#include "AsciiComponentClass.h"
#include "Utf8ComponentClass.h"
#include "ComponentAssertion.h"
#include "ComponentAtomicGroup.h"
#include "ComponentBackReference.h"
#include "ComponentBoundary.h"
#include "ComponentClass.h"
#include "ComponentCondReference.h"
#include "ComponentRepeat.h"
#include "ComponentSequence.h"
#include "ComponentVisitor.h"
#include "ComponentWordBoundary.h"
#include "ConstComponentVisitor.h"
#include "parse_error.h"
#include "shortcut_literal.h"
#include "grey.h"
#include "nfagraph/ng.h"
#include "compiler/compiler.h"
#include "util/ue2string.h"
#include "ue2common.h"
#include <stack>
using namespace std;
namespace ue2 {
class ConstructLiteralVisitor : public ConstComponentVisitor {
public:
~ConstructLiteralVisitor() override;
struct NotLiteral {};
void pre(const AsciiComponentClass &c) override {
const CharReach &cr = c.cr;
const size_t width = cr.count();
if (width == 1) {
lit.push_back(cr.find_first(), false);
} else if (width == 2 && cr.isCaselessChar()) {
lit.push_back(cr.find_first(), true);
} else {
throw NotLiteral();
}
}
void pre(const ComponentRepeat &c) override {
if (c.m_min == 0 || c.m_min != c.m_max) {
throw NotLiteral();
}
if (c.m_max < ComponentRepeat::NoLimit && c.m_max > 32767) {
throw ParseError("Bounded repeat is too large.");
}
repeat_stack.push(lit.length());
}
void post(const ComponentRepeat &c) override {
assert(!repeat_stack.empty());
const ue2_literal suffix = lit.substr(repeat_stack.top());
repeat_stack.pop();
for (unsigned i = 1; i < c.m_min; i++) {
lit += suffix;
}
}
void pre(const ComponentSequence &) override {
}
void pre(const ComponentAlternation &) override { throw NotLiteral(); }
void pre(const ComponentAssertion &) override { throw NotLiteral(); }
void pre(const ComponentAtomicGroup &) override { throw NotLiteral(); }
void pre(const ComponentBackReference &) override { throw NotLiteral(); }
void pre(const ComponentBoundary &) override { throw NotLiteral(); }
void pre(const ComponentByte &) override { throw NotLiteral(); }
void pre(const ComponentCondReference &) override { throw NotLiteral(); }
void pre(const ComponentEmpty &) override { throw NotLiteral(); }
void pre(const ComponentEUS &) override { throw NotLiteral(); }
void pre(const ComponentWordBoundary &) override { throw NotLiteral(); }
void pre(const UTF8ComponentClass &) override { throw NotLiteral(); }
void during(const AsciiComponentClass &) override {}
void during(const ComponentAlternation &) override {}
void during(const ComponentAssertion &) override {}
void during(const ComponentAtomicGroup &) override {}
void during(const ComponentBackReference &) override {}
void during(const ComponentBoundary &) override {}
void during(const ComponentByte &) override {}
void during(const ComponentCondReference &) override {}
void during(const ComponentEmpty &) override {}
void during(const ComponentEUS &) override {}
void during(const ComponentRepeat &) override {}
void during(const ComponentSequence &) override {}
void during(const ComponentWordBoundary &) override {}
void during(const UTF8ComponentClass &) override {}
void post(const AsciiComponentClass &) override {}
void post(const ComponentAlternation &) override {}
void post(const ComponentAssertion &) override {}
void post(const ComponentAtomicGroup &) override {}
void post(const ComponentBackReference &) override {}
void post(const ComponentBoundary &) override {}
void post(const ComponentByte &) override {}
void post(const ComponentCondReference &) override {}
void post(const ComponentEmpty &) override {}
void post(const ComponentEUS &) override {}
void post(const ComponentSequence &) override {}
void post(const ComponentWordBoundary &) override {}
void post(const UTF8ComponentClass &) override {}
ue2_literal lit;
stack<size_t> repeat_stack; };
ConstructLiteralVisitor::~ConstructLiteralVisitor() {}
bool shortcutLiteral(NG &ng, const ParsedExpression &pe) {
assert(pe.component);
if (!ng.cc.grey.allowLiteral) {
return false;
}
const auto &expr = pe.expr;
if (expr.min_offset || expr.max_offset != MAX_OFFSET || expr.min_length ||
expr.edit_distance || expr.hamm_distance) {
DEBUG_PRINTF("extended params not allowed\n");
return false;
}
ConstructLiteralVisitor vis;
try {
assert(pe.component);
pe.component->accept(vis);
assert(vis.repeat_stack.empty());
} catch (const ConstructLiteralVisitor::NotLiteral&) {
DEBUG_PRINTF("not a literal\n");
return false;
}
const ue2_literal &lit = vis.lit;
if (lit.empty()) {
DEBUG_PRINTF("empty literal\n");
return false;
}
if (expr.highlander && lit.length() <= 1) {
DEBUG_PRINTF("not shortcutting SEP literal\n");
return false;
}
DEBUG_PRINTF("constructed literal %s\n", dumpString(lit).c_str());
return ng.addLiteral(lit, expr.index, expr.report, expr.highlander,
expr.som, expr.quiet);
}
}