1 files changed, 33 insertions, 47 deletions

diff --git a/arabluatex_fullvoc.lua b/arabluatex_fullvoc.lua
index 6f690f8..cf33a3a 100644
--- a/arabluatex_fullvoc.lua
+++ b/arabluatex_fullvoc.lua
@@ -95,27 +95,15 @@ hamzafv = {
    {a="([%s%(%[%<%-])(')([ua])", b="%1أ%3"},
    {a="([%s%(%[%<%-])(')(i)", b="%1إ%3"},
    -- final
-   -- ^say'aN, .zim'aN, mi'aT are special orthographies
-   {a="(%^sa%.?[yY])(\"?%|?)(%')(aN)", b="%1%2ئ%4"},
-   {a="(.zi?m)(%')(aN)", b="%1ئ%3"},
+   -- mi'aT is special orthography (unlike ^say'aN and .zim'aN):
+   -- {a="(%^sa%.?[yY])(\"?%|?)(%')(aN)", b="%1%2ئ%4"}, --new
+   -- {a="(.zi?m)(%')(aN)", b="%1ئ%3"}, --new
    {a="(mi)(%')(a[Tt])", b="%1ائ%3"},
    {a="(mi)(%')(aN%_?[AY])", b="%1أ%3"},
-   -- hamzah + pr. suffix: the carrier passes into yāʾ before ī so
-   -- take out the following two lines:
-   -- {a="(')(I%p?)$", b="إ%2"}, -- affix pr. 1st person sg.
-   -- {a="(')(I)(%p?%s)", b="إ%2%3"}, -- affix pr. 1st person sg.
-   --
-   -- The Munjid says that such words as radI'aN do not have the
-   -- hamzah alone on the line so replace the following two lines with
-   -- the next six:
-   -- {a="([^uai])(')([uai]N?%p?)$", b="%1ء%3"},
-   -- {a="([^uai])(')([uai]N?)(%p?%s)", b="%1ء%3%4"},
-   {a="([Iy])(')(aN%p?)$", b="%1ئ%3"},
-   {a="([Iy])(')(aN)(%p?%s)", b="%1ئ%3%4"},
-   {a="([^uai])(')(aN)(%p?)$", b="%1ء%3ا%4"},
-   {a="([^uai])(')(aN)(%p?%s)", b="%1ء%3ا%4"},
-   {a="([^uai])(')([ui]N?%p?)$", b="%1ء%3"},
-   {a="([^uai])(')([ui]N?)(%p?%s)", b="%1ء%3%4"},
+   -- final hamzah is on the line after a letter of prolongation or a
+   -- consonant with sukūn
+   {a="([^Auai])(')(\"?[uai]N?)(%p?)$", b="%1ء%3%4"}, --new
+   {a="([^Auai])(')(\"?[uai]N?)(%p?%s)", b="%1ء%3%4"},
 -- u
    {a="(u)(')([uai]N?%p?)$", b="%1ؤ%3"},
    {a="(u)(')([uai]N?)(%p?%s)", b="%1ؤ%3%4"},
@@ -141,7 +129,7 @@ hamzafv = {
    {a="(i)(')(%p?%s)", b="%1ئْ%3"},
 --
    -- middle
-   {a="(U)(')", b="%1ء"},
+   {a="([Uw])(')", b="%1ء"}, --new
    {a="([Iy])(')", b="%1ئ"},
    {a="([^uai])(')([uU])", b="%1ؤ%3"},
    {a="([^uai])(')(%_?[aAY])", b="%1أ%3"},
@@ -235,27 +223,25 @@ hamzafveasy = { -- differences marked below with 'easy'
    {a="([%s%(%[%<%-])(')([ua])", b="%1أ%3"},
    {a="([%s%(%[%<%-])(')(i)", b="%1إ%3"},
    -- final
-   -- ^say'aN, .zim'aN, mi'aT are special orthographies
-   {a="(%^sa%.?[yY])(\"?%|?)(%')(aN)", b="%1%2ئ%4"},
-   {a="(.zi?m)(%')(aN)", b="%1ئ%3"},
+   -- mi'aT is special orthography (unlike ^say'aN and .zim'aN)
+   -- {a="(%^sa%.?[yY])(\"?%|?)(%')(aN)", b="%1%2ئ%4"}, --new
+   -- {a="(.zi?m)(%')(aN)", b="%1ئ%3"}, --new
    {a="(mi)(%')(a[Tt])", b="%1ائ%3"},
    {a="(mi)(%')(aN%_?[AY])", b="%1أ%3"},
-   -- hamzah + pr. suffix: the carrier passes into yāʾ before ī so
-   -- take out the following two lines:
-   -- {a="(')(I%p?)$", b="إ%2"}, -- affix pr. 1st person sg.
