1 files changed, 71 insertions, 1 deletions

diff --git a/arabluatex_trans.lua b/arabluatex_trans.lua
index e12eaa0..7820cd1 100644
--- a/arabluatex_trans.lua
+++ b/arabluatex_trans.lua
@@ -631,10 +631,78 @@ finaltrloc = {
 
 -- arabica
 
+hamzatrarabica = { -- ≠ from hamzatrloc: initial hamza has to be held
+   -- 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)$", b="%1ʾ%3"},
+   {a="(A)(')(i)(%W)", b="%1ʾ%3%4"},
+   {a="(A)(')(i)", b="%1ʾ%3"}, -- historic madda
+   {a="(A)(')", b="%1ʾ"}, -- historic madda
+   -- initial (needs both ^ and %W patterns):
+   -- hold it for now (see below, beginning of digraphs table)
+   {a="^(')([ua])", b="@%2"},
+   {a="^(')(i)", b="@%2"},
+   {a="(%W)(')([ua])", b="%1@%3"},
+   {a="(%W)(')(i)", b="%1@%3"},
+   -- final
+   {a="([Iy])(')(aN)$", 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"}
+}
+
 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
+   {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"},
@@ -682,6 +750,8 @@ trigraphstrarabica = { -- trigraphs or more
 }
 
 digraphstrarabica = {
+   {a="([uai]%-)(\"?[uai])", b="%1"}, -- hyphen + initial alif without hamza
+   {a="@", b=""}, -- remove the tag before the former hamza
    -- discard the ʾiʿrāb hyphen (begin)
    {a="(%-)(\"?[UI]na)(%p?%s)", b="%2%3"},
    {a="(%-)(\"?[UI]na)(%p?)$", b="%2%3"},