Language Initial
This commit is contained in:
IlyaShurupov
Ilya Shurupov
parent
73897f11df
commit
b8f2380c60
47 changed files with 831 additions and 14 deletions
525
Language/old/Tokenizer/public/AutomataGraph.h
Normal file
525
Language/old/Tokenizer/public/AutomataGraph.h
Normal file
|
|
@ -0,0 +1,525 @@
|
|||
|
||||
#pragma once
|
||||
|
||||
#include "List.hpp"
|
||||
#include "Buffer.hpp"
|
||||
#include "Buffer2D.hpp"
|
||||
#include "Utils.hpp"
|
||||
#include "Map.hpp"
|
||||
|
||||
#ifdef ENV_OS_WINDOWS
|
||||
#undef min
|
||||
#undef max
|
||||
#endif
|
||||
|
||||
namespace tp {
|
||||
|
||||
extern ModuleManifest gModuleTokenizer;
|
||||
|
||||
template <typename tAlphabetType, typename tStateType, tStateType tNoStateVal, tStateType tFailedStateVal>
|
||||
class TransitionMatrix;
|
||||
|
||||
template <typename tAlphabetType, typename tStateType, tStateType tNoStateVal, tStateType tFailedStateVal>
|
||||
class DFA;
|
||||
|
||||
// Non-Deterministic Finite-State Automata
|
||||
template <typename tAlphabetType, typename tStateType, tStateType tNoStateVal, tStateType tFailedStateVal>
|
||||
class NFA {
|
||||
static_assert(TypeTraits<tAlphabetType>::isIntegral, "tAlphabetType must be enumerable.");
|
||||
|
||||
public:
|
||||
struct Vertex;
|
||||
|
||||
private:
|
||||
struct Edge {
|
||||
Vertex* mVertex = nullptr;
|
||||
bool mConsumesSymbol = false;
|
||||
Range<tAlphabetType> mAcceptingRange;
|
||||
bool mAcceptsAll = false;
|
||||
bool mExclude = false;
|
||||
|
||||
bool isTransition(const tAlphabetType& symbol) {
|
||||
if (symbol == 0) return false;
|
||||
if (!mConsumesSymbol || mAcceptsAll) return true;
|
||||
bool const in_range = (symbol >= mAcceptingRange.mBegin && symbol <= mAcceptingRange.mEnd);
|
||||
return in_range != mExclude;
|
||||
}
|
||||
};
|
||||
|
||||
public:
|
||||
struct Vertex {
|
||||
List<Edge> edges;
|
||||
tStateType termination_state = tNoStateVal;
|
||||
#ifdef ENV_BUILD_DEBUG
|
||||
ualni debug_idx = 0;
|
||||
#endif
|
||||
ualni flag = 0;
|
||||
};
|
||||
|
||||
private:
|
||||
friend DFA<tAlphabetType, tStateType, tNoStateVal, tFailedStateVal>;
|
||||
|
||||
List<Vertex> mVertices;
|
||||
Vertex* mStart = nullptr;
|
||||
|
||||
public:
|
||||
NFA() = default;
|
||||
|
||||
Vertex* addVertex() {
|
||||
auto node = mVertices.newNode();
|
||||
#ifdef ENV_BUILD_DEBUG
|
||||
node->data.debug_idx = mVertices.length() + 1;
|
||||
#endif
|
||||
mVertices.pushBack(node);
|
||||
return &node->data;
|
||||
}
|
||||
|
||||
void addTransition(Vertex* from, Vertex* to, Range<tAlphabetType> range, bool consumes, bool accepts_all, bool exclude) {
|
||||
Edge edge;
|
||||
edge.mVertex = to;
|
||||
edge.mConsumesSymbol = consumes;
|
||||
edge.mAcceptingRange = range;
|
||||
edge.mExclude = exclude;
|
||||
edge.mAcceptsAll = accepts_all;
|
||||
from->edges.pushBack(edge);
|
||||
}
|
||||
|
||||
void setStartVertex(Vertex* start) {
|
||||
mStart = start;
|
||||
}
|
||||
|
||||
[[nodiscard]] Vertex* getStartVertex() const {
|
||||
return mStart;
|
||||
}
|
||||
|
||||
void setVertexState(Vertex* vertex, tStateType state) {
|
||||
vertex->termination_state = state;
|
||||
}
|
||||
|
||||
[[nodiscard]] bool isValid() const {
|
||||
if (!mStart) {
|
||||
return false;
|
||||
}
|
||||
return true;
|
||||
}
|
||||
|
||||
Range<tAlphabetType> getAlphabetRange() const {
|
||||
tAlphabetType start = 0, end = 0;
|
||||
Range<tAlphabetType> all_range(std::numeric_limits<tAlphabetType>::min(), std::numeric_limits<tAlphabetType>::max());
|
||||
bool first = true;
|
||||
|
||||
for (auto vertex : mVertices) {
|
||||
for (auto edge : vertex.data().edges) {
|
||||
if (!edge.data().mConsumesSymbol) continue;
|
||||
auto const& tran_range = edge.data().mAcceptingRange;
|
||||
if (edge.data().mAcceptsAll || edge.data().mExclude) return all_range;
|
||||
if (tran_range.mBegin < start || first) start = tran_range.mBegin;
|
||||
if (tran_range.mEnd > end || first) end = tran_range.mEnd;
|
||||
first = false;
|
||||
}
|
||||
}
|
||||
|
||||
return Range<tAlphabetType>( start, end + 1 );
|
||||
}
|
||||
|
||||
// vertices that are reachable from initial set with no input consumption (E-transitions)
|
||||
// does not include initial set
|
||||
void closure(const List<Vertex*>& set, List<Vertex*>& closure, ualni unique_call_id) {
|
||||
List<Vertex*> marked;
|
||||
marked = set;
|
||||
|
||||
while (marked.length()) {
|
||||
auto first = marked.first()->data;
|
||||
if (first->flag != unique_call_id) {
|
||||
first->flag = unique_call_id;
|
||||
closure.pushBack(first);
|
||||
}
|
||||
for (auto edge : first->edges) {
|
||||
if (!edge.data().mConsumesSymbol && edge.data().mVertex->flag != unique_call_id) {
|
||||
