MORPH
SIGN IN SIGN UP
kanaka / mal UNCLAIMED

mal - Make a Lisp

0 0 0 Assembly
Normal View History Raw
c++11: step 1
2015-03-26 22:25:50 +11:00
#include "MAL.h"
#include "Types.h"
#include <regex>
typedef std::regex Regex;
static const Regex intRegex("^[-+]?\\d+$");
static const Regex closeRegex("[\\)\\]}]");
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
static const Regex whitespaceRegex("[\\s,]+|;.*");
static const Regex tokenRegexes[] = {
Regex("~@"),
Regex("[\\[\\]{}()'`~^@]"),
Regex("\"(?:\\\\.|[^\\\\\"])*\""),
Regex("[^\\s\\[\\]{}('\"`,;)]+"),
};
c++11: step 1
2015-03-26 22:25:50 +11:00
class Tokeniser
{
public:
Tokeniser(const String& input);
String peek() const {
Split ASSERT into ASSERT and MAL_CHECK/MAL_FAIL * ASSERT is to check for internal errors * MAL_CHECK / MAL_FAIL is to check mal code errors
2015-03-28 22:54:26 +11:00
ASSERT(!eof(), "Tokeniser reading past EOF in peek\n");
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
return m_token;
c++11: step 1
2015-03-26 22:25:50 +11:00
}
String next() {
Split ASSERT into ASSERT and MAL_CHECK/MAL_FAIL * ASSERT is to check for internal errors * MAL_CHECK / MAL_FAIL is to check mal code errors
2015-03-28 22:54:26 +11:00
ASSERT(!eof(), "Tokeniser reading past EOF in next\n");
c++11: step 1
2015-03-26 22:25:50 +11:00
String ret = peek();
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
nextToken();
c++11: step 1
2015-03-26 22:25:50 +11:00
return ret;
}
bool eof() const {
return m_iter == m_end;
}
private:
void skipWhitespace();
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
void nextToken();
bool matchRegex(const Regex& regex);
typedef String::const_iterator StringIter;
c++11: step 1
2015-03-26 22:25:50 +11:00
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
String m_token;
StringIter m_iter;
StringIter m_end;
c++11: step 1
2015-03-26 22:25:50 +11:00
};
Tokeniser::Tokeniser(const String& input)
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
: m_iter(input.begin())
, m_end(input.end())
c++11: step 1
2015-03-26 22:25:50 +11:00
{
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
nextToken();
c++11: step 1
2015-03-26 22:25:50 +11:00
}
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
bool Tokeniser::matchRegex(const Regex& regex)
c++11: step 1
2015-03-26 22:25:50 +11:00
{
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
if (eof()) {
return false;
}
std::smatch match;
auto flags = std::regex_constants::match_continuous;
if (!std::regex_search(m_iter, m_end, match, regex, flags)) {
return false;
}
Split ASSERT into ASSERT and MAL_CHECK/MAL_FAIL * ASSERT is to check for internal errors * MAL_CHECK / MAL_FAIL is to check mal code errors
2015-03-28 22:54:26 +11:00
ASSERT(match.size() == 1, "Should only have one submatch, not %lu\n",
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
match.size());
Split ASSERT into ASSERT and MAL_CHECK/MAL_FAIL * ASSERT is to check for internal errors * MAL_CHECK / MAL_FAIL is to check mal code errors
2015-03-28 22:54:26 +11:00
ASSERT(match.position(0) == 0, "Need to match first character\n");
ASSERT(match.length(0) > 0, "Need to match a non-empty string\n");
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
// Don't advance m_iter now, do it after we've consumed the token in
// next(). If we do it now, we hit eof() when there's still one token left.
