Merge pull request 'feature/read_file' (#13) from feature/read_file into develop

Reviewed-on: #13

.gitignore

@@ -2,3 +2,4 @@
 .vscode
 builddir
 *.kb
+*.obk

meson.build

@@ -11,19 +11,20 @@ project('obelisk',
 
 llvm = dependency('llvm', version: '>= 14.0.0', modules : ['core', 'target', 'mcjit', 'nativecodegen'], required : true, method: 'config-tool')
 
-doxygen = find_program('doxygen', required : false)
-
 cdata = configuration_data()
 cdata.set('VERSION', meson.project_version())
 
+docs_enabled = get_option('docs')
+if docs_enabled
+    doxygen = find_program('doxygen', required : false)
+    if doxygen.found()
         if find_program('dot', required : false).found()
             cdata.set('HAVE_DOT', 'YES')
         else
             cdata.set('HAVE_DOT', 'NO')
         endif
-
-if doxygen.found()
         subdir('doc')
     endif
+endif
 
 subdir('src')

meson_options.txt

@@ -0,0 +1,5 @@
+option('docs',
+    type: 'boolean',
+    value: true,
+    description: 'Build documentation for obelisk'
+)

src/lexer.cpp

@@ -2,20 +2,33 @@
 
 #include <iostream>
 
+obelisk::Lexer::Lexer(const std::string& sourceFile)
+{
+    fileStream_.open(sourceFile, std::ifstream::in);
+    if (!fileStream_)
+    {
+        throw obelisk::LexerException("could not open source file " + sourceFile);
+    }
+}
+
+obelisk::Lexer::~Lexer()
+{
+    fileStream_.close();
+    fileStream_.clear();
+}
+
 int obelisk::Lexer::getToken()
 {
-    static int lastChar = ' ';
-
     while (isspace(lastChar))
     {
-        lastChar = std::getc(stdin);
+        lastChar = fileStream_.get();
     }
 
     if (isalpha(lastChar))
     {
         eraseIdentifier();
         appendIdentifier(lastChar);
-        while (isalnum((lastChar = getchar())))
+        while (isalnum((lastChar = fileStream_.get())))
         {
             appendIdentifier(lastChar);
         }
@@ -63,7 +76,7 @@ int obelisk::Lexer::getToken()
                 firstPeriod = true;
             }
             numberStr += lastChar;
-            lastChar = getchar();
+            lastChar = fileStream_.get();
         }
         while (isdigit(lastChar) || lastChar == '.');
 
@@ -83,7 +96,7 @@ int obelisk::Lexer::getToken()
     }
     else if (lastChar == '/')
     {
-        lastChar = getchar();
+        lastChar = fileStream_.get();
         if (lastChar == '/')
         {
             commentLine(&lastChar);
@@ -101,7 +114,7 @@ int obelisk::Lexer::getToken()
     }
 
     int thisChar = lastChar;
-    lastChar     = getchar();
+    lastChar     = fileStream_.get();
     return thisChar;
 }
 
@@ -109,7 +122,7 @@ void obelisk::Lexer::commentLine(int* lastChar)
 {
     do
     {
-        *lastChar = getchar();
+        *lastChar = fileStream_.get();
     }
     while (*lastChar != EOF && *lastChar != '\n' && *lastChar != '\r');
 }

src/lexer.h

@@ -1,6 +1,7 @@
 #ifndef OBELISK_LEXER_H
 #define OBELISK_LEXER_H
 
+#include <fstream>
 #include <string>
 
 namespace obelisk
@@ -12,6 +13,12 @@ namespace obelisk
     class Lexer
     {
         private:
+            int lastChar = ' ';
+            /**
+             * @brief The stream of the source file being read.
+             *
+             */
+            std::ifstream fileStream_;
             /**
              * @brief The last found identifier.
              *
@@ -21,7 +28,7 @@ namespace obelisk
              * @brief The last found number.
              *
              */
-            double numberValue_;
+            double numberValue_ = 0;
 
             /**
              * @brief Set the identifier.
@@ -113,6 +120,19 @@ namespace obelisk
                 kTokenString     = -9
             };
 