marked.pushBack(edge.data().mVertex);
|
||||
}
|
||||
}
|
||||
marked.popFront();
|
||||
}
|
||||
}
|
||||
|
||||
// vertices that are reachable from initial set with symbol transition
|
||||
void move(const List<Vertex*>& set, List<Vertex*>& reachable, tAlphabetType symbol, ualni unique_call_id) {
|
||||
for (auto vertex : set) {
|
||||
for (auto edge : vertex->edges) {
|
||||
if (!edge.data().mConsumesSymbol) continue;
|
||||
bool transition = edge.data().isTransition(symbol);
|
||||
if (transition && edge.data().mVertex->flag != unique_call_id) {
|
||||
edge.data().mVertex->flag = unique_call_id;
|
||||
reachable.pushBack(edge.data().mVertex);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
};
|
||||
|
||||
// Deterministic Finite-State Automata
|
||||
template <typename tAlphabetType, typename tStateType, tStateType tNoStateVal, tStateType tFailedStateVal>
|
||||
class DFA {
|
||||
static_assert(TypeTraits<tAlphabetType>::isIntegral, "tAlphabetType must be enumerable.");
|
||||
friend TransitionMatrix<tAlphabetType, tStateType, tNoStateVal, tFailedStateVal>;
|
||||
|
||||
struct Vertex {
|
||||
struct Edge {
|
||||
Vertex* vertex = nullptr;
|
||||
tAlphabetType transition_code = nullptr;
|
||||
};
|
||||
|
||||
List<Edge> edges;
|
||||
tStateType termination_state = tNoStateVal;
|
||||
bool marked = false;
|
||||
};
|
||||
|
||||
List<Vertex> mVertices;
|
||||
Vertex* mStart = nullptr;
|
||||
const Vertex* mIter = nullptr;
|
||||
Range<tAlphabetType> mAlphabetRange;
|
||||
bool mTrapState = false;
|
||||
|
||||
typedef typename NFA<tAlphabetType, tStateType, tNoStateVal, tFailedStateVal>::Vertex NState;
|
||||
|
||||
public:
|
||||
|
||||
explicit DFA(NFA<tAlphabetType, tStateType, tNoStateVal, tFailedStateVal>& nfa) {
|
||||
if (!nfa.isValid()) {
|
||||
return;
|
||||
}
|
||||
|
||||
mAlphabetRange = nfa.getAlphabetRange();
|
||||
|
||||
struct DStateKey {
|
||||
const List<NState*>* nStates;
|
||||
bool operator==(const DStateKey& in) const {
|
||||
if (nStates->length() != in.nStates->length()) {
|
||||
return false;
|
||||
}
|
||||
// FIXME : make linear time
|
||||
for (auto state : *nStates) {
|
||||
bool found = false;
|
||||
for (auto in_state : *in.nStates) {
|
||||
if (state.data() == in_state.data()) {
|
||||
found = true;
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
if (!found) {
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
return true;
|
||||
}
|
||||
|
||||
static ualni dStateHashFunc(DStateKey key) {
|
||||
alni out = 0;
|
||||
for (auto state : *key.nStates) {
|
||||
out += alni(state.data());
|
||||
}
|
||||
return out;
|
||||
};
|
||||
};
|
||||
|
||||
// all NFA states that are reachable from initial DFA State for specific symbol in alphabet
|
||||
struct DState {
|
||||
struct DTransition {
|
||||
DState* state = nullptr;
|
||||
tAlphabetType accepting_code;
|
||||
};
|
||||
|
||||
List<NState*> nStates;
|
||||
List<DTransition> transitions;
|
||||
|
||||
Vertex* dVertex= nullptr; // relevant DFA vertex
|
||||
|
||||
#ifdef ENV_BUILD_DEBUG
|
||||
ualni debug_idx = 0;
|
||||
#endif
|
||||
};
|
||||
|
||||
// includes closure of NFA start state by definition
|
||||
auto start_state = new DState();
|
||||
nfa.closure({ nfa.getStartVertex() }, start_state->nStates, ualni(start_state));
|
||||
|
||||
Map<DStateKey, DState*, DefaultAllocator, DStateKey::dStateHashFunc, 256> dStates;
|
||||
|
||||
dStates.put({ &start_state->nStates }, start_state);
|
||||
|
||||
List<DState*> working_set = { start_state };
|
||||
|
||||
// while there is items to work with
|
||||
auto currentDState = working_set.first();
|
||||
while (currentDState) {
|
||||
|
||||
// check all possible transitions for any symbol
|
||||
for (auto symbol : mAlphabetRange) {
|
||||
|
||||
List<NState*> reachableNStates;
|
||||
|
||||
nfa.move(currentDState->data->nStates, reachableNStates, symbol, ualni(currentDState + symbol));
|
||||
nfa.closure(reachableNStates, reachableNStates, ualni(currentDState + symbol));
|
||||
|
||||
if (!reachableNStates.length()) {
|
||||
continue;
|
||||
}
|
||||
|
||||
DState* targetDState = nullptr;
|
||||
|
||||
// check if set of all reachable NFA states already forms existing DFA state
|
||||
auto idx = dStates.presents({ &reachableNStates});
|
||||
if (idx) {
|
||||
targetDState = dStates.getSlotVal(idx);
|
||||
}
|
||||
|
||||
if (!targetDState) {
|
||||
// register new DFA state
|
||||
targetDState = new DState();
|
||||
|
||||
#ifdef ENV_BUILD_DEBUG
|
||||
targetDState->debug_idx = dStates.size();
|
||||
#endif
|
||||
|
||||
targetDState->nStates = reachableNStates;
|
||||
|
||||
// append to working stack
|
||||
working_set.pushBack(targetDState);
|
||||
dStates.put({ &targetDState->nStates }, targetDState);
