boardd: UTF-8 conversion improvement.

Handle in-character ANSI color escapes.
Also add optional extended ANSI color codes to preserve information.


git-svn-id: http://opensvn.csie.org/pttbbs/trunk@5820 63ad8ddf-47c3-0310-b6dd-a9e9d9715204

1 files changed, 108 insertions, 4 deletions

diff --git a/pttbbs/daemon/boardd/boardd.c b/pttbbs/daemon/boardd/boardd.c
index e7933d6e..c372e962 100644
--- a/pttbbs/daemon/boardd/boardd.c
+++ b/pttbbs/daemon/boardd/boardd.c
@@ -167,6 +167,47 @@ answer_key(struct evbuffer *buf, const char *key)
     }
 }
 
+#ifdef EXTENDED_INCHAR_ANSI
+static int
+move_string_end(char **buf)
+{
+    int n = 0;
+    while (**buf != '\0') {
+	(*buf)++;
+	n++;
+    }
+    return n;
+}
+
+// Make extended ANSI control code
+//   fg, bg, bright are the original color code (eg. 30, 42, 1)
+//   provide -1 means no change
+//   all -1 means reset
+static void
+make_ext_ansi_ctrl(char *buf, int size, int fg, int bg, int bright)
+{
+    int sep = 0;
+    strncpy(buf, "\033[", size);
+    size -= move_string_end(&buf);
+    if (bright >= 0) {
+	snprintf(buf, size, "%s11%d", sep ? ";" : "", bright);
+	size -= move_string_end(&buf);
+	sep = 1;
+    }
+    if (fg >= 0) {
+	snprintf(buf, size, "%s1%d", sep ? ";" : "", fg);
+	size -= move_string_end(&buf);
+	sep = 1;
+    }
+    if (bg >= 0) {
+	snprintf(buf, size, "%s1%d", sep ? ";" : "", bg);
+	size -= move_string_end(&buf);
+	sep = 1;
+    }
+    snprintf(buf, size, "m");
+}
+#endif
+
 #ifdef CONVERT_TO_UTF8
 
 // Converts given evbuffer contents to UTF-8 and returns the new buffer.
@@ -175,7 +216,7 @@ answer_key(struct evbuffer *buf, const char *key)
 struct evbuffer *
 evbuffer_b2u(struct evbuffer *source)
 {
-    unsigned char c[2];
+    unsigned char c[16];
     int out = 0;
 
     if (evbuffer_get_length(source) == 0)
@@ -194,9 +235,72 @@ evbuffer_b2u(struct evbuffer *source)
 	    out++;
 	} else {
 	    // Big5
+	    int todrain = 2;
+
+	    // Handle in-character colors
+	    int fg = -1, bg = -1, bright = -1;
+	    int n = evbuffer_copyout(source, c, sizeof(c));
+	    if (n < 2)
+		break;
+	    while (c[1] == '\033') {
+		c[n - 1] = '\0';
+
+		// At least have \033[m
+		if (n < 4 || c[2] != '[')
+		    break;
+
+		unsigned char *p = c + 3;
+		if (*p == 'm') {
+		    // ANSI reset
+		    fg = 7;
+		    bg = 0;
+		    bright = 0;
+		}
+		while (1) {
+		    int v = (int) strtol((char *)p, (char **)&p, 10);
+		    if (*p != 'm' && *p != ';')
+			break;
+
+		    if (v == 0)
+			bright = 0;
+		    else if (v == 1)
+			bright = 1;
+		    else if (v >= 30 && v <= 37)
+			fg = v;
+		    else if (v >= 40 && v <= 47)
+			bg = v;
+
+		    if (*p == 'm')
+			break;
+		    p++;
+		}
+		if (*p != 'm') {
+		    // Skip malicious or unsupported codes
+		    fg = bg = bright = -1;
+		    break;
+		} else {
+		    evbuffer_drain(source, p - c + 1);
+		    todrain = 1; // We keep a byte on buffer, so fix offset
+		    n = evbuffer_copyout(source, c + 1, sizeof(c) - 1);
+		    if (n < 1)
+			break;
+		    n++;
+		}
+	    }
+#ifdef EXTENDED_INCHAR_ANSI
+	    // Output control codes before the Big5 character
+	    if (fg >= 0 || bg >= 0 || bright >= 0) {
+		char ansicode[24];
+		make_ext_ansi_ctrl(ansicode, sizeof(ansicode), fg, bg, bright);
+		int dlen = evbuffer_add_printf(destination, ansicode, strlen(ansicode));
+		if (dlen < 0)
+		    break;
+		out += dlen;
+	    }
+#endif
 
-	    // Copy both bytes from source buffer
-	    if (evbuffer_copyout(source, c, 2) != 2)
+	    // n may be changed, check again
+	    if (n < 2)
 		break;
 
 	    uint8_t utf8[4];
@@ -207,7 +311,7 @@ evbuffer_b2u(struct evbuffer *source)
 		break;
 
 	    // Remove DBCS character from source buffer
-	    evbuffer_drain(source, 2);
+	    evbuffer_drain(source, todrain);
 	    out += len;
 	}
     }