1 files changed, 39 insertions, 17 deletions

diff --git a/arabluatex_voc.lua b/arabluatex_voc.lua
index 61e4c4a..02c3feb 100644
--- a/arabluatex_voc.lua
+++ b/arabluatex_voc.lua
@@ -201,6 +201,12 @@ hamza = {
    {a="(A)(')(u)", b="aآؤ%3"}, -- historic madda
    {a="(A)(')", b="aآء"}, -- historic madda
    -- initial (needs both ^ and %W patterns)
+   -- 'aw: the diphthong is to be resoved into 'awi' (next 4 lines)
+   {a="^('aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
+   {a="(%W)('aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1%2i%3%4"},
+   {a="^('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"},
+   {a="(%W)('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"},
+   -- then the 'initial' rules for the remaining cases
    {a="^(')([ua])", b="أ%2"},
    {a="^(')(i)", b="إ%2"},
    {a="(%W)(')([ua])", b="%1أ%3"},
@@ -292,6 +298,12 @@ hamzaeasy = { -- differences marked below with 'easy'
    {a="(A)(')", b="aاء"}, -- historic madda
    --easy (end)
    -- initial (needs both ^ and %W patterns)
+   -- 'aw: the diphthong is to be resoved into 'awi' (next 4 lines)
+   {a="^('aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
+   {a="(%W)('aw)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1%2i%3%4"},
+   {a="^('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"},
+   {a="(%W)('aw)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"},
+   -- then the 'initial' rules for the remaining cases
    {a="^(')([ua])", b="أ%2"},
    {a="^(')(i)", b="إ%2"},
    {a="(%W)(')([ua])", b="%1أ%3"},
@@ -415,27 +427,32 @@ tanwineasy = { -- 'easy' requires some lines to be taken out:
 trigraphs = { -- trigraphs or more
    -- 'llatI / 'llad_I
    {a="^'ll(a)([%_]?[dt])", b="الّ%1%2"},
-   {a="([%p%s%-])'ll(a)([%_]?[dt])", b="%1الّ%2%3"}, --p
+   {a="([%(%[%|%<%s%-])'ll(a)([%_]?[dt])", b="%1الّ%2%3"}, --p
+   -- law: the diphthong is to be resoved into 'awi' (next 4 lines)
+   {a="^(law)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
+   {a="(%W)(law)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1%2i%3%4"},
+   {a="^(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"},
+   {a="(%W)(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"},
    -- al- + lām
    {a="^(a)l%-(l)", b="ا%1ل%2%2"},
-   {a="([%p%s%-])(a)l%-(l)", b="%1ا%2ل%3%3"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-(l)", b="%1ا%2ل%3%3"}, --p
    -- al- + solar consonant
    {a="^(a)l%-([%_%^%.]?[tdrzsn])", b="ا%1ل%2%2"},
-   {a="([%p%s%-])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3%3"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3%3"}, --p
    -- assim. art. + solar consonant
    {a="^(a)([%_%^%.]?[tdrzsn])%-", b="ا%1ل%2"},
-   {a="([%p%s%-])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل%3"}, --p
+   {a="([%(%[%|%<%s%-])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل%3"}, --p
    -- al- + initial unstable hamza
    {a="^(a)l%-(\")([uai])", b="ا%1ل%3ٱ"},
-   {a="([%p%s%-])(a)l%-(\")([uai])", b="%1ا%2ل%4ٱ"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-(\")([uai])", b="%1ا%2ل%4ٱ"}, --p
    {a="^(a)l%-([uai])", b="ا%1ل%2ا"},
-   {a="([%p%s%-])(a)l%-([uai])", b="%1ا%2ل%3ا"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-([uai])", b="%1ا%2ل%3ا"}, --p
