Parse quoted values

3 files changed, 130 insertions, 14 deletions

diff --git a/examples/conf.pom b/examples/conf.pom
index cc8404f..ba10cca 100644
--- a/examples/conf.pom
+++ b/examples/conf.pom
@@ -11,6 +11,7 @@ Hello = 5
 [number]
 	one = 1
 	two = 2
-	three = 3
+	three = "é
+	yippee"
 []
 thing=yup
diff --git a/pom.c b/pom.c
index 9cf4f98..abeff17 100644
--- a/pom.c
+++ b/pom.c
@@ -341,6 +341,13 @@ parser_append_(struct parser *parser, void *ptr, size_t elem_size, size_t *pcoun
 #define parser_append_one(parser, field) \
 	parser_append(parser, field, 1)
 
+// append a character to parser->string_data
+static void
+parser_append_char(struct parser *parser, char c) {
+	char *pc = parser_append_one(parser, string_data);
+	if (pc) *pc = c;
+}
+
 static void parser_error(struct parser *parser, PRINTF_FORMAT_STRING const char *fmt, ...) ATTRIBUTE_PRINTF(2, 3);
 static void
 parser_error(struct parser *parser, const char *fmt, ...) {
@@ -571,10 +578,123 @@ check_valid_key(struct parser *parser, const char *key) {
 	}
 }
 
+static int
+parse_hex_digit(char c) {
+	if (c < '0') return -1;
+	if (c <= '9') return c - '0';
+	c &= 0xdf;
+	if (c < 'A') return -1;
+	if (c <= 'F') return c - 'A' + 10;
+	return -1;
+}
+
+// parse escape sequence in *p_str, advancing *p_str past it.
+static void
+parse_escape_sequence(struct parser *parser, const char **p_str) {
+	const char *str = *p_str;
+	switch (*str++) {
+	invalid_sequence: {
+		int len = (int)(str - *p_str);
+		parser_error(parser, "Invalid escape sequence: \\%.*s", len, *p_str);
+		return;
+		} break;
+	case 'n':
+		parser_append_char(parser, '\n');
+		break;
+	case 't':
+		parser_append_char(parser, '\t');
+		break;
+	case 'r':
+		parser_append_char(parser, '\r');
+		break;
+	case '\\':
+		parser_append_char(parser, '\\');
+		break;
+	case '"':
+		parser_append_char(parser, '"');
+		break;
+	case '\'':
+		parser_append_char(parser, '\'');
+		break;
+	case '`':
+		parser_append_char(parser, '`');
+		break;
+	case ',':
+		parser_append_char(parser, '\\');
+		parser_append_char(parser, ',');
+		break;
+	case 'x': {
+		int dig1 = parse_hex_digit(*str++);
+		if (dig1 < 0) goto invalid_sequence;
+		int dig2 = parse_hex_digit(*str++);
+		if (dig2 < 0) goto invalid_sequence;
+		int value = dig1 << 4 | dig2;
+		if (value == 0 || value > 0x7f) goto invalid_sequence;
+		parser_append_char(parser, value);
+		} break;
+	case 'u': {
+		if (*str++ != '{') goto invalid_sequence;
+		uint_fast32_t value = 0;
+		char c;
+		while ((c = *str++) != '}') {
+			int digit = parse_hex_digit(c);
+			if (digit < 0) goto invalid_sequence;
+			value <<= 4;
+			value |= digit;
+			if (value > 0x10ffff) goto invalid_sequence;
+		}
+		if (value >= 0xd800 && value <= 0xdfff)
+			goto invalid_sequence; // utf-16 surrogate
+		if (value < 0x80) {
+			// ASCII
+			parser_append_char(parser, value);
+		} else if (value < 0x800) {
+			// two-byte sequence
+			parser_append_char(parser, 0xc0 | value >> 6);
+			parser_append_char(parser, 0x80 | (value & 63));
+		} else if (value < 0x10000) {
+			// three-byte sequence
+			parser_append_char(parser, 0xe0 | value >> 12);
+			parser_append_char(parser, 0x80 | ((value >> 6) & 63));
+			parser_append_char(parser, 0x80 | (value & 63));
+		} else {
+			// four-byte sequence
+			parser_append_char(parser, 0xf0 | value >> 18);
+			parser_append_char(parser, 0x80 | ((value >> 12) & 63));
+			parser_append_char(parser, 0x80 | ((value >> 6) & 63));
+			parser_append_char(parser, 0x80 | (value & 63));
+		}
+		} break;
+	default:
+		goto invalid_sequence;
+	}
+	*p_str = str;
+}
+
 static void
 parse_quoted_value(struct parser *parser, const char *first_line) {
-	// TODO
-	abort();
+	const char *line = first_line;
+	char delimiter = *line++;
+	assert(delimiter == '"' || delimiter == '`');
+	while (!parser->eof && !parser->out_of_memory) {
+		char c;
+		while ((c = *line++)) {
+			if (c == delimiter)
+				goto finish;
+			if (c == '\\') {
+				parse_escape_sequence(parser, &line);
+			} else {
+				parser_append_char(parser, c);
+			}
+		}
+		parser_read_line(parser);
+		char *newline = parser_append_one(parser, string_data);
+		if (!newline) return;
+		*newline = '\n';
+		line = parser->line.array;
+	}
+finish:;
+	parser_append_char(parser, 0);
 }
 
 static void
@@ -647,10 +767,12 @@ parse_line(struct parser *parser) {
 		char *value = &line[value_start_idx];
 		strip_trailing_accepted_spaces(value);
 		size_t value_sz = strlen(value) + 1;
-		memcpy(parser_append(parser, string_data, value_sz),
-			value, value_sz);
+		char *value_out = parser_append(parser, string_data, value_sz);
+		if (!value_out) return;
+		memcpy(value_out, value, value_sz);
 	}
 	struct parser_item *item = parser_append_one(parser, items);
+	if (!item) return;
 	item->key = key_idx;
 	item->value = value_idx;
 	item->line = parser->line_number;
diff --git a/pre-commit.sh b/pre-commit.sh
index 6841365..85253dd 100755
--- a/pre-commit.sh
+++ b/pre-commit.sh
@@ -1,11 +1,4 @@
 #!/bin/sh
-if sed --version | grep -q 'GNU sed'; then
-	for file in pom.c pom.h; do
-		# Remove trailing white space
-		# (But only if file actually has trailing white space
-		#  we don't want to mess up last-modified-times otherwise)
-		grep -q '\s\s*$' $file && sed -i 's/\s\s*$//' $file
-	done
-fi
+
+# Ensure no doxygen errors
 which doxygen >/dev/null 2>/dev/null && { doxygen || exit 1; }
-git add -u