xC: refactor parsing of IRC formatting

1 files changed, 83 insertions, 83 deletions

diff --git a/xC.c b/xC.c
index e714c65..96d54b5 100644
--- a/xC.c
+++ b/xC.c
@@ -2961,6 +2961,7 @@ attr_printer_apply (struct attr_printer *self,
 
 	attr_printer_reset (self);
 
+	// TEXT_MONOSPACE is unimplemented, for obvious reasons
 	if (text_attrs)
 		attr_printer_tputs (self, tparm (set_attributes,
 			0,   // standout
@@ -3153,8 +3154,6 @@ formatter_add_item (struct formatter *self, struct formatter_item template_)
 	FORMATTER_ADD_ITEM ((self), ATTR, .attribute = ATTR_RESET)
 #define FORMATTER_ADD_TEXT(self, text_) \
 	FORMATTER_ADD_ITEM ((self), TEXT, .text = (text_))
-#define FORMATTER_ADD_SIMPLE(self, attribute_) \
-	FORMATTER_ADD_ITEM ((self), SIMPLE, .attribute = TEXT_ ## attribute_)
 
 // - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
 
@@ -3225,6 +3224,82 @@ irc_parse_mirc_color (const char *s, uint8_t *fg, uint8_t *bg)
 	return s;
 }
 
+// - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
+
+struct irc_char_attrs
+{
+	uint8_t fg, bg;                     ///< {Fore,back}ground colour or 99
+	uint8_t attributes;                 ///< TEXT_* flags, except TEXT_BLINK
+	uint8_t starts_at_boundary;         ///< Possible to split here?
+};
+
+static void
+irc_serialize_char_attrs (const struct irc_char_attrs *attrs, struct str *out)
+{
+	soft_assert (attrs->fg < 100 && attrs->bg < 100);
+
+	if (attrs->fg != 99 || attrs->bg != 99)
+	{
+		str_append_printf (out, "\x03%u", attrs->fg);
+		if (attrs->bg != 99)
+			str_append_printf (out, ",%02u", attrs->bg);
+	}
+	if (attrs->attributes & TEXT_BOLD)        str_append_c (out, '\x02');
+	if (attrs->attributes & TEXT_ITALIC)      str_append_c (out, '\x1d');
+	if (attrs->attributes & TEXT_UNDERLINE)   str_append_c (out, '\x1f');
+	if (attrs->attributes & TEXT_INVERSE)     str_append_c (out, '\x16');
+	if (attrs->attributes & TEXT_CROSSED_OUT) str_append_c (out, '\x1e');
+	if (attrs->attributes & TEXT_MONOSPACE)   str_append_c (out, '\x11');
+}
+
+static int
+irc_parse_attribute (char c)
+{
+	switch (c)
+	{
+	case '\x02' /* ^B */: return TEXT_BOLD;
+	case '\x11' /* ^Q */: return TEXT_MONOSPACE;
+	case '\x16' /* ^V */: return TEXT_INVERSE;
+	case '\x1d' /* ^] */: return TEXT_ITALIC;
+	case '\x1e' /* ^^ */: return TEXT_CROSSED_OUT;
+	case '\x1f' /* ^_ */: return TEXT_UNDERLINE;
+	case '\x0f' /* ^O */: return -1;
+	}
+	return 0;
+}
+
+// The text needs to be NUL-terminated, and a valid UTF-8 string
+static struct irc_char_attrs *
+irc_analyze_text (const char *text, size_t len)
+{
+	struct irc_char_attrs *attrs = xcalloc (len, sizeof *attrs),
+		blank = { .fg = 99, .bg = 99, .starts_at_boundary = true },
+		next = blank, cur = next;
+
+	for (size_t i = 0; i != len; cur = next)
+	{
+		const char *start = text;
+		hard_assert (utf8_decode (&text, len - i) >= 0);
+
+		int attribute = irc_parse_attribute (*start);
+		if (*start == '\x03')
+			text = irc_parse_mirc_color (text, &next.fg, &next.bg);
+		else if (attribute > 0)
+			next.attributes ^= attribute;
+		else if (attribute < 0)
+			next = blank;
+
+		while (start++ != text)
+		{
+			attrs[i++] = cur;
+			cur.starts_at_boundary = false;
+		}
+	}
+	return attrs;
+}
+
+// - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
+
 static const char *
 formatter_parse_mirc_color (struct formatter *self, const char *s)
 {
@@ -3261,24 +3336,15 @@ formatter_parse_mirc (struct formatter *self, const char *s)
 			str_reset (&buf);
 		}
 