-   -- {a="(')(I)(%p?%s)", b="إ%2%3"}, -- affix pr. 1st person sg.
-   --
+   -- easy (begin)
    -- The Munjid says that such words as radI'aN do not have the
-   -- hamzah alone on the line so replace the following two lines with
-   -- the next two:
-   -- {a="([^uai])(')([uai]N?%p?)$", b="%1ء%3"},
-   -- {a="([^uai])(')([uai]N?)(%p?%s)", b="%1ء%3%4"},
-   {a="([Iy])(')(aN%p?)$", b="%1ئ%3"},
-   {a="([Iy])(')(aN)(%p?%s)", b="%1ئ%3%4"},
-   {a="([^uai])(')(aN)(%p?)$", b="%1ء%3ا%4"},
-   {a="([^uai])(')(aN)(%p?%s)", b="%1ء%3ا%4"},
-   {a="([^uai])(')([ui]N?%p?)$", b="%1ء%3"},
-   {a="([^uai])(')([ui]N?)(%p?%s)", b="%1ء%3%4"},
+   -- hamzah alone on the line, so take out the following two lines
+   -- (final hamzah is on the line after a letter of prolongation or a
+   -- consonant with sukūn)
+   -- {a="([^Auai])(')(\"?[uai]N?)(%p?)$", b="%1ء%3%4"}, --new
+   -- {a="([^Auai])(')(\"?[uai]N?)(%p?%s)", b="%1ء%3%4"},
+   {a="([^Auai])(')(\"?aN)(%p?)$", b="%1ئ%3%4"}, --new
+   {a="([^Auai])(')(\"?aN)(%p?%s)", b="%1ئ%3%4"}, --new
+   {a="([^uai])(')(\"?a)(%p?)$", b="%1ء%3%4"}, --new
+   {a="([^uai])(')(\"?a)(%p?%s)", b="%1ء%3%4"}, --new
+   {a="([^uai])(')(\"?[ui]N?)(%p?)$", b="%1ء%3%4"}, --new
+   {a="([^uai])(')(\"?[ui]N?)(%p?%s)", b="%1ء%3%4"}, --new
+   --easy (end)
 -- u
    {a="(u)(')([uai]N?%p?)$", b="%1ؤ%3"},
    {a="(u)(')([uai]N?)(%p?%s)", b="%1ؤ%3%4"},
@@ -281,7 +267,7 @@ hamzafveasy = { -- differences marked below with 'easy'
    {a="(i)(')(%p?%s)", b="%1ئْ%3"},
 --
    -- middle
-   {a="(U)(')", b="%1ء"},
+   {a="([Uw])(')", b="%1ء"}, --new
    {a="([Iy])(')", b="%1ئ"},
    {a="([^uai])(')([uU])", b="%1ؤ%3"},
    {a="([^uai])(')(%_?[aAY])", b="%1أ%3"},
@@ -321,7 +307,7 @@ tanwinfv = { -- with assimilations (\SetArbDflt*)
    {a="%-?(aN)(_A)(%s)([rlmnwy])", b="ًى%3%4%4"},
    {a="%-?(aN)(Y)(%s)([rlmnwy])", b="ًى%3%4%4"},
    {a="(T)%-?(aN)(%s)([rlmnwy])", b="%1ً%3%4%4"},
-   {a="(ء)%-?(aN)(%s)([rlmnwy])", b="%1ً%3%4%4"},
+   {a="(ء)%-?(aN)(%s)([rlmnwy])", b="%1%2%3%4%4"}, --new
    {a="([^TA])%-?(aN)(%s)([rlmnwy])", b="%1ًا%3%4%4"},
    {a="%-?(iNI?)(%s)([rlmnwy])", b="ٍ%2%3%3"},
    -- assimilations (end)
@@ -333,7 +319,7 @@ tanwinfv = { -- with assimilations (\SetArbDflt*)
