
   local sets = f (libname)
+
+  local realalien = alien
+  if libname == "rex_posix" and not lib.flags ().STARTEND and alien then
+    alien = nil
+    io.stderr:write ("Cannot run posix tests with alien without REG_STARTEND\n")
+  end
+
   local n = 0 -- number of failures
   for _, set in ipairs (sets) do
     if verbose then
@@ -33,6 +47,7 @@ local function test_library (libname, setfile, verbose)
   if verbose then
     print ""
   end
+  alien = realalien
   return n
 end