1 files changed, 34 insertions, 12 deletions

diff --git a/arabluatex_voc.lua b/arabluatex_voc.lua
index 878f351..b7550e5 100644
--- a/arabluatex_voc.lua
+++ b/arabluatex_voc.lua
@@ -72,6 +72,8 @@ raw = {
 }
 
 hamza = {
+   -- next line for ʾiʿrāb hyphen
+   {a="(')(%-)([uaiUAI])", b="%1%3"},
    -- hard coded hamza
    {a="|\"'", b="ء"},
    {a="A\"'", b="آ"},
@@ -91,7 +93,8 @@ hamza = {
    {a="'i'([^uaiUAI])", b="إI%1"},   
    -- madda (historic writing below)
    {a="'a'([^uaiUAI])", b="آ%1"},
-   {a="'a?A([%_%^%.]?[%`%'btjghdrzsfqklmnywAY])", b="آ%1"},
+--tmp   {a="'a?A([%_%^%.]?[%`%'btjghdrzsfqklmnywAY])", b="آ%1"},
+   {a="([^UIui])'a?A([%_%^%.]?[%`%'btjghdrzsfqklmnywAY])", b="%1آ%2"},
    {a="(A)(')(uN?)$", b="aآء%3"},
    {a="(A)(')(uN?)(%W)", b="aآء%3%4"},
    {a="(A)(')(iN?)$", b="aآء%3"},
@@ -111,7 +114,7 @@ hamza = {
    {a="([^uai])(')([uai]N?)$", b="%1ء%3"},
    {a="([^uai])(')([uai]N?)(%W)", b="%1ء%3%4"},
 -- u
-   {a="(u)(')([uai]?N)$", b="%1ؤ%3"},
+   {a="(u)(')([uai]N?)$", b="%1ؤ%3"},
    {a="(u)(')([uai]N?)(%W)", b="%1ؤ%3%4"},
    {a="(u)(')$", b="%1ؤ"},
    {a="(u)(')(%W)", b="%1ؤ%3"},
@@ -155,6 +158,8 @@ hamza = {
 }
 
 hamzaeasy = { -- differences marked below with 'easy'
+   -- next line for ʾiʿrāb hyphen
+   {a="(')(%-)([uaiUAI])", b="%1%3"},
    -- hard coded hamza
    {a="|\"'", b="ء"},
    {a="A\"'", b="آ"},
@@ -174,14 +179,17 @@ hamzaeasy = { -- differences marked below with 'easy'
    {a="'i'([^uaiUAI])", b="إI%1"},   
    -- madda (historic writing below)
    {a="'a'([^uaiUAI])", b="آ%1"},
-   {a="'a?A([%_%^%.]?[%`%'btjghdrzsfqklmnywAY])", b="آ%1"},
---easy   {a="(A)(')(uN?)$", b="aآء%3"},
---easy   {a="(A)(')(uN?)(%W)", b="aآء%3%4"},
---easy   {a="(A)(')(iN?)$", b="aآء%3"},
---easy   {a="(A)(')(iN?)(%W)", b="aآء%3%4"},
---easy   {a="(A)(')(i)", b="aآئ%3"}, -- historic madda
---easy   {a="(A)(')(u)", b="aآؤ%3"}, -- historic madda
---easy   {a="(A)(')", b="aآء"}, -- historic madda
+--tmp   {a="'a?A([%_%^%.]?[%`%'btjghdrzsfqklmnywAY])", b="آ%1"},
+   {a="([^UIui])'a?A([%_%^%.]?[%`%'btjghdrzsfqklmnywAY])", b="%1آ%2"},
+   --easy (begin)
+   {a="(A)(')(uN?)$", b="aاء%3"},
+   {a="(A)(')(uN?)(%W)", b="aاء%3%4"},
+   {a="(A)(')(iN?)$", b="aاء%3"},
+   {a="(A)(')(iN?)(%W)", b="aاء%3%4"},
+   {a="(A)(')(i)", b="aائ%3"}, -- historic madda
+   {a="(A)(')(u)", b="aاؤ%3"}, -- historic madda
+   {a="(A)(')", b="aاء"}, -- historic madda
+   --easy (end)
    -- initial (needs both ^ and %W patterns)
    {a="^(')([ua])", b="أ%2"},
    {a="^(')(i)", b="إ%2"},
@@ -194,7 +202,7 @@ hamzaeasy = { -- differences marked below with 'easy'
    {a="([^uai])(')([uai]N?)$", b="%1ء%3"},
    {a="([^uai])(')([uai]N?)(%W)", b="%1ء%3%4"},
 -- u
-   {a="(u)(')([uai]?N)$", b="%1ؤ%3"},
+   {a="(u)(')([uai]N?)$", b="%1ؤ%3"},
    {a="(u)(')([uai]N?)(%W)", b="%1ؤ%3%4"},
    {a="(u)(')$", b="%1ؤ"},
    {a="(u)(')(%W)", b="%1ؤ%3"},
@@ -424,11 +432,25 @@ trigraphseasy = { -- differences marked below with 'easy'
 }
 
 digraphs = {
+   -- ʾiʿrāb: straight double quote must be discarded
+   {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"},
+   -- ʾiʿrāb (end)
    -- initial straight double quote gives a connective ʾalif
    {a="^\"[uai]", b="ٱ"},
    {a="([%s%-])\"[uai]", b="%1ٱ"},
    {a="(aW)(%s)([uai])", b="awuا%2%3"},
-   {a="(%-)([uai])", b="%1ا%2"}, -- hyphen + initial alif without hamza
+--tmp-out   {a="(%-)([uai])", b="%1ا%2"}, -- hyphen + initial alif without hamza
+   --tmp-added hyphen + initial alif without hamza:
+   {a="(%-)([uai])([%^%_%.%`]?)(%a)", b="%1ا%2%3%4"},
    {a="^([uai])", b="ا%1"},      -- initial alif without hamza
    {a="(%s)([uai])", b="%1ا%2"}, -- initial alif without hamza
    {a="%-%-", b="ـ"},