add spell checking

src/dsymbol.c

@@ -1,6 +1,6 @@
 
 // Compiler implementation of the D programming language
-// Copyright (c) 1999-2009 by Digital Mars
+// Copyright (c) 1999-2010 by Digital Mars
 // All Rights Reserved
 // written by Walter Bright
 // http://www.digitalmars.com
@@ -13,6 +13,7 @@
 #include <assert.h>
 
 #include "rmem.h"
+#include "speller.h"
 
 #include "mars.h"
 #include "dsymbol.h"
@@ -332,6 +333,27 @@ Dsymbol *Dsymbol::search(Loc loc, Identifier *ident, int flags)
     return NULL;
 }
 
+/***************************************************
+ * Search for symbol with correct spelling.
+ */
+
+void *symbol_search_fp(void *arg, const char *seed)
+{
+    Dsymbol *s = (Dsymbol *)arg;
+    Identifier id(seed, 0);
+    Module::clearCache();
+    s = s->search(0, &id, 4|2);
+    return s;
+}
+
+Dsymbol *Dsymbol::search_correct(Identifier *ident)
+{
+    if (global.gag)
+	return NULL;		// don't do it for speculative compiles; too time consuming
+
+    return (Dsymbol *)speller(ident->toChars(), &symbol_search_fp, this, idchars);
+}
+
 /***************************************
  * Search for identifier id as a member of 'this'.
  * id may be a template instance.

src/dsymbol.h

@@ -1,6 +1,6 @@
 
 // Compiler implementation of the D programming language
-// Copyright (c) 1999-2009 by Digital Mars
+// Copyright (c) 1999-2010 by Digital Mars
 // All Rights Reserved
 // written by Walter Bright
 // http://www.digitalmars.com
@@ -147,6 +147,7 @@ struct Dsymbol : Object
     virtual void semantic3(Scope *sc);
     virtual void inlineScan();
     virtual Dsymbol *search(Loc loc, Identifier *ident, int flags);
+    Dsymbol *search_correct(Identifier *id);
     Dsymbol *searchX(Loc loc, Scope *sc, Identifier *id);
     virtual int overloadInsert(Dsymbol *s);
 #ifdef _DH

src/expression.c

@@ -2091,7 +2091,21 @@ Expression *IdentifierExp::semantic(Scope *sc)
 	}
 	return e->semantic(sc);
     }
-    error("undefined identifier %s", ident->toChars());
+#if DMDV2
+    if (ident == Id::ctfe)
+    {  // Create the magic __ctfe bool variable
+       VarDeclaration *vd = new VarDeclaration(loc, Type::tbool, Id::ctfe, NULL);
+       Expression *e = new VarExp(loc, vd);
+       e = e->semantic(sc);
+       return e;
+    }
+#endif
+
+    s = sc->search_correct(ident);
+    if (s)
+	error("undefined identifier %s, did you mean %s %s?", ident->toChars(), s->kind(), s->toChars());
+    else
+	error("undefined identifier %s", ident->toChars());
     type = Type::terror;
     return this;
 }

src/module.c

@@ -864,6 +864,7 @@ void Module::gensymfile()
 
 int Module::needModuleInfo()
 {
+    //printf("needModuleInfo() %s, %d, %d\n", toChars(), needmoduleinfo, global.params.cov);
     return needmoduleinfo || global.params.cov;
 }
 
@@ -902,6 +903,13 @@ Dsymbol *Module::symtabInsert(Dsymbol *s)
     return Package::symtabInsert(s);
 }
 
+void Module::clearCache()
+{
+    for (int i = 0; i < amodules.dim; i++)
+    {	Module *m = (Module *)amodules.data[i];
+	m->searchCacheIdent = NULL;
+    }
+}
 
 /*******************************************
  * Can't run semantic on s now, try again later.

src/module.h

@@ -141,6 +141,7 @@ struct Module : Package
     void deleteObjFile();
     void addDeferredSemantic(Dsymbol *s);
     static void runDeferredSemantic();
+    static void clearCache();
     int imports(Module *m);
 
