Added octal/hex number support.

This commit is contained in:
yhirose 2015-02-22 22:23:08 -05:00
parent 16a61f2cf8
commit 64e7274b1d
3 changed files with 74 additions and 5 deletions

View File

@ -10,6 +10,7 @@ The PEG syntax is well described on page 2 in the [document](http://pdos.csail.m
* `<` ... `>` (Anchor operator) * `<` ... `>` (Anchor operator)
* `$<` ... `>` (Capture operator) * `$<` ... `>` (Capture operator)
* `~` (Ignore operator) * `~` (Ignore operator)
* `\x??` (Hex number char)
How to use How to use
---------- ----------

View File

@ -1119,8 +1119,9 @@ private:
g["Range"] <= cho(seq(g["Char"], chr('-'), g["Char"]), g["Char"]); g["Range"] <= cho(seq(g["Char"], chr('-'), g["Char"]), g["Char"]);
g["Char"] <= cho(seq(chr('\\'), cls("nrt'\"[]\\")), g["Char"] <= cho(seq(chr('\\'), cls("nrt'\"[]\\")),
seq(chr('\\'), cls("0-2"), cls("0-7"), cls("0-7")), // TODO: 0-2 should be 0-3. bug in the spec... seq(chr('\\'), cls("0-3"), cls("0-7"), cls("0-7")),
seq(chr('\\'), cls("0-7"), opt(cls("0-7"))), seq(chr('\\'), cls("0-7"), opt(cls("0-7"))),
seq(lit("\\x"), cls("0-9a-fA-F"), opt(cls("0-9a-fA-F"))),
seq(npd(chr('\\')), dot())); seq(npd(chr('\\')), dot()));
g["LEFTARROW"] <= seq(lit("<-"), g["Spacing"]); g["LEFTARROW"] <= seq(lit("<-"), g["Spacing"]);
@ -1342,6 +1343,58 @@ private:
return cxt.grammar; return cxt.grammar;
} }
bool is_hex(char c, int& v) {
if ('0' <= c && c <= '9') {
v = c - '0';
return true;
} else if ('a' <= c && c <= 'f') {
v = c - 'a' + 10;
return true;
} else if ('A' <= c && c <= 'F') {
v = c - 'A' + 10;
return true;
}
return false;
}
bool is_digit(char c, int& v) {
if ('0' <= c && c <= '9') {
v = c - '0';
return true;
}
return false;
}
std::tuple<char, int> parse_hex_number(const char* s, size_t l, size_t i) {
char ret = 0;
int n;
if (i < l && is_hex(s[i], n)) {
ret = n;
if (i + 1 < l && is_hex(s[i + 1], n)) {
ret = ret * 16 + n;
i++;
}
}
return std::make_tuple(ret, i);
}
std::tuple<char, int> parse_octal_number(const char* s, size_t l, size_t i) {
char ret = 0;
int n;
if (i < l && is_digit(s[i], n)) {
ret = n;
if (i + 1 < l && is_digit(s[i + 1], n)) {
ret = ret * 8 + n;
i++;
if (i + 1 < l && is_digit(s[i + 1], n)) {
ret = ret * 8 + n;
i++;
}
}
}
return std::make_tuple(ret, i);
}
std::string resolve_escape_sequence(const char* s, size_t l) { std::string resolve_escape_sequence(const char* s, size_t l) {
std::string r; std::string r;
r.reserve(l); r.reserve(l);
@ -1359,9 +1412,13 @@ private:
case '[': r += '['; break; case '[': r += '['; break;
case ']': r += ']'; break; case ']': r += ']'; break;
case '\\': r += '\\'; break; case '\\': r += '\\'; break;
case 'x': {
std::tie(ch, i) = parse_hex_number(s, l, i + 1);
r += ch;
break;
}
default: { default: {
// TODO: Octal number support std::tie(ch, i) = parse_octal_number(s, l, i + 1);
assert(false);
break; break;
} }
} }

View File

@ -183,6 +183,17 @@ TEST_CASE("Backtracking test", "[general]")
REQUIRE(count == 2); REQUIRE(count == 2);
} }
TEST_CASE("Octal/Hex value test", "[general]")
{
peglib::peg parser(
" ROOT <- '\132\x7a' "
);
auto ret = parser.parse("Zz");
REQUIRE(ret == true);
}
TEST_CASE("mutable lambda test", "[general]") TEST_CASE("mutable lambda test", "[general]")
{ {
vector<string> vec; vector<string> vec;
@ -539,8 +550,8 @@ TEST_CASE("PEG Char", "[peg]")
REQUIRE(exact(g, "Char", "\\]") == true); REQUIRE(exact(g, "Char", "\\]") == true);
REQUIRE(exact(g, "Char", "\\\\") == true); REQUIRE(exact(g, "Char", "\\\\") == true);
REQUIRE(exact(g, "Char", "\\000") == true); REQUIRE(exact(g, "Char", "\\000") == true);
REQUIRE(exact(g, "Char", "\\277") == true); REQUIRE(exact(g, "Char", "\\377") == true);
REQUIRE(exact(g, "Char", "\\377") == false); REQUIRE(exact(g, "Char", "\\477") == false);
REQUIRE(exact(g, "Char", "\\087") == false); REQUIRE(exact(g, "Char", "\\087") == false);
REQUIRE(exact(g, "Char", "\\079") == false); REQUIRE(exact(g, "Char", "\\079") == false);
REQUIRE(exact(g, "Char", "\\00") == true); REQUIRE(exact(g, "Char", "\\00") == true);