WORDS now searches transient vocab only.
[forth.jl.git] / src / lib_7_decompiler.4th
diff --git a/src/lib_7_decompiler.4th b/src/lib_7_decompiler.4th
new file mode 100644 (file)
index 0000000..505f7da
--- /dev/null
@@ -0,0 +1,151 @@
+\ Decompilation
+
+: .NAME
+        DUP @           ( get the flags/length byte )
+        F_LENMASK AND   ( mask out the flags - just want the length )
+
+        BEGIN
+                DUP 0>          ( length > 0? )
+        WHILE
+                SWAP 1+         ( addr len -- len addr+1 )
+                DUP @           ( len addr -- len addr char | get the next character)
+                DUP 32 >= OVER 127 <= AND IF
+                        EMIT    ( len addr char -- len addr | and print it)
+                ELSE
+                        BASE @ SWAP HEX
+                        ." \x" 0 .R
+                        BASE !
+                THEN
+                SWAP 1-         ( len addr -- addr len-1    | subtract one from length )
+        REPEAT
+        2DROP           ( len addr -- )
+;
+
+: ?HIDDEN
+        1+              ( skip over the link pointer )
+        @               ( get the flags/length byte )
+        F_HIDDEN AND    ( mask the F_HIDDEN flag and return it (as a truth value) )
+;
+
+: ?IMMEDIATE
+        1+              ( skip over the link pointer )
+        @               ( get the flags/length byte )
+        F_IMMED AND     ( mask the F_IMMED flag and return it (as a truth value) )
+;
+
+: SEE
+        BL WORD FIND    ( find the dictionary entry to decompile )
+
+        CR
+
+        0= IF
+                ." Word '" COUNT TYPE ." ' not found in dictionary."
+                EXIT
+        THEN
+
+        >LINK
+
+        ( Now we search again, looking for the next word in the dictionary.  This gives us
+          the length of the word that we will be decompiling.  (Well, mostly it does). )
+        HERE          ( address of the end of the last compiled word )
+        LATEST @        ( word last curr )
+        BEGIN
+                2 PICK          ( word last curr word )
+                OVER            ( word last curr word curr )
+                <>              ( word last curr word<>curr? )
+        WHILE                   ( word last curr )
+                NIP             ( word curr )
+                DUP @           ( word curr prev (which becomes: word last curr) )
+        REPEAT
+
+        DROP            ( at this point, the stack is: start-of-word end-of-word )
+        SWAP            ( end-of-word start-of-word )
+
+        DUP LINK> @ CASE
+                DOCOL OF
+                        \ Colon definition
+                        [CHAR] : EMIT SPACE DUP 1+ .NAME SPACE
+                        DUP ?IMMEDIATE IF ." IMMEDIATE " THEN CR
+                ENDOF
+                DOVAR OF
+                        \ Variable definition
+                        ." Variable " DUP 1+ .NAME CR
+                        2DROP EXIT
+                ENDOF
+                DOCON OF
+                        \ Constant definition
+                        ." Constant " DUP 1+ .NAME CR
+                        2DROP EXIT
+                ENDOF
+
+                \ Unknown codeword
+                ." Primitive or word with unrecognized codeword." CR 
+                DROP 2DROP EXIT
+        ENDCASE
+
+        4 SPACES
+
+        LINK> >BODY            ( get the data address, ie. points after DOCOL | end-of-word start-of-data )
+
+        ( now we start decompiling until we hit the end of the word )
+        BEGIN           ( end start )
+                2DUP >
+        WHILE
+                DUP @           ( end start codeword )
+
+                CASE
+                ['] LIT OF                ( is it LIT ? )
+                        1+ DUP @                ( get next word which is the integer constant )
+                        .                       ( and print it )
+                ENDOF
+                ['] LITSTRING OF          ( is it LITSTRING ? )
+                        [CHAR] S EMIT [CHAR] " EMIT SPACE ( print S"<space> )
+                        1+ DUP @                ( get the length word )
+                        SWAP 1+ SWAP            ( end start+1 length )
+                        2DUP TYPE               ( print the string )
+                        [CHAR] " EMIT SPACE          ( finish the string with a final quote )
+                        +                       ( end start+1+len, aligned )
+                        1-                     ( because we're about to add 4 below )
+                ENDOF
+                ['] 0BRANCH OF            ( is it 0BRANCH ? )
+                        ." 0BRANCH ( "
+                        1+ DUP @               ( print the offset )
+                        .
+                        ." ) "
+                ENDOF
+                ['] BRANCH OF             ( is it BRANCH ? )
+                        ." BRANCH ( "
+                        1+ DUP @               ( print the offset )
+                        .
+                        ." ) "
+                ENDOF
+                ['] ['] OF                  ( is it ['] ? )
+                        ." ['] "
+                        1+ DUP @               ( get the next codeword )
+                        >NAME                    ( and force it to be printed as a dictionary entry )
+                        .NAME SPACE
+                ENDOF
+                ['] EXIT OF               ( is it EXIT? )
+                        ( We expect the last word to be EXIT, and if it is then we don't print it
+                          because EXIT is normally implied by ;.  EXIT can also appear in the middle
+                          of words, and then it needs to be printed. )
+                        2DUP                    ( end start end start )
+                        1+                     ( end start end start+1 )
+                        <> IF                   ( end start | we're not at the end )
+                                ." EXIT "
+                        THEN
+                ENDOF
+                                        ( default case: )
+                        DUP                     ( in the default case we always need to DUP before using )
+                        >NAME                    ( look up the codeword to get the dictionary entry )
+                        .NAME SPACE               ( and print it )
+                ENDCASE
+
+                1+             ( end start+1 )
+        REPEAT
+
+        [CHAR] ; EMIT CR
+
+        2DROP           ( restore stack )
+;
+