arabluatex (5jul17)

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

1 files changed, 44 insertions, 8 deletions

diff --git a/Master/texmf-dist/tex/lualatex/arabluatex/arabluatex_trans.lua b/Master/texmf-dist/tex/lualatex/arabluatex/arabluatex_trans.lua
index 35f4b677cf6..404ae220a58 100644
--- a/Master/texmf-dist/tex/lualatex/arabluatex/arabluatex_trans.lua
+++ b/Master/texmf-dist/tex/lualatex/arabluatex/arabluatex_trans.lua
@@ -31,7 +31,8 @@ punctuationtr = {
 
 nulltr = {
    {a="%|", b=""},
-   {a="o[%S]-o", b=""}
+   {a="o[%S]-o", b=""},
+   {a="[%^%_](.)", b="<??>%1"}
 }
 
 -- cap
@@ -71,6 +72,17 @@ captr = {
 -- dmg
 
 hamzatrdmg = {
+   -- next lines for ʾalif alone
+   {a="(%.A)l%-(%^n)", b=".|l-%2"}, --additional (^n is lunar)
+   {a="([%(%[%|%<%s%-o])(%.A)l%-(%^n)", b="%1.|l-%3"}, --additional (^n is lunar) --p
+   {a="(%.A)l%-([%_%^%.]?[tdrzsnc])", b=".|%2-%2"},
+   {a="([%(%[%|%<%s%-o])(%.A)l%-([%_%^%.]?[tdrzsnc])", b="%1.|%3-%3"}, --p
+   {a="(%.A)([uai])l%-(%^n)", b="||%2l-%3"}, --additional (^n is lunar)
+   {a="([%(%[%|%<%s%-o])(%.A)([uai])l%-(%^n)", b="%1||%3l-%4"}, --additional (^n is lunar) --p
+   {a="(%.A)([uai])l%-([%_%^%.]?[tdrzsnc])", b="||%2%3-%3"},
+   {a="([%(%[%|%<%s%-o])(%.A)([uai])l%-([%_%^%.]?[tdrzsnc])", b="%1||%3%4-%4"}, --p
+   {a="(%.A)([^uai])", b=".|%2"},
+   {a="(%.A)([uai])", b="||%2"},
    -- hard coded hamza
    {a="|\"'", b="ʾ"},
    {a="A\"'", b="ʾA"},
@@ -108,8 +120,11 @@ hamzatrdmg = {
    -- then the 'initial' rules for the remaining cases
    {a="^(')([ua])", b="ʾ%2"},
    {a="^(')(i)", b="ʾ%2"},
-   {a="(%W)(')([ua])", b="%1ʾ%3"},
-   {a="(%W)(')(i)", b="%1ʾ%3"},
+   -- consider replacing initial %W with [%s%(%[%<%-]:
+   --   {a="(%W)(')([ua])", b="%1ʾ%3"},
+   --   {a="(%W)(')(i)", b="%1ʾ%3"},
+   {a="([%s%(%[%<%-])(')([ua])", b="%1ʾ%3"},
+   {a="([%s%(%[%<%-])(')(i)", b="%1ʾ%3"},
    -- final
    {a="([Iy])(')(aN)$", b="%1ʾ%3"},
    {a="([Iy])(')(aN)(%W)", b="%1ʾ%3%4"},
@@ -243,7 +258,7 @@ trigraphstrdmg = { -- trigraphs or more
    {a="'l%-(%^n)", b="'l-%1"}, -- ^n is lunar
    {a="'l%-([%_%^%.]?[tdrzsnc])", b="'%1-%1"},
    -- li-/la- + art. + lām
-   {a="l([ai])%-l%-(l)", b="l%1-%2%2"},
+   {a="l([ai])%-l%-(l)", b="l%1-%2-%2"},
    -- assim. art. with waṣla + solar consonant
    -- ('c' and '^n' are additional characters)
    {a="'(%^n)%-", b="'l-"}, -- ^n is lunar
@@ -365,6 +380,7 @@ digraphstrdmg = {
    {a="%.t", b="ṭ"},
    {a="%.z", b="ẓ"},
    {a="%.g", b="ġ"},
+   {a="%.y", b="y"},
    -- additional characters (begin)
    {a="%^c", b="č"},
    {a="%^z", b="ž"},
@@ -419,6 +435,8 @@ longvtrdmg = {
    {a="A", b="ā"},
    {a="U", b="ū"},
    {a="I", b="ī"},
+   {a="aY", b="ay"},
+   {a="iY", b="ī"},
    {a="Y", b="ạ̄"}
 }
 
