T32 per: fix tokenizer, repeat/copy impl, parse tree types

t32per/perparse.py

@@ -0,0 +1,90 @@
+
+from typing import *
+
+from perrepeat import *
+from pertoken import *
+
+
+T32GroupAttr = Literal['h', 'r', 's', 'w']
+T32FieldAttr = Literal['a','r','hex','event','setclr']
+T32MaskAttr = Literal['dec','hex','rhex']
+T32TreeVisibility = Literal['open','closed']
+
+
+class T32Field(NamedTuple):
+    attr: T32FieldAttr
+    qual: PerTokenQual
+    thing: int
+    bitrange: Tuple[int, int] # inclusive
+    name: str
+    description: str
+    values: Sequence[str]
+
+
+class T32Mask(NamedTuple):
+    attr: T32MaskAttr
+    qual: Tuple[PerTokenQual, PerTokenQual]
+    thing: int
+    bitrange: Tuple[int, int] # inclusive
+    mask: int
+    name: str
+    description: str
+    values: Sequence[str]
+
+
+class T32Line(NamedTuple):
+    hidden: bool
+    qual: PerTokenQual
+    offset: int
+    name: str
+    description: str
+    field:s Sequence[Union[T32Field, T32Mask]]
+
+
+class T32Group(NamedTuple):
+    attrs: T32GroupAttr
+    qual: PerTokenQual
+    address: Union[T32Address, T32AddressRange]
+    description: str
+    regs: Sequence[T32Line]
+    copy: bool = False
+
+
+class T32GroupRepeat(NamedTuple):
+    repeat: T32Repeat
+    width: int
+    groups: Sequence[T32Group]
+class T32GroupRepeatReplay(NamedTuple):
+    pass
+
+
+class T32Tree(NamedTuple):
+    base: T32Address
+    width: int
+    visibility: T32TreeVisibility
+    groups: Sequence[Union[T32Group, T32GroupRepeat, T32GroupRepeatReplay]]
+    subtrees: Sequence[Union[T32TreeRepeat, T32Tree]]
+
+
+class T32TreeRepeat(NamedTuple):
+    repeat: T32Repeat
+    subtrees: Sequence[Union[T32TreeRepeat, T32Tree]]
+
+
+class T32Per(NamedTuple):
+    title: str
+    props: str
+    author: str
+    changelog: Sequence[str]
+    manufacturer: str
+    doc: str
+    core: str
+    chip: Sequence[str]
+    copyright: str
+    description: str
+    keywords: str
+    date: str
+    id: str
+    config: Tuple[int, int]
+    trees: Union[T32Tree, T32TreeRepeat]
+

