implementation of loc transliteration tables

2 files changed, 201 insertions, 2 deletions

diff --git a/arabluatex.lua b/arabluatex.lua
index b716bea..d76a284 100644
--- a/arabluatex.lua
+++ b/arabluatex.lua
@@ -236,6 +236,38 @@ local function transdmg(str)
 return str
 end
 
+local function transloc(str)
+   str = string.gsub(str, "\\arb(%b{})", function(inside)
+   inside = string.sub(inside, 2, -2)
+      for i = 1,#hamzatrloc do
+         inside = string.gsub(inside, hamzatrloc[i].a, hamzatrloc[i].b)
+      end
+      for i = 1,#tanwintrloc do
+         inside  = string.gsub(inside, tanwintrloc[i].a, tanwintrloc[i].b)
+      end
+      for i = 1,#trigraphstrloc do
+         inside = string.gsub(inside, trigraphstrloc[i].a, trigraphstrloc[i].b)
+      end
+      for i = 1,#digraphstrloc do
+         inside = string.gsub(inside, digraphstrloc[i].a, digraphstrloc[i].b)
+      end
+      for i = 1,#singletrloc do
+         inside = string.gsub(inside, singletrloc[i].a, singletrloc[i].b)
+      end
+      for i = 1,#longvtrloc do
+         inside = string.gsub(inside, longvtrloc[i].a, longvtrloc[i].b)
+      end
+      for i = 1,#shortvtrloc do
+         inside = string.gsub(inside, shortvtrloc[i].a, shortvtrloc[i].b)
+      end
+      for i = 1,#finaltrloc do
+         inside = string.gsub(inside, finaltrloc[i].a, finaltrloc[i].b)
+      end
+      return string.format("\\txtrans{%s}", inside)
+  end)
+return str
+end
+
 function processvoc(str)
    str = "\\arb{".. str.."}"
    str = protectarb(str)
@@ -273,6 +305,8 @@ function processtrans(str, mode)
    str = holdcmd(str)
    if mode == "dmg" then
       str = transdmg(str)
+   elseif mode == "loc" then
+      str = transloc(str)
       else end
    str = unprotectarb(str)
 return str
diff --git a/arabluatex_trans.lua b/arabluatex_trans.lua
index e65c3eb..8a9aee8 100644
--- a/arabluatex_trans.lua
+++ b/arabluatex_trans.lua
@@ -22,6 +22,8 @@ Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
 02111-1307, USA.
 --]]
 
+-- dmg
+
 hamzatrdmg = {
    -- initial long u
    {a="%'%_U", b="ʾU"},
@@ -87,11 +89,11 @@ trigraphstrdmg = { -- trigraphs or more
    -- art. with waṣla + lām
    {a="'l%-(l)", b="'l-%1"},
    -- art. with waṣla + solar consonant
-   {a="'l%-([%_%^%.]?[tdrzsn])", b="'%1-"},
+   {a="'l%-([%_%^%.]?[tdrzsn])", b="'%1-%1"},
    -- li- + art. + lām
    {a="l(i)%-l%-(l)", b="l%1-l-%2%2"},
    -- assim. art. with waṣla + solar consonant
-   {a="'([%_%^%.]?[tdrzsn])%-", b="'%1-%1"},
+   {a="'([%_%^%.]?[tdrzsn])%-", b="'%1-"},
    -- li- + art. + solar consonant is a special orthography
    {a="l(i)%-l%-([%_%^%.]?[tdrzsn])", b="l%1-%2-%2"},
    -- li- + assim. art. + solar consonant is a special orthography
@@ -173,6 +175,169 @@ shortvtrdmg = {
         {a="i", b="i"}
 }
 
