AIR/cpp-peglib
1
0
Fork 0
mirror of https://github.com/yhirose/cpp-peglib.git synced 2024-12-22 20:05:31 +00:00
Code Issues Packages Projects Releases Wiki Activity

Work in progress.

Browse Source
This commit is contained in:
yhirose 2015-02-09 12:01:59 -05:00
parent 56304269d0
commit 9cb926f7a0
3 changed files with 144 additions and 100 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
example/calc2.cc
View File

@ -26,18 +26,12 @@ class Calculator
{ {
public: public:
Calculator() { Calculator() {
EXPRESSION = seq(TERM, zom(seq(TERM_OPERATOR, TERM))); EXPRESSION <= seq(TERM, zom(seq(TERM_OPERATOR, TERM))), reduce;
TERM = seq(FACTOR, zom(seq(FACTOR_OPERATOR, FACTOR))); TERM <= seq(FACTOR, zom(seq(FACTOR_OPERATOR, FACTOR))), reduce;
FACTOR = cho(NUMBER, seq(chr('('), EXPRESSION, chr(')'))); FACTOR <= cho(NUMBER, seq(chr('('), EXPRESSION, chr(')')));
TERM_OPERATOR = cls("+-"); TERM_OPERATOR <= cls("+-"), [](const char* s, size_t l) { return (char)*s; };
FACTOR_OPERATOR = cls("*/"); FACTOR_OPERATOR <= cls("*/"), [](const char* s, size_t l) { return (char)*s; };
NUMBER = oom(cls("0-9")); NUMBER <= oom(cls("0-9")), [](const char* s, size_t l) { return stol(string(s, l), nullptr, 10); };
actions[EXPRESSION] = reduce;
actions[TERM] = reduce;
actions[TERM_OPERATOR] = [](const char* s, size_t l) { return (char)*s; };
actions[FACTOR_OPERATOR] = [](const char* s, size_t l) { return (char)*s; };
actions[NUMBER] = [](const char* s, size_t l) { return stol(string(s, l), nullptr, 10); };
} }
bool execute(const char* s, long& v) const { bool execute(const char* s, long& v) const {

120
peglib.h
View File

@ -553,12 +553,14 @@ class Grouping
{ {
public: public:
Grouping(const std::shared_ptr<Rule>& rule) : rule_(rule) {} Grouping(const std::shared_ptr<Rule>& rule) : rule_(rule) {}
Grouping(const std::shared_ptr<Rule>& rule, std::function<void(const char* s, size_t l)> match) : rule_(rule), match_(match) {}
template <typename T> template <typename T>
Match parse(const char* s, size_t l, const SemanticActions<T>* sa, SemanticValues<T>* sv) const; Match parse(const char* s, size_t l, const SemanticActions<T>* sa, SemanticValues<T>* sv) const;
private: private:
std::shared_ptr<Rule> rule_; std::shared_ptr<Rule> rule_;
std::function<void(const char* s, size_t l)> match_;
}; };
class NonTerminal class NonTerminal
@ -892,8 +894,9 @@ public:
return static_cast<void*>(this); return static_cast<void*>(this);
} }
void operator=(const std::shared_ptr<Rule>& rule) { Definition& operator<=(const std::shared_ptr<Rule>& rule) {
set_rule(rule); set_rule(rule);
return *this;
} }
template <typename T> template <typename T>
@ -915,8 +918,21 @@ public:
return parse(s, strlen(s), sa, val); return parse(s, strlen(s), sa, val);
} }
template <typename T>
bool parse(const char* s, size_t l, T& val) const {
SemanticValues<Any> sv;
auto m = rule_->parse<Any>(s, l, nullptr, &sv);
return m.ret && m.len == l;
}
template <typename T>
bool parse(const char* s, T& val) const {
return parse(s, strlen(s), val);
}
bool parse(const char* s, size_t l) const { bool parse(const char* s, size_t l) const {
auto m = rule_->parse<int>(s, l, nullptr, nullptr); SemanticValues<Any> sv;
auto m = rule_->parse<Any>(s, l, nullptr, &sv);
return m.ret && m.len == l; return m.ret && m.len == l;
