implementation of ʾiʿrāb ġayr munṣarif (dmg mode)

1 files changed, 48 insertions, 12 deletions

diff --git a/arabluatex_fullvoc.lua b/arabluatex_fullvoc.lua
index 577fff7..991db22 100644
--- a/arabluatex_fullvoc.lua
+++ b/arabluatex_fullvoc.lua
@@ -23,6 +23,8 @@ along with this program.  If not, see
 --]]
 
 hamzafv = {
+   -- next line for ʾiʿrāb hyphen
+   {a="(')(%-)([uaiUAI])", b="%1%3"},
    -- hard coded hamza
    {a="|\"'", b="ء"},
    {a="A\"'", b="آ"},
@@ -51,7 +53,8 @@ hamzafv = {
    {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"},
@@ -71,7 +74,7 @@ hamzafv = {
    {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="%1ؤْ"},
    {a="(u)(')(%W)", b="%1ؤْ%3"},
@@ -115,6 +118,8 @@ hamzafv = {
 }
 
 hamzafveasy = { -- 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"},
@@ -143,14 +148,17 @@ hamzafveasy = { -- 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"},
@@ -163,7 +171,7 @@ hamzafveasy = { -- 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="%1ؤْ"},
    {a="(u)(')(%W)", b="%1ؤْ%3"},
@@ -389,6 +397,18 @@ trigraphsfveasy = { -- trigraphs or more (see 'easy' tag below for the diffs)
 }
 
 digraphsfv = {
+   -- iʿrāb: 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?)$", b="%2%3"},
+   -- ʾiʿrāb (end)
    -- initial straight double quote gives a connective ʾalif
    {a="^\"[uai]", b="ٱ"},
    {a="([%s%-])\"[uai]", b="%1ٱ"},
@@ -436,9 +456,11 @@ digraphsfv = {
    {a="ww", b="وّ"},
    {a="yy", b="يّ"},
    -- sukūn begin
+   --tmp: next line is added
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])%-(%_?[uaiUAIY])", b="%1%2"},
    {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])$", b="%1ْ"},
    {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])([%s])", b="%1ْ%2"},
-   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])([%_]?[^%_uaiUAIًٌٍ])", b="%1ْ%2"},
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])%-(%_?[^uaiUAIًٌٍ])", b="%1ْ%2"},
    -- take out sukūn in cases of assimilation
    {a="(n)(ْ)(%s)([روي])", b="%1%3%4"},
    {a="(n)(ْ)(%s)([ل])", b="%1%3%4"},
@@ -474,6 +496,18 @@ digraphsfv = {
 }
 
 digraphsfveasy = { -- see the diffenrences under 'easy' marker below
+   -- iʿrāb: 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?)$", b="%2%3"},
+   -- ʾiʿrāb (end)
    -- initial straight double quote gives a connective ʾalif
    {a="^\"[uai]", b="ٱ"},
    {a="([%s%-])\"[uai]", b="%1ٱ"},
@@ -524,6 +558,8 @@ digraphsfveasy = { -- see the diffenrences under 'easy' marker below
    -- first take out every previously generated sukūn by hamza rules,
    -- so there be no need to edit them:
    {a="ْ", b=""}, 
+   --tmp: next line is added
+--   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])%-([uai])", b="%1%2"},
 --   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])$", b="%1ْ"},
 --   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])([%s])", b="%1ْ%2"},
 --   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])([%_]?[^%_uaiUAIًٌٍ])", b="%1ْ%2"},