From fde6a756d1655091a1d77cec734329b061728b17 Mon Sep 17 00:00:00 2001
From: Lee Thomason <leethomason@gmail.com>
Date: Sat, 14 Jan 2012 18:08:12 -0800
Subject: [PATCH] another rev of text parsing.

---
 tinyxml2.cpp |  67 +++++++++++++++++++++++++++++++++++++++------------
 tinyxml2.h   |   8 +++---
 tinyxml2.suo | Bin 26624 -> 27648 bytes
 xmltest.cpp  |   7 +++---
 4 files changed, 59 insertions(+), 23 deletions(-)

diff --git a/tinyxml2.cpp b/tinyxml2.cpp
index cfe2969..ae6a747 100644
--- a/tinyxml2.cpp
+++ b/tinyxml2.cpp
@@ -7,6 +7,12 @@
 
 using namespace tinyxml2;
 
+static const char LINE_FEED				= (char)0x0a;				// all line endings are normalized to LF
+static const char LF = LINE_FEED;
+static const char CARRIAGE_RETURN		= (char)0x0d;			// CR gets filtered out
+static const char CR = CARRIAGE_RETURN;
+
+
 // --------- CharBuffer ----------- //
 /*static*/ CharBuffer* CharBuffer::Construct( const char* in )
 {
@@ -91,25 +97,54 @@ const char* XMLNode::ParseText( char* p, const char* endTag, char** next )
 {
 	TIXMLASSERT( endTag && *endTag );
 
-	char* start = SkipWhiteSpace( p );
-	if ( !start )
-		return 0;
-
-	char endChar = *endTag;
-	p = start;
-	int length = strlen( endTag );
+	char* start = p;
+	char* q = p;		// q (target) <= p (src) in same buffer.
+	char  endChar = *endTag;
+	int   length = strlen( endTag );	
+	char* nextTag = 0;
 
+	// Inner loop of text parsing.
 	while ( *p ) {
-		if ( *p == endChar ) {
-			if ( strncmp( p, endTag, length ) == 0 ) {
-				*p = 0;
-				*next = p + length;
-				return start;
-			}
+		if ( *p == endChar && strncmp( p, endTag, length ) == 0 ) {
+			*q = 0;
+			nextTag = p + length;
+			break;
 		}
-		++p;
+		else if ( *p == CR ) {
+			// CR-LF pair becomes LF
+			// CR alone becomes LF
+			// LF-CR becomes LF
+			if ( *(p+1) == LF ) {
+				p += 2;
+			}
+			else {
+				++p;
+			}
+			*q = LF;
+		}
+		else if ( *p == LF ) {
+			if ( *(p+1) == CR ) {
+				p += 2;
+			}
+			else {
+				++p;
+			}
+			*q = LF;
+		}
+		else {
+			*q = *p;
+			++p;
+		}
+		++q;
 	}	
-	return 0;
+
+	// Error? If we don't have a text tag, something went wrong. (Although 
+	// what the nextTag points at may be null.)
+	if ( nextTag == 0 ) {
+		return 0;
+	}
+	*next = nextTag;
+	return start;
 }
 
 
@@ -129,7 +164,7 @@ XMLComment::~XMLComment()
 void XMLComment::Print( FILE* fp, int depth )
 {
 	XMLNode::Print( fp, depth );
-	fprintf( fp, "<!-- %s -->\n", value );
+	fprintf( fp, "<!--%s-->\n", value );
 }
 
 
diff --git a/tinyxml2.h b/tinyxml2.h
index 2ff58c9..23f21a5 100644
--- a/tinyxml2.h
+++ b/tinyxml2.h
@@ -68,13 +68,13 @@ protected:
 		}
 		return false;
 	}
+	inline static int IsUTF8Continuation( char p ) { return p & 0x80; }
 
 	/* Parses text. (Not a text node.)
 	   - [ ] EOL normalization.
-	   - [x] Trim leading whitespace
-	   - [ ] Trim trailing whitespace.
-	   - [ ] Leaves inner whitespace
-	   - [ ] Inserts one space between lines.
+	   - [X] Do not trim leading whitespace
+	   - [X] Do not trim trailing whitespace.
+	   - [X] Leaves inner whitespace
 	*/
 	const char* ParseText( char* in, const char* endTag, char** next );
 
