linkparser 1.0.4 → 1.1.0

data/Rakefile.local

@@ -1,5 +1,7 @@
 #!rake
 
+require 'rake/extensiontask'
+
 # C extension constants
 EXT_MAKEFILE  = EXTDIR + 'Makefile'
 EXT_SOURCES   = FileList[ EXTDIR + '*.c' ]
@@ -11,55 +13,45 @@ EXT_SO        = EXTDIR + "linkparser_ext.#{CONFIG['DLEXT']}"
 #####################################################################
 
 # Make both the default task and the spec task depend on building the extension
-task :local => :build
-task :spec => :build
+task :local => :compile
+task :spec => :compile
 namespace :spec do
-	task :doc   => [ :build ]
-	task :quiet => [ :build ]
-	task :html  => [ :build ]
-	task :text  => [ :build ]
+	task :doc   => [ :compile ]
+	task :quiet => [ :compile ]
+	task :html  => [ :compile ]
+	task :text  => [ :compile ]
 end
 
-desc "Make the Makefile for the C extension"
-file EXT_MAKEFILE.to_s => EXT_SOURCES do
-	log "Configuring linkparser C extension"
-	in_subdirectory( EXTDIR ) do
-		ruby 'extconf.rb'
-	end
+Rake::ExtensionTask.new( 'linkparser_ext', GEMSPEC ) do |ext|
+	ext.ext_dir = 'ext'
+	ext.lib_dir = 'ext'
+	ext.source_pattern = "*.{c,h}"
+	ext.cross_compile = true
+	ext.cross_platform = 'i386-mswin32'
 end
-CLOBBER.include( EXTDIR + 'mkmf.log', EXT_SO )
 
-desc "Build the C extension"
-task :build => [ EXT_MAKEFILE.to_s, *EXT_SOURCES ] do
-	in_subdirectory( EXTDIR ) do
-		sh 'make'
-	end
+task :debug do
+	$DEBUG = true
+	ENV['DEBUG'] = '1'
 end
 
+# Make both the default task and the spec task depend on building the extension
+namespace :spec do
 
-desc "Rebuild the C extension"
-task :rebuild => [ :clean, :build ]
+	desc "Run specs under gdb"
+	task :gdb => [ :compile ] do |task|
+		require 'tempfile'
 
+	    cmd_parts = ['run']
+	    cmd_parts << '-Ilib:ext'
+	    cmd_parts << '/usr/bin/spec'
+	    cmd_parts += SPEC_FILES.collect { |fn| %["#{fn}"] }
+	    cmd_parts += COMMON_SPEC_OPTS + ['-f', 's', '-c']
 
-task :clean do
-	if EXT_MAKEFILE.exist?
-		in_subdirectory( EXTDIR ) do
-			sh 'make clean'
-		end
-	end
-end	
+		script = Tempfile.new( 'spec-gdbscript' )
+		script.puts( cmd_parts.join(' ') )
+		script.flush
 
-task :clobber do
-	if EXT_MAKEFILE.exist?
-		in_subdirectory( EXTDIR ) do
-			sh 'make distclean'
-		end
+		run 'gdb', '-x', script.path, RUBY
 	end
-end	
-CLOBBER.include( EXT_MAKEFILE )
-
-task :debug do
-	log "Setting $DEBUG to true"
-	$DEBUG = true
 end
-

data/ext/dictionary.c

@@ -1,6 +1,6 @@
 /*
  *  dict.c - Ruby LinkParser - Dictionary Class
- *  $Id: dictionary.c 48 2008-12-19 18:30:33Z deveiant $
+ *  $Id: dictionary.c,v 53ec62029ee4 2010/10/12 15:45:26 ged $
  *  
  *  Authors:
  *    * Michael Granger <ged@FaerieMUD.org>
@@ -16,19 +16,45 @@
  *  Memory management functions
  * -------------------------------------------------- */
 
+
+/*
+ * Allocation function
+ */
+static struct rlink_dictionary *
+rlink_dictionary_alloc() {
+	struct rlink_dictionary *ptr = ALLOC( struct rlink_dictionary );
+
+	ptr->dict	= NULL;
+
+	debugMsg(( "Initialized an rlink_dictionary <%p>", ptr ));
+	return ptr;
+}
+
+
 /*
  * Free function
  */
 static void
-rlink_dict_gc_free( Dictionary dict ) {
-	if ( dict ) dictionary_delete( dict );
+rlink_dict_gc_free( struct rlink_dictionary *ptr ) {
+	if ( ptr ) {
+		debugMsg(( "Freeing Dictionary <%p>", ptr ));
+		if ( ptr->dict )
+			dictionary_delete( ptr->dict );
+
+		ptr->dict = NULL;
+
+		xfree( ptr );
+		ptr = NULL;
+	} else {
+		debugMsg(( "Not freeing already freed Dictionary." ));
+	}
 }
 
