First test infrastructure

run_tests.py

@@ -0,0 +1,42 @@
+#!/usr/bin/python3
+import unittest, subprocess, filecmp, argparse, os.path, os, shutil
+
+REFERENCE_DIR = os.path.join(os.path.abspath(os.path.dirname(__file__)), "tests")
+
+REGEXES_DATA = [     #TODO
+        [r"[a-zA-Z0-9]*"]
+        ]
+
+def make_test(regexes, idx):
+    def test(self):
+        p = subprocess.Popen([os.path.join(args.builddir, "src", "Lexesis-test"), "test_%s" % idx, data_dir], stdin=subprocess.PIPE)
+        p.communicate(bytes("\n".join(regexes), "utf-8"))
+        self.assertEqual(0, p.returncode)
+
+        self.assertTrue(filecmp.cmp(os.path.join(REFERENCE_DIR, "test_%s.re" % idx), os.path.join(data_dir, "test_%s.re" % idx)))
+        self.assertTrue(filecmp.cmp(os.path.join(REFERENCE_DIR, "test_%s.enfa" % idx), os.path.join(data_dir, "test_%s.enfa" % idx)))
+        self.assertTrue(filecmp.cmp(os.path.join(REFERENCE_DIR, "test_%s.mssc" % idx), os.path.join(data_dir, "test_%s.mssc" % idx)))
+        self.assertTrue(filecmp.cmp(os.path.join(REFERENCE_DIR, "test_%s.min" % idx), os.path.join(data_dir, "test_%s.min" % idx)))
+
+    return test
+
+class Tests(unittest.TestCase):
+    pass
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--builddir", metavar="builddir", default=os.path.join(os.path.abspath(os.path.dirname(__file__)), "build"), required=False)
+    args = parser.parse_args()
+
+    data_dir = os.path.join(args.builddir, "test_data")
+    try:
+        shutil.rmtree(data_dir)
+    except Exception as err:
+        pass
+    os.mkdir(data_dir)
+
+    for i, regexes in enumerate(REGEXES_DATA):
+        t = make_test(regexes, i)
+        setattr(Tests, "test_%s" % i, t)
+
+    unittest.main()

src/CMakeLists.txt

@@ -5,8 +5,7 @@ add_library(Lexesis-backends
     backends/cpp.cpp
     )
 
-add_executable(Lexesis
-    main.cpp
+add_library(lxs
     automata.cpp
     backend.cpp
     backendmanager.cpp
@@ -15,7 +14,18 @@ add_executable(Lexesis
     inputparser.cpp
     template.cpp
     )
-target_link_libraries(Lexesis Lexesis-backends mstch::mstch)
+
+add_executable(Lexesis
+    main.cpp
+    )
+target_link_libraries(Lexesis Lexesis-backends lxs mstch::mstch)
+
+if (CMAKE_BUILD_TYPE MATCHES Debug)
+    add_executable(Lexesis-test
+        test.cpp
+        )
+    target_link_libraries(Lexesis-test Lexesis-backends lxs mstch::mstch)
+endif()
 
