From d08f1bc7d17dda9b826240397bc60db2e59dd21b Mon Sep 17 00:00:00 2001
From: ilor <kailoran@gmail.com>
Date: Thu, 25 Nov 2010 14:54:08 +0100
Subject: [PATCH] add any-parsing to Parser class, use it in wcclparser's new
main
---
libwccl/ops/functions.h | 2 +-
libwccl/parser/ANTLRParserResult.h | 28 +++++-
libwccl/parser/Parser.cpp | 55 +++++++++++
libwccl/parser/Parser.h | 6 ++
wcclparser/CMakeLists.txt | 12 +++
wcclparser/main.cpp | 141 +++++++++++++++++++++++++++++
6 files changed, 240 insertions(+), 4 deletions(-)
create mode 100644 wcclparser/main.cpp
diff --git a/libwccl/ops/functions.h b/libwccl/ops/functions.h
index ea3b55d..f2e4842 100644
--- a/libwccl/ops/functions.h
+++ b/libwccl/ops/functions.h
@@ -30,7 +30,7 @@ public:
* @note May be incomplete and/or contain internal info.
*/
virtual std::string raw_name() const = 0;
-protected:
+//protected:
/**
* Base type returned after application of function
* (shared pointer to a const Value)
diff --git a/libwccl/parser/ANTLRParserResult.h b/libwccl/parser/ANTLRParserResult.h
index c60f43b..7845f86 100644
--- a/libwccl/parser/ANTLRParserResult.h
+++ b/libwccl/parser/ANTLRParserResult.h
@@ -6,16 +6,38 @@
#include <libwccl/variables.h>
#include <libwccl/ops/functions.h>
+
template<class T>
-class ANTLRParserResult
+class ANTLRParserResult;
+
+class ANTLRParserResultBase
{
public:
- ANTLRParserResult() {
- this->variables.reset(new Wccl::Variables());
+ ANTLRParserResultBase()
+ : variables(new Wccl::Variables())
+ {
+ }
+
+ boost::shared_ptr<Wccl::Variables> variables;
+
+ virtual boost::shared_ptr<Wccl::FunctionBase> get_op_base() const = 0;
+};
+
+template<class T>
+class ANTLRParserResult : public ANTLRParserResultBase
+{
+public:
+ ANTLRParserResult()
+ {
}
boost::shared_ptr<Wccl::Variables> variables;
boost::shared_ptr<Wccl::Function<T> > op;
+
+ boost::shared_ptr<Wccl::FunctionBase> get_op_base() const
+ {
+ return op;
+ }
};
#endif // ANTLRPARSERRESULT_H
diff --git a/libwccl/parser/Parser.cpp b/libwccl/parser/Parser.cpp
index 35b5591..6f69c9b 100644
--- a/libwccl/parser/Parser.cpp
+++ b/libwccl/parser/Parser.cpp
@@ -108,3 +108,58 @@ boost::shared_ptr<ANTLRParserResult<Wccl::TSet> > Parser::parseSymSetOperator(
return parser.parse_sym_set_operator(this->tagset);
}
+
+// ----------------------------------------------------------------------------
+/**
+ * @desc Parse any operator contained in a std::string. Converts the string to
+ * to a stream and calls parseAnyOperator with it.
+ * @arg str operator string
+ * @return the parsed operator via a shared pointer
+ */
+boost::shared_ptr<ANTLRParserResultBase> Parser::parseAnyOperator(
+ const std::string& str) const
+{
+ std::stringstream ss (std::stringstream::in | std::stringstream::out);
+ ss << str;
+
+ return this->parseAnyOperator(ss);
+}
+
+/**
+ * @desc Parse any operator. Runs parse_*_operator rules in sequence
+ * in the parser grammar until one succedes, or all fail. Rethrows
+ * the exception returned by the last parse_* attempt.
+ * @arg istr input stream with the operator
+ * @return the parsed operator via a shared pointer
+ */
+boost::shared_ptr<ANTLRParserResultBase> Parser::parseAnyOperator(
+ std::istream& istr) const
+{
+ ANTLRLexer lexer(istr);
+ ANTLRParser parser(lexer);
+
+ boost::shared_ptr<ANTLRParserResultBase> result;
+ if (!result) {
+ try {
+ result = parser.parse_sym_set_operator(this->tagset);
+ } catch (antlr::ANTLRException) {
+ // ignore, try another type
+ }
+ }
+ if (!result) {
+ try {
+ result = parser.parse_string_operator(this->tagset);
+ } catch (antlr::ANTLRException) {
+ // ignore, try another type
+ }
+ }
+ if (!result) {
+ try {
+ result = parser.parse_predicates(this->tagset);
+ } catch (antlr::ANTLRException) {
+ throw;
+ }
+ }
+ assert(result);
+ return result;
+}
diff --git a/libwccl/parser/Parser.h b/libwccl/parser/Parser.h
index 10709d4..872adec 100644
--- a/libwccl/parser/Parser.h
+++ b/libwccl/parser/Parser.h
@@ -45,6 +45,12 @@ public:
boost::shared_ptr<ANTLRParserResult<Wccl::TSet> >
parseSymSetOperator(std::istream&) const;
+ // ---------------------------------------------------------------------------
+ // methods for parsing any operators
+ boost::shared_ptr<ANTLRParserResultBase>
+ parseAnyOperator(const std::string&) const;
+ boost::shared_ptr<ANTLRParserResultBase>
+ parseAnyOperator(std::istream&) const;