} }
@ -928,6 +944,12 @@ public:
std::function<void (const char* s, size_t l)> match; std::function<void (const char* s, size_t l)> match;
template <typename F>
void operator,(F fn) {
action = fn;
}
SemanticActionAdaptor<Any> action;
private: private:
friend class DefinitionReference; friend class DefinitionReference;
@ -1027,7 +1049,11 @@ Match NotPredicate::parse(const char* s, size_t l, const SemanticActions<T>* sa,
template <typename T> template <typename T>
Match Grouping::parse(const char* s, size_t l, const SemanticActions<T>* sa, SemanticValues<T>* sv) const { Match Grouping::parse(const char* s, size_t l, const SemanticActions<T>* sa, SemanticValues<T>* sv) const {
assert(rule_); assert(rule_);
return rule_->parse<T>(s, l, sa, sv); auto m = rule_->parse<T>(s, l, sa, sv);
if (m.ret && match_) {
match_(s, m.len);
}
return m;
} }
template <typename T> template <typename T>
@ -1047,17 +1073,19 @@ Match NonTerminal::parse(const char* s, size_t l, const SemanticActions<T>* sa,
outer_->match(s, m.len); outer_->match(s, m.len);
} }
if (sv) {
typedef std::function<T (const char* s, size_t l, const std::vector<T>& v, const std::vector<std::string>& n)> Action; typedef std::function<T (const char* s, size_t l, const std::vector<T>& v, const std::vector<std::string>& n)> Action;
Action action; Action action;
if (sa) { if (outer_->action) {
action = outer_->action;
} else if (sa) {
auto it = sa->find(outer_); auto it = sa->find(outer_);
if (it != sa->end()) { if (it != sa->end()) {
action = it->second; action = it->second;
} }
} }
if (sv) {
sv->names.push_back(outer_->name); sv->names.push_back(outer_->name);
auto val = reduce<T>(s, m.len, chldsv->values, chldsv->names, action); auto val = reduce<T>(s, m.len, chldsv->values, chldsv->names, action);
sv->values.push_back(val); sv->values.push_back(val);
@ -1160,6 +1188,10 @@ inline std::shared_ptr<Rule> grp(const std::shared_ptr<Rule>& rule) {
return std::make_shared<Rule>(Grouping(rule)); return std::make_shared<Rule>(Grouping(rule));
} }
inline std::shared_ptr<Rule> grp(const std::shared_ptr<Rule>& rule, std::function<void (const char* s, size_t l)> match) {
return std::make_shared<Rule>(Grouping(rule, match));
}
inline std::shared_ptr<Rule> ref(const std::map<std::string, Definition>& grammar, const std::string& name) { inline std::shared_ptr<Rule> ref(const std::map<std::string, Definition>& grammar, const std::string& name) {
return std::make_shared<Rule>(DefinitionReference(grammar, name)); return std::make_shared<Rule>(DefinitionReference(grammar, name));
} }
@ -1175,52 +1207,52 @@ Grammar make_peg_grammar()
Grammar g; Grammar g;
// Setup PEG syntax parser // Setup PEG syntax parser
g["Grammar"] = seq(g["Spacing"], oom(g["Definition"]), g["EndOfFile"]); g["Grammar"] <= seq(g["Spacing"], oom(g["Definition"]), g["EndOfFile"]);
g["Definition"] = seq(g["Identifier"], g["LEFTARROW"], g["Expression"]); g["Definition"] <= seq(g["Identifier"], g["LEFTARROW"], g["Expression"]);
g["Expression"] = seq(g["Sequence"], zom(seq(g["SLASH"], g["Sequence"]))); g["Expression"] <= seq(g["Sequence"], zom(seq(g["SLASH"], g["Sequence"])));
g["Sequence"] = zom(g["Prefix"]); g["Sequence"] <= zom(g["Prefix"]);
