done implementing 'easy' rules set

1 files changed, 176 insertions, 0 deletions

diff --git a/arabluatex_voc.lua b/arabluatex_voc.lua
index 5eafca5..4f5399f 100644
--- a/arabluatex_voc.lua
+++ b/arabluatex_voc.lua
@@ -154,6 +154,89 @@ hamza = {
    {a="(i)(')([^uaiUAI])", b="%1ئ%3"}
 }
 
+hamzaeasy = { -- differences marked below with 'easy'
+   -- hard coded hamza
+   {a="|\"'", b="ء"},
+   {a="A\"'", b="آ"},
+   {a="[au]\"'", b="أ"},
+   {a="w\"'", b="ؤ"},
+   {a="i\"'", b="إ"},
+   {a="y\"'", b="ئ"},
+   -- hamza takes tašdīd too
+   {a="''([Uu])", b="ؤؤ%1"},
+   {a="''([Aa])", b="أأ%1"},
+   {a="''([Ii])", b="ئئ%1"},
+   -- initial long u and i (for a, see below)
+   {a="%'%_U", b="أU"},
+   {a="%'%_I", b="إI"},
+   -- taḫfīfu 'l-hamza
+   {a="'u'([^uaiUAI])", b="أU%1"},
+   {a="'i'([^uaiUAI])", b="إI%1"},   
+   -- madda (historic writing below)
+   {a="'a'([^uaiUAI])", b="آ%1"},
+   {a="'a?A([%_%^%.]?[%`%'btjghdrzsfqklmnywAY])", b="آ%1"},
+--easy   {a="(A)(')(uN?)$", b="aآء%3"},
+--easy   {a="(A)(')(uN?)(%W)", b="aآء%3%4"},
+--easy   {a="(A)(')(iN?)$", b="aآء%3"},
+--easy   {a="(A)(')(iN?)(%W)", b="aآء%3%4"},
+--easy   {a="(A)(')(i)", b="aآئ%3"}, -- historic madda
+--easy   {a="(A)(')(u)", b="aآؤ%3"}, -- historic madda
+--easy   {a="(A)(')", b="aآء"}, -- historic madda
+   -- initial (needs both ^ and %W patterns)
+   {a="^(')([ua])", b="أ%2"},
+   {a="^(')(i)", b="إ%2"},
+   {a="(%W)(')([ua])", b="%1أ%3"},
+   {a="(%W)(')(i)", b="%1إ%3"},
+   -- final
+   -- ^say'aN and .zim'aN are special orthographies
+   {a="(%^say)(%')(aN)", b="%1ئ%3"},
+   {a="(.zi?m)(%')(aN)", b="%1ئ%3"},
+   {a="([^uai])(')([uai]N?)$", b="%1ء%3"},
+   {a="([^uai])(')([uai]N?)(%W)", b="%1ء%3%4"},
+-- u
+   {a="(u)(')([uai]?N)$", b="%1ؤ%3"},
+   {a="(u)(')([uai]N?)(%W)", b="%1ؤ%3%4"},
+   {a="(u)(')$", b="%1ؤ"},
+   {a="(u)(')(%W)", b="%1ؤ%3"},
+-- a
+   {a="(a)(')(A)$", b="%1آ"},
+   {a="(a)(')(A)(%W)", b="%1آ%4"},
+   {a="(a)(')([u]N?)$", b="%1أ%3"},
+   {a="(a)(')([u]N?)(%W)", b="%1أ%3%4"},
+   {a="(a)(')(a)$", b="%1أ%3"},
+   {a="(a)(')(a)(%W)", b="%1أ%3%4"},
+   {a="(a)(')(aN)$", b="%1أً"},
+   {a="(a)(')(aN)(%W)", b="%1أً%4"},
+   {a="(a)(')([i]N?)$", b="%1إ%3"},
+   {a="(a)(')([i]N?)(%W)", b="%1إ%3%4"},
+   {a="(a)(')$", b="%1أ"},
+   {a="(a)(')(%W)", b="%1أ%3"},
+-- i
+   {a="(i)(')([uai]N?)$", b="%1ئ%3"},
+   {a="(i)(')([uai]N?)(%W)", b="%1ئ%3%4"},
+   {a="(i)(')$", b="%1ئ"},
+   {a="(i)(')(%W)", b="%1ئ%3"},
+--
+   -- middle
+   {a="(U)(')", b="%1ء"},
+   {a="([Iy])(')", b="%1ئ"},
+   {a="([^uai])(')([uU])", b="%1ؤ%3"},
+   {a="([^uai])(')([aA])", b="%1أ%3"},
+   {a="([^uai])(')([iI])", b="%1ئ%3"},
+   {a="(u)(')([uU])", b="%1ؤ%3"},
+   {a="(u)(')([aA])", b="%1ؤ%3"},
+   {a="(u)(')([iI])", b="%1ئ%3"},
+   {a="(a)(')([aA])", b="%1أ%3"},
+   {a="(a)(')([uU])", b="%1ؤ%3"},
+   {a="(a)(')([iI])", b="%1ئ%3"},
+   {a="(i)(')([aA])", b="%1ئ%3"},
+   {a="(i)(')([uU])", b="%1ئ%3"},
+   {a="(i)(')([iI])", b="%1ئ%3"},
+   {a="(a)(')([^uaiUAI])", b="%1أ%3"},
+   {a="(u)(')([^uaiUAI])", b="%1ؤ%3"},
+   {a="(i)(')([^uaiUAI])", b="%1ئ%3"}
+}
+
 tanwin = {
    {a="uNU", b="ٌو"},
    {a="aNU", b="ًوا"},
@@ -187,6 +270,39 @@ tanwin = {
    {a="(iN)", b="ٍ"}
 }
 
