Use the stack buffer for attribute names

This makes the yxml_t struct a bit smaller and removes the hardcoded
limit on the length of an attribute name.

Reduces the compiled size by 200 bytes for some reason I don't
understand (better code re-use?). Doesn't seem to affect the
performance.

4 files changed, 63 insertions, 40 deletions

diff --git a/yxml-states b/yxml-states
index 0c3bc27..b284546 100644
--- a/yxml-states
+++ b/yxml-states
@@ -149,6 +149,6 @@ elem3      '>' selfclose misc2
 attr0      Name attrname attr0; SP attrnameend attr1; '=' attrnameend attr2
 attr1      SP attr1; '=' attr2
 attr2      SP attr2; '\''|'"' $quote attr3
-attr3      AttValue setattrval attr3; '&' refstart attr4; $quote elem2
+attr3      AttValue setattrval attr3; '&' refstart attr4; $quote attrvalend elem2
 attr4      Ref ref attr4; '\x3b' refend attr3
 
diff --git a/yxml.c b/yxml.c
index cf3726c..fc5283c 100644
--- a/yxml.c
+++ b/yxml.c
@@ -131,11 +131,11 @@ static inline int yxml_retmisc(yxml_t *x, unsigned ch) {
 }
 
 
-static int yxml_elemstart(yxml_t *x, unsigned ch) {
+static int yxml_pushstack(yxml_t *x, char **res, unsigned ch) {
 	if(x->stacklen+2 >= x->stacksize)
 		return YXML_ESTACK;
 	x->stacklen++;
-	x->elem = (char *)x->stack+x->stacklen;
+	*res = (char *)x->stack+x->stacklen;
 	x->stack[x->stacklen] = ch;
 	x->stacklen++;
 	x->stack[x->stacklen] = 0;
@@ -143,7 +143,7 @@ static int yxml_elemstart(yxml_t *x, unsigned ch) {
 }
 
 
-static inline int yxml_elemname(yxml_t *x, unsigned ch) {
+static int yxml_pushstackc(yxml_t *x, unsigned ch) {
 	if(x->stacklen+1 >= x->stacksize)
 		return YXML_ESTACK;
 	x->stack[x->stacklen] = ch;
@@ -153,6 +153,23 @@ static inline int yxml_elemname(yxml_t *x, unsigned ch) {
 }
 
 
+static void yxml_popstack(yxml_t *x) {
+	do
+		x->stacklen--;
+	while(x->stack[x->stacklen]);
+}
+
+
+static inline int yxml_elemstart(yxml_t *x, unsigned ch) {
+	return yxml_pushstack(x, &x->elem, ch);
+}
+
+
+static inline int yxml_elemname(yxml_t *x, unsigned ch) {
+	return yxml_pushstackc(x, ch);
+}
+
+
 static inline int yxml_elemnameend(yxml_t *x, unsigned ch) {
 	return YXML_OPEN;
 }
@@ -161,10 +178,7 @@ static inline int yxml_elemnameend(yxml_t *x, unsigned ch) {
 /* Also used in yxml_elemcloseend(), since this function just removes the last
  * element from the stack and returns CLOSE and EOD when appropriate. */
 static int yxml_selfclose(yxml_t *x, unsigned ch) {
-	do
-		x->stacklen--;
-	while(x->stack[x->stacklen]);
-
+	yxml_popstack(x);
 	if(x->stacklen) {
 		x->elem = (char *)x->stack+x->stacklen-1;
 		while(*(x->elem-1))
@@ -192,20 +206,12 @@ static inline int yxml_elemcloseend(yxml_t *x, unsigned ch) {
 
 
 static inline int yxml_attrstart(yxml_t *x, unsigned ch) {
-	x->attrlen = 1;
-	x->attr[0] = ch;
-	x->attr[1] = 0;
-	return YXML_OK;
+	return yxml_pushstack(x, &x->attr, ch);
 }
 
 
 static inline int yxml_attrname(yxml_t *x, unsigned ch) {
-	if(x->attrlen >= YXML_MAX_ATTRNAME)
-		return YXML_EATTR;
-	x->attr[x->attrlen] = ch;
-	x->attrlen++;
-	x->attr[x->attrlen] = 0;
-	return YXML_OK;
+	return yxml_pushstackc(x, ch);
 }
 
 
@@ -214,6 +220,12 @@ static inline int yxml_attrnameend(yxml_t *x, unsigned ch) {
 }
 
 
+static inline int yxml_attrvalend(yxml_t *x, unsigned ch) {
+	yxml_popstack(x);
+	return YXML_OK;
+}
+
+
 static inline int yxml_attrsend(yxml_t *x, unsigned ch) {
 	return YXML_EOA;
 }
@@ -352,7 +364,7 @@ yxml_ret_t yxml_parse(yxml_t *x, int _ch) {
 		}
 		if(x->quote == ch) {
 			x->state = YXMLS_elem2;
-			return YXML_OK;
+			return yxml_attrvalend(x, ch);
 		}
 		break;
 	case YXMLS_attr4:
diff --git a/yxml.c.in b/yxml.c.in
index c6c2bd0..45e0468 100644
--- a/yxml.c.in
+++ b/yxml.c.in
@@ -74,11 +74,11 @@ static inline int yxml_retmisc(yxml_t *x, unsigned ch) {
 }
 
