From df38d2270241927e5436a0bc36ffb99044a8f4fb Mon Sep 17 00:00:00 2001
From: KatolaZ <katolaz@freaknet.org>
Date: Wed, 8 Dec 2021 21:44:11 +0000
Subject: count tag childrend to fix lack of SEP emission on data with leading
 whitespaces

---
 xml2tsv.c | 36 ++++++++++++++++++++++++------------
 1 file changed, 24 insertions(+), 12 deletions(-)

diff --git a/xml2tsv.c b/xml2tsv.c
index 342d900..27984bc 100644
--- a/xml2tsv.c
+++ b/xml2tsv.c
@@ -25,9 +25,16 @@
 
 /* tag stack */
 
+typedef struct {
+	char s[STR_MAX];
+	int ref;
+} taginfo_t;
+
+
+
 typedef struct {
 	int top;
-	char st[DEPTH_MAX][STR_MAX];
+	taginfo_t st[DEPTH_MAX];
 } tstack_t;
 
 int stack_empty(tstack_t *t){
@@ -37,22 +44,27 @@ int stack_empty(tstack_t *t){
 int stack_push(tstack_t *t, const char *c){
 	if (t->top < DEPTH_MAX){
 		t->top ++;
-		strncpy(t->st[t->top], c, STR_MAX - 1);
-		t->st[t->top][STR_MAX - 1] = '\0';
+		strncpy(t->st[t->top].s, c, STR_MAX - 1);
+		t->st[t->top].s[STR_MAX - 1] = '\0';
+		t->st[t->top].ref = 0;
+		if (t->top){
+			/* Increment the ref counter of the parent tag*/
+			t->st[t->top-1].ref += 1;
+		}
 		return 0;
 	}
 	return -1;
 }
 
-char* stack_pop(tstack_t *t){
+taginfo_t* stack_pop(tstack_t *t){
 	if (!stack_empty(t))
-		return t->st[t->top--];
+		return t->st + t->top--;
 	return NULL;
 }
 
-char* stack_peek(tstack_t *t){
+taginfo_t* stack_peek(tstack_t *t){
 	if (!stack_empty(t))
-		return t->st[t->top];
+		return t->st + t->top;
 	return NULL;
 }
 
@@ -102,7 +114,7 @@ void print_cur_str(tstack_t *t){
 	int i;
 	for (i=0; i<=t->top; i++){
 		putchar('/');
-		fputs(t->st[i], stdout);
+		fputs(t->st[i].s, stdout);
 	}
 }
 
@@ -110,7 +122,7 @@ void print_cur_str_fp(FILE *f, tstack_t *t){
 	int i;
 	for (i=0; i<=t->top; i++){
 		fputc('/', f);
-		fputs(t->st[i], f);
+		fputs(t->st[i].s, f);
 	}
 }
 
@@ -164,7 +176,7 @@ xmlcdata(XMLParser *x, const char *d, size_t dl)
 void
 xmldata(XMLParser *x, const char *d, size_t dl)
 {
-	if (strcspn(d, " \t\n") && emitsep){
+	if (stack_peek(&st) || (strcspn(d, " \t\n") && emitsep)){
 		putchar(SEP);
 		emitsep = FALSE;
 	}
@@ -202,7 +214,7 @@ xmltagend(XMLParser *x, const char *t, size_t tl, int isshort)
 	if (stack_empty(&st)){
 		fprintf(stderr, "Error: tag-end '%s' before any open tag\n", t);
 	}
-	tag = stack_pop(&st);
+	tag = stack_pop(&st)->s;
 	if (strcmp(t, tag)){
 		fprintf(stderr, "Error: tag-end '%s' closes tag '%s'\n", t, tag);
 	}
@@ -212,7 +224,7 @@ void
 xmltagstart(XMLParser *x, const char *t, size_t tl)
 {
 	if (stack_push(&st, t)){
-		fprintf(stderr, "Error: stack full. Ignoring tag '%s' (parent tag: '%s')\n", t, stack_peek(&st));
+		fprintf(stderr, "Error: stack full. Ignoring tag '%s' (parent tag: '%s')\n", t, stack_peek(&st)->s);
 		return;
 	}
 	putchar('\n');
-- 
cgit v1.2.3