+tanwineasy = { -- 'easy' requires some lines to be taken out:
+   {a="uNU", b="ٌو"},
+   {a="aNU", b="ًوا"},
+   {a="iNU", b="ٍو"},
+   -- assimilations (begin)
+--   {a="(uN)(%s)([rlmnwy])", b="ٌ%2%3%3"},
+--   {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="([^TA])(aN)(%s)([rlmnwy])", b="%1ًا%3%4%4"},
+--   {a="(iN)(%s)([rlmnwy])", b="ٍ%2%3%3"},
+   -- assimilations (end)
+   -- quoted tanwīn (begin)
+   {a="(\"uN)", b=""},
+   {a="(B)(\"aN)", b="%1"},
+   {a="(\"aN)(_A)", b="ى"},
+   {a="(\"aN)(Y)", b="ى"},
+   {a="(T)(\"aN)", b="%1"},
+   {a="(ء)(\"aN)", b="%1"},
+   {a="([^TA])(\"aN)", b="%1ا"},
+   {a="(\"iN)", b=""},
+   -- quoted tanwīn (end)
+   {a="(uN)", b="ٌ"},
+   {a="(B)(aN)", b="%1ً"},
+   {a="(aN)(_A)", b="ًى"},
+   {a="(aN)(Y)", b="ًى"},
+   {a="(T)(aN)", b="%1ً"},
+   {a="(ء)(aN)", b="%1ً"},
+   {a="([^TA])(aN)", b="%1ًا"},
+   {a="(iN)", b="ٍ"}
+}
+
 trigraphs = { -- trigraphs or more
    -- 'llatI / 'llad_I
    {a="^'ll(a)([%_]?[dt])", b="الّ%1%2"},
@@ -247,6 +363,66 @@ trigraphs = { -- trigraphs or more
    {a="(n)(%s)([rlmnwy])", b="%1%2%3%3"}
 }
 
+trigraphseasy = { -- differences marked below with 'easy'
+   -- 'llatI / 'llad_I
+   {a="^'ll(a)([%_]?[dt])", b="الّ%1%2"},
+   {a="([%s%-])'ll(a)([%_]?[dt])", b="%1الّ%2%3"},
+   -- al- + lām (easy)
+   {a="^(a)l%-(l)", b="ا%1ل%2"},
+   {a="([%s%-])(a)l%-(l)", b="%1ا%2ل%3"},
+   -- al- + solar consonant (easy)
+   {a="^(a)l%-([%_%^%.]?[tdrzsn])", b="ا%1ل%2"},
+   {a="([%s%-])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3"},
+   -- assim. art. + solar consonant (easy)
+   {a="^(a)([%_%^%.]?[tdrzsn])%-", b="ا%1ل"},
+   {a="([%s%-])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل"},
+   -- al- + initial unstable hamza
+   {a="^(a)l%-(\")([uai])", b="ا%1لٱ%3"},
+   {a="([%s%-])(a)l%-(\")([uai])", b="%1ا%2لٱ%4"},
+   {a="^(a)l%-([uai])", b="ا%1لا%2"},
+   {a="([%s%-])(a)l%-([uai])", b="%1ا%2لا%3"},
+   -- li-/la- + art. + initial unstable hamza is a special orthography
+   {a="l([ai])%-l%-(\")([uai])", b="ل%1لٱ%3"},
+   {a="l([ai])%-l%-([uai])", b="ل%1لا%2"},
+   -- al- + lunar consonant (i.e. what remains)
+   {a="^(a)l%-", b="ا%1ل"},
+   {a="([%s%-])(a)l%-", b="%1ا%2ل"},
+   -- diphthongs to be resolved before ʾalif conjunctionis
+   {a="(aW)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="awuا%2%3"},
+   {a="(aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1u%2%3"},
+   {a="(ay)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
+   -- art. with waṣla + lām (easy)
+   {a="'l%-(l)", b="ال%1"},
+   -- art. with waṣla + solar consonant (easy)
+   {a="'l%-([%_%^%.]?[tdrzsn])", b="ال%1"},
+   -- li-/la- + art. + lām (easy)
+   {a="l([ai])%-l%-(l)", b="ل%1%2"},
+   -- assim. art. with waṣla + solar consonant (easy)
+   {a="'([%_%^%.]?[tdrzsn])%-", b="ال"},
+   -- li-/la- + art. + solar consonant is a special orthography (easy)
+   {a="l([ai])%-l%-([%_%^%.]?[tdrzsn])", b="ل%1ل%2"},
+   -- li-/la + assim. art. + solar consonant is a special orthography (easy)
+   {a="l([ai])%-([%_%^%.]?[tdrzsn])%-([%_%^%.]?[tdrzsn])", b="ل%1ل%3"},
+   -- art. with waṣla + initial unstable hamza
+   {a="'l%-(\")([uai])", b="الٱ%2"},
+   {a="'l%-([uai])", b="الا%1"},
+   -- art. with waṣla + lunar consonant (i.e. what remains)
+   {a="'l%-", b="ال"},
+   -- the silent wāw
+   {a="uU$", b="uو"},
+   {a="uU(%W)", b="uو%1"},
+   {a="aU$", b="aو"},
+   {a="aU(%W)", b="aو%1"},
+   {a="iU$", b="iو"},
+   {a="iU(%W)", b="iو%1"},
+   -- words ending in -āT with silent wāw/yāʾ
+   {a="(_a)UA", b="%1وا"},
+   {a="(_a)U", b="%1و"},
+   {a="(_a)I", b="%1ي"},
+   -- assimilations
+--easy   {a="(n)(%s)([rlmnwy])", b="%1%2%3%3"}
+}
+
 digraphs = {
    -- initial straight double quote gives a connective ʾalif
    {a="^\"[uai]", b="ٱ"},