t32per/perrepeat.py

@@ -0,0 +1,128 @@
+
+from typing import *
+
+
+class T32SpacedAddress(NamedTuple):
+    addrspace: str
+    address: int
+
+    def __add__(self, other):
+        if isinstance(other, int):
+            return T32SpacedAddress(self.addrspace, self.address+other)
+        elif isinstance(other, T32SpacedAddress):
+            assert self.addrspace == other.addrspace
+            return T32SpacedAddress(self.addrspace, self.address+other.address)
+        else:
+            raise TypeError("cannot add %s to T32SpacedAddress"%type(other))
+
+    def __repr__(self):
+        return "T32SpacedAddress(%s, %s)"%(repr(self.addrspace),hex(self.address))
+    def __str__(self):
+        return "%s:%s"%(self.addrspace,hex(self.address))
+
+
+T32Address = Union[str, int, T32SpacedAddress]
+T32RepeatVarType = Literal['strings','list','increment']
+
+
+class T32RepeatVar(NamedTuple):
+    type: T32RepeatVarType
+    args: Sequence[T32Address]
+
+
+class T32AddressRange(NamedTuple):
+    base: T32Address
+    size: int
+
+
+class T32Repeat(NamedTuple):
+    count: int
+    vars: Sequence[T32RepeatVar]
+
+    def get_value(self, repind: int, varind: int):
+        assert repind >= 0 and repind < self.count, "%d vs %s"%(repind,self)
+        assert varind >= 0 and varind < len(self.vars), "%d vs %s"%(varind,self)
+        v = self.vars[varind]
+        if v.type == 'increment':
+            vvv = v.args[0] + v.args[1]*repind
+            #print("repind %s -> %s"%(repind,vvv), self)
+            return v.args[0] + v.args[1]*repind
+        else:
+            #print("repind %s -> %s"%(repind,v.args[repind]), self)
+            return v.args[repind]
+
+    def eval(self, repind: int, s: str) -> str:
+        if len(self.vars) == 0: return s
+        if "$" not in s: return s
+
+        for i in range(self.count):
+            t = "$%d" % (i+1)
+            if t in s:
+                s = s.replace(t, str(self.get_value(repind, i)))
+        return s
+
+
+
+def parse_spacedAddress(s: str) -> T32SpacedAddress:
+    sp = s.split(':')
+    assert len(sp) == 2
+    return T32SpacedAddress(sp[0], parse_int(sp[1], False))
+
+
+def parse_int(s: str, spaced: bool = True):
+    assert s is not None and len(s) > 0, "'%s'"%s
+    if ':' in s:
+        return parse_spacedAddress(s)
+    elif s[-1] == '.':
+        return int(s[:-1])
+    elif s.startswith('0x') or s.startswith('0X'):
+        return int(s[2:], 16)
+    else:
+        return int(s, 0) # shrug
+
+
+def parse_repeatvar(s: str) -> T32RepeatVar:
+    assert (s[0] == '(' and s[-1] == ')'), "badly formatted repeatvar: %s"%s
+    s = s[1:-1]
+    sl = s.lower()
+
+    typ = None
+    rest = None
+    if sl.startswith('strings '):
+        typ = 'strings'
+        rest = []
+        start = len(typ)+1
+        instr = False
+        for i in range(start, len(s)):
+            if s[i] == ' ' and not instr:
+                continue
+            if s[i] == '"':
+                if not instr:
+                    instr = True
+                    start = i+1
+                else:
+                    rest.append(s[start:i])
+                    instr = False
+    elif sl.startswith('list '):
+        typ = 'list'
+        rest = sl[len(typ)+1:].strip().split()
+    elif sl.startswith('increment '):
+        typ = 'increment'
+        rest = sl[len(typ)+1:].strip().split()
+        assert len(rest) == 2, "bad increment repeatvar args: %s"%s
+        rest = [parse_int(x) for x in rest]
+    else:
+        assert False, "bad repeatvar type: %s"%s
+
+    return T32RepeatVar(typ, rest)
+
+
+def parse_repeat(args: Sequence[str]) -> T32Repeat:
+    assert len(args) >= 1
+    count = parse_int(args[0], False)
+    vars = tuple(parse_repeatvar(x) for x in args[1:])
+    for v in vars:
+        if v.type != 'increment':
+            assert len(v.args) == count
+    return T32Repeat(count, vars)
+

t32per/pertoken.py

@@ -3,6 +3,8 @@
 from enum import Enum
 from typing import *
 
+from perrepeat import *
+
 class PerTokenName(Enum):
     Title = "; @Title"
     Props = "; @Props"
@@ -18,16 +20,16 @@ class PerTokenName(Enum):
     Keywords = "; @Keywords"
     Date = "; @Date"
     #HeaderSep = "; -------"
-    Id = "; @Id"
+    Id = "; $Id"
 
     CONFIG = "config" # what is this
     BASE = "base" # base address, base addrspace:expr
     # expr can be int or (d.l(addrspace:off)) -> ???
     WIDTH = "width" # what is this
-    SAVEINDEX = "saveindex" # what is this
+    #SAVEINDEX = "saveindex" # what is this
 
     TREE_OPEN = "tree.open"
-    TREE_CLOSE = "tree.close" # like tree.open
+    TREE_CLOSE = "tree.close" # like tree.open, determines visibility
     TREE_END = "tree.end"
     TREE = "tree"
 
@@ -51,7 +53,7 @@ class PerTokenName(Enum):
     RBITFLD = "rbitfld"
     HEXFLD = "hexfld"
     #RHEXFLD? WHEXFLD??
-    EVENTFLD = "eventfld"
+    EVENTFLD = "eventfld" # == read-clear?
     SETCLRFLD = "setclrfld"
     # masks in registers?
     # hexmask.qual IDK start--end MASK "CODE,description"
@@ -71,8 +73,14 @@ class PerTokenName(Enum):
     # IDK: entry, read, wait, saveindex, saveout, set, getx, register
     # EH: include, endian.{be,le} (-> conditional endianness)
     # TODO: copy: copy from previous group? (->derivedFrom whee)
+    # copy: copy all from previous group
+    # (tree+group define peripheral, can be multiple groups per tree tho. 'base' part of tree, optional)
     # TODO: repeat{,.end}: repeat macro stuff (cf. dimIndexElement)
+    # repeat <repno> <list1> <list2> <...>
+    # stuff -> $1,$2 are variables
+    # repeat.end
     # TODO: repeat.replay: copy+replay
+    # repeat.replay: copy previous repeat
 
 PER_TOKEN_IGNORE = {
     'assert','autoindent.on','autoindent.off','button','elif','else','endif',
@@ -122,49 +130,110 @@ class PerToken(NamedTuple):
     args: List[str] = []
 