     // Back end

src/mtype.c

@@ -592,7 +592,7 @@ Expression *Type::getProperty(Loc loc, Identifier *ident)
     else if (ident == Id::size)
     {
 	error(loc, ".size property should be replaced with .sizeof");
-	e = new IntegerExp(loc, size(loc), Type::tsize_t);
+	e = new ErrorExp();
     }
     else if (ident == Id::alignof)
     {
@@ -630,8 +630,16 @@ Expression *Type::getProperty(Loc loc, Identifier *ident)
     }
     else
     {
-	error(loc, "no property '%s' for type '%s'", ident->toChars(), toChars());
-	e = new IntegerExp(loc, 1, Type::tint32);
+	Dsymbol *s = NULL;
+	if (ty == Tstruct || ty == Tclass || ty == Tenum || ty == Ttypedef)
+	    s = toDsymbol(NULL);
+	if (s)
+	    s = s->search_correct(ident);
+	if (s)
+	    error(loc, "no property '%s' for type '%s', did you mean '%s'?", ident->toChars(), toChars(), s->toChars());
+	else
+	    error(loc, "no property '%s' for type '%s'", ident->toChars(), toChars());
+	e = new ErrorExp();
     }
     return e;
 }

src/root/speller.c

@@ -0,0 +1,98 @@
+
+#include <stdio.h>
+#include <string.h>
+#include <stdlib.h>
+
+#include "speller.h"
+
+const char idchars[] = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789_";
+
+/**************************************************
+ * Looks for correct spelling.
+ * Currently only looks a 'distance' of one from the seed[].
+ * This does an exhaustive search, so can potentially be very slow.
+ * Input:
+ *	seed		wrongly spelled word
+ *	fp		search function
+ *	fparg		argument to search function
+ *	charset		character set
+ * Returns:
+ *	NULL		no correct spellings found
+ *	void*		value returned by fp() for first possible correct spelling
+ */
+
+void *speller(const char *seed, fp_speller_t fp, void *fparg, const char *charset)
+{
+    size_t seedlen = strlen(seed);
+    if (!seedlen)
+	return NULL;
+
+    char *buf = (char *)alloca(seedlen + 2);	// leave space for extra char
+    if (!buf)
+	return NULL;				// no matches
+
+    /* Deletions */
+    memcpy(buf, seed + 1, seedlen);
+    for (int i = 0; i < seedlen; i++)
+    {
+	//printf("del buf = '%s'\n", buf);
+	void *p = (*fp)(fparg, buf);
+	if (p)
+	    return p;
+
+	buf[i] = seed[i];
+    }
+
+    /* Transpositions */
+    memcpy(buf, seed, seedlen + 1);
+    for (int i = 0; i + 1 < seedlen; i++)
+    {
+	// swap [i] and [i + 1]
+	buf[i] = seed[i + 1];
+	buf[i + 1] = seed[i];
+
+	//printf("tra buf = '%s'\n", buf);
+	void *p = (*fp)(fparg, buf);
+	if (p)
+	    return p;
+
+	buf[i] = seed[i];
+    }
+
+    if (charset && *charset)
+    {
+	/* Substitutions */
+	memcpy(buf, seed, seedlen + 1);
+	for (int i = 0; i < seedlen; i++)
+	{
+	    for (const char *s = charset; *s; s++)
+	    {
+		buf[i] = *s;
+
+		//printf("sub buf = '%s'\n", buf);
+		void *p = (*fp)(fparg, buf);
+		if (p)
+		    return p;
+	    }
+	    buf[i] = seed[i];
+	}
+
+	/* Insertions */
+	memcpy(buf + 1, seed, seedlen + 1);
+	for (int i = 0; i <= seedlen; i++)	// yes, do seedlen+1 iterations
+	{
+	    for (const char *s = charset; *s; s++)
+	    {
+		buf[i] = *s;
+
+		//printf("ins buf = '%s'\n", buf);
+		void *p = (*fp)(fparg, buf);
+		if (p)
+		    return p;
+	    }
+	    buf[i] = seed[i];	// going past end of seed[] is ok, as we hit the 0
+	}
+    }
+
+    return NULL;		// didn't find any corrections
+}

src/root/speller.h

@@ -0,0 +1,7 @@
+
+typedef void *(fp_speller_t)(void *, const char *);
+
+extern const char idchars[];
+
+void *speller(const char *seed, fp_speller_t fp, void *fparg, const char *charset);
+

src/scope.c

@@ -1,5 +1,5 @@
 
-// Copyright (c) 1999-2005 by Digital Mars
+// Copyright (c) 1999-2010 by Digital Mars
 // All Rights Reserved
 // written by Walter Bright
 // http://www.digitalmars.com
@@ -11,6 +11,7 @@
 #include <assert.h>
 