+            /**
+             * @brief Construct a new Lexer object.
+             *
+             * @param[in] sourceFile The source file to read.
+             */
+            Lexer(const std::string& sourceFile);
+
+            /**
+             * @brief Destroy the Lexer object.
+             *
+             */
+            ~Lexer();
+
             /**
              * @brief Gets the next token in the source code.
              *

src/meson.build

@@ -1,3 +1,10 @@
+conf_data = configuration_data()
+conf_data.set('version', meson.project_version())
+configure_file(input : 'version.h.in',
+    output : 'version.h',
+    configuration : conf_data
+)
+
 obelisk_sources = files(
     'obelisk.cpp',
     'lexer.cpp',

src/models/fact.cpp

@@ -7,8 +7,8 @@ const char* obelisk::Fact::createTable()
         CREATE TABLE "fact" (
             "id"           INTEGER NOT NULL UNIQUE,
             "left_entity"  INTEGER NOT NULL,
-            "right_entity" INTEGER NOT NULL,
             "verb"         INTEGER NOT NULL,
+            "right_entity" INTEGER NOT NULL,
             "is_true"      INTEGER NOT NULL DEFAULT 0,
             PRIMARY KEY("id" AUTOINCREMENT),
             UNIQUE("left_entity", "right_entity", "verb")

src/obelisk.cpp

@@ -2,20 +2,20 @@
 #include "lexer.h"
 #include "obelisk.h"
 #include "parser.h"
+#include "version.h"
 
 #include <iomanip>
 #include <iostream>
 #include <limits>
 #include <memory>
 
-static int obelisk::mainLoop()
+int obelisk::mainLoop(const std::vector<std::string>& sourceFiles, const std::string& kbFile)
 {
-    auto parser = std::unique_ptr<obelisk::Parser> {new obelisk::Parser()};
     std::unique_ptr<obelisk::KnowledgeBase> kb;
 
     try
     {
-        kb = std::unique_ptr<obelisk::KnowledgeBase> {new obelisk::KnowledgeBase("cromer.kb")};
+        kb = std::unique_ptr<obelisk::KnowledgeBase> {new obelisk::KnowledgeBase(kbFile.c_str())};
     }
     catch (obelisk::KnowledgeBaseException& exception)
     {
@@ -23,8 +23,20 @@ static int obelisk::mainLoop()
         return EXIT_FAILURE;
     }
 
-    // Prime the first token.
-    fprintf(stderr, "ready> ");
+    size_t file = 0;
+    std::shared_ptr<obelisk::Lexer> lexer;
+    try
+    {
+        lexer = std::shared_ptr<obelisk::Lexer> {new obelisk::Lexer(sourceFiles[file++])};
+    }
+    catch (obelisk::LexerException& exception)
+    {
+        std::cout << exception.what() << std::endl;
+        return EXIT_FAILURE;
+    }
+    auto parser = std::unique_ptr<obelisk::Parser> {new obelisk::Parser(lexer)};
+
+    // prime the first token
     try
     {
         parser->getNextToken();
@@ -37,14 +49,29 @@ static int obelisk::mainLoop()
 
     while (true)
     {
-        fprintf(stderr, "ready> ");
         switch (parser->getCurrentToken())
         {
             case obelisk::Lexer::kTokenEof :
+                // end of source file found, create a new lexer and pass it to the parser to use
+                if (file >= sourceFiles.size())
+                {
                     return EXIT_SUCCESS;
-            case ';' : // ignore top-level semicolons.
-                std::cout << "Identifier: " << parser->getLexer()->getIdentifier() << std::endl;
-                std::cout << "Num: " << parser->getLexer()->getNumberValue() << std::endl;
+                }
+                try
+                {
+                    lexer = std::shared_ptr<obelisk::Lexer> {new obelisk::Lexer(sourceFiles[file++])};
+                    parser->setLexer(lexer);
+                    // prime the first token in the parser
+                    parser->getNextToken();
+                }
+                catch (obelisk::LexerException& exception)
+                {
+                    std::cout << exception.what() << std::endl;
+                    return EXIT_FAILURE;
+                }
+                break;
+            case ';' :
+                // semicolon found, the end of a statement
                 try
                 {
                     parser->getNextToken();
@@ -73,38 +100,57 @@ static int obelisk::mainLoop()
     return EXIT_SUCCESS;
 }
 
+static void obelisk::showUsage()
+{
+    std::cout << obelisk::usageMessage << std::endl;
+}
+
 int main(int argc, char** argv)
 {
-    for (int i = 1; i < argc; i++)
-    {
-        std::cout << argv[i] << std::endl;
-    }
+    std::vector<std::string> sourceFiles;
+    std::string kbFile = "obelisk.kb";
 