 
-# TODO: tokenize into data stream with useful arguments
 # TODO: data stream -> tree structure
 
-def find_tok(l: str, hdr: bool) -> PerTokenName:
+def find_tok(l: str, hdr: bool) -> Union[PerTokenName, Sequence[PerTokenQual]]:
     ll=l.lower() if hdr else l
     for h in (PER_TOKEN_HEADER if hdr else PER_TOKEN_BODY):
         #print("ll='%s', h='%s'"%(ll,h.value))
         if ll.startswith(h.value.lower() if hdr else h.value):
-            return h
+            if hdr: return h
+            rest = ll[len(h.value):]
+            if len(rest) == 0 or rest[0] == ' ':
+                return (h, None)
+            elif rest[0] == '.':
+                # quals follow
+                quals = []
+                cqualstart = 1
+                for i in range(1, len(rest)):
+                    if rest[i] == ' ':
+                        quals.append(rest[cqualstart:i].upper())
+                        break
+                    elif rest[i] == '.':
+                        if i != cqualstart + 1:
+                            quals.append(rest[cqualstart:i].upper())
+                        cqualstart = i+1
+                qs = tuple(PerTokenQual._member_map_[k] for k in quals)
+                return (h, qs[0] if len(qs) == 1 else qs)
+            else:
+                continue # not a match
     assert not hdr or l.split()[0] in PER_TOKEN_IGNORE, "Unknown token on line: %s"%l
     return None
 
 
+def split_str(s: str) -> List[str]:
+    r = []
+    start = 0
+    instr = False
+    inparen = 0
+    for i in range(len(s)):
+        if s[i] == '"':
+            if not instr and i > 0 and s[i-1] == '"' and start != i-1:
+                # sigh...
+                ss = s[start:i].strip()
+                if len(ss) > 0: r.append(ss)
+                start = i
+            instr = not instr
+        elif s[i] == '(' and not instr:
+            if inparen == 0 and start != i-1:
+                ss = s[start:i].strip()
+                if len(ss) > 0: r.append(ss)
+                start = i
+            inparen += 1
+        elif s[i] == ')' and not instr:
+            assert inparen > 0
+            inparen -= 1
+        elif s[i] == ' ' and start != i-1 and not instr and inparen == 0:
+            ss = s[start:i].strip()
+            if len(ss) > 0: r.append(ss)
+            start = i
+    if start < len(s):
+        ss = s[start:].strip()
+        if len(ss) > 0: r.append(ss)
+    return r
+
 def tokenize_body(f, l=None):
     prevtell = -1
+    emptycount = 0
+    #prevl = None
     while True:
         if l is None: l = f.readline().strip()
         if len(l) == 0:
             tell = f.tell()
-            if tell == prevtell: break # EOF
+            #print("empty, prevl:",prevl)
+            if tell == prevtell:
+                emptycount += 1
+                if emptycount == 3:
+                    break # EOF
+            else:
+                emptycount = 0
             prevtell = tell
+            l = None
             continue
+        #prevl = l
         ll = l.lower()
-        sp = l.split()
+        sp = split_str(l)#l.split()
         # regular token
         t = find_tok(ll, False)
         #print("t",t)
         if t is not None:
-            yield PerToken(t, args=sp[1:])
+            #print("regular token", t)
+            tt, qq = t
+            yield PerToken(tt, qual=qq, args=sp[1:])
             l = None
             continue
 
-        ht = sp[0].lower().split('.')
-        assert len(ht) > 0 or sp[0].lower() in PER_TOKEN_IGNORE, "Unknown token on line: %s"%l
-        t = find_tok(ht[0], False)
-        if t is not None:
-            assert len(ht) in {2,3},"bad qual %s in line %s"%(repr(ht),l)
-            quals = [PerTokenQual.__members__[k] for k in ht[1:]]
-            if len(quals) == 1: quals = quals[0]
-            else: quals = tuple(quals)
-            yield PerToken(t, qual=quals, args=sp[1:])
-            l = None
-            continue
+        #ht = sp[0].lower().split('.')
+        #assert len(ht) > 0 or sp[0].lower() in PER_TOKEN_IGNORE, "Unknown token on line: %s"%l
+        #t = find_tok(ht[0], False)
+        #if t is not None:
+        #    assert len(ht) in {2,3},"bad qual %s in line %s"%(repr(ht),l)
+        #    quals = [PerTokenQual.__members__[k] for k in ht[1:]]
+        #    if len(quals) == 1: quals = quals[0]
+        #    else: quals = tuple(quals)
+        #    yield PerToken(t, qual=quals, args=sp[1:])
+        #    l = None
+        #    continue
 
         l = None
 
@@ -182,11 +251,12 @@ def tokenize(f):
             prevtell = tell
             continue
         if l[0] != ';':
+            yield PerToken(curtok, args=curlines)
             yield from tokenize_body(f,l)
             break
         if l.startswith('; -------'): continue
 