|
||||
}
|
||||
|
||||
// add transition to DFA state
|
||||
currentDState->data->transitions.pushBack({targetDState, symbol });
|
||||
}
|
||||
|
||||
working_set.popFront();
|
||||
currentDState = working_set.first();
|
||||
}
|
||||
|
||||
// create own vertices
|
||||
for (auto node : dStates) {
|
||||
tStateType state = tNoStateVal;
|
||||
for (auto iter : node->val->nStates) {
|
||||
if (iter->termination_state != tNoStateVal) {
|
||||
state = iter->termination_state;
|
||||
break;
|
||||
}
|
||||
}
|
||||
node->val->dVertex = addVertex(state);
|
||||
}
|
||||
|
||||
// connect all vertices
|
||||
for (auto node : dStates) {
|
||||
for (auto edge : node->val->transitions) {
|
||||
addTransition(node->val->dVertex, edge.data().state->dVertex, edge.data().accepting_code);
|
||||
}
|
||||
}
|
||||
|
||||
// set the starting vertex
|
||||
mStart = start_state->dVertex;
|
||||
|
||||
// cleanup
|
||||
for (auto node : dStates) {
|
||||
delete node->val;
|
||||
}
|
||||
|
||||
collapseEquivalentVertices();
|
||||
mAlphabetRange = getAlphabetRange();
|
||||
}
|
||||
|
||||
tStateType move(tAlphabetType symbol) {
|
||||
if (mTrapState || !mIter) {
|
||||
return tNoStateVal;
|
||||
}
|
||||
|
||||
for (auto edge : mIter->edges) {
|
||||
if (edge.data().transition_code == symbol) {
|
||||
mIter = edge.data().vertex;
|
||||
return mIter->termination_state;
|
||||
}
|
||||
}
|
||||
|
||||
mTrapState = true;
|
||||
return tNoStateVal;
|
||||
}
|
||||
|
||||
void start() {
|
||||
mIter = mStart;
|
||||
mTrapState = false;
|
||||
}
|
||||
|
||||
[[nodiscard]] uhalni nVertices() const {
|
||||
return (uhalni) mVertices.length();
|
||||
}
|
||||
|
||||
[[nodiscard]] Range<tAlphabetType> getRange() const {
|
||||
return mAlphabetRange;
|
||||
}
|
||||
|
||||
private:
|
||||
[[nodiscard]] Range<tAlphabetType> getAlphabetRange() const {
|
||||
Range<tAlphabetType> out;
|
||||
for (auto vertex : mVertices) {
|
||||
vertex.data().marked = false;
|
||||
}
|
||||
|
||||
bool first = true;
|
||||
getAlphabetRangeUtil(mStart, &out, first);
|
||||
out.mEnd++;
|
||||
return out;
|
||||
}
|
||||
|
||||
void getAlphabetRangeUtil(Vertex* vert, Range<tAlphabetType>* out, bool& first) const {
|
||||
vert->marked = true;
|
||||
for (auto edge : vert->edges) {
|
||||
auto const code = edge.data().transition_code;
|
||||
|
||||
if (first) {
|
||||
*out = { code, code };
|
||||
first = false;
|
||||
}
|
||||
|
||||
if (code < out->mBegin) {
|
||||
out->mBegin = code;
|
||||
}
|
||||
if (code > out->mEnd) {
|
||||
out->mEnd = code;
|
||||
}
|
||||
|
||||
if (!edge.data().vertex->marked) {
|
||||
getAlphabetRangeUtil(edge.data().vertex, out, first);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
void collapseEquivalentVertices() {
|
||||
// TODO
|
||||
}
|
||||
|
||||
Vertex* addVertex(tStateType state) {
|
||||
auto node = mVertices.addNodeBack();
|
||||
node->data.termination_state = state;
|
||||
return &node->data;
|
||||
}
|
||||
|
||||
void addTransition(Vertex* from, Vertex* to, tAlphabetType transition_symbol) {
|
||||
from->edges.pushBack({ to, transition_symbol });
|
||||
}
|
||||
};
|
||||
|
||||
template <typename tAlphabetType, typename tStateType, tStateType tNoStateVal, tStateType tFailedStateVal>
|
||||
class TransitionMatrix {
|
||||
|
||||
static_assert(TypeTraits<tAlphabetType>::isIntegral, "tAlphabetType must be enumerable.");
|
||||
|
||||
Buffer2D<ualni> mTransitions;
|
||||
Buffer<tStateType> mStates;
|
||||
Range<tAlphabetType> mSymbolRange = { 0, 0 };
|
||||
|
||||
ualni mIter = 0;
|
||||
ualni mIterPrev = 0;
|
||||
ualni mStart = 0;
|
||||
|
||||
public:
|
||||
|
||||
TransitionMatrix() = default;
|
||||
|
||||
auto getStates() const { return &mStates; }
|
||||
auto getTransitions() const { return &mTransitions; }
|
||||
auto getStart() const { return mStart; }
|
||||
|
||||
void construct(const DFA<tAlphabetType, tStateType, tNoStateVal, tFailedStateVal>& dfa) {
|
||||
mSymbolRange = dfa.getRange();
|
||||
auto range_len = ualni(mSymbolRange.mEnd - mSymbolRange.mBegin);
|
||||
auto sizeX = range_len ? range_len : 1;
|
||||
auto sizeY = (ualni) (dfa.nVertices() + 1);
|
||||
|
||||
mTransitions.reserve({ sizeX, sizeY });
|
||||
mTransitions.assign(dfa.nVertices());
|
||||
mStates.reserve(sizeY);
|
||||
|
||||
ualni idx = 0;
|
||||
for (auto vertex : dfa.mVertices) {
|
||||
auto state = vertex.data().termination_state;
|
||||
mStates[idx] = state;
|
||||
idx++;
|
||||
}
|
||||
|
||||
mStates[dfa.nVertices()] = tFailedStateVal;
|
||||
|
||||
idx = 0;
|
||||
for (auto vertex : dfa.mVertices) {
|
||||
if (&vertex.data() == dfa.mStart) {
|
||||
mStart = mIter = mIterPrev = idx;
|
||||
}
|
||||
idx++;
|
||||
}
|
||||