-		switch (c)
-		{
-		case '\x02': FORMATTER_ADD_SIMPLE (self, BOLD);        break;
-		case '\x11': /* monospace, N/A */                      break;
-		case '\x1d': FORMATTER_ADD_SIMPLE (self, ITALIC);      break;
-		case '\x1e': FORMATTER_ADD_SIMPLE (self, CROSSED_OUT); break;
-		case '\x1f': FORMATTER_ADD_SIMPLE (self, UNDERLINE);   break;
-		case '\x16': FORMATTER_ADD_SIMPLE (self, INVERSE);     break;
-
-		case '\x03':
+		int attribute = irc_parse_attribute (c);
+		if (c == '\x03')
 			s = formatter_parse_mirc_color (self, s);
-			break;
-		case '\x0f':
+		else if (attribute > 0)
+			FORMATTER_ADD_ITEM (self, SIMPLE, .attribute = attribute);
+		else if (attribute < 0)
 			FORMATTER_ADD_RESET (self);
-			break;
-		default:
+		else
 			str_append_c (&buf, c);
-		}
 	}
 
 	if (buf.len)
@@ -8240,70 +8306,6 @@ irc_process_message (const struct irc_message *msg, struct server *s)
 // This is a rather basic algorithm; something like ICU with proper
 // locale specification would be needed to make it work better.
 
-struct irc_char_attrs
-{
-	uint8_t fg, bg;                     ///< {Fore,back}ground colour or 99
-	uint8_t attributes;                 ///< TEXT_* flags, except TEXT_BLINK
-	uint8_t starts_at_boundary;         ///< Possible to split here?
-};
-
-static void
-irc_serialize_char_attrs (const struct irc_char_attrs *attrs, struct str *out)
-{
-	soft_assert (attrs->fg < 100 && attrs->bg < 100);
-
-	if (attrs->fg != 99 || attrs->bg != 99)
-	{
-		str_append_printf (out, "\x03%u", attrs->fg);
-		if (attrs->bg != 99)
-			str_append_printf (out, ",%02u", attrs->bg);
-	}
-	if (attrs->attributes & TEXT_BOLD)        str_append_c (out, '\x02');
-	if (attrs->attributes & TEXT_ITALIC)      str_append_c (out, '\x1d');
-	if (attrs->attributes & TEXT_UNDERLINE)   str_append_c (out, '\x1f');
-	if (attrs->attributes & TEXT_INVERSE)     str_append_c (out, '\x16');
-	if (attrs->attributes & TEXT_CROSSED_OUT) str_append_c (out, '\x1e');
-	if (attrs->attributes & TEXT_MONOSPACE)   str_append_c (out, '\x11');
-}
-
-// The text needs to be NUL-terminated
-// TODO: try to deduplicate analogous code in formatter_parse_mirc()
-static struct irc_char_attrs *
-irc_analyze_text (const char *text, size_t len)
-{
-	struct irc_char_attrs *attrs = xcalloc (len, sizeof *attrs),
-		blank = { .fg = 99, .bg = 99, .starts_at_boundary = true },
-		next = blank, cur = next;
-
-	for (size_t i = 0; i != len; cur = next)
-	{
-		const char *start = text;
-		hard_assert (utf8_decode (&text, len - i) >= 0);
-		switch (*start)
-		{
-		case '\x02': next.attributes ^= TEXT_BOLD;        break;
-		case '\x11': next.attributes ^= TEXT_MONOSPACE;   break;
-		case '\x1d': next.attributes ^= TEXT_ITALIC;      break;
-		case '\x1e': next.attributes ^= TEXT_CROSSED_OUT; break;
-		case '\x1f': next.attributes ^= TEXT_UNDERLINE;   break;
-		case '\x16': next.attributes ^= TEXT_INVERSE;     break;
-
-		case '\x03':
-			text = irc_parse_mirc_color (text, &next.fg, &next.bg);
-			break;
-		case '\x0f':
-			next = blank;
-		}
-
-		while (start++ != text)
-		{
-			attrs[i++] = cur;
-			cur.starts_at_boundary = false;
-		}
-	}
-	return attrs;
-}
-
 static size_t
 wrap_text_for_single_line (const char *text, struct irc_char_attrs *attrs,
 	size_t text_len, size_t target_len, struct str *output)
@@ -8344,8 +8346,6 @@ wrap_text_for_single_line (const char *text, struct irc_char_attrs *attrs,
 	return eaten;
 }
 
-// - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
-
 // In practice, this should never fail at all, although it's not guaranteed
 static bool
 wrap_message (const char *message,