arabica: started implementing this standard: added new functions; now working on tables

2 files changed, 272 insertions, 4 deletions

diff --git a/arabluatex.lua b/arabluatex.lua
index 91932b7..722084c 100644
--- a/arabluatex.lua
+++ b/arabluatex.lua
@@ -494,6 +494,44 @@ local function transloc(str)
 return str
 end
 
+local function transarabica(str)
+   str = string.gsub(str, "\\arb(%b{})", function(inside)
+   inside = string.sub(inside, 2, -2)
+   for i = 1,#hamzatrarabica do
+      inside = string.gsub(inside, hamzatrarabica[i].a, hamzatrarabica[i].b)
+   end
+   for i = 1,#tanwintrarabica do
+      inside  = string.gsub(inside, tanwintrarabica[i].a, tanwintrarabica[i].b)
+   end
+   for i = 1,#trigraphstrarabica do
+      inside = string.gsub(inside, trigraphstrarabica[i].a, trigraphstrarabica[i].b)
+   end
+   for i = 1,#digraphstrarabica do
+      inside = string.gsub(inside, digraphstrarabica[i].a, digraphstrarabica[i].b)
+   end
+   for i = 1,#singletrarabica do
+      inside = string.gsub(inside, singletrarabica[i].a, singletrarabica[i].b)
+   end
+   for i = 1,#longvtrarabica do
+      inside = string.gsub(inside, longvtrarabica[i].a, longvtrarabica[i].b)
+   end
+   for i = 1,#shortvtrarabica do
+      inside = string.gsub(inside, shortvtrarabica[i].a, shortvtrarabica[i].b)
+   end
+   for i = 1,#finaltrarabica do
+      inside = string.gsub(inside, finaltrarabica[i].a, finaltrarabica[i].b)
+   end
+   for i = 1,#punctuationtr do
+      inside = string.gsub(inside, punctuationtr[i].a, punctuationtr[i].b)
+   end
+   for i = 1,#nulltr do
+      inside = string.gsub(inside, nulltr[i].a, nulltr[i].b)
+   end
+   return string.format("\\txtrans{%s}", inside)
+   end)
+return str
+end
+
 local function processbuckw(str)
    str = string.gsub(str, "\\arb(%b{})", function(inside)
    inside = string.sub(inside, 2, -2)
@@ -579,6 +617,8 @@ function processtrans(str, mode, rules, scheme)
       str = transdmg(str, rules)
    elseif mode == "loc" then
       str = transloc(str)
+   elseif mode == "arabica" then
+      str = transarabica(str)
    end
    str = unprotectarb(str)
 return str
diff --git a/arabluatex_trans.lua b/arabluatex_trans.lua
index 275e235..f6c90f7 100644
--- a/arabluatex_trans.lua
+++ b/arabluatex_trans.lua
@@ -496,7 +496,7 @@ trigraphstrloc = { -- trigraphs or more
    {a="^(a)l%-([uai])", b="%1l-%2"},
    {a="(%s)(a)l%-([uai])", b="%1%2l-%3"},
    -- li-/la- + art. + initial unstable hamza is a special orthography
-   {a="l([ai])%-l%-([uai])", b="l%1l-%2"},
+   {a="l([ai])%-l%-([uai])", b="l%1-l-%2"},
    -- al- + lunar consonant (i.e. what remains)
    {a="^(a)l%-", b="%1l-"},
    {a="(%s)(a)l%-", b="%1%2l-"},