|
||||
ualni vertexIdx = 0;
|
||||
for (auto vertex : dfa.mVertices) {
|
||||
for (auto edge : vertex.data().edges) {
|
||||
ualni vertex2Idx = 0;
|
||||
for (auto vertex2 : dfa.mVertices) {
|
||||
if (edge.data().vertex == &vertex2.data()) break;
|
||||
vertex2Idx++;
|
||||
}
|
||||
auto const code = edge.data().transition_code;
|
||||
mTransitions.set( { (ualni) (code - mSymbolRange.mBegin), (ualni) vertexIdx }, vertex2Idx);
|
||||
}
|
||||
vertexIdx++;
|
||||
}
|
||||
}
|
||||
|
||||
bool isTrapped() {
|
||||
return mStates[mIter] == tFailedStateVal;
|
||||
}
|
||||
|
||||
tStateType move(tAlphabetType symbol) {
|
||||
if (symbol >= mSymbolRange.mBegin && symbol < mSymbolRange.mEnd) {
|
||||
mIter = mTransitions.get({ (ualni) (symbol - mSymbolRange.mBegin), (ualni) mIter });
|
||||
}
|
||||
else {
|
||||
mIter = mStates.size() - 1;
|
||||
}
|
||||
|
||||
if (mIterPrev == mStart) {
|
||||
if (mStates[mIter] == tFailedStateVal) {
|
||||
reset();
|
||||
return tFailedStateVal;
|
||||
}
|
||||
else {
|
||||
mIterPrev = mIter;
|
||||
return tNoStateVal;
|
||||
}
|
||||
}
|
||||
else {
|
||||
if (mStates[mIter] == tFailedStateVal) {
|
||||
if (mStates[mIterPrev] != tNoStateVal) {
|
||||
auto out = mStates[mIterPrev];
|
||||
reset();
|
||||
return out;
|
||||
}
|
||||
else {
|
||||
reset();
|
||||
return tFailedStateVal;
|
||||
}
|
||||
}
|
||||
else {
|
||||
mIterPrev = mIter;
|
||||
return tNoStateVal;
|
||||
}
|
||||
}
|
||||
|
||||
mIterPrev = mIter;
|
||||
return mStates[mIter];
|
||||
}
|
||||
|
||||
void reset() {
|
||||
mIter = mStart;
|
||||
mIterPrev = mStart;
|
||||
}
|
||||
};
|
||||
}
|
||||
519
Language/old/Tokenizer/public/RegularExpression.h
Normal file
519
Language/old/Tokenizer/public/RegularExpression.h
Normal file
|
|
@ -0,0 +1,519 @@
|
|||
|
||||
#pragma once
|
||||
|
||||
#include "AutomataGraph.h"
|
||||
|
||||
namespace tp {
|
||||
extern ModuleManifest gModuleTokenizer;
|
||||
}
|
||||
|
||||
namespace tp::RegEx {
|
||||
|
||||
struct AstNode {
|
||||
enum Type {
|
||||
NONE,
|
||||
ANY,
|
||||
OR,
|
||||
IF,
|
||||
CLASS,
|
||||
COMPOUND,
|
||||
REPEAT,
|
||||
VAL,
|
||||
} mType = NONE;
|
||||
|
||||
AstNode() = default;
|
||||
virtual ~AstNode() = default;
|
||||
};
|
||||
|
||||
template <typename tAlphabetType>
|
||||
struct AstVal : public AstNode {
|
||||
explicit AstVal(tAlphabetType val) : mVal(val) { mType = VAL; }
|
||||
~AstVal() override = default;
|
||||
tAlphabetType mVal;
|
||||
};
|
||||
|
||||
struct AstCompound : public AstNode {
|
||||
AstCompound() { mType = COMPOUND; }
|
||||
~AstCompound() override {
|
||||
for (auto iter : mChilds) {
|
||||
delete iter.data();
|
||||
}
|
||||
mChilds.removeAll();
|
||||
}
|
||||
List<AstNode*> mChilds;
|
||||
};
|
||||
|
||||
struct AstAlternation : public AstNode {
|
||||
AstAlternation() { mType = OR; }
|
||||
~AstAlternation() override {
|
||||
delete mFirst;
|
||||
delete mSecond;
|
||||
}
|
||||
AstNode* mFirst = nullptr;
|
||||
AstNode* mSecond = nullptr;
|
||||
};
|
||||
|
||||
struct AstIf : public AstNode {
|
||||
AstIf() { mType = IF; }
|
||||
~AstIf() override { delete mNode; }
|
||||
AstNode* mNode = nullptr;
|
||||
};
|
||||
|
||||
struct AstAny : public AstNode {
|
||||
AstAny() { mType = ANY; }
|
||||
~AstAny() override = default;
|
||||
};
|
||||
|
||||
struct AstRepetition : public AstNode {
|
||||
AstRepetition() { mType = REPEAT; }
|
||||
~AstRepetition() override { delete mNode; }
|
||||
AstNode* mNode = nullptr;
|
||||
bool mPlus = false;
|
||||
};
|
||||
|
||||
template <typename tAlphabetType>
|
||||
struct AstClass : public AstNode {
|
||||
AstClass() { mType = CLASS; }
|
||||
~AstClass() override { mRanges.removeAll(); }
|
||||
List<Range<tAlphabetType>> mRanges;
|
||||
bool mExclude = false;
|
||||
};
|
||||
|
||||
struct ParseError {
|
||||
const char* description = nullptr;
|
||||
uhalni offset = 0;
|
||||
[[nodiscard]] bool isError() const { return description != nullptr; }
|
||||
};
|
||||
|
||||
template <typename tAlphabetType, typename tStateType, tStateType tNoStateVal>
|
||||
class Parser {
|
||||
|
||||
enum TokType : uint1 {
|
||||
TOK_COMPOUND_START = 0,
|
||||
TOK_COMPOUND_END,
|
||||
TOK_CLASS_START,
|
||||
TOK_CLASS_END,
|
||||
TOK_CLASS_START_EXCLUDE,
|
||||
TOK_CLASS_END_EXCLUDE,
|
||||
TOK_OR,
|
||||
TOK_IF,
|
||||
TOK_ANY,
|
||||
TOK_REPEAT,
|
||||
TOK_REPEAT_PLUS,
|
||||
TOK_HYPHEN,
|
||||
TOK_SPECIALS_END_,
|
||||
TOK_VAL,
|
||||
TOK_NONE,
|
||||
};
|
||||
|
||||
tAlphabetType SpecialSymbols[TOK_SPECIALS_END_] = {
|
||||
'(', ')', '[', ']', '{', '}', '|', '?', '.', '*', '+', '-',
|
||||
};
|
||||
|
||||
tAlphabetType mEscapeSymbol = '\\';
|
||||
|
||||
struct Token {
|
||||
TokType type;
|
||||