-    // This can be used to store a double as 2 floats in the database, then restore it back to a double.
-    // Inspired by Godot's double precision on the GPU to render large worlds.
-    /*try
+    while (true)
     {
-        float first;
-        float second;
-        double var = 0.123456789012345;
-
-        obelisk::KnowledgeBase* kb = new obelisk::KnowledgeBase("cromer.kb");
-        kb->getFloat(first, second, var);
-        std::cout << std::setprecision(std::numeric_limits<double>::digits10)
-                  << "Double: " << var << std::endl
-                  << "First: " << first << std::endl
-                  << "Second: " << second << std::endl;
-        var = 0.0;
-        kb->getDouble(var, first, second);
-        std::cout << std::setprecision(std::numeric_limits<double>::digits10)
-                  << "Double: " << var << std::endl
-                  << "First: " << first << std::endl
-                  << "Second: " << second << std::endl;
-    }
-    catch (obelisk::KnowledgeBaseException& exception)
+        int option_index = 0;
+        switch (getopt_long(argc, argv, "k:hv", obelisk::long_options, &option_index))
         {
+            case 'k' :
+                kbFile = std::string(optarg);
+                continue;
+            case 'h' :
+                obelisk::showUsage();
+                return EXIT_SUCCESS;
+                break;
+            case 'v' :
+                std::cout << "obelisk " << obelisk::version << std::endl;
+                return EXIT_SUCCESS;
+                break;
+            default :
+                obelisk::showUsage();
                 return EXIT_FAILURE;
-    }*/
+                break;
 
-    return obelisk::mainLoop();
+            case -1 :
+                break;
+        }
+
+        break;
+    }
+
+    if (optind < argc)
+    {
+        while (optind < argc)
+        {
+            sourceFiles.push_back(argv[optind++]);
+        }
+    }
+
+    if (sourceFiles.size() == 0)
+    {
+        obelisk::showUsage();
+        return EXIT_FAILURE;
+    }
+
+    return obelisk::mainLoop(sourceFiles, kbFile);
 }

src/obelisk.h

@@ -1,15 +1,46 @@
+#include <getopt.h>
+
 /**
- * @brief The obelisk namespace contains everything needed to compile obelisk
+ * @brief The obelisk namespace contains everything needed to compile obelisk.
  * code.
  *
  */
 namespace obelisk
 {
+    /**
+     * @brief The usage messsage displayed during help or incorrect usage.
+     *
+     */
+    std::string usageMessage = R"(Usage: obelisk [OPTION]... [FILE]...
+Compile the obelisk source FILE(s) into knoweldge base and library.
+
+Options:
+  -h, --help            shows this help/usage message
+  -k, --kb=FILENAME     output knowldege base filename
+  -v, --version         shows the version of obelisk)";
+
+    /**
+     * @brief The command line arguments that obelisk accepts.
+     *
+     */
+    static struct option long_options[] = {
+        {"help",    no_argument,       0, 'h'},
+        {"kb",      required_argument, 0, 'k'},
+        {"version", no_argument,       0, 'v'},
+        {0,         0,                 0, 0  }
+    };
+
+    /**
+     * @brief Prints out the usage of obelisk to the stdin.
+     *
+     */
+    static void showUsage();
+
     /**
      * @brief This is the main loop for obelisk.
      * This loop handles lexing and parsing of obelisk source code.
      *
      * @return int Returns EXIT_SUCCESS or EXIT_FAILURE.
      */
-    static int mainLoop();
+    int mainLoop(const std::vector<std::string> &sourceFiles, const std::string &kbFile);
 } // namespace obelisk

src/parser.cpp

@@ -11,16 +11,17 @@
 #include <string>
 #include <vector>
 
-obelisk::Parser::Parser()
-{
-    lexer_ = std::unique_ptr<obelisk::Lexer> {new obelisk::Lexer()};
-}
-
-std::unique_ptr<obelisk::Lexer>& obelisk::Parser::getLexer()
+std::shared_ptr<obelisk::Lexer> obelisk::Parser::getLexer()
 {
     return lexer_;
 }
 
+void obelisk::Parser::setLexer(std::shared_ptr<obelisk::Lexer> lexer)
+{
+    lexer_        = lexer;
+    currentToken_ = 0;
+}
+
 int obelisk::Parser::getNextToken()
 {
     try

src/parser.h

@@ -15,8 +15,8 @@ namespace obelisk
     class Parser
     {
         private:
-            std::unique_ptr<obelisk::Lexer> lexer_;
-            int currentToken_;
+            std::shared_ptr<obelisk::Lexer> lexer_;
+            int currentToken_ = 0;
 
             void setCurrentToken(int currentToken);
 
@@ -37,9 +37,13 @@ namespace obelisk
             void parseFact(std::vector<obelisk::Fact>& facts);
 
         public:
-            Parser();
+            Parser(std::shared_ptr<obelisk::Lexer> lexer) :
+                lexer_(lexer)
+            {
+            }
 
-            std::unique_ptr<obelisk::Lexer>& getLexer();
+            std::shared_ptr<obelisk::Lexer> getLexer();
+            void setLexer(std::shared_ptr<obelisk::Lexer> lexer);
 
             int getCurrentToken();
 

src/version.h.in

@@ -0,0 +1,8 @@
+namespace obelisk
+{
+    /**
+     * @brief The current version of obelisk.
+     *
+     */
+    std::string version = "@version@";
+} // namespace obelisk