zasm: read each argument immediately insto an argspec

Previously, we would go the other way around: match all available
argspecs to raw argument strings. It's much better to go the other way
around because we can "reduce" each argument much sooner. That will be
important when we start to support constants and expressions.

apps/zasm/zasm.asm

@@ -86,6 +86,22 @@ toWord:
 	ret
 
 
+; Read arg from (HL) into argspec at (DE)
+; HL is advanced to the next word. Z is set if there's a next word.
+readArg:
+	push	de
+	ld	de, tmpVal
+	call	readWord
+	push	hl
+	ld	hl, tmpVal
+	call	matchArg
+	pop	hl
+	pop	de
+	ld	(de), a
+	call	toWord
+	ret
+
+; Read line from (HL) into (curWord), (curArg1) and (curArg2)
 readLine:
 	push	de
 	xor	a
@@ -97,71 +113,91 @@ readLine:
 	call	toWord
 	jr	nz, .end
 	ld	de, curArg1
-	call	readWord
-	call	toWord
+	call	readArg
 	jr	nz, .end
 	ld	de, curArg2
-	call	readWord
+	call	readArg
 .end:
 	pop	de
 	ret
 
-; match argument string at (HL) with argspec A.
-; Set Z/NZ on match
+; Returns length of string at (HL) in A.
+strlen:
+	push	bc
+	push	hl
+	ld	bc, 0
+	ld	a, 0	; look for null char
+.loop:
+	cpi
+	jp	z, .found
+	jr	.loop
+.found:
+	; How many char do we have? the (NEG BC)-1, which started at 0 and
+	; decreased at each CPI call. In this routine, we stay in the 8-bit
+	; realm, so C only.
+	ld	a, c
+	neg
+	dec	a
+	pop	hl
+	pop	bc
+	ret
+
+; find argspec for string at (HL). Returns matching argspec in A.
+; Return value 1 holds a special meaning: arg is not empty, but doesn't match
+; any argspec (A == 0 means arg is empty). A return value of 1 means an error.
 matchArg:
+	call	strlen
+	cp	0
+	ret	z		; empty string? A already has our result: 0
+
 	push	bc
 	push	de
-	push	ix
+	push	hl
+
+	cp	1
+	jr	z, .matchsingle	; Arg is one char? We have a "single" type.
 
-	cp	0
-	jr	z, .matchnone
-	; Let's see if our argspec is a "single" one.
-	ex	hl, de		; For "simple" cmp, we don't need HL. But we'll
-				; need it later.
-	ld	hl, argspecsSingle
-	ld	bc, ARGSPEC_SINGLE_CNT
-.loop1:
-	cpi
-	jr	z, .matchsingle		; our argspec in the "single" list
-	jp	po, .loop1end
-	jr	.loop1
-.loop1end:
 	; Not a "single" arg. Do the real thing then.
-	ex	hl, de		; now we need HL back...
 	ld	de, argspecTbl
+	; DE now points the the "argspec char" part of the entry, but what
+	; we're comparing in the loop is the string next to it. Let's offset
+	; DE by one so that the loop goes through strings.
+	inc	de
 	ld	b, ARGSPEC_TBL_CNT
-.loop2:
-	ld	ixl, a
-	ld	a, (de)
-	cp	ixl
+.loop1:
+	ld	a, 4
+	call	JUMP_STRNCMP
 	jr	z, .found		; got it!
 	ld	a, 5
 	call	JUMP_ADDDE
-	ld	a, ixl
-	djnz	.loop2
+	djnz	.loop1
 	; exhausted? we have a problem os specifying a wrong argspec. This is
 	; an internal consistency error.
+	ld	a, 1
 	jr	.end
 .found:
-	; found the matching argspec row. Let's compare the strings now.
-	inc	de	; the string starts on the 2nd byte of the row
-	ld	a, 4
-	call	JUMP_STRNCMP	; Z is set
+	; found the matching argspec row. Our result is one byte left of DE.
+	dec	de
+	ld	a, (de)
 	jr	.end
 
 .matchsingle:
-	; single match is easy: compare A with (HL). They must be equal.
-	ex	hl, de
-	ld	b, a
 	ld	a, (hl)
-	cp	b	; Z set if A == B
+	ld	hl, argspecsSingle
+	ld	bc, ARGSPEC_SINGLE_CNT
+.loop2:
+	cpi
+	jr	z, .end		; found! our result is already in A. go straight
+				; to end.
+	jp	po, .loop2notfound
+	jr	.loop2
+.loop2notfound:
+	; something's wrong. error
+	ld	a, 1
 	jr	.end
 
-.matchnone:
-	ld	a, (hl)
-	cp	0	; arg must be null to match
 .end:
-	pop	ix
+	pop	hl
 	pop	de
 	pop	bc
 	ret
@@ -178,13 +214,11 @@ matchPrimaryRow:
 	; name matches, let's see the rest
 	ld	ixh, d
 	ld	ixl, e
-	ld	hl, curArg1
-	ld	a, (ix+4)
-	call	matchArg
+	ld	a, (curArg1)
+	cp	(ix+4)
 	jr	nz, .end
-	ld	hl, curArg2
-	ld	a, (ix+5)
-	call	matchArg
+	ld	a, (curArg2)
+	cp	(ix+5)
 .end:
 	pop	ix
 	pop	hl
@@ -243,30 +277,8 @@ argspecTbl:
 	.db	'p', "(SP)"
 
 ; This is a list of primary instructions (single upcode) that lead to a
-; constant (no group code to insert).
-; That doesn't mean that they don't take any argument though. For example,
-; "DEC IX" leads to a special upcode. These kind of constants are indicated
-; as a single byte to save space. Meaning:
+; constant (no group code to insert). Format:
 ;
-; All single char registers (A/B/C etc) -> themselves
-; HL -> h
-; (HL) -> l
-; DE -> d
-; (DE) -> e
-; BC -> b
-; (BC) -> c
-; IX -> X
-; (IX) -> x
-; IY -> Y
-; (IY) -> y
-; AF -> a
-; AF' -> f
-; SP -> s
-; (SP) -> p
-; None -> 0
-;
-; This is a sorted list of "primary" (single byte) instructions along with
-; metadata
 ; 4 bytes for the name (fill with zero)
 ; 1 byte for arg constant
 ; 1 byte for 2nd arg constant
@@ -303,8 +315,15 @@ instTBlPrimary:
 ; enough space for 4 chars and a null
 curWord:
 	.db	0, 0, 0, 0, 0
+
+; Args are 3 bytes: argspec, then values of numerical constants (when that's
+; appropriate)
 curArg1:
-	.db	0, 0, 0, 0, 0
+	.db	0, 0, 0
 curArg2:
+	.db	0, 0, 0
+
+; space for tmp stuff
+tmpVal:
 	.db	0, 0, 0, 0, 0