tAlphabetType val;
|
||||
};
|
||||
|
||||
const tAlphabetType* mSource = nullptr;
|
||||
uhalni mOffset = 0;
|
||||
Token mCurToken;
|
||||
uhalni mTokLength = 0;
|
||||
|
||||
public:
|
||||
|
||||
ParseError mError;
|
||||
|
||||
// regular expression must be a zero termination string
|
||||
AstCompound* parse(const tAlphabetType* regex) {
|
||||
mSource = regex;
|
||||
return parseRegEx();
|
||||
}
|
||||
|
||||
private:
|
||||
|
||||
AstCompound* parseRegEx() {
|
||||
auto out = new AstCompound();
|
||||
for (AstNode* node = parseElement(); node; node = parseElement()) {
|
||||
out->mChilds.pushBack(node);
|
||||
}
|
||||
if (!out->mChilds.length()) {
|
||||
genError("Expected A Expression");
|
||||
}
|
||||
if (mError.description) {
|
||||
delete out;
|
||||
return nullptr;
|
||||
}
|
||||
return out;
|
||||
}
|
||||
|
||||
AstNode* parseElement() {
|
||||
AstNode* out = nullptr;
|
||||
switch (readTok().type) {
|
||||
case TOK_COMPOUND_START: out = parseCompound(); break;
|
||||
case TOK_CLASS_START: out = parseClass(); break;
|
||||
case TOK_CLASS_START_EXCLUDE: out = parseClass(true); break;
|
||||
case TOK_ANY: out = parseAny(); break;
|
||||
case TOK_VAL: out = parseVal(); break;
|
||||
case TOK_NONE: { discardTok(); return nullptr; };
|
||||
default: break;
|
||||
}
|
||||
if (!out) {
|
||||
discardTok();
|
||||
return nullptr;
|
||||
}
|
||||
switch (readTok().type) {
|
||||
case TOK_OR: out = parseAlternation(out); break;
|
||||
case TOK_REPEAT: out = parseRepetition(out); break;
|
||||
case TOK_REPEAT_PLUS: out = parseRepetition(out, true); break;
|
||||
case TOK_IF: out = parseIf(out); break;
|
||||
case TOK_NONE: break;
|
||||
default: { discardTok(); }
|
||||
}
|
||||
return out;
|
||||
}
|
||||
|
||||
AstCompound* parseCompound() {
|
||||
auto out = new AstCompound();
|
||||
for (AstNode* node = parseElement(); node; node = parseElement()) {
|
||||
out->mChilds.pushBack(node);
|
||||
}
|
||||
if (readTok().type != TOK_COMPOUND_END) {
|
||||
genError("Expected Compound End");
|
||||
}
|
||||
if (mError.description) {
|
||||
delete out;
|
||||
return nullptr;
|
||||
}
|
||||
return out;
|
||||
}
|
||||
|
||||
AstClass<tAlphabetType>* parseClass(bool exclude = false) {
|
||||
auto out = new AstClass<tAlphabetType>();
|
||||
out->mExclude = exclude;
|
||||
auto& ranges = out->mRanges;
|
||||
|
||||
readTok();
|
||||
|
||||
READ_VAL:
|
||||
if (mCurToken.type != TOK_VAL) {
|
||||
delete out;
|
||||
genError("Expected A Value");
|
||||
return nullptr;
|
||||
}
|
||||
char range_start = mCurToken.val;
|
||||
|
||||
readTok();
|
||||
if (mCurToken.type != TOK_HYPHEN) {
|
||||
delete out;
|
||||
genError("Expected A Range");
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
readTok();
|
||||
if (mCurToken.type != TOK_VAL) {
|
||||
delete out;
|
||||
genError("Expected A Value");
|
||||
return nullptr;
|
||||
}
|
||||
char range_end = mCurToken.val;
|
||||
|
||||
ranges.pushBack({ range_start, range_end });
|
||||
|
||||
readTok();
|
||||
if ((mCurToken.type == TOK_CLASS_END && !exclude) || (mCurToken.type == TOK_CLASS_END_EXCLUDE && exclude)) {
|
||||
return out;
|
||||
}
|
||||
else {
|
||||
goto READ_VAL;
|
||||
}
|
||||
}
|
||||
|
||||
AstAny* parseAny() {
|
||||
return new AstAny();
|
||||
}
|
||||
|
||||
AstVal<tAlphabetType>* parseVal() {
|
||||
auto out = new AstVal<tAlphabetType>(mCurToken.val);
|
||||
return out;
|
||||
}
|
||||
|
||||
AstAlternation* parseAlternation(AstNode* left) {
|
||||
auto right = parseElement();
|
||||
if (!right) {
|
||||
genError("Expected Alternation right Side");
|
||||
delete left;
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
auto out = new AstAlternation();
|
||||
|
||||
out->mFirst = left;
|
||||
out->mSecond = right;
|
||||
|
||||
return out;
|
||||
}
|
||||
|
||||
AstRepetition* parseRepetition(AstNode* left, bool plus = false) {
|
||||
auto out = new AstRepetition();
|
||||
out->mNode = left;
|
||||
out->mPlus = plus;
|
||||
return out;
|
||||
}
|
||||
|
||||
AstIf* parseIf(AstNode* left) {
|
||||
auto out = new AstIf();
|
||||
out->mNode = left;
|
||||
return out;
|
||||
}
|
||||
|
||||
void genError(const char* desc) {
|
||||
mError = { desc, mOffset };
|
||||
}
|
||||
|
||||
Token& readTok() {
|
||||
|
||||
const tAlphabetType* crs = mSource + mOffset;
|
||||
|
||||
// zero termination string
|
||||
if (*crs == 0) {
|
||||
mCurToken.type = TOK_NONE;
|
||||
return mCurToken;
|
||||
}
|
||||
|
||||
mTokLength = 1;
|
||||
mCurToken.type = TOK_VAL;
|
||||
mCurToken.val = crs[0];
|
||||
|
||||
if (crs[0] == mEscapeSymbol) {
|
||||
mCurToken.val = crs[1];
|
||||
mTokLength = 2;
|
||||
}
|
||||
else {
|
||||
for (uhalni tok = 0; tok < TOK_SPECIALS_END_; tok++) {
|
||||
