From 6a3a62a41abc81a82d6cb6c5491d5ac778256150 Mon Sep 17 00:00:00 2001
From: Thomas Voss <mail@thomasvoss.com>
Date: Sun, 5 May 2024 15:35:27 +0200
Subject: Rename freadrune() to u8fgetr() + improvements

---
 lib/mbio/freadrune.c | 49 -------------------------------------------------
 lib/mbio/u8fgetr.c   | 44 ++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 44 insertions(+), 49 deletions(-)
 delete mode 100644 lib/mbio/freadrune.c
 create mode 100644 lib/mbio/u8fgetr.c

(limited to 'lib/mbio')

diff --git a/lib/mbio/freadrune.c b/lib/mbio/freadrune.c
deleted file mode 100644
index 5c24f52..0000000
--- a/lib/mbio/freadrune.c
+++ /dev/null
@@ -1,49 +0,0 @@
-#include <stdio.h>
-#include <string.h>
-
-#include "macros.h"
-#include "mbio.h"
-#include "mbstring.h"
-#include "rune.h"
-
-#define RETURN_INVAL \
-	do { \
-		*ch = RUNE_ERROR; \
-		return 3; \
-	} while (false)
-
-int
-freadrune(rune *ch, FILE *stream)
-{
-	int c, n = 0;
-	char8_t buf[U8_LEN_MAX];
-
-	if ((c = fgetc(stream)) == EOF)
-		goto eof_or_err;
-
-	buf[0] = (char8_t)c;
-	n = u8byte1(c) ? 0 : u8byte2(c) ? 1 : u8byte3(c) ? 2 : u8byte4(c) ? 3 : 4;
-
-	if (n == 0) {
-		*ch = buf[0];
-		return 1;
-	} else if (n == 4)
-		RETURN_INVAL;
-
-	for (int i = 0; i < n; i++) {
-		if ((c = fgetc(stream)) == EOF)
-			goto eof_or_err;
-		if (!u8bytec(c))
-			RETURN_INVAL;
-		buf[i + 1] = c;
-	}
-
-	return u8tor(ch, buf);
-
-eof_or_err:
-	if (ferror(stream))
-		return MBERR;
-	if (n == 0)
-		return MBEOF;
-	RETURN_INVAL;
-}
diff --git a/lib/mbio/u8fgetr.c b/lib/mbio/u8fgetr.c
new file mode 100644
index 0000000..f880a9e
--- /dev/null
+++ b/lib/mbio/u8fgetr.c
@@ -0,0 +1,44 @@
+#include <stdio.h>
+#include <string.h>
+
+#include "macros.h"
+#include "mbio.h"
+#include "mbstring.h"
+#include "rune.h"
+
+/* Width in bytes of U+FFFD REPLACEMENT CHARACTER */
+constexpr int ERR_WIDTH = 3;
+
+int
+u8fgetr(rune *ch, FILE *stream)
+{
+	int c, n = 0;
+	char8_t buf[U8_LEN_MAX];
+
+	if ((c = fgetc(stream)) == EOF)
+		goto eof;
+
+	n = u8byte1(c) ? 0 : u8byte2(c) ? 1 : u8byte3(c) ? 2 : u8byte4(c) ? 3 : -1;
+	if (n == -1)
+		goto err;
+
+	buf[0] = (char8_t)c;
+	for (int i = 0; i < n; i++) {
+		/* If fgetc() returns EOF we must have malformed UTF-8 unless the EOF
+		   was the result of a reading error. */
+		if ((c = fgetc(stream)) == EOF || !u8bytec(c)) {
+			if (ferror(stream))
+				goto eof;
+			goto err;
+		}
+		buf[i + 1] = c;
+	}
+
+	return u8tor(ch, buf);
+eof:
+	*ch = MBEOF;
+	return 0;
+err:
+	*ch = RUNE_ERROR;
+	return ERR_WIDTH;
+}
-- 
cgit v1.2.3