1 files changed, 35 insertions, 21 deletions

diff --git a/arabluatex_trans.lua b/arabluatex_trans.lua
index e7087c0..ef55a2b 100644
--- a/arabluatex_trans.lua
+++ b/arabluatex_trans.lua
@@ -95,6 +95,12 @@ hamzatrdmg = {
    {a="(A)(')(i)", b="%1ʾ%3"}, -- historic madda
    {a="(A)(')", b="%1ʾ"}, -- historic madda
    -- initial (needs both ^ and %W patterns)
+   -- 'aw: the diphthong is to be resoved into 'awi' (next 4 lines)
+   {a="^('aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
+   {a="(%W)('aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1%2i%3%4"},
+   {a="^('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"},
+   {a="(%W)('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"},
+   -- then the 'initial' rules for the remaining cases
    {a="^(')([ua])", b="ʾ%2"},
    {a="^(')(i)", b="ʾ%2"},
    {a="(%W)(')([ua])", b="%1ʾ%3"},
@@ -127,6 +133,7 @@ hamzatrdmg = {
 }
 
 tanwintrdmg = {
+   {a="%-?([uai]NU)(%s)([ui])", b="\\arbup{un%3}%2'"},
    {a="%-?uNU", b="\\arbup{un}"},
    {a="%-?aNU", b="\\arbup{an}"},
    {a="%-?iNU", b="\\arbup{in}"},
@@ -165,24 +172,29 @@ tanwintrdmg = {
 trigraphstrdmg = { -- trigraphs or more
    -- 'llatI / 'llad_I
    {a="^'ll(a)([%_]?[dt])", b="'ll%1%2"},
-   {a="([%p%s])'ll(a)([%_]?[dt])", b="%1'll%2%3"}, --p
+   {a="([%(%[%|%<%s])'ll(a)([%_]?[dt])", b="%1'll%2%3"}, --p
+   -- law: the diphthong is to be resoved into 'awi' (next 4 lines)
+   {a="^(law)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
+   {a="(%W)(law)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1%2i%3%4"},
+   {a="^(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"},
+   {a="(%W)(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"},
    -- al- + lām
    {a="^(a)l%-(l)", b="%1l-%2"},
-   {a="([%p%s%-])(a)l%-(l)", b="%1%2l-%3"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-(l)", b="%1%2l-%3"}, --p
    -- al- + solar consonant
    {a="^(a)l%-([%_%^%.]?[tdrzsn])", b="%1%2-%2"},
-   {a="([%p%s%-])(a)l%-([%_%^%.]?[tdrzsn])", b="%1%2%3-%3"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-([%_%^%.]?[tdrzsn])", b="%1%2%3-%3"}, --p
    -- assim. art. + solar consonant
    {a="^(a)([%_%^%.]?[tdrzsn])%-", b="%1%2-"},
-   {a="([%p%s%-])(a)([%_%^%.]?[tdrzsn])%-", b="%1%2%3-"}, --p
+   {a="([%(%[%|%<%s%-])(a)([%_%^%.]?[tdrzsn])%-", b="%1%2%3-"}, --p
    -- al- + initial unstable hamza
    {a="^(a)l%-(\"?[uai])", b="%1l-%2"},
-   {a="([%p%s%-])(a)l%-(\"?[uai])", b="%1%2l-%3"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-(\"?[uai])", b="%1%2l-%3"}, --p
    -- 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="([%p%s%-])(a)l%-", b="%1%2l-"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-", b="%1%2l-"}, --p
    -- 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"},
@@ -231,14 +243,15 @@ digraphstrdmg = {
    {a="(%W)(\"?[uai])", b="%1%2"},      -- initial alif without hamza
    -- this is not necessary, take out for now:
 -- {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="(%_A)(%s)(\"?[uai])", b="ạ%2'"},
-   {a="(Y)(%s)(\"?[uai])", b="ạ%2'"},
-   {a="(%_a)(%s)(\"?[uai])", b="a%2'"},
-   {a="(A)(%s)(\"?[uai])", b="a%2'"},
-   {a="([%_]?[Uu])(%s)(\"?[uai])", b="u%2'"},
-   {a="([%_]?[Ii])(%s)(\"?[uai])", b="i%2'"},
+   {a="(aW)(%s)([%(%[%|%<]?)(\"?[uai])", b="awu%2%3%4"}, --p
+   {a="(UA)(%s)([%(%[%|%<]?)(\"?[uai])", b="u%2%3'"}, --p
+   {a="([^%_][uai])(%s)([%(%[%|%<]?)(\"?[uai])", b="%1%2%3'"}, --p
+   {a="(%_A)(%s)([%(%[%|%<]?)(\"?[uai])", b="ạ%2%3'"}, --p
+   {a="(Y)(%s)([%(%[%|%<]?)(\"?[uai])", b="ạ%2%3'"}, --p
+   {a="(%_a)(%s)([%(%[%|%<]?)(\"?[uai])", b="a%2%3'"}, --p
+   {a="(A)(%s)([%(%[%|%<]?)(\"?[uai])", b="a%2%3'"}, --p
+   {a="([%_]?[Uu])(%s)([%(%[%|%<]?)(\"?[uai])", b="u%2%3'"}, --p
+   {a="([%_]?[Ii])(%s)([%(%[%|%<]?)(\"?[uai])", b="i%2%3'"}, --p
    -- ʾiʿrāb hyphen (begin)
    {a="(%-)(\"?[UI]na)(%p?%s)", b="\\arbup{%2}%3"},
    {a="(%-)(\"?[UI]na)(%p?)$", b="\\arbup{%2}%3"},
@@ -253,13 +266,14 @@ digraphstrdmg = {
    -- ʾiʿrāb hyphen (end)
    -- shorten long vowels preceding ʾalif conjunctionis
    {a="(U)(A)", b="U"},
-   {a="(aW)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="awu%2%3"},
-   {a="(%_a)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="a%2%3"},
-   {a="(%_A)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="ạ%2%3"},
-   {a="(A)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="a%2%3"},
-   {a="(Y)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="ạ%2%3"},
-   {a="([%_]?[Uu])(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="u%2%3"},
-   {a="([%_]?[Ii])(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="i%2%3"},
+   --p (next 7 lines, just after %s)
+   {a="(aW)(%s)([%(%[%|%<]?['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="awu%2%3"},
+   {a="(%_a)(%s)([%(%[%|%<]?['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="a%2%3"},
+   {a="(%_A)(%s)([%(%[%|%<]?['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="ạ%2%3"},
+   {a="(A)(%s)([%(%[%|%<]?['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="a%2%3"},
+   {a="(Y)(%s)([%(%[%|%<]?['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="ạ%2%3"},
+   {a="([%_]?[Uu])(%s)([%(%[%|%<]?['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="u%2%3"},
+   {a="([%_]?[Ii])(%s)([%(%[%|%<]?['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="i%2%3"},
    {a="%-%-", b=""},
 --   {a="T([^uai])", b="%1"},
    {a="T(%p?%s)", b="h%1"},