g["Prefix"] = seq(opt(cho(g["AND"], g["NOT"])), g["Suffix"]); g["Prefix"] <= seq(opt(cho(g["AND"], g["NOT"])), g["Suffix"]);
g["Suffix"] = seq(g["Primary"], opt(cho(g["QUESTION"], g["STAR"], g["PLUS"]))); g["Suffix"] <= seq(g["Primary"], opt(cho(g["QUESTION"], g["STAR"], g["PLUS"])));
g["Primary"] = cho(seq(g["Identifier"], npd(g["LEFTARROW"])), g["Primary"] <= cho(seq(g["Identifier"], npd(g["LEFTARROW"])),
seq(g["OPEN"], g["Expression"], g["CLOSE"]), seq(g["OPEN"], g["Expression"], g["CLOSE"]),
g["Literal"], g["Class"], g["DOT"]); g["Literal"], g["Class"], g["DOT"]);
g["Identifier"] = seq(g["IdentifierContent"], g["Spacing"]); g["Identifier"] <= seq(g["IdentCont"], g["Spacing"]);
g["IdentifierContent"] = seq(g["IdentStart"], zom(g["IdentCont"])); g["IdentCont"] <= seq(g["IdentStart"], zom(g["IdentRest"]));
g["IdentStart"] = cls("a-zA-Z_"); g["IdentStart"] <= cls("a-zA-Z_");
g["IdentCont"] = cho(g["IdentStart"], cls("0-9")); g["IdentRest"] <= cho(g["IdentStart"], cls("0-9"));
g["Literal"] = cho(seq(cls("'"), g["SingleQuotesContent"], cls("'"), g["Spacing"]), g["Literal"] <= cho(seq(cls("'"), g["SQCont"], cls("'"), g["Spacing"]),
seq(cls("\""), g["DoubleQuotesContent"], cls("\""), g["Spacing"])); seq(cls("\""), g["DQCont"], cls("\""), g["Spacing"]));
g["SingleQuotesContent"] = zom(seq(npd(cls("'")), g["Char"])); g["SQCont"] <= zom(seq(npd(cls("'")), g["Char"]));
g["DoubleQuotesContent"] = zom(seq(npd(cls("\"")), g["Char"])); g["DQCont"] <= zom(seq(npd(cls("\"")), g["Char"]));
g["Class"] = seq(chr('['), g["ClassContent"], chr(']'), g["Spacing"]); g["Class"] <= seq(chr('['), g["ClassCont"], chr(']'), g["Spacing"]);
g["ClassContent"] = zom(seq(npd(chr(']')), g["Range"])); g["ClassCont"] <= zom(seq(npd(chr(']')), g["Range"]));
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")), seq(chr('\\'), cls("0-2"), 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(npd(chr('\\')), any())); seq(npd(chr('\\')), any()));
g["LEFTARROW"] = seq(lit("<-"), g["Spacing"]); g["LEFTARROW"] <= seq(lit("<-"), g["Spacing"]);
g["SLASH"] = seq(chr('/'), g["Spacing"]); g["SLASH"] <= seq(chr('/'), g["Spacing"]);
g["AND"] = seq(chr('&'), g["Spacing"]); g["AND"] <= seq(chr('&'), g["Spacing"]);
g["NOT"] = seq(chr('!'), g["Spacing"]); g["NOT"] <= seq(chr('!'), g["Spacing"]);
g["QUESTION"] = seq(chr('?'), g["Spacing"]); g["QUESTION"] <= seq(chr('?'), g["Spacing"]);
g["STAR"] = seq(chr('*'), g["Spacing"]); g["STAR"] <= seq(chr('*'), g["Spacing"]);
g["PLUS"] = seq(chr('+'), g["Spacing"]); g["PLUS"] <= seq(chr('+'), g["Spacing"]);
g["OPEN"] = seq(chr('('), g["Spacing"]); g["OPEN"] <= seq(chr('('), g["Spacing"]);
g["CLOSE"] = seq(chr(')'), g["Spacing"]); g["CLOSE"] <= seq(chr(')'), g["Spacing"]);
g["DOT"] = seq(chr('.'), g["Spacing"]); g["DOT"] <= seq(chr('.'), g["Spacing"]);
g["Spacing"] = zom(cho(g["Space"], g["Comment"])); g["Spacing"] <= zom(cho(g["Space"], g["Comment"]));
g["Comment"] = seq(chr('#'), zom(seq(npd(g["EndOfLine"]), any())), g["EndOfLine"]); g["Comment"] <= seq(chr('#'), zom(seq(npd(g["EndOfLine"]), any())), g["EndOfLine"]);
g["Space"] = cho(chr(' '), chr('\t'), g["EndOfLine"]); g["Space"] <= cho(chr(' '), chr('\t'), g["EndOfLine"]);
g["EndOfLine"] = cho(lit("\r\n"), chr('\n'), chr('\r')); g["EndOfLine"] <= cho(lit("\r\n"), chr('\n'), chr('\r'));