if (SpecialSymbols[tok] == mCurToken.val) {
|
||||
mCurToken.type = TokType(tok);
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
mOffset += mTokLength;
|
||||
return mCurToken;
|
||||
}
|
||||
|
||||
void discardTok() {
|
||||
mOffset -= mTokLength;
|
||||
mTokLength = 0;
|
||||
}
|
||||
};
|
||||
|
||||
template <typename tStateType>
|
||||
struct CompileError {
|
||||
ParseError mParseError;
|
||||
uhalni mRuleIndex = 0;
|
||||
tStateType mRuleState;
|
||||
const char* description = nullptr;
|
||||
[[nodiscard]] bool isError() const { return description; }
|
||||
};
|
||||
|
||||
template <typename tAlphabetType, typename tStateType, tStateType tNoStateVal, tStateType tFailedStateVal>
|
||||
class Compiler {
|
||||
|
||||
typedef NFA<tAlphabetType, tStateType, tNoStateVal, tFailedStateVal> Graph;
|
||||
typedef typename Graph::Vertex Vertex;
|
||||
typedef Parser<tAlphabetType, tStateType, tNoStateVal> Parser;
|
||||
|
||||
struct Node {
|
||||
Vertex* left = nullptr;
|
||||
Vertex* right = nullptr;
|
||||
};
|
||||
|
||||
private:
|
||||
Graph* mGraph = nullptr;
|
||||
|
||||
public:
|
||||
CompileError<tStateType> mError;
|
||||
|
||||
Node compile(Graph& graph, const tAlphabetType* regex, tStateType state) {
|
||||
mGraph = &graph;
|
||||
return compileUtil(regex, state);
|
||||
}
|
||||
|
||||
Node compile(Graph& aGraph, InitialierList<Pair<const tAlphabetType*, tStateType>> aRules) {
|
||||
mGraph = &aGraph;
|
||||
|
||||
auto left = mGraph->addVertex();
|
||||
auto right = mGraph->addVertex();
|
||||
|
||||
halni idx = 0;
|
||||
for (auto rule: aRules) {
|
||||
|
||||
auto node = compileUtil(rule.head, rule.tail);
|
||||
|
||||
if (!(node.left && node.right)) {
|
||||
mError.mRuleIndex = idx;
|
||||
return {};
|
||||
}
|
||||
|
||||
transitionAny(left, node.left);
|
||||
transitionAny(node.right, right);
|
||||
|
||||
idx++;
|
||||
}
|
||||
|
||||
mGraph->setStartVertex(left);
|
||||
|
||||
return { left, right };
|
||||
}
|
||||
|
||||
private:
|
||||
|
||||
Node compileUtil(const tAlphabetType* regex, tStateType state) {
|
||||
Parser parser;
|
||||
auto astNode = parser.parse(regex);
|
||||
if (parser.mError.isError()) {
|
||||
mGraph->setStartVertex(nullptr);
|
||||
mError.description = "Parsing Of Regular Expression Failed";
|
||||
mError.mRuleState = state;
|
||||
mError.mParseError = parser.mError;
|
||||
return {};
|
||||
}
|
||||
|
||||
auto node = compileNode(astNode, nullptr, nullptr);
|
||||
delete astNode;
|
||||
|
||||
mGraph->setVertexState(node.right, state);
|
||||
mGraph->setStartVertex(node.left);
|
||||
|
||||
return node;
|
||||
}
|
||||
|
||||
Node compileVal(AstVal<tAlphabetType>* val, Vertex* aLeft = nullptr, Vertex* aRight = nullptr) {
|
||||
auto left = aLeft ? aLeft : mGraph->addVertex();
|
||||
auto right = aRight ? aRight : mGraph->addVertex();
|
||||
transitionVal(left, right, val->mVal);
|
||||
return { left, right };
|
||||
}
|
||||
|
||||
Node compileAlternation(AstAlternation* alt, Vertex* aLeft = nullptr, Vertex* aRight = nullptr) {
|
||||
auto first_node = compileNode(alt->mFirst, aLeft, aRight);
|
||||
auto second_node = compileNode(alt->mSecond);
|
||||
transitionAny(first_node.left, second_node.left);
|
||||
transitionAny(second_node.right, first_node.right);
|
||||
return first_node;
|
||||
}
|
||||
|
||||
Node compileAny(AstAny*, Vertex* aLeft = nullptr, Vertex* aRight = nullptr) {
|
||||
auto left = aLeft ? aLeft : mGraph->addVertex();
|
||||
auto right = aRight ? aRight : mGraph->addVertex();
|
||||
transitionAny(left, right, true);
|
||||
return { left, right };
|
||||
}
|
||||
|
||||
Node compileRepeat(AstRepetition* repeat, Vertex* aLeft = nullptr, Vertex* aRight = nullptr) {
|
||||
if (repeat->mPlus) {
|
||||
auto middle = mGraph->addVertex();
|
||||
|
||||
auto left_node = compileNode(repeat->mNode, aLeft, middle);
|
||||
|
||||
auto right_node = compileNode(repeat->mNode, middle, aRight);
|
||||
transitionAny(right_node.right, right_node.left);
|
||||
transitionAny(right_node.left, right_node.right);
|
||||
|
||||
return { left_node.left, right_node.right };
|
||||
}
|
||||
else {
|
||||
auto node = compileNode(repeat->mNode, aLeft, aRight);
|
||||
transitionAny(node.right, node.left);
|
||||
transitionAny(node.left, node.right);
|
||||
return node;
|
||||
}
|
||||
}
|
||||
|
||||
Node compileIf(AstIf* ifNode, Vertex* aLeft = nullptr, Vertex* aRight = nullptr) {
|
||||
auto node = compileNode(ifNode->mNode, aLeft, aRight);
|
||||
transitionAny(node.left, node.right);
|
||||
return node;
|
||||
}
|
||||
|
||||
Node compileClass(AstClass<tAlphabetType>* node, Vertex* aLeft = nullptr, Vertex* aRight = nullptr) {
|
||||
auto left = aLeft ? aLeft : mGraph->addVertex();
|
