from openocd contrib
authorDobrica Pavlinusic <dpavlin@rot13.org>
Sun, 12 Jun 2016 21:19:36 +0000 (23:19 +0200)
committerDobrica Pavlinusic <dpavlin@rot13.org>
Sun, 12 Jun 2016 21:19:36 +0000 (23:19 +0200)
svf2xsvf.py [new file with mode: 0755]

diff --git a/svf2xsvf.py b/svf2xsvf.py
new file mode 100755 (executable)
index 0000000..91ca37e
--- /dev/null
@@ -0,0 +1,729 @@
+#!/usr/bin/python3
+
+# Copyright 2008, SoftPLC Corporation  http://softplc.com
+# Dick Hollenbeck dick@softplc.com
+
+
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU General Public License
+# as published by the Free Software Foundation; either version 2
+# of the License, or (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, you may find one here:
+# http://www.gnu.org/licenses/old-licenses/gpl-2.0.html
+# or you may search the http://www.gnu.org website for the version 2 license,
+# or you may write to the Free Software Foundation, Inc.,
+# 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA
+
+
+# A python program to convert an SVF file to an XSVF file.  There is an
+# option to include comments containing the source file line number from the origin
+# SVF file before each outputted XSVF statement.
+#
+# We deviate from the XSVF spec in that we introduce a new command called
+# XWAITSTATE which directly flows from the SVF RUNTEST command.  Unfortunately
+# XRUNSTATE was ill conceived and is not used here.  We also add support for the
+# three Lattice extensions to SVF: LCOUNT, LDELAY, and LSDR.  The xsvf file
+# generated from this program is suitable for use with the xsvf player in
+# OpenOCD with my modifications to xsvf.c.
+#
+# This program is written for python 3.0, and it is not easy to change this
+# back to 2.x.  You may find it easier to use python 3.x even if that means
+# building it.
+
+
+import re
+import sys
+import struct
+
+
+# There are both ---<Lexer>--- and ---<Parser>--- sections to this program
+
+
+if len( sys.argv ) < 3:
+    print("usage %s <svf_filename> <xsvf_filename>" % sys.argv[0])
+    exit(1)
+
+
+inputFilename = sys.argv[1]
+outputFilename = sys.argv[2]
+
+doCOMMENTs = True       # Save XCOMMENTs in the output xsvf file
+#doCOMMENTs = False       # Save XCOMMENTs in the output xsvf file
+
+# pick your file encoding
+file_encoding = 'ISO-8859-1'
+#file_encoding = 'utf-8'
+
+
+xrepeat = 0             # argument to XREPEAT, gives retry count for masked compares
+
+
+#-----< Lexer >---------------------------------------------------------------
+
+StateBin = (RESET,IDLE,
+    DRSELECT,DRCAPTURE,DRSHIFT,DREXIT1,DRPAUSE,DREXIT2,DRUPDATE,
+    IRSELECT,IRCAPTURE,IRSHIFT,IREXIT1,IRPAUSE,IREXIT2,IRUPDATE) = range(16)
+
+# Any integer index into this tuple will be equal to its corresponding StateBin value
+StateTxt = ("RESET","IDLE",
+    "DRSELECT","DRCAPTURE","DRSHIFT","DREXIT1","DRPAUSE","DREXIT2","DRUPDATE",
+    "IRSELECT","IRCAPTURE","IRSHIFT","IREXIT1","IRPAUSE","IREXIT2","IRUPDATE")
+
+
+(XCOMPLETE,XTDOMASK,XSIR,XSDR,XRUNTEST,hole0,hole1,XREPEAT,XSDRSIZE,XSDRTDO,
+    XSETSDRMASKS,XSDRINC,XSDRB,XSDRC,XSDRE,XSDRTDOB,XSDRTDOC,
+    XSDRTDOE,XSTATE,XENDIR,XENDDR,XSIR2,XCOMMENT,XWAIT,XWAITSTATE,
+    LCOUNT,LDELAY,LSDR,XTRST) = range(29)
+
+#Note: LCOUNT, LDELAY, and LSDR are Lattice extensions to SVF and provide a way to loop back
+# and check a completion status, essentially waiting on a part until it signals that it is done.
+# For example below: loop 25 times, each time through the loop do a LDELAY (same as a true RUNTEST)
+# and exit loop when LSDR compares match.
+"""
+LCOUNT 25;
+! Step to DRPAUSE give 5 clocks and wait for 1.00e+000 SEC.
+LDELAY DRPAUSE 5 TCK   1.00E-003 SEC;
+! Test for the completed status. Match means pass.
+! Loop back to LDELAY line if not match and loop count less than 25.