 
 /*
  * Object validity checker. Returns the data pointer.
  */
-static Dictionary
+static struct rlink_dictionary *
 check_dict( VALUE self ) {
 	Check_Type( self, T_DATA );
 
@@ -36,7 +62,7 @@ check_dict( VALUE self ) {
 		rb_raise( rb_eTypeError, "wrong argument type %s (expected LinkParser::Dictionary)",
 				  rb_class2name(CLASS_OF( self )) );
     }
-	
+
 	return DATA_PTR( self );
 }
 
@@ -44,21 +70,21 @@ check_dict( VALUE self ) {
 /*
  * Fetch the data pointer and check it for sanity.
  */
-static Dictionary
+static struct rlink_dictionary *
 get_dict( VALUE self ) {
-	Dictionary dict = check_dict( self );
+	struct rlink_dictionary *ptr = check_dict( self );
 
-	if ( !dict )
+	if ( !ptr )
 		rb_raise( rb_eRuntimeError, "uninitialized Dictionary" );
 
-	return dict;
+	return ptr;
 }
 
 
 /* 
  * Get the Dictionary behind the LinkParser::Dictionary +object+ specified.
- */ 
-Dictionary
+ */
+struct rlink_dictionary *
 rlink_get_dict( VALUE obj ) {
 	return get_dict( obj );
 }
@@ -93,35 +119,67 @@ rlink_make_oldstyle_dict( VALUE dict_file, VALUE pp_file, VALUE cons_file, VALUE
 	SafeStringValue( pp_file    );
 	SafeStringValue( cons_file  );
 	SafeStringValue( affix_file );
-	
+
 	return dictionary_create(
-		STR2CSTR( dict_file  ),
-		STR2CSTR( pp_file    ),
-		STR2CSTR( cons_file  ),
-		STR2CSTR( affix_file )
+		StringValuePtr(dict_file ),
+		StringValuePtr(pp_file   ),
+		StringValuePtr(cons_file ),
+		StringValuePtr(affix_file)
 	);
 }
 
 
 /*
  *  call-seq:
- *    LinkParser::Dictionary.new( options={} )
- *    LinkParser::Dictionary.new( language, options={} )
- *    LinkParser::Dictionary.new( dict, pp, ck, affix, option={} )
+ *      LinkParser::Dictionary.new( *args )                        -> dict
+ *
+ * Create a new LinkParser::Dictionary.
+ * 
+ * The preferred way to set up the dictionary is to call it with no
+ * arguments, which will look for a dictionary with the same language
+ * as the current environment. Alternatively, a fixed language can be
+ * specified by specifying an ISO639 language code, for example,
+ * <tt>LinkParser::Dictionary.new( :en )</tt>.
  * 
- *  Create a new LinkParser::Dictionary with data files for the given +language+, or
- *  using the specified data files.
+ * Explicit dictionary file names can be also specified, like so:
+ *     
+ *     Dictionary.new( dict_file,
+ *                     post_process_file,
+ *                     constituent_knowledge_file,
+ *                     affix_file )
+ *     
+ * This mode of dictionary construction is not recommended for new 
+ * development, and is intended for advanced users only. To create the 
+ * dictionary, the Dictionary looks in the current directory and the data 
+ * directory for the files +dict_file+, +post_process_file+,
+ * +constituent_knowledge_file+, and +affix_file+. The last three entries
+ * may be omitted. If +dict_file+ is a fully specified path name, then
+ * the other file names, which need not be fully specified, will be
+ * prefixed by the directory specified by +dict_file+.
+ * 
+ * In any case, a Hash of options can be specified which will be used
+ * as default ParseOption attributes for any sentences created from
+ * it.
  *
+ *  Examples:
+ *      dict = LinkParser::Dictionary.new
+ *      
+ *      dict = LinkParser::Dictionary.new( :de )
+ *      
+ *      dict = LinkParser::Dictionary.new( '/var/data/custom_dicts/4.2.dict' )
+ *      
+ * 
  */
 static VALUE
 rlink_dict_initialize( int argc, VALUE *argv, VALUE self ) {
 	if ( !check_dict(self) ) {
 		int i = 0;
+		struct rlink_dictionary *ptr = NULL;
 		Dictionary dict = NULL;
 		VALUE arg1, arg2, arg3, arg4, arg5 = Qnil;
 		VALUE lang = Qnil;
 		VALUE opthash = Qnil;
-		
+
 		switch( i = rb_scan_args(argc, argv, "05", &arg1, &arg2, &arg3, &arg4, &arg5) ) {
 		  /* Dictionary.new */
 		  case 0:
@@ -155,10 +213,10 @@ rlink_dict_initialize( int argc, VALUE *argv, VALUE self ) {
 			dict = rlink_make_oldstyle_dict( arg1, arg2, arg3, arg4 );
 			opthash = arg5;
 			break;
-		
-		  /* Anything else is an error */	
+
+		  /* Anything else is an error */
 		  default:
-			rb_raise( rb_eArgError, 
+			rb_raise( rb_eArgError,
 				"wrong number of arguments (%d for 0,1,2,4, or 5)", i );
 		}
 