@@ -505,13 +505,13 @@ trigraphstrloc = { -- trigraphs or more
    -- art. with waṣla + solar consonant
    {a="'l%-([%_%^%.]?[tdrzsn])", b="al-%1"},
    -- li-/la- + art. + lām
-   {a="l([ai])%-l%-(l)", b="l%1-%2"},
+   {a="l([ai])%-l%-(l)", b="l%1-l-%2"},
    -- assim. art. with waṣla + solar consonant
    {a="'([%_%^%.]?[tdrzsn])%-", b="al-"},
    -- li-/la- + art. + solar consonant is a special orthography
-   {a="l([ai])%-l%-([%_%^%.]?[tdrzsn])", b="l%1l-%2"},
+   {a="l([ai])%-l%-([%_%^%.]?[tdrzsn])", b="l%1-l-%2"},
    -- li-/la- + assim. art. + solar consonant is a special orthography
-   {a="l([ai])%-([%_%^%.]?[tdrzsn])%-([%_%^%.]?[tdrzsn])", b="l%1l-%3"},
+   {a="l([ai])%-([%_%^%.]?[tdrzsn])%-([%_%^%.]?[tdrzsn])", b="l%1-l-%3"},
    -- art. with waṣla + initial unstable hamza
    {a="'l%-([uai])", b="al-%1"},   
    -- art. with waṣla + lunar consonant (i.e. what remains)
@@ -622,3 +622,231 @@ shortvtrloc = {
 finaltrloc = {
    {a="ʾ", b="'"},
 }
+
+-- arabica
+
+hamzatrarabica = { --UNTOUCHED/LOC
+   -- hard coded hamza
+   {a="|\"'", b="ʾ"},
+   {a="A\"'", b="ʾA"},
+   {a="[au]\"'", b="ʾ"},
+   {a="w\"'", b="ʾ"},
+   {a="i\"'", b="ʾ"},
+   {a="y\"'", b="ʾ"},
+   -- hamza takes tašdīd too
+   {a="''([Uu])", b="ʾʾ%1"},
+   {a="''([Aa])", b="ʾʾ%1"},
+   {a="''([Ii])", b="ʾʾ%1"},
+   -- initial long u and i (for a, see below)
+   {a="%'%_U", b="U"},
+   {a="%'%_I", b="I"},
+   -- taḫfīfu 'l-hamza
+   {a="^'u'([^uaiUAI])", b="U%1"},
+   {a="(%W)'u'([^uaiUAI])", b="%1U%2"},
+   {a="'u'([^uaiUAI])", b="ʾU"},
+   {a="^'i'([^uaiUAI])", b="I%1"},
+   {a="(%W)'i'([^uaiUAI])", b="%1I%2"},
+   {a="'i'([^uaiUAI])", b="ʾI"},
+   -- madda (historic writing below)
+   {a="^(')(A)", b="%2"},
+   {a="(%W)(')(A)", b="%1%3"},   
+   {a="^'a'([^uaiUAI])", b="A%1"},
+   {a="(%W)'a'([^uaiUAI])", b="%1A%2"},   
+   {a="'a'([^uaiUAI])", b="A%1"},
+   {a="^'a?A", b="A"},
+   {a="(%W)'a?A", b="%1A"},
+   {a="'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"}
+}
+
+tanwintrarabica = { --UNTOUCHED/LOC
+   {a="%-?uNU", b="un"},
+   {a="%-?aNU", b="an"},
+   {a="%-?iNU", b="in"},
+   {a="%-?(\"?At)%-?([ui])N", b="%1%2n"},
+   {a="%-?([ui])N", b="%1n"},
+   {a="%-?(aN)(_A)", b="an"},
+   {a="%-?(aN)(Y)", b="an"},
+   {a="(T)%-?(\"?aN)", b="tan"},
+   {a="([^TA])%-?(\"?aN)", b="%1an"}
+}
+
+-- new
+trigraphstrarabica = { -- trigraphs or more
+   -- 'llatI / 'llad_I
+   {a="^'ll(a)([%_]?[dt])", b="ll%1%2"},
+   {a="([%(%[%|%<%s])'ll(a)([%_]?[dt])", b="%1ll%2%3"}, --p
+   -- 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-"},
+   {a="(%s)(a)([%_%^%.]?[tdrzsn])%-", b="%1%2l-"},
+   -- al- + initial unstable hamza
+   {a="^(a)l%-([uai])", b="%1l-%2"},
+   {a="(%s)(a)l%-([uai])", b="%1%2l-%3"},
