arabluatex (5jul17)

git-svn-id: svn://tug.org/texlive/trunk@44754 c570f23f-e606-0410-a88d-b1316a301751

1 files changed, 44 insertions, 16 deletions

diff --git a/Master/texmf-dist/tex/lualatex/arabluatex/arabluatex_fullvoc.lua b/Master/texmf-dist/tex/lualatex/arabluatex/arabluatex_fullvoc.lua
index 99efb752791..919a35ed2a3 100644
--- a/Master/texmf-dist/tex/lualatex/arabluatex/arabluatex_fullvoc.lua
+++ b/Master/texmf-dist/tex/lualatex/arabluatex/arabluatex_fullvoc.lua
@@ -25,6 +25,12 @@ along with this program.  If not, see
 hamzafv = {
    -- next line for ʾiʿrāb hyphen
    {a="(')(%-)([uaiUAI])", b="%1%3"},
+   -- next lines for ʾalif alone
+   {a="(%.A)([uai]?)l%-(%^n)", b="ا%2ل%3"}, --additional (^n is lunar)
+   {a="([%(%[%|%<%s%-o])(%.A)([uai]?)l%-(%^n)", b="%1ا%3%4"}, --additional (^n is lunar) --p
+   {a="(%.A)([uai]?)l%-([%_%^%.]?[tdrzsnc])", b="ا%2ل%3%3"},
+   {a="([%(%[%|%<%s%-o])(%.A)([uai]?)l%-([%_%^%.]?[tdrzsnc])", b="%1ا%3ل%4%4"}, --p
+   {a="%.A", b="ا"},
    -- hard coded hamza
    {a="|\"'", b="ء"},
    {a="A\"'", b="آ"},
@@ -78,14 +84,19 @@ hamzafv = {
    -- then the 'initial' rules for the remaining cases
    {a="^(')([ua])", b="أ%2"},
    {a="^(')(i)", b="إ%2"},
-   {a="(%W)(')([ua])", b="%1أ%3"},
-   {a="(%W)(')(i)", b="%1إ%3"},
+   -- consider replacing initial %W with [%s%(%[%<%-]:
+   --   {a="(%W)(')([ua])", b="%1أ%3"},
+   --   {a="(%W)(')(i)", b="%1إ%3"},
+   {a="([%s%(%[%<%-])(')([ua])", b="%1أ%3"},
+   {a="([%s%(%[%<%-])(')(i)", b="%1إ%3"},
    -- final
    -- ^say'aN and .zim'aN are special orthographies
-   {a="(%^say)(%')(aN)", b="%1ئ%3"},
+   {a="(%^sa%.?[yY])(\"?%|?)(%')(aN)", b="%1%2ئ%4"},
    {a="(.zi?m)(%')(aN)", b="%1ئ%3"},
-   {a="(')(I%p?)$", b="إ%2"}, -- affix pr. 1st person sg.
-   {a="(')(I)(%p?%s)", b="إ%2%3"}, -- affix pr. 1st person sg.
+   -- hamzah + pr. suffix: the carrier passes into yāʾ before ī so
+   -- take out the following two lines:
+   -- {a="(')(I%p?)$", b="إ%2"}, -- affix pr. 1st person sg.
+   -- {a="(')(I)(%p?%s)", b="إ%2%3"}, -- affix pr. 1st person sg.
    {a="([^uai])(')([uai]N?%p?)$", b="%1ء%3"},
    {a="([^uai])(')([uai]N?)(%p?%s)", b="%1ء%3%4"},
 -- u
@@ -135,6 +146,12 @@ hamzafv = {
 hamzafveasy = { -- differences marked below with 'easy'
    -- next line for ʾiʿrāb hyphen
    {a="(')(%-)([uaiUAI])", b="%1%3"},
+   -- next lines for ʾalif alone
+   {a="(%.A)([uai]?)l%-(%^n)", b="ا%2ل%3"}, --additional (^n is lunar)
+   {a="([%(%[%|%<%s%-o])(%.A)([uai]?)l%-(%^n)", b="%1ا%3%4"}, --additional (^n is lunar) --p
+   {a="(%.A)([uai]?)l%-([%_%^%.]?[tdrzsnc])", b="ا%2ل%3%3"},
+   {a="([%(%[%|%<%s%-o])(%.A)([uai]?)l%-([%_%^%.]?[tdrzsnc])", b="%1ا%3ل%4%4"}, --p
+   {a="%.A", b="ا"},
    -- hard coded hamza
    {a="|\"'", b="ء"},
    {a="A\"'", b="آ"},
@@ -190,14 +207,19 @@ hamzafveasy = { -- differences marked below with 'easy'
    -- then the 'initial' rules for the remaining cases
    {a="^(')([ua])", b="أ%2"},
    {a="^(')(i)", b="إ%2"},
-   {a="(%W)(')([ua])", b="%1أ%3"},
-   {a="(%W)(')(i)", b="%1إ%3"},
+   -- consider replacing initial %W with [%s%(%[%<%-]:
+   --   {a="(%W)(')([ua])", b="%1أ%3"},
+   --   {a="(%W)(')(i)", b="%1إ%3"},
+   {a="([%s%(%[%<%-])(')([ua])", b="%1أ%3"},
+   {a="([%s%(%[%<%-])(')(i)", b="%1إ%3"},
    -- final
    -- ^say'aN and .zim'aN are special orthographies
-   {a="(%^say)(%')(aN)", b="%1ئ%3"},
+   {a="(%^sa%.?[yY])(\"?%|?)(%')(aN)", b="%1%2ئ%4"},
    {a="(.zi?m)(%')(aN)", b="%1ئ%3"},
-   {a="(')(I%p?)$", b="إ%2"}, -- affix pr. 1st person sg.
-   {a="(')(I)(%p?%s)", b="إ%2%3"}, -- affix pr. 1st person sg.
+   -- hamzah + pr. suffix: the carrier passes into yāʾ before ī so
+   -- take out the following two lines:
+   -- {a="(')(I%p?)$", b="إ%2"}, -- affix pr. 1st person sg.
+   -- {a="(')(I)(%p?%s)", b="إ%2%3"}, -- affix pr. 1st person sg.
    {a="([^uai])(')([uai]N?%p?)$", b="%1ء%3"},