-        if l.startswith('; @') and l[3] != ' ':
+        if (l.startswith('; @') or l.startswith('; $')) and l[3] != ' ':
             # new token!
             # flush old one
             if curtok is not None:
@@ -198,21 +268,100 @@ def tokenize(f):
             curlines.append(l[3:].strip())
 
 
+def repeat_apply(r: T32Repeat, l: List[PerToken]):
+    for i in range(r.count):
+        for tok in l:
+            #print("==>",tok,r)
+            if tok.args is None or len(tok.args) == 0 or len(r.vars) == 0 or \
+                    all("$" not in a for a in tok.args):
+                yield tok
+            else:
+                yield PerToken(tok.name, tok.qual, [r.eval(i,x) for x in tok.args])
+
+
 # flatten copy and repeat statements
-def tokenize_flatten(itor):
-    yield from itor # TODO
+# actually no let's not, we can turn this into derivedFrom statements, which is probably better
+# eh, we can leave it in for easier processing earlier on
+def tokenize_flatten_repeat(itor: Iterable[PerToken]):
+    currep = None
+    currvar = None
+    lastrvar = None
+    for tok in itor:
+        if tok.name == PerTokenName.REPEAT:
+            assert currvar is None
+            currvar = parse_repeat(tok.args)
+            currep = []
+        elif tok.name == PerTokenName.REPEAT_END:
+            yield from repeat_apply(currvar, currep)
+            lastrvar = currvar
+            currvar = None
+        elif tok.name == PerTokenName.REPEAT_REPLAY:
+            assert lastrvar is not None
+            yield from repeat_apply(lastrvar, currep)
+        elif currvar is not None:
+            currep.append(tok)
+        else:
+            yield tok
+def tokenize_flatten_copy(itor: Iterable[PerToken]):
+    curgrp = None
+    curgcmd = None
+    lastgrp = None
+    lastgcmd = None
+    in_grp = False
+
+    for tok in itor:
+        if tok.name in {PerTokenName.GROUP,PerTokenName.HGROUP,
+                        PerTokenName.RGROUP,PerTokenName.SGROUP,
+                        PerTokenName.WGROUP}:
+            lastgrp = curgrp
+            lastgcmd = curgcmd
+            curgrp = []
+            curgcmd = tok
+            in_grp = True
+            yield tok
+        elif tok.name == PerTokenName.TREE_END:
+            # group must end at tree end
+            in_grp = False
+            yield tok
+        elif tok.name == PerTokenName.COPY:
+            assert lastgrp is not None
+            yield from lastgrp
+            # be able to do multiple copies
+            in_grp = False
+            curgrp = lastgrp
+        elif in_grp:
+            curgrp.append(tok)
+        else:
+            yield tok
+
+
+
+def tokenize_flatten(itor: Iterable[PerToken]):
+    yield from tokenize_flatten_repeat(tokenize_flatten_copy(itor))
 
 
 if __name__ == '__main__':
-    import glob
-    for p in glob.glob('t/*/*.per'):
-        print(p)
-        with open(p,'r') as f:
-            for x in tokenize(f): pass
-
-    with open('t/arm/peram65xx.per','r') as f:
-        for x in tokenize(f): print(x)
-    with open('t/arm/perfm0p.per','r') as f:
-        for x in tokenize(f): print(x)
+    if False:
+        import glob
+        for p in glob.glob('t/*/*.per'):
+            print(p)
+            with open(p,'r') as f:
+                for x in tokenize(f): pass
+
+    #with open('t/arm/perfm0p.per','r') as f: # nothing special
+    #    for x in tokenize(f): print(x)
+    #with open('t/arm/peram65xx.per','r') as f: # general, also has repeat cmd
+    #    for x in tokenize_flatten(tokenize(f)): print(x)
+    with open('t/arm/perpxa.per','r') as f: # copy cmd
+        for x in tokenize_flatten(tokenize(f)): print(x)
+    #with open('t/arm/perpsoc4000.per','r') as f: # weird paren grammar
+    #    for x in tokenize(f): print(x)
+    # weird string literal grammar:
+    #with open('t/arm/perrzt1.per','r') as f: # at VLAN_PRIORITY0
+    #    for x in tokenize(f): print(x)
+    # arm/pertegrax1.per arm/pers32g2.per arm/perkinetisw.per arm/perrzt1.per arm/perstm32f7x.per arm/perxavier.per arm/pertegrax2.per
+    # counterexample/-test: arm/percortexa73a53.per
+    #with open('t/arm/percortexa73a53.per','r') as f:
+    #    for x in tokenize(f): print(x)