    {a="%-?(\"aN)(_A)", b="ى"},
    {a="%-?(\"aN)(Y)", b="ى"},
    {a="(T)%-?(\"aN)", b="%1"},
-   {a="(ء)%-?(\"aN)", b="%1"},
+   {a="([��آ])(��)%-?(\"aN)", b="%1%2"}, --new
    {a="([^TA])%-?(\"aN)", b="%1ا"},
    {a="%-?(\"iNI?)", b=""},
    -- quoted tanwīn (end)
@@ -342,7 +328,7 @@ tanwinfv = { -- with assimilations (\SetArbDflt*)
    {a="%-?(aN)(_A)", b="ًى"},
    {a="%-?(aN)(Y)", b="ًى"},
    {a="(T)%-?(aN)", b="%1ً"},
-   {a="(ء)%-?(aN)", b="%1ً"},
+   {a="([��آ])(��)%-?(aN)", b="%1%2ً"}, --new
    {a="([^TA])%-?(aN)", b="%1ًا"},
    {a="%-?(iNI?)", b="ٍ"}
 }
@@ -368,7 +354,7 @@ tanwinfveasy = { -- no assimilations (see below)
 --easy   {a="%-?(aN)(_A)(%s)([rlmnwy])", b="ًى%3%4%4"},
 --easy   {a="%-?(aN)(Y)(%s)([rlmnwy])", b="ًى%3%4%4"},
 --easy   {a="(T)%-?(aN)(%s)([rlmnwy])", b="%1ً%3%4%4"},
---easy   {a="(ء)%-?(aN)(%s)([rlmnwy])", b="%1ً%3%4%4"},
+--easy   {a="(ء)%-?(aN)(%s)([rlmnwy])", b="%1%2%3%4%4"}, --new
 --easy   {a="([^TA])%-?(aN)(%s)([rlmnwy])", b="%1ًا%3%4%4"},
 --easy   {a="%-?(iNI?)(%s)([rlmnwy])", b="ٍ%2%3%3"},
    -- assimilations (end)
@@ -380,7 +366,7 @@ tanwinfveasy = { -- no assimilations (see below)
    {a="%-?(\"aN)(_A)", b="ى"},
    {a="%-?(\"aN)(Y)", b="ى"},
    {a="(T)%-?(\"aN)", b="%1"},
-   {a="(ء)%-?(\"aN)", b="%1"},
+   {a="([��آ])(��)%-?(\"aN)", b="%1%2"}, --new
    {a="([^TA])%-?(\"aN)", b="%1ا"},
    {a="%-?(\"iNI?)", b=""},
    -- quoted tanwīn (end)
@@ -389,7 +375,7 @@ tanwinfveasy = { -- no assimilations (see below)
    {a="%-?(aN)(_A)", b="ًى"},
    {a="%-?(aN)(Y)", b="ًى"},
    {a="(T)%-?(aN)", b="%1ً"},
-   {a="(ء)%-?(aN)", b="%1ً"},
+   {a="([��آ])(��)%-?(aN)", b="%1%2ً"}, --new
    {a="([^TA])%-?(aN)", b="%1ًا"},
    {a="%-?(iNI?)", b="ٍ"}
 }
@@ -554,7 +540,7 @@ trigraphsfveasy = { -- trigraphs or more (see 'easy' tag below for the diffs)
 }
 
 digraphsfvidgham = {
-   -- iʿrāb: straight double quote must be discarded
+   -- 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"},
@@ -808,7 +794,7 @@ digraphsfv = {
 }
 
 digraphsfveasy = { -- see the differences under 'easy' marker below
-   -- iʿrāb: straight double quote must be discarded
+   -- 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"},