+LSDR  1 TDI  (0)
+        TDO  (1);
+"""
+
+#XTRST is an opcode Xilinx seemed to have missed and it comes from the SVF TRST statement.
+
+LineNumber = 1
+
+def s_ident(scanner, token): return ("ident", token.upper(), LineNumber)
+
+def s_hex(scanner, token):
+    global LineNumber
+    LineNumber = LineNumber + token.count('\n')
+    token = ''.join(token.split())
+    return ("hex", token[1:-1], LineNumber)
+
+def s_int(scanner, token): return ("int", int(token), LineNumber)
+def s_float(scanner, token): return ("float", float(token), LineNumber)
+#def s_comment(scanner, token): return ("comment", token, LineNumber)
+def s_semicolon(scanner, token): return ("semi", token, LineNumber)
+
+def s_nl(scanner,token):
+    global LineNumber
+    LineNumber = LineNumber + 1
+    #print( 'LineNumber=', LineNumber, file=sys.stderr )
+    return None
+
+#2.00E-002
+
+scanner = re.Scanner([
+    (r"[a-zA-Z]\w*", s_ident),
+#    (r"[-+]?[0-9]+[.]?[0-9]*([eE][-+]?[0-9]+)?", s_float),
+    (r"[-+]?[0-9]+(([.][0-9eE+-]*)|([eE]+[-+]?[0-9]+))", s_float),
+    (r"\d+", s_int),
+    (r"\(([0-9a-fA-F]|\s)*\)", s_hex),
+    (r"(!|//).*$", None),
+    (r";", s_semicolon),
+    (r"\n",s_nl),
+    (r"\s*", None),
+    ],
+    re.MULTILINE
+    )
+
+# open the file using the given encoding
+file = open( sys.argv[1], encoding=file_encoding )
+
+# read all svf file input into string "input"
+input = file.read()
+
+file.close()
+
+# Lexer:
+# create a list of tuples containing (tokenType, tokenValue, LineNumber)
+tokens = scanner.scan( input )[0]
+
+input = None    # allow gc to reclaim memory holding file
+
+#for tokenType, tokenValue, ln in tokens: print( "line %d: %s" % (ln, tokenType), tokenValue )
+
+
+#-----<parser>-----------------------------------------------------------------
+
+tokVal = tokType = tokLn = None
+
+tup = iter( tokens )
+
+def nextTok():
+    """
+    Function to read the next token from tup into tokType, tokVal, tokLn (linenumber)
+    which are globals.
+    """
+    global tokType, tokVal, tokLn, tup
+    tokType, tokVal, tokLn = tup.__next__()
+
+
+class ParseError(Exception):
+    """A class to hold a parsing error message"""
+    def __init__(self, linenumber, token, message):
+        self.linenumber = linenumber
+        self.token = token
+        self.message = message
+    def __str__(self):
+        global inputFilename
+        return "Error in file \'%s\' at line %d near token %s\n %s" % (
+                   inputFilename, self.linenumber, repr(self.token), self.message)
+
+
+class MASKSET(object):
+    """
+    Class MASKSET holds a set of bit vectors, all of which are related, will all
+    have the same length, and are associated with one of the seven shiftOps:
+    HIR, HDR, TIR, TDR, SIR, SDR, LSDR. One of these holds a mask, smask, tdi, tdo, and a
+    size.
+    """
+    def __init__(self, name):
+        self.empty()
+        self.name = name
+
+    def empty(self):
+        self.mask = bytearray()
+        self.smask = bytearray()
+        self.tdi = bytearray()
+        self.tdo = bytearray()
+        self.size = 0
+
+    def syncLengths( self, sawTDI, sawTDO, sawMASK, sawSMASK, newSize ):
+        """
+        Set all the lengths equal in the event some of the masks were
+        not seen as part of the last change set.
+        """
+        if self.size == newSize:
+            return
+
+        if newSize == 0:
+            self.empty()
+            return
+
+        # If an SIR was given without a MASK(), then use a mask of all zeros.
+        # this is not consistent with the SVF spec, but it makes sense because
+        # it would be odd to be testing an instruction register read out of a
+        # tap without giving a mask for it.  Also, lattice seems to agree and is
+        # generating SVF files that comply with this philosophy.
