long vowels, diphthongs and consonants with ǧazmah before ʾalif conjunctionis

3 files changed, 32 insertions, 2 deletions

diff --git a/arabluatex_fullvoc.lua b/arabluatex_fullvoc.lua
index 1129911..e6cdb42 100644
--- a/arabluatex_fullvoc.lua
+++ b/arabluatex_fullvoc.lua
@@ -138,6 +138,10 @@ trigraphsfv = { -- trigraphs or more
    -- al- + lunar consonant (i.e. what remains)
    {a="^(a)l%-", b="ا%1لْ"},
    {a="(%s)(a)l%-", b="%1ا%2لْ"},
+   -- diphthongs to be resolved before ʾalif conjunctionis
+   {a="(aW)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="awuا%2%3"},
+   {a="(aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1u%2%3"},
+   {a="(ay)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
    -- art. with waṣla + lām
    {a="'l%-(l)", b="ٱل%1%1"},
    -- art. with waṣla + solar consonant
@@ -171,6 +175,10 @@ trigraphsfv = { -- trigraphs or more
 
 -- this is new
 digraphsfv = {
+   -- diphthongs to be resolved before ʾalif conjunctionis
+   {a="(aW)(%s)([uai])", b="awuا%2ٱ"},
+   {a="(aw)(%s)([uai])", b="%1u%2ٱ"},
+   {a="(ay)(%s)([uai])", b="%1i%2ٱ"},
    {a="([uai]%-)([uai])", b="%1ٱ"}, -- hyphen + initial alif without hamza
    {a="([%_]?[uaiUAIY])(%s)([uai])", b="%1%2ٱ"}, -- initial alif without hamza
    {a="^([uai])", b="ا%1"},      -- initial alif without hamza
@@ -233,6 +241,7 @@ digraphsfv = {
    {a="%.g", b="غ"},
    {a="(U)(A)", b="%1ا"},
    {a="WA", b="وْا"},
+   {a="(a)W\"", b="%1وا"},
    {a="(a)W", b="%1وْا"},
    {a="_A", b="aى"},
    {a="_u", b="ٗ"},
diff --git a/arabluatex_trans.lua b/arabluatex_trans.lua
index f7274c1..cdcc654 100644
--- a/arabluatex_trans.lua
+++ b/arabluatex_trans.lua
@@ -109,6 +109,9 @@ trigraphstrdmg = { -- trigraphs or more
    -- al- + lunar consonant (i.e. what remains)
    {a="^(a)l%-", b="%1l-"},
    {a="(%s)(a)l%-", b="%1%2l-"},
+   -- diphthongs to be resolved before ʾalif conjunctionis
+   {a="(aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1u%2%3"},
+   {a="(ay)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
    -- art. with waṣla + lām
    {a="'l%-(l)", b="'l-%1"},
    -- art. with waṣla + solar consonant
@@ -144,6 +147,14 @@ tanwintrdmg = {
    {a="uNU", b="un"},
    {a="aNU", b="an"},
    {a="iNU", b="in"},
+   -- tanwīn preceding ʾalif conjunctionis
+   {a="(uN)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="uni%2%3"},
+   {a="(aN)(_A)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="ani%3%4"},
+   {a="(aN)(Y)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="ani%3%4"},
+   {a="(T)(aN)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="tani%3%4"},
+   {a="([^TA])(aN)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1ani%3%4"},
+   {a="(iN)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="ini%2%3"},
+   --
    {a="uN", b="un"},
    {a="(aN)(_A)", b="an"},
    {a="(aN)(Y)", b="an"},
@@ -155,12 +166,15 @@ tanwintrdmg = {
 digraphstrdmg = {
    {a="([uai]%-)([uai])", b="%1'"}, -- hyphen + initial alif without hamza
    {a="^([uai])", b="%1"},      -- initial alif without hamza
---   {a="([%_]?[uaiUAIY])(%s)([uai])", b="%1%2'"}, -- initial alif without hamza
+-- {a="([%_]?[uaiUAIY])(%s)([uai])", b="%1%2'"}, -- initial alif without hamza
+   {a="(aW)(%s)([uai])", b="awu%2%3"},
    {a="([^%_][uai])(%s)([uai])", b="%1%2'"},
    {a="([%_]?[AYa])(%s)([uai])", b="a%2'"},
    {a="([%_]?[Uu])(%s)([uai])", b="u%2'"},
    {a="([%_]?[Ii])(%s)([uai])", b="i%2'"},
    -- shorten long vowels preceding ʾalif conjunctionis
+   {a="(U)(A)", b="U"},
+   {a="(aW)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="awu%2%3"},
    {a="([%_]?[AYa])(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="a%2%3"},
    {a="([%_]?[Uu])(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="u%2%3"},
    {a="([%_]?[Ii])(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="i%2%3"},
@@ -179,7 +193,8 @@ digraphstrdmg = {
    {a="%.t", b="ṭ"},
    {a="%.z", b="ẓ"},
    {a="%.g", b="ġ"},
-   {a="(U)(A)", b="ū"},
+   -- the following needs to be moved above shortening rules
+--   {a="(U)(A)", b="ū"},
    {a="WA", b="w"},
    {a="(a)W", b="%1w"},
    {a="_A", b="ā"},
diff --git a/arabluatex_voc.lua b/arabluatex_voc.lua
index b6dbd12..4d3c656 100644
--- a/arabluatex_voc.lua
+++ b/arabluatex_voc.lua
@@ -169,6 +169,10 @@ trigraphs = { -- trigraphs or more
    -- al- + lunar consonant (i.e. what remains)
    {a="^(a)l%-", b="ا%1ل"},
    {a="(%s)(a)l%-", b="%1ا%2ل"},
+   -- diphthongs to be resolved before ʾalif conjunctionis
+   {a="(aW)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="awuا%2%3"},
+   {a="(aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1u%2%3"},
+   {a="(ay)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
    -- art. with waṣla + lām
    {a="'l%-(l)", b="ال%1%1"},
    -- art. with waṣla + solar consonant
@@ -224,6 +228,7 @@ tanwin = {
 }
 
 digraphs = {
+   {a="(aW)(%s)([uai])", b="awuا%2%3"},
    {a="(%-)([uai])", b="%1ا%2"}, -- hyphen + initial alif without hamza
    {a="^([uai])", b="ا%1"},      -- initial alif without hamza
    {a="(%s)([uai])", b="%1ا%2"}, -- initial alif without hamza
@@ -274,6 +279,7 @@ digraphs = {
    {a="%.g", b="غ"},
    {a="(U)(A)", b="%1ا"},
    {a="WA", b="وا"},
+   {a="(a)W\"", b="%1وْا"},
    {a="(a)W", b="%1وا"},
    {a="_A", b="aى"},
    {a="_u", b="ٗ"},