    {a="([^uai])(')([uai]N?)(%p?%s)", b="%1ء%3%4"},
 -- u
@@ -565,6 +587,7 @@ digraphsfvidgham = {
    {a="hh", b="هّ"},
    {a="ww", b="وّ"},
    {a="yy", b="يّ"},
+   {a="%.y%.y", b="ىّ"},
    -- additional characters + šaddah (begin)
    {a="pp", b="پّ"},
    {a="vv", b="ڤّ"},
@@ -575,10 +598,10 @@ digraphsfvidgham = {
    -- additional characters + šaddah (end)
    -- sukūn begin
    -- first, take out hyphen if any:
-   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])%-([uaiUAI])", b="%1%2"},
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY])%-([uaiUAI])", b="%1%2"},
    {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy]%p?)$", b="%1ْ"},
    {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])(%p?%s)", b="%1ْ%2"},
-   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])([%_]?[^%_uaiUAIYًٌٍ])", b="%1ْ%2"},
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY])([%_]?[^%_uaiUAIYًٌٍ])", b="%1ْ%2"},
    -- take out sukūn in cases of assimilation
    {a="(n)(ْ)(%s)(ر)", b="%1%3%4"},
    {a="(n)(ْ)(%s)(و)", b="%1%3%4"},
@@ -599,6 +622,7 @@ digraphsfvidgham = {
    {a="%.t", b="ط"},
    {a="%.z", b="ظ"},
    {a="%.g", b="غ"},
+   {a="%.y", b="ى"},
    -- additional characters (begin)
    {a="%^c", b="چ"},
    {a="%^z", b="ژ"},
@@ -688,6 +712,7 @@ digraphsfv = {
    {a="hh", b="هّ"},
    {a="ww", b="وّ"},
    {a="yy", b="يّ"},
+   {a="%.y%.y", b="ىّ"},
    -- additional characters + šaddah (begin)
    {a="pp", b="پّ"},
    {a="vv", b="ڤّ"},
@@ -698,10 +723,10 @@ digraphsfv = {
    -- additional characters + šaddah (end)
    -- sukūn begin
    -- first, take out hyphen if any:
-   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])%-([uaiUAI])", b="%1%2"},
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY])%-([uaiUAI])", b="%1%2"},
    {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy]%p?)$", b="%1ْ"},
    {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])(%p?%s)", b="%1ْ%2"},
-   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])([%_]?[^%_uaiUAIYًٌٍ])", b="%1ْ%2"},
+   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY])([%_]?[^%_uaiUAIYًٌٍ])", b="%1ْ%2"},
    -- take out sukūn in cases of assimilation
 --   {a="(n)(ْ)(%s)(ر)", b="%1%3%4"},
 --   {a="(n)(ْ)(%s)(و)", b="%1%3%4"},
@@ -722,6 +747,7 @@ digraphsfv = {
    {a="%.t", b="ط"},
    {a="%.z", b="ظ"},
    {a="%.g", b="غ"},
+   {a="%.y", b="ى"},
    -- additional characters (begin)
    {a="%^c", b="چ"},
    {a="%^z", b="ژ"},
@@ -811,6 +837,7 @@ digraphsfveasy = { -- see the differences under 'easy' marker below
    {a="hh", b="هّ"},
    {a="ww", b="وّ"},
    {a="yy", b="يّ"},
+   {a="%.y%.y", b="ىّ"},
    -- additional characters + šaddah (begin)
    {a="pp", b="پّ"},
    {a="vv", b="ڤّ"},
@@ -824,10 +851,10 @@ digraphsfveasy = { -- see the differences under 'easy' marker below
    -- so there be no need to edit them:
    {a="ْ", b=""}, 
    -- first, take out hyphen if any:
---   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])%-([uaiUAI])", b="%1%2"},
+--   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY])%-([uaiUAI])", b="%1%2"},
 --   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy]%p?)$", b="%1ْ"},
 --   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])(%p?%s)", b="%1ْ%2"},
---   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwy])([%_]?[^%_uaiUAIًٌٍ])", b="%1ْ%2"},
+--   {a="([%_%^%.]?[Bbtjghxdrzs%`fqklmnwyY])([%_]?[^%_uaiUAIًٌٍ])", b="%1ْ%2"},
    -- take out sukūn in cases of assimilation
 --   {a="(n)(ْ)(%s)(ر)", b="%1%3%4"},
 --   {a="(n)(ْ)(%s)(و)", b="%1%3%4"},
@@ -848,6 +875,7 @@ digraphsfveasy = { -- see the differences under 'easy' marker below
    {a="%.t", b="ط"},
    {a="%.z", b="ظ"},
    {a="%.g", b="غ"},
+   {a="%.y", b="ى"},
    -- additional characters (begin)
    {a="%^c", b="چ"},
    {a="%^z", b="ژ"},