From 5cb461a629973380fc58050bee2d8d893db7923d Mon Sep 17 00:00:00 2001
From: Hiltjo Posthuma <hiltjo@codemadness.org>
Date: Fri, 3 Jan 2020 16:37:46 +0100
Subject: some fixes and improvements

- print_no_cr: fix for modification of const buffer.
- improve buffering of data handlers using the start and end handlers.
- add XML entity conversion.
- some white-space fixes.
---
 xml2tsv.c | 47 ++++++++++++++++++++++++++++++-----------------
 1 file changed, 30 insertions(+), 17 deletions(-)

diff --git a/xml2tsv.c b/xml2tsv.c
index 5725531..ddfa22a 100644
--- a/xml2tsv.c
+++ b/xml2tsv.c
@@ -36,13 +36,13 @@ int stack_push(tstack_t *t, const char *c){
 		return 0;
 	}
 	return -1;
-}	
+}
 
 char* stack_pop(tstack_t *t){
 	if (t->top >= 0)
 		return t->st[t->top--];
 	return NULL;
-} 
+}
 
 char* stack_peek(tstack_t *t){
 	if (t->top >= 0)
@@ -61,17 +61,15 @@ void stack_init(tstack_t *t){
 
 /* utility functions */
 
-void print_no_cr(FILE *f, const char *c){
-	char *tmp = c;
-	while (c != NULL){
-		tmp = strchr(c, '\n');
-		if (tmp != NULL)
-			*tmp = '\0';
-		fprintf(f, "%s", c);
-		if (tmp != NULL)
-			c = tmp + 1;
-		else
-			c = NULL;
+void print_no_cr(FILE *f, const char *s){
+	const char *tmp = s;
+	size_t len;
+	while (*tmp != '\0'){
+		len = strcspn(tmp, "\n");
+		fwrite(tmp, 1, len, f);
+		tmp += len;
+		if (tmp[len] == '\n')
+			tmp++;
 	}
 }
 
@@ -93,14 +91,20 @@ void
 xmlattr(XMLParser *x, const char *t, size_t tl, const char *a, size_t al,
         const char *v, size_t vl)
 {
-	printf("\t%s=%s", a, v);
+	printf("%s", v);
 }
 
 void
 xmlattrentity(XMLParser *x, const char *t, size_t tl, const char *a, size_t al,
               const char *v, size_t vl)
 {
-	printf("attrentity: %s\n", a);
+	char buf[16];
+	int n;
+
+	if ((n = xml_entitytostr(v, buf, sizeof(buf))) > 0)
+		xmlattr(x, t, tl, a, al, buf, (size_t)n);
+	else
+		xmlattr(x, t, tl, a, al, v, vl);
 }
 
 void
@@ -111,17 +115,19 @@ xmlattrend(XMLParser *x, const char *t, size_t tl, const char *a, size_t al)
 void
 xmlattrstart(XMLParser *x, const char *t, size_t tl, const char *a, size_t al)
 {
+	printf("\t%s=", a);
 }
 
 void
 xmlcdatastart(XMLParser *x)
 {
+	printf("\t");
 }
 
 void
 xmlcdata(XMLParser *x, const char *d, size_t dl)
 {
-	printf("\t%s", d);
+	print_no_cr(stdout, d);
 }
 
 void
@@ -147,7 +153,6 @@ xmlcommentend(XMLParser *x)
 void
 xmldata(XMLParser *x, const char *d, size_t dl)
 {
-	printf("\t");
 	print_no_cr(stdout, d);
 }
 
@@ -159,11 +164,19 @@ xmldataend(XMLParser *x)
 void
 xmldataentity(XMLParser *x, const char *d, size_t dl)
 {
+	char buf[16];
+	int n;
+
+	if ((n = xml_entitytostr(d, buf, sizeof(buf))) > 0)
+		xmldata(x, buf, (size_t)n);
+	else
+		xmldata(x, d, dl);
 }
 
 void
 xmldatastart(XMLParser *x)
 {
+	printf("\t");
 }
 
 void
-- 
cgit v1.2.3