+-- loc
+
+hamzatrloc = {
+   -- initial long u
+   {a="%'%_U", b="U"},
+   -- madda (historic writing below)
+   {a="^(')(A)", b="%2"},
+   {a="(%W)(')(A)", b="%1%3"},
+   {a="'A", b="ʾA"},
+   {a="(A)(')(i)$", b="%1ʾ%3"},
+   {a="(A)(')(i)(%W)", b="%1ʾ%3%4"},
+   {a="(A)(')(i)", b="%1ʾ%3"}, -- historic madda
+   {a="(A)(')", b="%1ʾ"}, -- historic madda
+   -- initial (needs both ^ and %W patterns)
+   {a="^(')([ua])", b="%2"},
+   {a="^(')(i)", b="%2"},
+   {a="(%W)(')([ua])", b="%1%3"},
+   {a="(%W)(')(i)", b="%1%3"},
+   -- final
+   {a="([Iy])(')(aN)$", b="%1ʾ%3"},
+   {a="([Iy])(')(aN)(%W)", b="%1ʾ%3%4"},
+   {a="([^uai])(')([uai]N?)$", b="%1ʾ%3"},
+   {a="([^uai])(')([uai]N?)(%W)", b="%1ʾ%3%4"},
+   {a="([UI])(')([uai])$", b="%1ʾ%3"},
+   {a="([UI])(')([uai])(%W)", b="%1ʾ%3%4"},
+   -- middle
+   {a="(U)(')", b="%1ʾ"},
+   {a="([Iy])(')", b="%1ʾ"},
+   {a="([^uai])(')([uU])", b="%1ʾ%3"},
+   {a="([^uai])(')([aA])", b="%1ʾ%3"},
+   {a="([^uai])(')([iI])", b="%1ʾ%3"},
+   {a="(u)(')([uU])", b="%1ʾ%3"},
+   {a="(u)(')([aA])", b="%1ʾ%3"},
+   {a="(u)(')([iI])", b="%1ʾ%3"},
+   {a="(a)(')([aA])", b="%1ʾ%3"},
+   {a="(a)(')([uU])", b="%1ʾ%3"},
+   {a="(a)(')([iI])", b="%1ʾ%3"},
+   {a="(i)(')([aA])", b="%1ʾ%3"},
+   {a="(i)(')([uU])", b="%1ʾ%3"},
+   {a="(i)(')([iI])", b="%1ʾ%3"},
+   {a="(a)(')([^uaiUAI])", b="%1ʾ%3"},
+   {a="(u)(')([^uaiUAI])", b="%1ʾ%3"},
+   {a="(i)(')([^uaiUAI])", b="%1ʾ%3"}
+}
+
+trigraphstrloc = { -- trigraphs or more
+   -- 'llatI / 'llad_I
+   {a="^'ll(a)([%_]?[dt])", b="all%1%2"},
+   {a="(%s)'ll(a)([%_]?[dt])", b="%1all%2%3"},
+   -- al- + lām
+   {a="^(a)l%-(l)", b="%1l-%2"},
+   {a="(%s)(a)l%-(l)", b="%1%2l-%3"},
+   -- al- + solar consonant
+   {a="^(a)l%-([%_%^%.]?[tdrzsn])", b="%1l-%2"},
+   {a="(%s)(a)l%-([%_%^%.]?[tdrzsn])", b="%1%2l-%3"},
+   -- assim. art. + solar consonant
+   {a="^(a)([%_%^%.]?[tdrzsn])%-", b="%1l-%2"},
+   {a="(%s)(a)([%_%^%.]?[tdrzsn])%-", b="%1%2l-%3"},
+   -- al- + initial unstable hamza
+   {a="^(a)l%-([uai])", b="%1l-%2"},
+   {a="(%s)(a)l%-([uai])", b="%1%2l-%3"},
+   -- li- + art. + initial unstable hamza is a special orthography
+   {a="l(i)%-l%-([uai])", b="l%1l-%2"},
+   -- al- + lunar consonant (i.e. what remains)
+   {a="^(a)l%-", b="%1l-"},
+   {a="(%s)(a)l%-", b="%1%2l-"},
+   -- art. with waṣla + lām
+   {a="'l%-(l)", b="al-%1"},
+   -- art. with waṣla + solar consonant
+   {a="'l%-([%_%^%.]?[tdrzsn])", b="al-%1"},