 
-static int yxml_elemstart(yxml_t *x, unsigned ch) {
+static int yxml_pushstack(yxml_t *x, char **res, unsigned ch) {
 	if(x->stacklen+2 >= x->stacksize)
 		return YXML_ESTACK;
 	x->stacklen++;
-	x->elem = (char *)x->stack+x->stacklen;
+	*res = (char *)x->stack+x->stacklen;
 	x->stack[x->stacklen] = ch;
 	x->stacklen++;
 	x->stack[x->stacklen] = 0;
@@ -86,7 +86,7 @@ static int yxml_elemstart(yxml_t *x, unsigned ch) {
 }
 
 
-static inline int yxml_elemname(yxml_t *x, unsigned ch) {
+static int yxml_pushstackc(yxml_t *x, unsigned ch) {
 	if(x->stacklen+1 >= x->stacksize)
 		return YXML_ESTACK;
 	x->stack[x->stacklen] = ch;
@@ -96,6 +96,23 @@ static inline int yxml_elemname(yxml_t *x, unsigned ch) {
 }
 
 
+static void yxml_popstack(yxml_t *x) {
+	do
+		x->stacklen--;
+	while(x->stack[x->stacklen]);
+}
+
+
+static inline int yxml_elemstart(yxml_t *x, unsigned ch) {
+	return yxml_pushstack(x, &x->elem, ch);
+}
+
+
+static inline int yxml_elemname(yxml_t *x, unsigned ch) {
+	return yxml_pushstackc(x, ch);
+}
+
+
 static inline int yxml_elemnameend(yxml_t *x, unsigned ch) {
 	return YXML_OPEN;
 }
@@ -104,10 +121,7 @@ static inline int yxml_elemnameend(yxml_t *x, unsigned ch) {
 /* Also used in yxml_elemcloseend(), since this function just removes the last
  * element from the stack and returns CLOSE and EOD when appropriate. */
 static int yxml_selfclose(yxml_t *x, unsigned ch) {
-	do
-		x->stacklen--;
-	while(x->stack[x->stacklen]);
-
+	yxml_popstack(x);
 	if(x->stacklen) {
 		x->elem = (char *)x->stack+x->stacklen-1;
 		while(*(x->elem-1))
@@ -135,20 +149,12 @@ static inline int yxml_elemcloseend(yxml_t *x, unsigned ch) {
 
 
 static inline int yxml_attrstart(yxml_t *x, unsigned ch) {
-	x->attrlen = 1;
-	x->attr[0] = ch;
-	x->attr[1] = 0;
-	return YXML_OK;
+	return yxml_pushstack(x, &x->attr, ch);
 }
 
 
 static inline int yxml_attrname(yxml_t *x, unsigned ch) {
-	if(x->attrlen >= YXML_MAX_ATTRNAME)
-		return YXML_EATTR;
-	x->attr[x->attrlen] = ch;
-	x->attrlen++;
-	x->attr[x->attrlen] = 0;
-	return YXML_OK;
+	return yxml_pushstackc(x, ch);
 }
 
 
@@ -157,6 +163,12 @@ static inline int yxml_attrnameend(yxml_t *x, unsigned ch) {
 }
 
 
+static inline int yxml_attrvalend(yxml_t *x, unsigned ch) {
+	yxml_popstack(x);
+	return YXML_OK;
+}
+
+
 static inline int yxml_attrsend(yxml_t *x, unsigned ch) {
 	return YXML_EOA;
 }
diff --git a/yxml.h b/yxml.h
index f8a4648..6e70298 100644
--- a/yxml.h
+++ b/yxml.h
@@ -23,14 +23,13 @@
 #include <stdint.h>
 #include <stddef.h>
 
-#define YXML_MAX_ATTRNAME 126
 #define YXML_MAX_REF 7 /* Must be >=7 in order for the hack in yxml_refend() to work */
 
 
 typedef enum {
 	YXML_EREF    = -5, /* Invalid character or entity reference (&whatever;) */
 	YXML_ECLOSE  = -4, /* Close tag does not match open tag (<Tag> .. </OtherTag>) */
-	YXML_ESTACK  = -3, /* Stack overflow (too deeply nested tags or too long element name) */
+	YXML_ESTACK  = -3, /* Stack overflow (too deeply nested tags or too long element/attribute name) */
 	YXML_EATTR   = -2, /* Too long attribute name                    */
 	YXML_ESYN    = -1, /* Syntax error (unexpected byte)             */
 	YXML_OK      =  0, /* Character consumed, no new token present   */
@@ -76,7 +75,7 @@ typedef struct {
 
 	/* Currently opened attribute name, zero-length if not in an attribute.
 	 * Changed after YXML_ATTR. */
-	char attr[YXML_MAX_ATTRNAME+1];
+	char *attr;
 
 	/* Line number, byte offset within that line, and total bytes read. These
 	 * values refer to the position _after_ the last byte given to
@@ -88,7 +87,7 @@ typedef struct {
 
 	/* PRIVATE */
 	int state;
-	unsigned char *stack; /* Stack of element names, separated by \0. Also starts with a \0. */
+	unsigned char *stack; /* Stack of element names + attribute name, separated by \0. Also starts with a \0. */
 	size_t stacksize, stacklen;
 	size_t attrlen;
 	unsigned quote;