utf8.offset

5 files changed, 120 insertions, 2 deletions

diff --git a/logo.sketch b/logo.sketch
index 73045fd..a98d6a7 100644
--- a/logo.sketch
+++ b/logo.sketch
diff --git a/src/linit.js b/src/linit.js
index f9f8296..4d50f8b 100644
--- a/src/linit.js
+++ b/src/linit.js
@@ -9,6 +9,7 @@ const lcorolib = require('./lcorolib.js');
 const lmathlib = require('./lmathlib.js');
 const lstrlib  = require('./lstrlib.js');
 const ltablib  = require('./ltablib.js');
+const lutf8lib = require('./lutf8lib.js');
 const lualib   = require('./lualib.js');
 
 const loadedlibs = {
@@ -16,6 +17,7 @@ const loadedlibs = {
     [lualib.LUA_MATHLIBNAME]: lmathlib.luaopen_math,
     [lualib.LUA_STRLIBNAME]:  lstrlib.luaopen_string,
     [lualib.LUA_TABLIBNAME]:  ltablib.luaopen_table,
+    [lualib.LUA_UTF8LIBNAME]: lutf8lib.luaopen_utf8,
     "_G":                     lbaselib.luaopen_base
 };
 
diff --git a/src/lstrlib.js b/src/lstrlib.js
index 2f89777..3db3e5e 100644
--- a/src/lstrlib.js
+++ b/src/lstrlib.js
@@ -5,8 +5,6 @@ const assert  = require('assert');
 const lua     = require('./lua.js');
 const lapi    = require('./lapi.js');
 const lauxlib = require('./lauxlib.js');
-const CT      = lua.constant_types;
-const TS      = lua.thread_status;
 
 const str_len = function(L) {
     lapi.lua_pushinteger(L, lauxlib.luaL_checkstring(L, 1).length);
diff --git a/src/lutf8lib.js b/src/lutf8lib.js
new file mode 100644
index 0000000..bc43d54
--- /dev/null
+++ b/src/lutf8lib.js
@@ -0,0 +1,79 @@
+"use strict";
+
+const assert  = require('assert');
+
+const lua     = require('./lua.js');
+const lapi    = require('./lapi.js');
+const lauxlib = require('./lauxlib.js');
+
+
+const iscont = function(p) {
+    return p & 0xC0 === 0x80;
+};
+
+/* translate a relative string position: negative means back from end */
+const u_posrelat = function(pos, len) {
+    if (pos >= 0) return pos;
+    else if (0 - pos > len) return 0;
+    else return len + pos + 1;
+};
+
+/*
+** offset(s, n, [i])  -> index where n-th character counting from
+**   position 'i' starts; 0 means character at 'i'.
+*/
+const byteoffset = function(L) {
+    let s = lauxlib.luaL_checkstring(L, 1);
+    let n = lauxlib.luaL_checkinteger(L, 2);
+    let posi = n >= 0 ? 1 : s.length + 1;
+    posi = u_posrelat(lauxlib.luaL_optinteger(L, 3, posi), s.length);
+    lauxlib.luaL_argcheck(L, 1 <= posi && --posi <= s.length, 3, "position ot ouf range");
+
+    if (n === 0) {
+        /* find beginning of current byte sequence */
+        while (posi > 0 && iscont(s.slice(posi))) posi--;
+    } else {
+        if (iscont(s.slice(posi)))
+            lauxlib.luaL_error(L, "initial position is a continuation byte");
+
+        if (n < 0) {
+            while (n < 0 && posi > 0) {  /* move back */
+                do {  /* find beginning of previous character */
+                    posi--;
+                } while (posi > 0 && iscont(s.slice(posi)));
+                n++;
+            }
+        } else {
+            n--;  /* do not move for 1st character */
+            while (n > 0 && posi < s.length) {
+                do {  /* find beginning of next character */
+                    posi++;
+                } while (iscont(s.slice(posi)));  /* (cannot pass final '\0') */
+                n--;
+            }
+        }
+    }
+
+    if (n === 0)  /* did it find given character? */
+        lapi.lua_pushinteger(L, posi + 1);
+    else  /* no such character */
+        lapi.lua_pushnil(L);
+
+    return 1;
+};
+
+const funcs = {
+    "offset": byteoffset
+};
+
+/* pattern to match a single UTF-8 character */
+const UTF8PATT = "[\0-\x7F\xC2-\xF4][\x80-\xBF]*";
+
+const luaopen_utf8 = function(L) {
+    lauxlib.luaL_newlib(L, funcs);
+    lapi.lua_pushstring(L, UTF8PATT);
+    lapi.lua_setfield(L, -2, "charpattern");
+    return 1;
+};
+
+module.exports.luaopen_utf8 = luaopen_utf8;
\ No newline at end of file
diff --git a/tests/lutf8lib.js b/tests/lutf8lib.js
new file mode 100644
index 0000000..3006bd4
--- /dev/null
+++ b/tests/lutf8lib.js
@@ -0,0 +1,39 @@
+"use strict";
+
+const test    = require('tape');
+const tests   = require("./tests.js");
+
+const lapi    = require("../src/lapi.js");
+const lauxlib = require("../src/lauxlib.js");
+const linit   = require('../src/linit.js');
+
+test('utf8.offset', function (t) {
+    let luaCode = `
+        return utf8.offset("( ͡° ͜ʖ ͡° )", 5)
+    `, L;
+    
+    t.plan(3);
+
+    t.doesNotThrow(function () {
+
+        L = lauxlib.luaL_newstate();
+
+        linit.luaL_openlibs(L);
+
+        lauxlib.luaL_loadstring(L, luaCode);
+
+    }, "Lua program loaded without error");
+
+    t.doesNotThrow(function () {
+
+        lapi.lua_call(L, 0, -1);
+
+    }, "Lua program ran without error");
+
+    t.strictEqual(
+        lapi.lua_tointeger(L, -1),
+        7,
+        "Correct element(s) on the stack"
+    );
+
+});
\ No newline at end of file