||||
auto right = aRight ? aRight : mGraph->addVertex();
|
||||
|
||||
if (node->mRanges.length() == 1) {
|
||||
auto const& range = node->mRanges.first()->data;
|
||||
transitionRange(left, right, { range.mBegin, range.mEnd }, node->mExclude);
|
||||
return { left, right };
|
||||
}
|
||||
|
||||
for (auto range : node->mRanges) {
|
||||
auto middle = mGraph->addVertex();
|
||||
transitionRange(left, middle, { range.data().mBegin, range.data().mEnd }, node->mExclude);
|
||||
transitionAny(middle, right);
|
||||
}
|
||||
return { left, right };
|
||||
}
|
||||
|
||||
Node compileCompound(AstCompound* compound, Vertex* aLeft = nullptr, Vertex* aRight = nullptr) {
|
||||
Vertex* left = nullptr;
|
||||
Vertex* rigth = nullptr;
|
||||
|
||||
ualni idx = 0;
|
||||
for (auto child : compound->mChilds) {
|
||||
auto pass_left = idx == 0 ? aLeft : rigth;
|
||||
auto pass_right = idx == compound->mChilds.length() - 1 ? aRight : nullptr;
|
||||
auto node = compileNode(child.data(), pass_left, pass_right);
|
||||
if (!left) left = node.left;
|
||||
rigth = node.right;
|
||||
idx++;
|
||||
}
|
||||
|
||||
return { left, rigth };
|
||||
}
|
||||
|
||||
Node compileNode(AstNode* node, Vertex* aLeft = nullptr, Vertex* aRight = nullptr) {
|
||||
switch (node->mType) {
|
||||
case AstNode::CLASS: return compileClass((AstClass<tAlphabetType>*)node, aLeft, aRight);
|
||||
case AstNode::COMPOUND: return compileCompound((AstCompound*)node, aLeft, aRight);
|
||||
case AstNode::IF: return compileIf((AstIf*)node, aLeft, aRight);
|
||||
case AstNode::REPEAT: return compileRepeat((AstRepetition*)node, aLeft, aRight);
|
||||
case AstNode::ANY: return compileAny((AstAny*)node, aLeft, aRight);
|
||||
case AstNode::OR: return compileAlternation((AstAlternation*)node, aLeft, aRight);
|
||||
case AstNode::VAL: return compileVal((AstVal<tAlphabetType>*)node, aLeft, aRight);
|
||||
case AstNode::NONE:
|
||||
break;
|
||||
}
|
||||
ASSERT(0)
|
||||
return {};
|
||||
}
|
||||
|
||||
void transitionAny(Vertex* from, Vertex* to, bool consumes = false) {
|
||||
mGraph->addTransition(from, to, {}, consumes, true, false);
|
||||
}
|
||||
|
||||
void transitionVal(Vertex* from, Vertex* to, tAlphabetType val) {
|
||||
mGraph->addTransition(from, to, { val, val }, true, false, false);
|
||||
}
|
||||
|
||||
void transitionRange(Vertex* from, Vertex* to, Range<tAlphabetType> range, bool exclude) {
|
||||
mGraph->addTransition(from, to, range, true, false, exclude);
|
||||
}
|
||||
};
|
||||
|
||||
template <typename tAlphabetType, typename tStateType, tStateType tNoStateVal, tStateType tFailedStateVal>
|
||||
CompileError<tStateType> compile(NFA<tAlphabetType, tStateType, tNoStateVal, tFailedStateVal>& out, const tAlphabetType* regex, tStateType state) {
|
||||
Compiler<tAlphabetType, tStateType, tNoStateVal, tFailedStateVal> compiler;
|
||||
compiler.compile(out, regex, state);
|
||||
return compiler.mError;
|
||||
}
|
||||
|
||||
template <typename tAlphabetType, typename tStateType, tStateType tNoStateVal, tStateType tFailedStateVal>
|
||||
CompileError<tStateType> compile(NFA<tAlphabetType, tStateType, tNoStateVal, tFailedStateVal>& out, const InitialierList<Pair<const tAlphabetType*, tStateType>>& rules) {
|
||||
Compiler<tAlphabetType, tStateType, tNoStateVal, tFailedStateVal> compiler;
|
||||
compiler.compile(out, rules);
|
||||
return compiler.mError;
|
||||
}
|
||||
}
|
||||
184
Language/old/Tokenizer/public/Tokenizer.hpp
Normal file
184
Language/old/Tokenizer/public/Tokenizer.hpp
Normal file
|
|
@ -0,0 +1,184 @@
|
|||
|
||||
#pragma once
|
||||
|
||||
#include "RegularExpression.h"
|
||||
#include "Strings.hpp"
|
||||
|
||||
namespace tp {
|
||||
|
||||
extern ModuleManifest gModuleTokenizer;
|
||||
|
||||
template <typename tAlphabetType, typename tTokType, tTokType tNoTokVal, tTokType tFailedTokVal>
|
||||
class Tokenizer {
|
||||
|
||||
TransitionMatrix<tAlphabetType, tTokType, tNoTokVal, tFailedTokVal> mTransitionMatrix;
|
||||
|
||||
RegEx::CompileError<tTokType> mError;
|
||||
|
||||
bool scanFailed() { return mTransitionMatrix.isTrapped(); }
|
||||
|
||||
public:
|
||||
// Some useful RE to be reused
|
||||
static constexpr const tAlphabetType* etherRE = "\n|\t| |\r";
|
||||
static constexpr const tAlphabetType* intRE = "((\\-)|(\\+))?[0-9]+i?";
|
||||
static constexpr const tAlphabetType* floatRE = R"(((\-)|(\+))?([0-9]+)(\.)([0-9]*)?f?)";
|
||||
static constexpr const tAlphabetType* commentBlockRE = R"((/\*){\*-\*}*(\*/))";
|
||||
static constexpr const tAlphabetType* stringRE = R"("{"-"}*")";
|
||||
static constexpr const tAlphabetType* idRE = "([a-z]|[A-Z]|_)+([a-z]|[A-Z]|[0-9]|_)*";
|
||||
|
||||
public:
|
||||
Tokenizer() { MODULE_SANITY_CHECK(gModuleTokenizer) }