g["EndOfFile"] = npd(any()); g["EndOfFile"] <= npd(any());
// Set definition names // Set definition names
for (auto& x: g) { for (auto& x: g) {
@ -1244,7 +1276,7 @@ std::shared_ptr<Grammar> make_grammar(const char* syntax, std::string& start)
sa[peg["Definition"]] = [&](const std::vector<Any>& v) { sa[peg["Definition"]] = [&](const std::vector<Any>& v) {
const auto& name = v[0].get<std::string>(); const auto& name = v[0].get<std::string>();
(*grammar)[name] = v[2].get<std::shared_ptr<Rule>>(); (*grammar)[name] <= v[2].get<std::shared_ptr<Rule>>();
(*grammar)[name].name = name; (*grammar)[name].name = name;
if (start.empty()) { if (start.empty()) {
@ -1315,24 +1347,24 @@ std::shared_ptr<Grammar> make_grammar(const char* syntax, std::string& start)
} }
}; };
sa[peg["IdentifierContent"]] = [](const char*s, size_t l) { sa[peg["IdentCont"]] = [](const char*s, size_t l) {
return std::string(s, l); return std::string(s, l);
}; };
sa[peg["Literal"]] = [](const std::vector<Any>& v) { sa[peg["Literal"]] = [](const std::vector<Any>& v) {
return lit(v[0].get<std::string>().c_str()); return lit(v[0].get<std::string>().c_str());
}; };
sa[peg["SingleQuotesContent"]] = [](const char*s, size_t l) { sa[peg["SQCont"]] = [](const char*s, size_t l) {
return std::string(s, l); return std::string(s, l);
}; };
sa[peg["DoubleQuotesContent"]] = [](const char*s, size_t l) { sa[peg["DQCont"]] = [](const char*s, size_t l) {
return std::string(s, l); return std::string(s, l);
}; };
sa[peg["Class"]] = [](const std::vector<Any>& v) { sa[peg["Class"]] = [](const std::vector<Any>& v) {
return cls(v[0].get<std::string>().c_str()); return cls(v[0].get<std::string>().c_str());
}; };
sa[peg["ClassContent"]] = [](const char*s, size_t l) { sa[peg["ClassCont"]] = [](const char*s, size_t l) {
return std::string(s, l); return std::string(s, l);
}; };

72
test/test.cc
View File

@ -35,21 +35,40 @@ TEST_CASE("String capture test", "[general]")
using namespace peglib; using namespace peglib;
using namespace std; using namespace std;
TEST_CASE("String capture test with match", "[general]") TEST_CASE("String capture test2", "[general]")
{
{
vector<string> tags;
Definition ROOT, TAG, TAG_NAME, WS;
ROOT <= seq(WS, zom(TAG));
TAG <= seq(chr('['), TAG_NAME, chr(']'), WS);
TAG_NAME <= oom(seq(npd(chr(']')), any())), [&](const char* s, size_t l) { tags.push_back(string(s, l)); };
WS <= zom(cls(" \t"));
auto ret = ROOT.parse(" [tag1] [tag:2] [tag-3] ");
REQUIRE(ret == true);
REQUIRE(tags.size() == 3);
REQUIRE(tags[0] == "tag1");
REQUIRE(tags[1] == "tag:2");
REQUIRE(tags[2] == "tag-3");
}
REQUIRE(VARINT_COUNT == 0);
}
TEST_CASE("String capture test with embedded match action", "[general]")
{ {
{ {
Definition ROOT, TAG, TAG_NAME, WS; Definition ROOT, TAG, TAG_NAME, WS;
ROOT = seq(WS, zom(TAG));
TAG = seq(chr('['), TAG_NAME, chr(']'), WS);
TAG_NAME = oom(seq(npd(chr(']')), any()));
WS = zom(cls(" \t"));
vector<string> tags; vector<string> tags;
TAG_NAME.match = [&](const char* s, size_t l) { ROOT <= seq(WS, zom(TAG));
tags.push_back(string(s, l)); TAG <= seq(chr('['), grp(TAG_NAME, [&](const char* s, size_t l) { tags.push_back(string(s, l)); }), chr(']'), WS);
}; TAG_NAME <= oom(seq(npd(chr(']')), any()));
WS <= zom(cls(" \t"));
auto ret = ROOT.parse(" [tag1] [tag:2] [tag-3] "); auto ret = ROOT.parse(" [tag1] [tag:2] [tag-3] ");