+   -- li-/la- + art. + initial unstable hamza is a special orthography
+   {a="l([ai])%-l%-([uai])", b="l%1-l-%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="l-%1"},
+   -- art. with waṣla + solar consonant
+   {a="'l%-([%_%^%.]?[tdrzsn])", b="l-%1"},
+   -- li-/la- + art. + lām
+   {a="l([ai])%-l%-(l)", b="l%1-l-%2"},
+   -- assim. art. with waṣla + solar consonant
+   {a="'([%_%^%.]?[tdrzsn])%-", b="l-"},
+   -- li-/la- + art. + solar consonant is a special orthography
+   {a="l([ai])%-l%-([%_%^%.]?[tdrzsn])", b="l%1-l-%2"},
+   -- li-/la- + assim. art. + solar consonant is a special orthography
+   {a="l([ai])%-([%_%^%.]?[tdrzsn])%-([%_%^%.]?[tdrzsn])", b="l%1-l-%3"},
+   -- art. with waṣla + initial unstable hamza
+   {a="'l%-([uai])", b="l-%1"},   
+   -- art. with waṣla + lunar consonant (i.e. what remains)
+   {a="'l%-", b="l-"},
+   -- the silent wāw
+   {a="uU$", b="u"},
+   {a="uU(%W)", b="u%1"},
+   {a="aU$", b="a"},
+   {a="aU(%W)", b="a%1"},
+   {a="iU$", b="i"},
+   {a="iU(%W)", b="i%1"},
+   -- words ending in -āT with silent wāw/yāʾ
+   {a="(_a)UA", b="A"},
+   {a="(_a)U", b="A"},
+   {a="(_a)I", b="A"}
+}
+
+digraphstrarabica = {
+   -- discard the ʾiʿrāb hyphen (begin)
+   {a="(%-)(\"?[UI]na)(%p?%s)", b="%2%3"},
+   {a="(%-)(\"?[UI]na)(%p?)$", b="%2%3"},
+   {a="(%-)(\"?At[ui])(%p?%s)", b="%2%3"},
+   {a="(%-)(\"?At[ui])(%p?)$", b="%2%3"},
+   {a="(%-)(\"?Ani)(%p?%s)", b="%2%3"},
+   {a="(%-)(\"?Ani)(%p?)$", b="%2%3"},
+   {a="(%-)(\"?ayni)(%p?%s)", b="%2%3"},
+   {a="(%-)(\"?ayni)(%p?)$", b="%2%3"},
+   {a="(%-)([uai])(%p?%s)", b="%2%3"},
+   {a="(%-)([uai])(%p?)$", b="%2%3"},
+   -- discard the ʾiʿrāb hyphen (end)
+   {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$", b="ī"},
+   {a="iyy(%W)", b="ī%1"},
+   {a="iyy", b="īy"},
+   {a="([tkdsg])(h)", b="%1'%2"},
+   --   {a="T([^uai])", b="h%1"},
+   {a="([a%']l%-)(.-)T([%(%[%|%<%s])(al%-)", b="%1%2h%3%4"}, --p
+   {a="T([%(%[%|%<%s])(al%-)", b="t%1%2"}, --p
+   {a="T$", 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="w"},
+   {a="(a)W", b="%1w"},
+   {a="_A", b="á"},
+   {a="_u", b="ū"},
+   {a="_a", b="ā"},
+   {a="_i", b="ī"},
+   {a="%.b", b="b"},
+   {a="%.f", b="f"},
+   {a="%.q", b="q"},
+   {a="%.k", b="k"},
+   {a="%.n", b="n"},
+   {a="%^d", b="d"}
+}
+
+-- new
+singletrarabica = {
+   {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="ʿ"},
+   {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="\"", b=""},
+   {a="B", b=""}
+}
+
+longvtrarabica = {
+   {a="A", b="ā"},
+   {a="U", b="ū"},
+   {a="I", b="ī"},
+   {a="Y", b="á"},
+}
+
+shortvtrarabica = {
+   {a="u", b="u"},
+   {a="a", b="a"},
+   {a="i", b="i"}
+}
+
+-- new
+finaltrarabica = {
+--   {a="ʾ", b="'"},
+}