@@ -431,6 +449,9 @@ shortvtrdmg = {
 -- loc
 
 hamzatrloc = {
+   -- next lines for ʾalif alone
+   {a="(%.A)([^uai])", b=".|%2"},
+   {a="(%.A)([uai])", b="||%2"},
    -- hard coded hamza
    {a="|\"'", b="ʾ"},
    {a="A\"'", b="ʾA"},
@@ -468,8 +489,11 @@ hamzatrloc = {
    -- initial (needs both ^ and %W patterns)
    {a="^(')([ua])", b="%2"},
    {a="^(')(i)", b="%2"},
-   {a="(%W)(')([ua])", b="%1%3"},
-   {a="(%W)(')(i)", b="%1%3"},
+   -- consider replacing initial %W with [%s%(%[%<%-]:
+   --   {a="(%W)(')([ua])", b="%1%3"},
+   --   {a="(%W)(')(i)", b="%1%3"},
+   {a="([%s%(%[%<%-])(')([ua])", b="%1%3"},
+   {a="([%s%(%[%<%-])(')(i)", b="%1%3"},
    -- final
    {a="([Iy])(')(aN)$", b="%1ʾ%3"},
    {a="([Iy])(')(aN)(%W)", b="%1ʾ%3%4"},
@@ -610,6 +634,7 @@ digraphstrloc = {
    {a="%.t", b="ṭ"},
    {a="%.z", b="ẓ"},
    {a="%.g", b="gh"},
+   {a="%.y", b="y"},
    -- additional characters (begin)
    {a="%^c", b="ch"},
    {a="%^z", b="zh"},
@@ -663,6 +688,8 @@ longvtrloc = {
    {a="A", b="ā"},
    {a="U", b="ū"},
    {a="I", b="ī"},
+   {a="aY", b="ay"},
+   {a="iY", b="ī"},
    {a="Y", b="á"},
 }
 
@@ -679,6 +706,9 @@ finaltrloc = {
 -- arabica
 
 hamzatrarabica = { -- ≠ from hamzatrloc: initial hamza has to be held
+   -- next lines for ʾalif alone
+   {a="(%.A)([^uai])", b=".|%2"},
+   {a="(%.A)([uai])", b="||%2"},
    -- hard coded hamza
    {a="|\"'", b="ʾ"},
    {a="A\"'", b="ʾA"},
@@ -717,8 +747,11 @@ hamzatrarabica = { -- ≠ from hamzatrloc: initial hamza has to be held
    -- 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"},
+   -- consider replacing initial %W with [%s%(%[%<%-]:
+   --   {a="(%W)(')([ua])", b="%1@%3"},
+   --   {a="(%W)(')(i)", b="%1@%3"},
+   {a="([%s%(%[%<%-])(')([ua])", b="%1@%3"},
+   {a="([%s%(%[%<%-])(')(i)", b="%1@%3"},
    -- final
    {a="([Iy])(')(aN)$", b="%1ʾ%3"},
    {a="([Iy])(')(aN)(%W)", b="%1ʾ%3%4"},
@@ -846,6 +879,7 @@ digraphstrarabica = {
    {a="%.t", b="ṭ"},
    {a="%.z", b="ẓ"},
    {a="%.g", b="ġ"},
+   {a="%.y", b="y"},
    -- additional characters (begin)
    {a="%^c", b="č"},
    {a="%^z", b="ž"},
@@ -896,6 +930,8 @@ singletrarabica = {
 }
 
 longvtrarabica = {
+   {a="aY", b="ay"},
+   {a="iY", b="ī"},
    {a="[AY]", b="ā"},
    {a="U", b="ū"},
    {a="I", b="ī"}