    -- li-/la- + art. + initial unstable hamza is a special orthography
    {a="l([ai])%-l%-(\")([uai])", b="ل%1ل%3ٱ"},
    {a="l([ai])%-l%-([uai])", b="ل%1ل%2ا"},
    -- al- + lunar consonant (i.e. what remains)
    {a="^(a)l%-", b="ا%1ل"},
-   {a="([%p%s%-])(a)l%-", b="%1ا%2ل"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-", b="%1ا%2ل"}, --p
    -- 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"},
@@ -480,27 +497,32 @@ trigraphseasy = { -- differences marked below with 'easy'
    {a="l%-l_ah", b="l-ll_ah"},
    -- 'llatI / 'llad_I
    {a="^'ll(a)([%_]?[dt])", b="الّ%1%2"},
-   {a="([%p%s%-])'ll(a)([%_]?[dt])", b="%1الّ%2%3"}, --p
+   {a="([%(%[%|%<%s%-])'ll(a)([%_]?[dt])", b="%1الّ%2%3"}, --p
+   -- law: the diphthong is to be resoved into 'awi' (next 4 lines)
+   {a="^(law)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1i%2%3"},
+   {a="(%W)(law)(%s)(['][%_%^%.]?[l'btjghxdrzs`fqkmnwy]%-)", b="%1%2i%3%4"},
+   {a="^(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1i%2%3"},
+   {a="(%W)(law)(%s)([%(%[%|%<]?\"?[uai])", b="%1%2i%3%4"},
    -- al- + lām (easy)
    {a="^(a)l%-(l)", b="ا%1ل%2"},
-   {a="([%p%s%-])(a)l%-(l)", b="%1ا%2ل%3"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-(l)", b="%1ا%2ل%3"}, --p
    -- al- + solar consonant (easy)
    {a="^(a)l%-([%_%^%.]?[tdrzsn])", b="ا%1ل%2"},
-   {a="([%p%s%-])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-([%_%^%.]?[tdrzsn])", b="%1ا%2ل%3"}, --p
    -- assim. art. + solar consonant (easy)
    {a="^(a)([%_%^%.]?[tdrzsn])%-", b="ا%1ل"},
-   {a="([%p%s%-])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل"}, --p
+   {a="([%(%[%|%<%s%-])(a)([%_%^%.]?[tdrzsn])%-", b="%1ا%2ل"}, --p
    -- al- + initial unstable hamza
    {a="^(a)l%-(\")([uai])", b="ا%1ل%3ٱ"},
-   {a="([%p%s%-])(a)l%-(\")([uai])", b="%1ا%2ل%4ٱ"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-(\")([uai])", b="%1ا%2ل%4ٱ"}, --p
    {a="^(a)l%-([uai])", b="ا%1ل%2ا"},
-   {a="([%p%s%-])(a)l%-([uai])", b="%1ا%2ل%3ا"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-([uai])", b="%1ا%2ل%3ا"}, --p
    -- li-/la- + art. + initial unstable hamza is a special orthography
    {a="l([ai])%-l%-(\")([uai])", b="ل%1ل%3ٱ"},
    {a="l([ai])%-l%-([uai])", b="ل%1ل%2ا"},
    -- al- + lunar consonant (i.e. what remains)
    {a="^(a)l%-", b="ا%1ل"},
-   {a="([%p%s%-])(a)l%-", b="%1ا%2ل"}, --p
+   {a="([%(%[%|%<%s%-])(a)l%-", b="%1ا%2ل"}, --p
    -- 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"},
@@ -550,12 +572,12 @@ digraphs = {
    -- ʾiʿrāb (end)
    -- initial straight double quote gives a connective ʾalif
    {a="^\"[uai]", b="ٱ"},
-   {a="([%p%s%-])\"[uai]", b="%1ٱ"}, --p
+   {a="([%(%[%|%<%s%-])\"[uai]", b="%1ٱ"}, --p
    {a="(aW)(%s)([uai])", b="awuا%2%3"},
    -- hyphen + initial alif without hamza:
    {a="(%-)([uai])([%^%_%.%`]?)(%a)", b="%1ا%3%4"},
-   {a="^([uai])", b="ا%1"},      -- initial alif without hamza
-   {a="([%p%s])([uai])", b="%1ا"}, -- initial alif without hamza --p
+   {a="^([%(%[%|%<]?)([uai])", b="%1ا%2"}, -- initial alif without hamza --p
+   {a="(%s)([%(%[%|%<]?)([uai])", b="%1%2ا"}, -- initial alif without hamza --p
    {a="%-%-", b="ـ"},
    {a="ؤؤ", b="ؤّ"},
    {a="أأ", b="أّ"},