private:
const Corpus2::Tagset &tagset;
};
diff --git a/wcclparser/CMakeLists.txt b/wcclparser/CMakeLists.txt
index 8aa7e3f..4918095 100644
--- a/wcclparser/CMakeLists.txt
+++ b/wcclparser/CMakeLists.txt
@@ -1,9 +1,21 @@
PROJECT( parser )
+find_package(Libedit)
+if (Libedit_FOUND)
+ message(STATUS "Building with libedit")
+ add_definitions( -DHAVE_LIBEDIT )
+ set(LIBS ${LIBS} ${Libedit_LIBRARIES})
+endif (Libedit_FOUND)
+
include_directories( ${CMAKE_SOURCE_DIR} )
add_definitions(-DLIBWCCL_WCCLPARSER_DATA_DIR="${PROJECT_SOURCE_DIR}/")
+add_executable(wcclparser
+ main.cpp
+)
+target_link_libraries (wcclparser wccl ${Boost_LIBRARIES} antlr ${LIBS})
+
# String operator
add_executable(parser-strop
strop_main.cpp
diff --git a/wcclparser/main.cpp b/wcclparser/main.cpp
new file mode 100644
index 0000000..37c910c
--- /dev/null
+++ b/wcclparser/main.cpp
@@ -0,0 +1,141 @@
+#include <cstdlib>
+
+#include <libwccl/values/strset.h>
+#include <libwccl/parser/Parser.h>
+
+#include <boost/bind.hpp>
+
+#include <antlr/MismatchedTokenException.hpp>
+// ----------------------------------------------------------------------------
+
+//#ifdef HAVE_LIBEDIT
+#include <histedit.h>
+//#endif
+
+/**
+ * @desc A simple command line tester for testing operators
+ */
+
+
+namespace {
+ const char* _prompt = "Enter any operator expression: ";
+}
+
+void std_read_loop(boost::function<bool (const std::string&)>& line_cb)
+{
+ while (std::cin.good()) {
+ std::string s;
+ getline(std::cin, s);
+ if (line_cb(s)) {
+ return;
+ }
+ }
+}
+
+#ifdef HAVE_LIBEDIT
+const char* query_prompt(EditLine*) {
+ return _prompt;
+}
+
+void libedit_read_loop(boost::function<bool (const std::string&)>& line_cb)
+{
+ EditLine *el = el_init("wccl-parser", stdin, stdout, stderr);
+ el_set(el, EL_PROMPT, &query_prompt);
+ el_set(el, EL_EDITOR, "emacs");
+ History* myhistory = history_init();
+ if (myhistory == NULL) {
+ std::cerr << "EditLine history init error\n";
+ el_end(el);
+ std_read_loop(line_cb);
+ return;
+ }
+ HistEvent ev;
+ history(myhistory, &ev, H_SETSIZE, 1024);
+ el_set(el, EL_HIST, history, myhistory);
+
+ bool more = true;
+ while (more) {
+ int count;
+ const char *line = el_gets(el, &count); // line gets a trailing \n
+ if (line == NULL || line[0] == 0) {
+ more = false;
+ } else {
+ std::string s(line, strlen(line) - 1);
+ if (line_cb(s)) {
+ more = false;
+ } else {
+ history(myhistory, &ev, H_ENTER, line);
+ }
+ }
+ }
+
+ history_end(myhistory);
+ el_end(el);
+}
+#endif
+
+bool process_line(const std::string& line, Parser& parser)
+{
+ if (line.empty() || line == "exit" || line == "quit") {
+ return true;
+ } else if (line == "clear" || line == "cls") {
+ if (system("clear")) {}
+ return false;
+ }
+
+ boost::shared_ptr<const Wccl::Value> retVal;
+ boost::shared_ptr<ANTLRParserResultBase> retOp;
+ boost::shared_ptr<Corpus2::Sentence> sentence;
+ Wccl::SentenceContext sc(sentence);
+
+ try {
+ retOp = parser.parseAnyOperator(line);
+ if (retOp) {
+ Wccl::FunExecContext cx(sc, retOp->variables);
+ retVal = retOp->get_op_base()->apply_internal(cx);
+ if (retVal) {
+ std::cerr << "Parsed expression: " << retVal->to_raw_string()
+ << std::endl;
+ } else {
+ std::cerr << "Problem while parsing -- "
+ << "retVal is NULL!" << std::endl;
+ }
+ } else {
+ std::cerr << "Problem while parsing -- "
+ << "parser returned NULL!" << std::endl;
+ }
+ } catch (antlr::MismatchedTokenException &e) {
+ std::cerr << e.getMessage() << std::endl;
+ } catch (Wccl::InvalidVariableName &e) {
+ std::cerr << "Wccl::InvalidVariableName " << e.info() << std::endl;
+ } catch (Wccl::VariableTypeMismatch &e) {
+ std::cerr << "Wccl::VariableTypeMismatch " << e.info() << std::endl;
+ } catch (Wccl::WcclError& e) {
+ std::cerr << "Wccl::WcclError " << e.info() << std::endl;
+ } catch (PwrNlp::PwrNlpError& e) {
+ std::cerr << "PwrNlp::PwrNlpError " << e.info() << std::endl;
+ } catch (antlr::ANTLRException& e) {
+ std::cerr << "Antlr error " << e.getMessage() << std::endl;
+ }
+ return false;
+}
+
+int main()
+{
+ Corpus2::Tagset tagset;
+ Parser parser(tagset);
+
+ if (system("clear")) {
+ //
+ }
+
+ boost::function<bool (const std::string&)> f;
+ f = boost::bind(&process_line, _1, boost::ref(parser));
+#ifdef HAVE_LIBEDIT
+ libedit_read_loop(f);
+#else
+ std_read_loop(f);
+#endif
+
+ return 0;
+}
--
GitLab