@@ -166,17 +224,20 @@ rlink_dict_initialize( int argc, VALUE *argv, VALUE self ) {
 		if ( !dict && i < 4 ) {
 			if ( RTEST(lang) ) {
 				SafeStringValue( lang );
-				dict = dictionary_create_lang( STR2CSTR(lang) );
+				dict = dictionary_create_lang( StringValuePtr(lang) );
 			} else {
 				dict = dictionary_create_default_lang();
 			}
 		}
-		
+
 		/* If the dictionary still isn't created, there was an error
 		   creating it */
 		if ( !dict ) rlink_raise_lp_error();
 
-		DATA_PTR( self ) = dict;
+		debugMsg(( "Created dictionary %p", dict ));
+		DATA_PTR( self ) = ptr = rlink_dictionary_alloc();
+
+		ptr->dict = dict;
 
 		/* If they passed in an options hash, save it for later. */
 		if ( RTEST(opthash) ) rb_iv_set( self, "@options", opthash );
@@ -199,13 +260,14 @@ rlink_dict_initialize( int argc, VALUE *argv, VALUE self ) {
  *  connector in the dictionary. This is useful for designing a parsing
  *  algorithm that progresses in stages, first trying the cheap connectors.
  */
-static VALUE 
+static VALUE
 rlink_get_max_cost( VALUE self ) {
-	Dictionary dict = get_dict( self );
-	int cost = dictionary_get_max_cost( dict );
-	
+	struct rlink_dictionary *ptr = get_dict( self );
+
+	int cost = dictionary_get_max_cost( ptr->dict );
+
 	debugMsg(( "Max cost is: %d", cost ));
-	
+
 	return INT2NUM( cost );
 }
 
@@ -219,7 +281,7 @@ rlink_get_max_cost( VALUE self ) {
  *  LinkParser::Sentence. If you specify an +options+ hash, its values will override
  *  those of the Dictionary's for the resulting Sentence.
  */
-static VALUE 
+static VALUE
 rlink_parse( int argc, VALUE *argv, VALUE self ) {
 	VALUE input_string, options, sentence;
 	VALUE args[2];
@@ -231,13 +293,13 @@ rlink_parse( int argc, VALUE *argv, VALUE self ) {
 	args[0] = input_string;
 	args[1] = self;
 	sentence = rb_class_new_instance( 2, args, rlink_cSentence );
-	
+
 	/* Now call #parse on it */
 	if ( i == 1 )
 		rb_funcall( sentence, rb_intern("parse"), 0, 0 );
 	else
 		rb_funcall( sentence, rb_intern("parse"), 1, options );
-	
+
 	return sentence;
 }
 
@@ -254,9 +316,13 @@ rlink_parse( int argc, VALUE *argv, VALUE self ) {
  */
 void
 rlink_init_dict() {
+#ifdef FOR_RDOC
+	rlink_mLinkParser = rb_define_module( "LinkParser" );
+#endif
+
 	rlink_cDictionary = rb_define_class_under( rlink_mLinkParser, "Dictionary",
 	 	rb_cObject );
-	
+
 	rb_define_alloc_func( rlink_cDictionary, rlink_dict_s_alloc );
 	rb_define_method( rlink_cDictionary, "initialize", rlink_dict_initialize, -1 );
 

data/ext/extconf.rb

@@ -2,54 +2,101 @@
 
 require 'mkmf'
 