 install(TARGETS Lexesis
     RUNTIME DESTINATION bin

src/test.cpp

@@ -0,0 +1,55 @@
+#include "Lexesis/automata.h"
+#include "Lexesis/driver.h"
+#include "Lexesis/re.h"
+
+#include <cassert>
+#include <fstream>
+#include <iostream>
+#include <sstream>
+#include <vector>
+
+int main(int argc, char** argv) {
+    assert(argc == 3);
+
+    std::vector<std::shared_ptr<lxs::RE> > regexes;
+    std::string testname = argv[1];
+    std::string outputdir = argv[2];
+
+    std::string line;
+    std::ofstream reOut(outputdir + "/" + testname + ".re");
+    while (getline(std::cin, line)) {
+        std::shared_ptr<lxs::RE> re = lxs::parseRE(line);
+        reOut << re->toRe() << std::endl;
+        regexes.push_back(re);
+    }
+    reOut.close();
+
+    std::vector<lxs::ENFA> enfas;
+    for (std::size_t i = 0; i < regexes.size(); i++) {
+        lxs::ENFA enfa;
+        regexes[i]->toENFA(enfa, 0);
+        enfa.numStates++;
+        enfa.starting = 0;
+        enfa.priority[*enfa.accepting.begin()] = i;
+        enfa.acceptingToken[*enfa.accepting.begin()] = std::to_string(i);
+        enfas.push_back(enfa);
+    }
+
+    lxs::ENFA merged = lxs::merge(enfas);
+    
+    std::ofstream enfaOut(outputdir + "/" + testname + ".enfa");
+    enfaOut << lxs::toDot(merged) << std::endl;
+    enfaOut.close();
+
+    lxs::DFA dfa = lxs::mssc(merged);
+    std::ofstream msscOut(outputdir + "/" + testname + ".mssc");
+    msscOut << lxs::toDot(dfa) << std::endl;
+    msscOut.close();
+
+    lxs::DFA min = lxs::minimize(dfa);
+    std::ofstream minOut(outputdir + "/" + testname + ".min");
+    minOut << lxs::toDot(min) << std::endl;
+    minOut.close();
+
+    return 0;
+}

tests/test_0.enfa

@@ -0,0 +1,78 @@
+digraph {
+rankdir=LR
+in [shape=point style=invis]
+0 [label="0" color=yellow]
+1 [label="1"]
+2 [label="2"]
+3 [label="3"]
+4 [label="4" color=green shape=doublecircle]
+2 -> 3 [label="0"]
+2 -> 3 [label="1"]
+2 -> 3 [label="2"]
+2 -> 3 [label="3"]
+2 -> 3 [label="4"]
+2 -> 3 [label="5"]
+2 -> 3 [label="6"]
+2 -> 3 [label="7"]
+2 -> 3 [label="8"]
+2 -> 3 [label="9"]
+2 -> 3 [label="A"]
+2 -> 3 [label="B"]
+2 -> 3 [label="C"]
+2 -> 3 [label="D"]
+2 -> 3 [label="E"]
+2 -> 3 [label="F"]
+2 -> 3 [label="G"]
+2 -> 3 [label="H"]
+2 -> 3 [label="I"]
+2 -> 3 [label="J"]
+2 -> 3 [label="K"]
+2 -> 3 [label="L"]
+2 -> 3 [label="M"]
+2 -> 3 [label="N"]
+2 -> 3 [label="O"]
+2 -> 3 [label="P"]
+2 -> 3 [label="Q"]
+2 -> 3 [label="R"]
+2 -> 3 [label="S"]
+2 -> 3 [label="T"]
+2 -> 3 [label="U"]
+2 -> 3 [label="V"]
+2 -> 3 [label="W"]
+2 -> 3 [label="X"]
+2 -> 3 [label="Y"]
+2 -> 3 [label="Z"]
+2 -> 3 [label="a"]
+2 -> 3 [label="b"]
+2 -> 3 [label="c"]
+2 -> 3 [label="d"]
+2 -> 3 [label="e"]
+2 -> 3 [label="f"]
+2 -> 3 [label="g"]
+2 -> 3 [label="h"]
+2 -> 3 [label="i"]
+2 -> 3 [label="j"]
+2 -> 3 [label="k"]
+2 -> 3 [label="l"]
+2 -> 3 [label="m"]
+2 -> 3 [label="n"]
+2 -> 3 [label="o"]
+2 -> 3 [label="p"]
+2 -> 3 [label="q"]
+2 -> 3 [label="r"]
+2 -> 3 [label="s"]
+2 -> 3 [label="t"]
+2 -> 3 [label="u"]
+2 -> 3 [label="v"]
+2 -> 3 [label="w"]
+2 -> 3 [label="x"]
+2 -> 3 [label="y"]
+2 -> 3 [label="z"]
+0 -> 1 [label="ε"]
+1 -> 2 [label="ε"]
+1 -> 4 [label="ε"]
+3 -> 2 [label="ε"]
+3 -> 4 [label="ε"]
+in -> 0
+}
+

tests/test_0.re

@@ -0,0 +1 @@
+([0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz])*