+   -- li- + art. + lām
+   {a="l(i)%-l%-(l)", b="l%1l-%2"},
+   -- assim. art. with waṣla + solar consonant
+   {a="'([%_%^%.]?[tdrzsn])%-", b="al-"},
+   -- li- + art. + solar consonant is a special orthography
+   {a="l(i)%-l%-([%_%^%.]?[tdrzsn])", b="l%1l-%2"},
+   -- li- + assim. art. + solar consonant is a special orthography
+   {a="l(i)%-([%_%^%.]?[tdrzsn])%-([%_%^%.]?[tdrzsn])", b="l%1l-%3"},
+   -- art. with waṣla + initial unstable hamza
+   {a="'l%-([uai])", b="al-%1"},   
+   -- art. with waṣla + lunar consonant (i.e. what remains)
+   {a="'l%-", b="al-"}
+}
+
+tanwintrloc = {
+        {a="uN", b="un"},
+        {a="(aN)(_A)", b="an"},
+        {a="(aN)(Y)", b="an"},
+        {a="(T)(aN)", b="tan"},
+        {a="([^TA])(aN)", b="%1an"},
+        {a="iN", b="in"}
+}
+
+digraphstrloc = {
+   {a="(%-)([uai])", b="%1%2"}, -- hyphen + initial alif without hamza
+   {a="^([uai])", b="%1"},      -- initial alif without hamza
+   {a="(%s)([uai])", b="%1%2"}, -- initial alif without hamza
+   {a="%-%-", b=""},
+   {a="uww", b="ūw"},
+   {a="iyy", b="īy"},
+   {a="([tkdsg])(h)", b="%1'%2"},
+   {a="T([^uai])", b="h%1"},
+   {a="T$", b="h"},
+   {a="T(%W)", b="h%1"},
+   {a="_t", b="th"},
+   {a="%^g", b="j"},
+   {a="%.h", b="ḥ"},
+   {a="_h", b="kh"},
+   {a="_d", b="dh"},
+   {a="%^s", b="sh"},
+   {a="%.s", b="ṣ"},
+   {a="%.d", b="ḍ"},
+   {a="%.t", b="ṭ"},
+   {a="%.z", b="ẓ"},
+   {a="%.g", b="gh"},
+   {a="(U)(A)", b="ū"},
+   {a="WA", b="ū"},
+   {a="(a)W", b="%1w"},
+   {a="_A", b="á"},
+   {a="_u", b="ū"},
+   {a="_a", b="ā"},
+   {a="_i", b="ī"}
+}
+
+singletrloc = {
+        {a="b", b="b"},
+        {a="t", b="t"},
+        {a="j", b="j"},
+        {a="x", b="kh"},
+        {a="d", b="d"},
+        {a="r", b="r"},
+        {a="z", b="z"},
+        {a="s", b="s"},
+        {a="`", b="`"},
+        {a="f", b="f"},
+        {a="q", b="q"},
+        {a="k", b="k"},
+        {a="l", b="l"},
+        {a="m", b="m"},
+        {a="n", b="n"},
+        {a="h", b="h"},
+        {a="w", b="w"},
+        {a="y", b="y"},
+        {a="T", b="t"},
+        {a="Y", b="á"},
+        {a='"', b=''},
+}
+
+longvtrloc = {
+        {a="A", b="ā"},
+        {a="U", b="ū"},
+        {a="I", b="ī"}
+}
+
+shortvtrloc = {
+        {a="u", b="u"},
+        {a="a", b="a"},
+        {a="i", b="i"}
+}
+
+finaltrloc = {
+   {a="ʾ", b="'"},
+}
 
 -- return {
 -- raw = raw,