|
||||
|
||||
void build(const InitialierList<Pair<const tAlphabetType*, tTokType>>& rules) {
|
||||
NFA<tAlphabetType, tTokType, tNoTokVal, tFailedTokVal> nfa;
|
||||
|
||||
mError = RegEx::compile(nfa, rules);
|
||||
if (mError.isError()) {
|
||||
return;
|
||||
}
|
||||
|
||||
DFA<tAlphabetType, tTokType, tNoTokVal, tFailedTokVal> dfa(nfa);
|
||||
mTransitionMatrix.construct(dfa);
|
||||
}
|
||||
|
||||
[[nodiscard]] bool isBuild() const { return !mError.isError(); }
|
||||
|
||||
auto getMatrix() const { return &mTransitionMatrix; }
|
||||
|
||||
const RegEx::CompileError<tTokType>& getBuildError() { return mError; }
|
||||
|
||||
void resetMatrix() { mTransitionMatrix.reset(); }
|
||||
|
||||
tTokType advanceSymbol(tAlphabetType symbol) { return mTransitionMatrix.move(symbol); }
|
||||
|
||||
tTokType advanceToken(const tAlphabetType* source, ualni source_len, ualni* token_len) {
|
||||
tTokType out = tNoTokVal;
|
||||
*token_len = 0;
|
||||
for (ualni idx = 0; idx < source_len; idx++) {
|
||||
out = advanceSymbol(source[idx]);
|
||||
if (out != tNoTokVal) {
|
||||
*token_len = idx;
|
||||
break;
|
||||
}
|
||||
}
|
||||
return out;
|
||||
}
|
||||
|
||||
~Tokenizer() = default;
|
||||
};
|
||||
|
||||
template <typename tAlphabetType, typename tTokType, tTokType tNoTokVal, tTokType tFailedTokVal, tTokType tSourceEndTokVal>
|
||||
class SimpleTokenizer {
|
||||
public:
|
||||
typedef Tokenizer<tAlphabetType, tTokType, tNoTokVal, tFailedTokVal> tTokenizer;
|
||||
|
||||
private:
|
||||
tTokenizer mTokenizer;
|
||||
|
||||
const tAlphabetType* mSource = nullptr;
|
||||
ualni mLastTokLen = 0;
|
||||
ualni mSourceLen = 0;
|
||||
ualni mAdvancedOffset = 0;
|
||||
|
||||
public:
|
||||
struct Cursor {
|
||||
const tAlphabetType* mSource = nullptr;
|
||||
ualni mAdvancedOffset = 0;
|
||||
const tAlphabetType* str() { return mSource + mAdvancedOffset; }
|
||||
|
||||
struct Location2D {
|
||||
ualni line = 0;
|
||||
ualni character = 0;
|
||||
};
|
||||
|
||||
Location2D get2DLocation() {
|
||||
Location2D out;
|
||||
typedef typename StringLogic<tAlphabetType>::Index Idx;
|
||||
Buffer<Idx> offsets;
|
||||
StringLogic<tAlphabetType>::calcLineOffsets(mSource, StringLogic<tAlphabetType>::calcLength(mSource), offsets);
|
||||
for (ualni idx = 1; idx < offsets.size(); idx++) {
|
||||
if (offsets[idx] > mAdvancedOffset) {
|
||||
out.line = idx - 1;
|
||||
break;
|
||||
}
|
||||
}
|
||||
out.character = mAdvancedOffset - offsets[out.line];
|
||||
return out;
|
||||
}
|
||||
};
|
||||
|
||||
SimpleTokenizer() = default;
|
||||
auto getTokenizer() const { return &mTokenizer; }
|
||||
|
||||
void build(const InitialierList<Pair<const tAlphabetType*, tTokType>>& rules) { mTokenizer.build(rules); }
|
||||
|
||||
[[nodiscard]] bool isBuild() const { return mTokenizer.isBuild(); }
|
||||
|
||||
const RegEx::CompileError<tTokType>& getBuildError() { return mTokenizer.getBuildError(); }
|
||||
|
||||
void bindSource(const tAlphabetType* source) {
|
||||
mSource = source;
|
||||
while (mSource[mSourceLen])
|
||||
mSourceLen++;
|
||||
mSourceLen++;
|
||||
}
|
||||
|
||||
[[nodiscard]] bool isInputLeft() const {
|
||||
if (!mSource) {
|
||||
return false;
|
||||
}
|
||||
if (mSource[mAdvancedOffset] == 0) {
|
||||
return false;
|
||||
}
|
||||
return true;
|
||||
}
|
||||
|
||||
Cursor getCursor() const { return { mSource, mAdvancedOffset }; }
|
||||
|
||||
Cursor getCursorPrev() const { return { mSource, mAdvancedOffset - mLastTokLen }; }
|
||||
|
||||
void setCursor(const Cursor& crs) {
|
||||
mAdvancedOffset = crs.mAdvancedOffset;
|
||||
mLastTokLen = 0;
|
||||
}
|
||||
|
||||
tTokType readTok() {
|
||||
if (mSourceLen == mAdvancedOffset + 1) {
|
||||
return tSourceEndTokVal;
|
||||
}
|
||||
|
||||
tTokType out = mTokenizer.advanceToken(mSource + mAdvancedOffset, mSourceLen - mAdvancedOffset, &mLastTokLen);
|
||||
mAdvancedOffset += mLastTokLen;
|
||||
return out;
|
||||
}
|
||||
|
||||
tTokType lookupTok() {
|
||||
auto out = readTok();
|
||||
discardTok();
|
||||
return out;
|
||||
}
|
||||
|
||||
void discardTok() {
|
||||
mTokenizer.resetMatrix();
|
||||
mAdvancedOffset -= mLastTokLen;
|
||||
mLastTokLen = 0;
|
||||
}
|
||||
|
||||
void skipTok() { readTok(); }
|
||||
|
||||
void reset() {
|
||||
mAdvancedOffset = mLastTokLen = 0;
|
||||
mTokenizer.resetMatrix();
|
||||
}
|
||||
|
||||
[[nodiscard]] ualni lastTokLEn() const { return mLastTokLen; }
|
||||
|
||||
String extractVal() {
|
||||
auto crs = getCursorPrev();
|
||||
String out;
|
||||
out.resize(mLastTokLen);
|
||||
memCopy(out.write(), crs.str(), mLastTokLen);
|
||||
return out;
|
||||
}
|
||||
};
|
||||
}
|
||||
Loading…
Add table
Add a link
Reference in a new issue