-ADDITIONAL_LIBRARY_DIRS = %w[
-	/usr/local/lib
-	/opt/lib
-	/opt/local/lib
-]
-ADDITIONAL_INCLUDE_DIRS = %w[
-	/usr/local/include
-	/opt/include
-	/opt/local/include
-]
-
-$CFLAGS << ' -Wall' << ' -DDEBUG'
-
-def fail( *messages )
-	$stderr.puts( *messages )
-	exit 1
+if lgdir = with_config( 'link-grammar' )
+	ENV['PATH'] = "#{lgdir}/bin" + File::PATH_SEPARATOR + ENV['PATH']
+end
+
+### Read the output of a command using the fork+pipe syntax so execution errors 
+### propagate to Ruby.
+def read_cmd_output( *cmd )
+	output = IO.read( '|-' ) or exec( *cmd )
+	return output.chomp
+end
+
+pkgconfig = with_config( 'pkg-config' ) || 'pkg-config'
+pkgconfig = find_executable( pkgconfig ) or
+	fail "Couldn't find your pkg-config binary"
+
+$LDFLAGS << read_cmd_output( pkgconfig, '--libs-only-L', 'link-grammar' )
+$CFLAGS << read_cmd_output( pkgconfig, '--cflags', 'link-grammar' )
+
+# Sort out the universal vs. single-archicture build problems on MacOS X
+if RUBY_PLATFORM.include?( 'darwin' )
+	puts "MacOS X build: fixing architecture flags:"
+
+	# Only keep the '-arch <a>' flags present in both the cflags reported
+	# by pkg-config and those that Ruby specifies.
+	commonflags = nil
+	if ENV['ARCHFLAGS']
+		puts "  using the value in ARCHFLAGS environment variable (%p)." % [ ENV['ARCHFLAGS'] ]
+		commonflags = ENV['ARCHFLAGS']
+	elsif pkgconfig
+		puts "  finding flags common to both Ruby and link-grammar..."
+		archflags = []
+		pkgcflags = read_cmd_output( pkgconfig, '--cflags', 'link-grammar' )
+		pkgcflags.scan( /-arch\s+(\S+)/ ).each do |arch|
+			puts "  testing for architecture: %p" % [ arch ]
+			archflags << "-arch #{arch}" if Config::CONFIG['CFLAGS'].index("-arch #{arch}")
+		end
+
+		commonflags = archflags.join(' ')
+		puts "  common arch flags: %s" % [ commonflags ]
+	else
+		$stderr.puts %{
+		===========   WARNING   ===========
+		
+		You are building this extension on OS X without setting the 
+		ARCHFLAGS environment variable, and pkg-config wasn't found in 
+		your PATH. If you are seeing this message, that means that the 
+		build will probably fail.
+
+		If it does, you can correct this by either including the path 
+		to 'pkg-config' in your PATH or setting the environment variable 
+		ARCHFLAGS to '-arch <arch>' before building.
+
+		For example:
+		(in bash) $ export PATH=/opt/local/bin:$PATH                  
+		          $ export ARCHFLAGS='-arch x86_64'
+		(in tcsh) % set path = ( /opt/local/bin $PATH )
+		          % setenv ARCHFLAGS '-arch x86_64'
+
+		Then try building again.
+
+		===================================
+		}.gsub( /^\t+/, '  ' )
+	end
+
+	if commonflags
+		$CFLAGS.gsub!( /-arch\s+\S+ /, '' )
+		$LDFLAGS.gsub!( /-arch\s+\S+ /, '' )
+		CONFIG['LDSHARED'].gsub!( /-arch\s+\S+ /, '' )
+
+		$CFLAGS << ' ' << commonflags
+		$LDFLAGS << ' ' << commonflags
+		CONFIG['LDSHARED'] << ' ' << commonflags
+	end
 end
 
 
 dir_config( 'link-grammar' )
 
 
-find_library( "link-grammar", "dictionary_create", *ADDITIONAL_LIBRARY_DIRS ) or
+find_library( "link-grammar", "dictionary_create" ) or
 	fail( "Could not find Link-Grammar library", 
 	      "(http://www.abisource.com/projects/link-grammar/#download)." )
-find_header( "link-grammar/link-includes.h", *ADDITIONAL_INCLUDE_DIRS ) or
+find_header( "link-grammar/link-includes.h" ) or
 	fail( "Could not find link-includes.h" )
 
-unless have_func( "linkage_free_diagram", 'link-grammar/link-includes.h' )
+unless have_func( "linkage_is_fat", 'link-grammar/link-includes.h' )
 	message "Your link-grammar library is too old for this binding.\n",
 	        "Please upgrade to the latest version posted here:\n",
 	        "  http://www.abisource.com/projects/link-grammar/#download\n",
 	        "and try again.\n"
-	fail "No linkage_free_diagram() in the installed link-grammar."
-end
-
-unless have_func( "linkage_get_current_sublinkage" )
-	message "Link grammar library is unpatched. Building anyways.\n"
-	message "  LinkParser::Linkage#current_sublinkage will not work,\n"
-	message "  but everything else will work okay.\n"
-end
-
-unless have_func( "linkgrammar_get_version" )
-	message "You are building with an older, unsupported version of the link-grammar\n"
-	message " library. If you encounter problems, please update it to 4.4.0 or later.\n"
-	message "Trying to build anyway.\n"
+	fail "No linkage_is_fat() in the installed link-grammar."
 end
 
 have_func( 'parse_options_get_spell_guess' )
+have_func( 'linkage_get_disjunct_str' )
 
 create_makefile( 'linkparser_ext' )
+