m_token = match.str(0);
return true;
c++11: step 1
2015-03-26 22:25:50 +11:00
}
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
void Tokeniser::nextToken()
c++11: step 1
2015-03-26 22:25:50 +11:00
{
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
m_iter += m_token.size();
c++11: step 1
2015-03-26 22:25:50 +11:00
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
skipWhitespace();
if (eof()) {
c++11: step 1
2015-03-26 22:25:50 +11:00
return;
}
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
for (auto &it : tokenRegexes) {
if (matchRegex(it)) {
return;
}
}
String mismatch(m_iter, m_end);
if (mismatch[0] == '"') {
Test uncaught throw, catchless try* . Fix 46 impls. Fixes made to: ada, c, chuck, clojure, coffee, common-lisp, cpp, crystal, d, dart, elm, erlang, es6, factor, fsharp, gnu-smalltalk, groovy, guile, haxe, hy, js, livescript, matlab, miniMAL, nasm, nim, objc, objpascal, ocaml, perl, perl6, php, plsql, ps, python, r, rpython, ruby, scheme, swift3, tcl, ts, vb, vimscript, wasm, yorick. Catchless try* test is an optional test. Not all implementations support catchless try* but a number were fixed so they at least don't crash on catchless try*.
2018-12-03 13:20:44 -06:00
MAL_CHECK(false, "expected '\"', got EOF");
c++11: step 1
2015-03-26 22:25:50 +11:00
}
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
else {
Test uncaught throw, catchless try* . Fix 46 impls. Fixes made to: ada, c, chuck, clojure, coffee, common-lisp, cpp, crystal, d, dart, elm, erlang, es6, factor, fsharp, gnu-smalltalk, groovy, guile, haxe, hy, js, livescript, matlab, miniMAL, nasm, nim, objc, objpascal, ocaml, perl, perl6, php, plsql, ps, python, r, rpython, ruby, scheme, swift3, tcl, ts, vb, vimscript, wasm, yorick. Catchless try* test is an optional test. Not all implementations support catchless try* but a number were fixed so they at least don't crash on catchless try*.
2018-12-03 13:20:44 -06:00
MAL_CHECK(false, "unexpected '%s'", mismatch.c_str());
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
}
c++11: step 1
2015-03-26 22:25:50 +11:00
}
void Tokeniser::skipWhitespace()
{
Fix Reader issues on Linux * use regex_search rather than regex_iterator * use multiple individual whitespace and token regexes, rather than one giant regex This appears to work on 14.10 and 15.04
2015-03-28 11:47:00 +11:00
while (matchRegex(whitespaceRegex)) {
m_iter += m_token.size();
c++11: step 1
2015-03-26 22:25:50 +11:00
}
}
static malValuePtr readAtom(Tokeniser& tokeniser);
static malValuePtr readForm(Tokeniser& tokeniser);
static void readList(Tokeniser& tokeniser, malValueVec* items,
const String& end);
static malValuePtr processMacro(Tokeniser& tokeniser, const String& symbol);
malValuePtr readStr(const String& input)
{
Tokeniser tokeniser(input);
if (tokeniser.eof()) {
throw malEmptyInputException();
}
return readForm(tokeniser);
}
static malValuePtr readForm(Tokeniser& tokeniser)
{
Test uncaught throw, catchless try* . Fix 46 impls. Fixes made to: ada, c, chuck, clojure, coffee, common-lisp, cpp, crystal, d, dart, elm, erlang, es6, factor, fsharp, gnu-smalltalk, groovy, guile, haxe, hy, js, livescript, matlab, miniMAL, nasm, nim, objc, objpascal, ocaml, perl, perl6, php, plsql, ps, python, r, rpython, ruby, scheme, swift3, tcl, ts, vb, vimscript, wasm, yorick. Catchless try* test is an optional test. Not all implementations support catchless try* but a number were fixed so they at least don't crash on catchless try*.
2018-12-03 13:20:44 -06:00
MAL_CHECK(!tokeniser.eof(), "expected form, got EOF");
c++11: step 1
2015-03-26 22:25:50 +11:00
String token = tokeniser.peek();
Split ASSERT into ASSERT and MAL_CHECK/MAL_FAIL * ASSERT is to check for internal errors * MAL_CHECK / MAL_FAIL is to check mal code errors
2015-03-28 22:54:26 +11:00
MAL_CHECK(!std::regex_match(token, closeRegex),
Test uncaught throw, catchless try* . Fix 46 impls. Fixes made to: ada, c, chuck, clojure, coffee, common-lisp, cpp, crystal, d, dart, elm, erlang, es6, factor, fsharp, gnu-smalltalk, groovy, guile, haxe, hy, js, livescript, matlab, miniMAL, nasm, nim, objc, objpascal, ocaml, perl, perl6, php, plsql, ps, python, r, rpython, ruby, scheme, swift3, tcl, ts, vb, vimscript, wasm, yorick. Catchless try* test is an optional test. Not all implementations support catchless try* but a number were fixed so they at least don't crash on catchless try*.
