arabluatex (5jul17)

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

1 files changed, 38 insertions, 11 deletions

diff --git a/Master/texmf-dist/tex/lualatex/arabluatex/arabluatex_voc.lua b/Master/texmf-dist/tex/lualatex/arabluatex/arabluatex_voc.lua
index 6c63a883c49..ca1aa99ca87 100644
--- a/Master/texmf-dist/tex/lualatex/arabluatex/arabluatex_voc.lua
+++ b/Master/texmf-dist/tex/lualatex/arabluatex/arabluatex_voc.lua
@@ -170,6 +170,12 @@ buckwalter = {
 hamza = {
    -- 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"},
@@ -214,14 +220,19 @@ hamza = {
    -- 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?)$", b="%1ء%3"},
    {a="([^uai])(')([uai]N?)(%p?%s)", b="%1ء%3%4"},
 -- u
@@ -271,6 +282,12 @@ hamza = {
 hamzaeasy = { -- differences marked below with 'easy'
    -- next line for ʾiʿrāb hyphen
    {a="(')(%-)([uaiUAI])", b="%1%3"},
+   -- next lines for ʾalif alone (easy)
+   {a="(%.A)([uai]?)l%-(%^n)", b="ا%2ل%3"}, --additional (^n is lunar)
+   {a="([%(%[%|%<%s%-o])(%.A)([uai]?)l%-(%^n)", b="%1ا%3%4"}, --additional (^n is lunar) --p
+   {a="(%.A)([uai]?)l%-([%_%^%.]?[tdrzsnc])", b="ا%2ل%3"},
+   {a="([%(%[%|%<%s%-o])(%.A)([uai]?)l%-([%_%^%.]?[tdrzsnc])", b="%1ا%3ل%4"}, --p
+   {a="%.A", b="ا"},
    -- hard coded hamza
    {a="|\"'", b="ء"},
    {a="A\"'", b="آ"},
@@ -317,14 +334,19 @@ hamzaeasy = { -- differences marked below with 'easy'
    -- then the 'initial' rules for the remaining cases
    {a="^(')([ua])", b="أ%2"},
    {a="^(')(i)", b="إ%2"},
-   {a="(%W)(')([ua])", b="%1أ%3"},
-   {a="(%W)(')(i)", b="%1إ%3"},
+   -- consider replacing initial %W with [%s%(%[%<%-]:
+   --   {a="(%W)(')([ua])", b="%1أ%3"},
+   --   {a="(%W)(')(i)", b="%1إ%3"},
+   {a="([%s%(%[%<%-])(')([ua])", b="%1أ%3"},
+   {a="([%s%(%[%<%-])(')(i)", b="%1إ%3"},
    -- final
    -- ^say'aN and .zim'aN are special orthographies
-   {a="(%^say)(%')(aN)", b="%1ئ%3"},
+   {a="(%^sa%.?[yY])(\"?%|?)(%')(aN)", b="%1%2ئ%4"},
    {a="(.zi?m)(%')(aN)", b="%1ئ%3"},
-   {a="(')(I%p?)$", 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
@@ -695,6 +717,7 @@ digraphs = {
    {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"},
@@ -714,6 +737,7 @@ digraphs = {
    {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"},
@@ -776,6 +800,8 @@ longv = {
    {a="A", b="َا"},
    {a="U", b="ُو"},
    {a="I", b="ِي"},
+   {a="aY", b="aى"},
+   {a="iY", b="iى"},
    {a="Y", b="aى"}
 }
 
@@ -816,5 +842,6 @@ punctuation = {
 null = {
    {a="%|", b=""},
    {a="([^0-9])(%-)", b="%1"},
-   {a="o[%S]-o", b=""}
+   {a="o[%S]-o", b=""},
+   {a="[%^%_](.)", b=">??<%1"}
 }