degesch: add a CTCP parser

1 files changed, 154 insertions, 0 deletions

diff --git a/degesch.c b/degesch.c
index 2cc7944..10a82c8 100644
--- a/degesch.c
+++ b/degesch.c
@@ -2239,6 +2239,160 @@ init_readline (void)
 	return 0;
 }
 
+// --- CTCP decoding -----------------------------------------------------------
+
+#define CTCP_M_QUOTE '\020'
+#define CTCP_X_DELIM '\001'
+#define CTCP_X_QUOTE '\\'
+
+struct ctcp_chunk
+{
+	LIST_HEADER (struct ctcp_chunk)
+
+	bool is_extended;                   ///< Is this a tagged extended message?
+	struct str tag;                     ///< The tag, if any
+	struct str text;                    ///< Message contents
+};
+
+static struct ctcp_chunk *
+ctcp_chunk_new (void)
+{
+	struct ctcp_chunk *self = xcalloc (1, sizeof *self);
+	str_init (&self->tag);
+	str_init (&self->text);
+	return self;
+}
+
+static void
+ctcp_chunk_destroy (struct ctcp_chunk *self)
+{
+	str_free (&self->tag);
+	str_free (&self->text);
+	free (self);
+}
+
+// - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
+
+static void
+ctcp_low_level_decode (const char *message, struct str *output)
+{
+	bool escape = false;
+	for (const char *p = message; *p; p++)
+	{
+		if (escape)
+		{
+			switch (*p)
+			{
+			case '0': str_append_c (output, '\0'); break;
+			case 'r': str_append_c (output, '\r'); break;
+			case 'n': str_append_c (output, '\n'); break;
+			default:  str_append_c (output, *p);
+			}
+			escape = false;
+		}
+		else if (*p == CTCP_M_QUOTE)
+			escape = true;
+		else
+			str_append_c (output, *p);
+	}
+}
+
+static void
+ctcp_intra_decode (const char *chunk, size_t len, struct str *output)
+{
+	bool escape = false;
+	for (size_t i = 0; i < len; i++)
+	{
+		char c = chunk[i];
+		if (escape)
+		{
+			if (c == 'a')
+				str_append_c (output, CTCP_X_DELIM);
+			else
+				str_append_c (output, c);
+			escape = false;
+		}
+		else if (c == CTCP_X_QUOTE)
+			escape = true;
+		else
+			str_append_c (output, c);
+	}
+}
+
+static void
+ctcp_parse_tagged (const char *chunk, size_t len, struct ctcp_chunk *output)
+{
+	// We may search for the space before doing the higher level decoding,
+	// as it doesn't concern space characters at all
+	size_t tag_end = len;
+	for (size_t i = 0; i < len; i++)
+		if (chunk[i] == ' ')
+		{
+			tag_end = i;
+			break;
+		}
+
+	output->is_extended = true;
+	ctcp_intra_decode (chunk, tag_end, &output->tag);
+	if (tag_end++ != len)
+		ctcp_intra_decode (chunk + tag_end, len - tag_end, &output->text);
+}
+
+static struct ctcp_chunk *
+ctcp_parse (const char *message)
+{
+	struct str m;
+	str_init (&m);
+	ctcp_low_level_decode (message, &m);
+
+	struct ctcp_chunk *result = NULL, *result_tail = NULL;
+
+	size_t start = 0;
+	bool in_ctcp = false;
+	for (size_t i = 0; i < m.len; i++)
+	{
+		char c = m.str[i];
+		if (c != CTCP_X_DELIM)
+			continue;
+
+		// Remember the current state
+		size_t my_start = start;
+		bool my_is_ctcp = in_ctcp;
+
+		start = i + 1;
+		in_ctcp = !in_ctcp;
+
+		// Skip empty chunks
+		if (my_start == i)
+			continue;
+
+		struct ctcp_chunk *chunk = ctcp_chunk_new ();
+		if (my_is_ctcp)
+			ctcp_parse_tagged (m.str + my_start, i - my_start, chunk);
+		else
+			ctcp_intra_decode (m.str + my_start, i - my_start, &chunk->text);
+		LIST_APPEND_WITH_TAIL (result, result_tail, chunk);
+	}
+
+	// Finish the last text part.  We ignore unended tagged chunks.
+	if (!in_ctcp && start != m.len)
+	{
+		struct ctcp_chunk *chunk = ctcp_chunk_new ();
+		ctcp_intra_decode (m.str + start, m.len - start, &chunk->text);
+		LIST_APPEND_WITH_TAIL (result, result_tail, chunk);
+	}
+
+	str_free (&m);
+	return result;
+}
+
+static void
+ctcp_destroy (struct ctcp_chunk *list)
+{
+	LIST_FOR_EACH (struct ctcp_chunk, iter, list)
+		ctcp_chunk_destroy (iter);
+}
+
 // --- Input handling ----------------------------------------------------------
 
 // TODO: we will need a proper mode parser; to be shared with kike