From 869878ec3e73735796eff54dad21e41a3195965d Mon Sep 17 00:00:00 2001
From: jonsd <jonsd@d46cf337-b52f-0410-862d-fd96e6ae7743>
Date: Mon, 5 Oct 2009 12:20:07 +0000
Subject: [PATCH] [1.41.09] Ignore characters U+200c, U+200d (zero width
 non-joiner, zero width joiner).

git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@224 d46cf337-b52f-0410-862d-fd96e6ae7743
---
 dictsource/hu_list  |  32 ++++++++++++++++----------------
 dictsource/hu_rules |  13 ++++++++++---
 phsource/j2/j2e     | Bin 26051 -> 26051 bytes
 src/readclause.cpp  |  12 ++++++++++++
 4 files changed, 38 insertions(+), 19 deletions(-)

diff --git a/dictsource/hu_list b/dictsource/hu_list
index 43193a8a..242bf630 100755
--- a/dictsource/hu_list
+++ b/dictsource/hu_list
@@ -43,7 +43,7 @@ _2M3    ke:tmillia:R2d
 _0M4      billio:
 _1M4      billio:
 _2M4      ke:tbillio:
-_dpt      ||_!'Ege:s
+_dpt      ||Ege:s //Not need the pause before this word
 
 _0Z1	tizEd      // tenths
 _0Z2	sa:zAd     // hundredths
@@ -415,11 +415,11 @@ ny	En^
 /	pER2		$max3
 =	EJEnlY:		$max3
 #	kEt:Y:SkER2Est		$max3
-@	kukAts
-&	e:S
+@	kukAts	$max3
+&	e:S	$max3
 €	EuR2o
 ¶	bEkEzde:S
-§	pAR2AgR2AfuS
+§	s'AkAs
 ~	tildE
 °	fokjEl
 Ft	foR2int
@@ -429,21 +429,21 @@ Ft	foR2int
 _,	vEss2Y:
 _;	pontoSvEss2Y:
 _:	kEt:Y:Spont
-_!	fElkia:lto:
-_?	ke:R2dY:
-_-	kYtY:jEl	$max3
-__	Ala:hu:za:S
+_!	fElkia:lto:jEl
+_?	ke:R2dY:jEl
+_-	m'inus	$max3
+__	Ala:hu:za:S	$max3
 _'	ApostR2o:f
 _–	nAJkYtY:
 _—	kviR2tmi:nus
-_"	ide:zY:
-_(	za:R2o:
-_)	za:R2o:
-_[	s'YglEtES||z'a:R2o:
-_]	s'YglEtES||z'a:R2o:
-_{	k'AptSoS||z'a:R2o:
-_}	k'AptSoS||z'a:R2o:
-_…	ha:R2_mASpont
+_"	ide:zY:jEl	$max3
+_(	b'Alza:R2o:jEl	$max3
+_)	j'obbza:R2o:jEl	$max3
+_[	n^'ito:sYglEtES	$max3
+_]	z'a:R2o:sYglEtES	$max3
+_{	k'EzdY:kAptSoS	$max3
+_}	v'e:gkAptSoS	$max3
+_…	pontpontpont	$max3
 _•	di:spont
 
 
diff --git a/dictsource/hu_rules b/dictsource/hu_rules
index 157c4511..7e5f0eb9 100644
--- a/dictsource/hu_rules
+++ b/dictsource/hu_rules
@@ -165,7 +165,10 @@ _) javítsd  jAvi:tSd
         k        k
         kk       k:
         kk (C    kk
-     _) köz      kYz
+_) köztársaság k'Ysta:R2SASa:g
+_) központ  k'Yspont
+
+_) köz      kYz
 _) község  kYSSe:g
 _) készség  ke:SSe:g
 _) kétség  ke:tS:e:g
@@ -340,7 +343,9 @@ szöve) ts (ég  tS:
 romlo) tts tS:
 á) t (sor t
 á) t (sé t
+ne) t (c t
 ne) t (se t
+ne) t (szo t
 
 lehe) ts (ég  tS:
 tehe) ts (ég  tS:
@@ -395,9 +400,11 @@ bi) z (tons s
 bi) z (tos s
 bron) z (sz z
 e) z  (t s
+ho) z (ta s
+ho) z (tá s
+inté) z (ked s
 költö) z (köd s
-kö) z (pont s
-kö) z (társaság s
+
 lefeje) z (te s
 lefeje) z (ték s
 pén) z (tár s
diff --git a/phsource/j2/j2e b/phsource/j2/j2e
index ac5edc56f08f065cbbb599f08706edef1dbfc844..b079fa608500043fee13f8bc573d851180c25760 100644
GIT binary patch
delta 107
zcmX?nn(^>y#trehj7*yobT^BIUgU^iKE`yPQH0?(Bcos~6BDxmM-Kx72LtaRCNbtJ
zE?LG7E(YH9Olz2ncrP*J@G@{sVJcvL$p4wqjGuwsnQ0Fb0}$xTG&8~2e`H-YXShTw
F0stki8jt`0

delta 107
zcmX?nn(^>y#trehjLe%8bT^BI9^i;zKE`yPQH0?(BZFWK6BDxmM-Kx72LtaRCNbtJ
zE?LG7E(YFpOzW76crP*J@G@{sVVc79kpDBI89xKN6O$7Y0}$xTG%z_c?E$m@$U1D!
IaEVp~02kmI9{>OV

diff --git a/src/readclause.cpp b/src/readclause.cpp
index 8765d724..b8bd7df7 100644
--- a/src/readclause.cpp
+++ b/src/readclause.cpp
@@ -70,6 +70,12 @@ static const char *punct_close = ")]}>;'\"";  // always pitch fall unless follow
 static const char *tone_punct_on = "\0016T";  // add reverberation, lower pitch
 static const char *tone_punct_off = "\001T";
 
+// ignore these characters
+static const unsigned short chars_ignore[] = {
+  0x200c,  // zero width non-joiner
+  0x200d,  // zero width joiner
+  0 };
+
 // punctuations symbols that can end a clause
 static const unsigned short punct_chars[] = {',','.','?','!',':',';',
   0x2013,  // en-dash
@@ -2190,6 +2196,12 @@ f_input = f_in;  // for GetC etc
 				c2 = ' ';
 			}
 
+			if(lookupwchar(chars_ignore,c1))
+			{
+				// ignore this character (eg. zero-width-non-joiner U+200C)
+				continue;
+			}
+
 			if(iswspace(c1))
 			{
 				char *p_word;