From 54d93cf2b4bc7bcef460109aeb7be0c77d910d12 Mon Sep 17 00:00:00 2001
From: Juho Hiltunen <jaacoppi@kapsi.fi>
Date: Sun, 23 Aug 2020 18:44:15 +0300
Subject: [PATCH] code cleanup: move ssml reference handling logic to a new
 function ParseSsmlReference()

It's unclear why c2 needs to be set after an entity reference.
---
 src/libespeak-ng/readclause.c | 29 +++--------------------------
 src/libespeak-ng/ssml.c       | 35 +++++++++++++++++++++++++++++++++++
 src/libespeak-ng/ssml.h       |  4 ++++
 3 files changed, 42 insertions(+), 26 deletions(-)

diff --git a/src/libespeak-ng/readclause.c b/src/libespeak-ng/readclause.c
index 9a7cb3a4..b45f4ff2 100644
--- a/src/libespeak-ng/readclause.c
+++ b/src/libespeak-ng/readclause.c
@@ -603,16 +603,6 @@ static void RemoveChar(char *p)
 	memset(p, ' ', utf8_in(&c, p));
 }
 
-static MNEM_TAB xml_char_mnemonics[] = {
-	{ "gt",   '>' },
-	{ "lt",   0xe000 + '<' },   // private usage area, to avoid confusion with XML tag
-	{ "amp",  '&' },
-	{ "quot", '"' },
-	{ "nbsp", ' ' },
-	{ "apos", '\'' },
-	{ NULL,   -1 }
-};
-
 int ReadClause(Translator *tr, char *buf, short *charix, int *charix_top, int n_buf, int *tone_type, char *voice_change)
 {
 	/* Find the end of the current clause.
@@ -641,7 +631,6 @@ int ReadClause(Translator *tr, char *buf, short *charix, int *charix_top, int n_
 	int phoneme_mode = 0;
 	int n_xml_buf;
 	int terminator;
-	int found;
 	bool any_alnum = false;
 	int punct_data = 0;
 	bool is_end_clause;
@@ -727,22 +716,10 @@ int ReadClause(Translator *tr, char *buf, short *charix, int *charix_top, int n_
 				c2 = GetC();
 				sprintf(ungot_string, "%s%c%c", &xml_buf2[0], c1, c2);
 
+				int found = -1;
 				if (c1 == ';') {
-					if (xml_buf2[0] == '#') {
-						// character code number
-						if (xml_buf2[1] == 'x')
-							found = sscanf(&xml_buf2[2], "%x", (unsigned int *)(&c1));
-						else
-							found = sscanf(&xml_buf2[1], "%d", &c1);
-					} else {
-						if ((found = LookupMnem(xml_char_mnemonics, xml_buf2)) != -1) {
-							c1 = found;
-							if (c2 == 0)
-								c2 = ' ';
-						}
-					}
-				} else
-					found = -1;
+					found = ParseSsmlReference(xml_buf2, &c1, &c2);
+				}
 
 				if (found <= 0) {
 					ungot_string_ix = 0;
diff --git a/src/libespeak-ng/ssml.c b/src/libespeak-ng/ssml.c
index de811d84..d4535f46 100644
--- a/src/libespeak-ng/ssml.c
+++ b/src/libespeak-ng/ssml.c
@@ -949,3 +949,38 @@ int ProcessSsmlTag(wchar_t *xml_buf, char *outbuf, int *outix, int n_outbuf, con
 	}
 	return 0;
 }
+
+static MNEM_TAB xml_entity_mnemonics[] = {
+	{ "gt",   '>' },
+	{ "lt",   0xe000 + '<' },   // private usage area, to avoid confusion with XML tag
+	{ "amp",  '&' },
+	{ "quot", '"' },
+	{ "nbsp", ' ' },
+	{ "apos", '\'' },
+	{ NULL,   -1 }
+};
+
+int ParseSsmlReference(char *ref, int *c1, int *c2) {
+	// Check if buffer *ref contains an XML character or entity reference
+	// if found, set *c1 to the replacement char
+	// change *c2 for entity references
+	// returns >= 0 on success
+
+	if (ref[0] == '#') {
+		// character reference
+		if (ref[1] == 'x')
+			return sscanf(&ref[2], "%x", c1);
+		else
+			return sscanf(&ref[1], "%d", c1);
+	} else { 
+		// entity reference
+		int found;
+		if ((found = LookupMnem(xml_entity_mnemonics, ref)) != -1) {
+			*c1 = found;
+			if (*c2 == 0)
+				*c2 = ' ';
+			return found;
+		}
+	}
+	return -1;
+}
diff --git a/src/libespeak-ng/ssml.h b/src/libespeak-ng/ssml.h
index 6c6235e2..bcc8a8aa 100644
--- a/src/libespeak-ng/ssml.h
+++ b/src/libespeak-ng/ssml.h
@@ -80,6 +80,10 @@ int ProcessSsmlTag(wchar_t *xml_buf,
                    int *n_param_stack,
                    int *speech_parameters);
 
+int ParseSsmlReference(char *ref,
+                       int *c1,
+                       int *c2);
+
 #ifdef __cplusplus
 }
 #endif