@ -69,8 +88,8 @@ TEST_CASE("Cyclic grammer test", "[general]")
Definition PARENT; Definition PARENT;
Definition CHILD; Definition CHILD;
PARENT = seq(CHILD); PARENT <= seq(CHILD);
CHILD = seq(PARENT); CHILD <= seq(PARENT);
} }
REQUIRE(VARINT_COUNT == 0); REQUIRE(VARINT_COUNT == 0);
@ -98,16 +117,14 @@ TEST_CASE("Calculator test", "[general]")
// Construct grammer // Construct grammer
Definition EXPRESSION, TERM, FACTOR, TERM_OPERATOR, FACTOR_OPERATOR, NUMBER; Definition EXPRESSION, TERM, FACTOR, TERM_OPERATOR, FACTOR_OPERATOR, NUMBER;
EXPRESSION = seq(TERM, zom(seq(TERM_OPERATOR, TERM))); EXPRESSION <= seq(TERM, zom(seq(TERM_OPERATOR, TERM)));
TERM = seq(FACTOR, zom(seq(FACTOR_OPERATOR, FACTOR))); TERM <= seq(FACTOR, zom(seq(FACTOR_OPERATOR, FACTOR)));
FACTOR = cho(NUMBER, seq(chr('('), EXPRESSION, chr(')'))); FACTOR <= cho(NUMBER, seq(chr('('), EXPRESSION, chr(')')));
TERM_OPERATOR = cls("+-"); TERM_OPERATOR <= cls("+-");
FACTOR_OPERATOR = cls("*/"); FACTOR_OPERATOR <= cls("*/");
NUMBER = oom(cls("0-9")); NUMBER <= oom(cls("0-9"));
// Setup actions // Setup actions
SemanticActions<Any> actions;
auto reduce = [](const vector<Any>& v) -> long { auto reduce = [](const vector<Any>& v) -> long {
long ret = v[0].get<long>(); long ret = v[0].get<long>();
for (auto i = 1u; i < v.size(); i += 2) { for (auto i = 1u; i < v.size(); i += 2) {
@ -122,6 +139,7 @@ TEST_CASE("Calculator test", "[general]")
return ret; return ret;
}; };
SemanticActions<Any> actions;
actions[EXPRESSION] = reduce; actions[EXPRESSION] = reduce;
actions[TERM] = reduce; actions[TERM] = reduce;
actions[TERM_OPERATOR] = [](const char* s, size_t l) { return *s; }; actions[TERM_OPERATOR] = [](const char* s, size_t l) { return *s; };
@ -178,7 +196,7 @@ TEST_CASE("Calculator test2", "[general]")
a[g["TERM"]] = reduce; a[g["TERM"]] = reduce;
a[g["TERM_OPERATOR"]] = [](const char* s, size_t l) { return *s; }; a[g["TERM_OPERATOR"]] = [](const char* s, size_t l) { return *s; };
a[g["FACTOR_OPERATOR"]] = [](const char* s, size_t l) { return *s; }; a[g["FACTOR_OPERATOR"]] = [](const char* s, size_t l) { return *s; };
a[g["NUMBER"]] = [&](const char* s, size_t l) { return stol(string(s, l), nullptr, 10); }; a[g["NUMBER"]] = [](const char* s, size_t l) { return stol(string(s, l), nullptr, 10); };
// Parse // Parse
Any val; Any val;
@ -336,15 +354,15 @@ TEST_CASE("PEG IdentStart", "[peg]")
REQUIRE(g["IdentStart"].parse("0") == false); REQUIRE(g["IdentStart"].parse("0") == false);
} }
TEST_CASE("PEG IdentCont", "[peg]") TEST_CASE("PEG IdentRest", "[peg]")
{ {
Grammar g = make_peg_grammar(); Grammar g = make_peg_grammar();
REQUIRE(g["IdentCont"].parse("_") == true); REQUIRE(g["IdentRest"].parse("_") == true);
REQUIRE(g["IdentCont"].parse("a") == true); REQUIRE(g["IdentRest"].parse("a") == true);
REQUIRE(g["IdentCont"].parse("Z") == true); REQUIRE(g["IdentRest"].parse("Z") == true);
REQUIRE(g["IdentCont"].parse("") == false); REQUIRE(g["IdentRest"].parse("") == false);
REQUIRE(g["IdentCont"].parse(" ") == false); REQUIRE(g["IdentRest"].parse(" ") == false);
REQUIRE(g["IdentCont"].parse("0") == true); REQUIRE(g["IdentRest"].parse("0") == true);
} }
TEST_CASE("PEG Literal", "[peg]") TEST_CASE("PEG Literal", "[peg]")