 #include "root.h"
+#include "speller.h"
 
 #include "mars.h"
 #include "init.h"
@@ -359,3 +360,27 @@ void Scope::setNoFree()
 	    //assert(0);
     }
 }
+
+
+/************************************************
+ * Given the failed search attempt, try to find
+ * one with a close spelling.
+ */
+
+void *scope_search_fp(void *arg, const char *seed)
+{
+    //printf("scope_search_fp('%s')\n", seed);
+    Scope *sc = (Scope *)arg;
+    Identifier id(seed, 0);
+    Module::clearCache();
+    Dsymbol *s = sc->search(0, &id, NULL);
+    return s;
+}
+
+Dsymbol *Scope::search_correct(Identifier *ident)
+{
+    if (global.gag)
+	return NULL;		// don't do it for speculative compiles; too time consuming
+
+    return (Dsymbol *)speller(ident->toChars(), &scope_search_fp, this, idchars);
+}

src/scope.h

@@ -110,6 +110,7 @@ struct Scope
     void mergeCallSuper(Loc loc, unsigned cs);
 
     Dsymbol *search(Loc loc, Identifier *ident, Dsymbol **pscopesym);
+    Dsymbol *search_correct(Identifier *ident);
     Dsymbol *insert(Dsymbol *s);
 
     ClassDeclaration *getClassScope();

src/template.c

@@ -3659,7 +3659,12 @@ TemplateDeclaration *TemplateInstance::findTemplateDeclaration(Scope *sc)
 	id = name;
 	s = sc->search(loc, id, &scopesym);
 	if (!s)
-	{   error("template '%s' is not defined", id->toChars());
+	{
+	    s = sc->search_correct(id);
+	    if (s)
+		error("template '%s' is not defined, did you mean %s?", id->toChars(), s->toChars());
+	    else
+		error("template '%s' is not defined", id->toChars());
 	    return NULL;
 	}
 #if LOG

src/win32.mak

@@ -96,7 +96,7 @@ OBJ8= go.obj gdag.obj gother.obj gflow.obj gloop.obj var.obj el.obj \
 # from ROOT
 
 ROOTOBJS= lstring.obj array.obj gnuc.obj man.obj rmem.obj port.obj root.obj \
-	stringtable.obj dchar.obj response.obj async.obj
+	stringtable.obj dchar.obj response.obj async.obj speller.obj
 
 OBJS= $(OBJ1) $(OBJ8) $(ROOTOBJS)
 
@@ -149,7 +149,8 @@ ROOTSRC= $(ROOT)\dchar.h $(ROOT)\dchar.c $(ROOT)\lstring.h \
 	$(ROOT)\rmem.h $(ROOT)\rmem.c $(ROOT)\port.h \
 	$(ROOT)\stringtable.h $(ROOT)\stringtable.c \
 	$(ROOT)\gnuc.h $(ROOT)\gnuc.c $(ROOT)\man.c $(ROOT)\port.c \
-	$(ROOT)\response.c $(ROOT)\async.h $(ROOT)\async.c
+	$(ROOT)\response.c $(ROOT)\async.h $(ROOT)\async.c \
+	$(ROOT)\speller.h $(ROOT)\speller.c
 
 MAKEFILES=win32.mak linux.mak osx.mak freebsd.mak solaris.mak
 
@@ -420,6 +421,9 @@ root.obj : $(ROOT)\root.c
 response.obj : $(ROOT)\response.c
 	$(CC) -c $(CFLAGS) $(ROOT)\response.c
 
+speller.obj : $(ROOT)\speller.h $(ROOT)\speller.c
+	$(CC) -c $(CFLAGS) $(ROOT)\speller.c
+
 stringtable.obj : $(ROOT)\stringtable.c
 	$(CC) -c $(CFLAGS) $(ROOT)\stringtable.c