2018-12-03 13:20:44 -06:00
"unexpected '%s'", token.c_str());
c++11: step 1
2015-03-26 22:25:50 +11:00
if (token == "(") {
tokeniser.next();
std::unique_ptr<malValueVec> items(new malValueVec);
readList(tokeniser, items.get(), ")");
return mal::list(items.release());
}
if (token == "[") {
tokeniser.next();
std::unique_ptr<malValueVec> items(new malValueVec);
readList(tokeniser, items.get(), "]");
return mal::vector(items.release());
}
if (token == "{") {
tokeniser.next();
Revert { X } -> ( hash-map X ) reader macro The macro breaks the step 2 tests.
2015-03-28 15:08:25 +11:00
malValueVec items;
readList(tokeniser, &items, "}");
return mal::hash(items.begin(), items.end(), false);
c++11: step 1
2015-03-26 22:25:50 +11:00
}
return readAtom(tokeniser);
}
static malValuePtr readAtom(Tokeniser& tokeniser)
{
struct ReaderMacro {
const char* token;
const char* symbol;
};
ReaderMacro macroTable[] = {
{ "@", "deref" },
{ "`", "quasiquote" },
{ "'", "quote" },
{ "~@", "splice-unquote" },
{ "~", "unquote" },
};
struct Constant {
const char* token;
malValuePtr value;
};
Replace table iterations with new for-loop syntax Didn't even know this existed!
2015-03-28 12:22:01 +11:00
Constant constantTable[] = {
c++11: step 1
2015-03-26 22:25:50 +11:00
{ "false", mal::falseValue() },
{ "nil", mal::nilValue() },
{ "true", mal::trueValue() },
};
String token = tokeniser.next();
if (token[0] == '"') {
return mal::string(unescape(token));
}
if (token[0] == ':') {
return mal::keyword(token);
}
if (token == "^") {
malValuePtr meta = readForm(tokeniser);
malValuePtr value = readForm(tokeniser);
// Note that meta and value switch places
return mal::list(mal::symbol("with-meta"), value, meta);
}
Replace table iterations with new for-loop syntax Didn't even know this existed!
2015-03-28 12:22:01 +11:00
for (auto &constant : constantTable) {
if (token == constant.token) {
return constant.value;
c++11: step 1
2015-03-26 22:25:50 +11:00
}
}
Replace table iterations with new for-loop syntax Didn't even know this existed!
2015-03-28 12:22:01 +11:00
for (auto &macro : macroTable) {
if (token == macro.token) {
return processMacro(tokeniser, macro.symbol);
c++11: step 1
2015-03-26 22:25:50 +11:00
}
}
if (std::regex_match(token, intRegex)) {
return mal::integer(token);
}
return mal::symbol(token);
}
static void readList(Tokeniser& tokeniser, malValueVec* items,
const String& end)
{
while (1) {
Test uncaught throw, catchless try* . Fix 46 impls. Fixes made to: ada, c, chuck, clojure, coffee, common-lisp, cpp, crystal, d, dart, elm, erlang, es6, factor, fsharp, gnu-smalltalk, groovy, guile, haxe, hy, js, livescript, matlab, miniMAL, nasm, nim, objc, objpascal, ocaml, perl, perl6, php, plsql, ps, python, r, rpython, ruby, scheme, swift3, tcl, ts, vb, vimscript, wasm, yorick. Catchless try* test is an optional test. Not all implementations support catchless try* but a number were fixed so they at least don't crash on catchless try*.
2018-12-03 13:20:44 -06:00
MAL_CHECK(!tokeniser.eof(), "expected '%s', got EOF", end.c_str());
c++11: step 1
2015-03-26 22:25:50 +11:00
if (tokeniser.peek() == end) {
tokeniser.next();
return;
}
items->push_back(readForm(tokeniser));
}
}
static malValuePtr processMacro(Tokeniser& tokeniser, const String& symbol)
{
return mal::list(mal::symbol(symbol), readForm(tokeniser));
}
Reference in New Issue Copy Permalink