diff --git a/tinyxml2.suo b/tinyxml2.suo
index 4fe777ea33304723a23a51ad6c297c7597c14a77..032681299f6592cb8b79076996fc2c691789f742 100644
GIT binary patch
delta 630
zcmY*UO-~b16n%I4rk&F1lv?_sC0ei>A<<eVDba;BnUF|pV-upaC~51Yd?dnFVAF*x
zj9pmniWoOtvmo;}8WX|9B^od;`~k*4;JDBQ^-f^|Omf~k_ndR@yVKf-)(_a8mVzn9
z*kP~NOVE8D>r&D`NNj{KitAD?dV|<?LYANtoa~`Eklujyy7c?|m;F8PkleZ;R{1aB
zj*X`G{t<cSH4{^A6FCLq#p4=vU%E4G_~)r!aI{W&#!@~@YKn>gf$}{(;AvqGaFDJt
z@h_0!z{L3Arbl&5XHhqVcI2xMWX3$~6^5HRD{6!6EW_|`te}RDqi)=LnFODVtMMt7
zn)+j05AFC1SQIOX_Y%&}O^C&07VvRZ!3<Z>s1AwU;W0qNj9C|-0~E!|r8R)K_>&5R
zy85|phU4}X4QnM$3{U<H;qJpIyrh*Xc&n1d$b+^9j7jvS^DX3^Z5_)`6URZurAiIE
zk7hAn;iTM&Jvj>Y&ow%!nHeYZmd#lB?x{%G;ixv3H>yT|;bK+Cl9>yh%CIo*G$+Nt
z%tx;hC*A#VVb|MW>3I%q{M5L;QA>l4BRq-unl8^#(897(z)s^D)|B#@JHvzc?OrJy
g+Y4_FZKc-7KHS`SwzM@niuI=veB01Oexe2c0f9u*jQ{`u

delta 1024
zcmb`F-%pZJ7{|}SmtO(`ihxN8Mw_OzmmdrBSE0a#rfcN@u@VC{8)eg0Db!f2i&krc
z=caB7yy>FWoNt@8wZ@9A+FB#8y6D2T)?YvuwI)4B#r}bwv(Ni{&+|OzyzhD5llnqa
zMY=p9QW*##8|8A@h`t#nM2&JYDhr~4TOzld<?J|fKu$WIj(p8<@tWoJC!trxkF*No
z`37N;`$4$@)DV*#<0hGSA&D7BQLxpFLDWZ)086^0)}A1W5W<pS^kc|4atb+(uy_Rh
z1abz6A~A%;r?x&txgslwpuq7$shXprdsO(X;5kZ#HqC=7r5KAf2-cV3GiQZT-6v^M
zLI_80LoVQ6sV`CSNi4|1Fa01C%)DeT&TdFx(J-J?pk$r-x7u)%t_ussmm;?d;|by+
z4&o<1;wCP9+KG+yZE7bfI}R`5QF%6V9#p(A)99p{Z<jWf`d5qEXS+!DgSz|{ESwJV
zGA_+?!h2IQg)0I5%ATu~#)O5Y1f>naPqRY3)W<FHwQzjcB$)i))sPF-K{jxg-ocZE
zAQf_TJ*M7H9!J9C@cY~@pWW8yb~$Z!yO+0le0GnG=RJJ?A>Qxz_dBm8yN71xqyCxr
z<a}=uM&yU3<5M^}5QT@q7HAKuDC-g1S&kpnN+s|v82PX7I5(=-Td|Q_Cd(h(R^G~y
z#F?-co|<$J3b!;=>Ik7I8647qcd5x_f%|PnOU-r(^LnHQVRg(l*I=U!LiN~diA;^5
z%6}C~6L+a^j_fC9ylQh~np{TJiWPWMc8qPXTCDAizbR^;_mQl1VD}nJvUtTf(8}5%
zrBhN0RA+#Qti9@d+>r^k13PP1rH;t*!>C1wcc;nXv&m#43Te$W{28*=b(2|gkxY^c
zq?f2jHCqagcnjo$``BjM6D+cA)9B@p9-Re6(898+N{0CwD1`kGkafZJ`WA3BcEB2E
I&hSV60-nwb6951J

diff --git a/xmltest.cpp b/xmltest.cpp
index 8df6ce0..c3f1b3b 100644
--- a/xmltest.cpp
+++ b/xmltest.cpp
@@ -7,6 +7,7 @@ using namespace tinyxml2;
 
 int main( int argc, const char* argv )
 {
+#if 0
 	{
 		static const char* test = "<!--hello world-->";
 
@@ -14,14 +15,14 @@ int main( int argc, const char* argv )
 		doc.Parse( test );
 		doc.Print( stdout );
 	}
-	/*
+#endif
 	{
-		static const char* test = "<hello></hello>";
+		static const char* test = "<!--hello world\n"
+			                      "          line 2\r-->";
 
 		XMLDocument doc;
 		doc.Parse( test );
 		doc.Print( stdout );
 	}
-	*/
 	return 0;
 }
\ No newline at end of file