+        if self.name == 'SIR' and not sawMASK:
+            self.mask = bytearray( newSize )
+
+        if newSize != len(self.mask):
+            self.mask = bytearray( newSize )
+            if self.name == 'SDR':  # leave mask for HIR,HDR,TIR,TDR,SIR zeros
+                for i in range( newSize ):
+                    self.mask[i] = 1
+
+        if newSize != len(self.tdo):
+            self.tdo = bytearray( newSize )
+
+        if newSize != len(self.tdi):
+            self.tdi = bytearray( newSize )
+
+        if newSize != len(self.smask):
+            self.smask = bytearray( newSize )
+
+        self.size = newSize
+#-----</MASKSET>-----
+
+
+def makeBitArray( hexString, bitCount ):
+    """
+    Converts a packed sequence of hex ascii characters into a bytearray where
+    each element in the array holds exactly one bit. Only "bitCount" bits are
+    scanned and these must be the least significant bits in the hex number. That
+    is, it is legal to have some unused bits in the must significant hex nibble
+    of the input "hexString". The string is scanned starting from the backend,
+    then just before returning we reverse the array. This way the append()
+    method can be used, which I assume is faster than an insert.
+    """
+    global tokLn
+    a = bytearray()
+    length = bitCount
+    hexString = list(hexString)
+    hexString.reverse()
+    #print(hexString)
+    for c in hexString:
+        if length <= 0:
+            break;
+        c = int(c, 16)
+        for mask in [1,2,4,8]:
+            if length <= 0:
+                break;
+            length = length - 1
+            a.append( (c & mask) != 0 )
+    if length > 0:
+        raise ParseError( tokLn, hexString, "Insufficient hex characters for given length of %d" % bitCount )
+    a.reverse()
+    #print(a)
+    return a
+
+
+def makeXSVFbytes( bitarray ):
+    """
+    Make a bytearray which is contains the XSVF bits which will be written
+    directly to disk.  The number of bytes needed is calculated from the size
+    of the argument bitarray.
+    """
+    bitCount = len(bitarray)
+    byteCount = (bitCount+7)//8
+    ba = bytearray( byteCount )
+    firstBit = (bitCount % 8) - 1
+    if firstBit == -1:
+        firstBit = 7
+    bitNdx = 0
+    for byteNdx in range(byteCount):
+        mask = 1<<firstBit
+        byte = 0
+        while mask:
+            if bitarray[bitNdx]:
+                byte |= mask;
+            mask = mask >> 1
+            bitNdx = bitNdx + 1
+        ba[byteNdx] = byte
+        firstBit = 7
+    return ba
+
+
+def writeComment( outputFile, shiftOp_linenum, shiftOp ):
+    """
+    Write an XCOMMENT record to outputFile
+    """
+    comment = "%s @%d\0" % (shiftOp, shiftOp_linenum)   # \0 is terminating nul
+    ba = bytearray(1)
+    ba[0] = XCOMMENT
+    ba += comment.encode()
+    outputFile.write( ba )
+
+
+def combineBitVectors( trailer, meat, header ):
+    """
+    Combine the 3 bit vectors comprizing a transmission.  Since the least
+    significant bits are sent first, the header is put onto the list last so
+    they are sent first from that least significant position.
+    """
+    ret = bytearray()
+    ret.extend( trailer )
+    ret.extend( meat )
+    ret.extend( header )
+    return ret
+
+
+def writeRUNTEST( outputFile, run_state, end_state, run_count, min_time, tokenTxt ):
+    """
+    Write the output for the SVF RUNTEST command.
+    run_count - the number of clocks
+    min_time - the number of seconds
+    tokenTxt - either RUNTEST or LDELAY
+    """
+    # convert from secs to usecs
+    min_time = int( min_time * 1000000)
+
+    # the SVF RUNTEST command does NOT map to the XSVF XRUNTEST command.  Check the SVF spec, then
+    # read the XSVF command.   They are not the same.  Use an XSVF XWAITSTATE to
+    # implement the required behavior of the SVF RUNTEST command.
+    if doCOMMENTs:
+        writeComment( output, tokLn, tokenTxt )
+
+    if tokenTxt == 'RUNTEST':
+        obuf = bytearray(11)
+        obuf[0] = XWAITSTATE
+        obuf[1] = run_state
+        obuf[2] = end_state
+        struct.pack_into(">i", obuf, 3, run_count )  # big endian 4 byte int to obuf
+        struct.pack_into(">i", obuf, 7, min_time )   # big endian 4 byte int to obuf
+        outputFile.write( obuf )
+    else:   # == 'LDELAY'
+        obuf = bytearray(10)
+        obuf[0] = LDELAY
+        obuf[1] = run_state
+        # LDELAY has no end_state
+        struct.pack_into(">i", obuf, 2, run_count )  # big endian 4 byte int to obuf
+        struct.pack_into(">i", obuf, 6, min_time )   # big endian 4 byte int to obuf
+        outputFile.write( obuf )
+
+
+output = open( outputFilename, mode='wb' )
+
+hir = MASKSET('HIR')
+hdr = MASKSET('HDR')
+tir = MASKSET('TIR')
+tdr = MASKSET('TDR')
+sir = MASKSET('SIR')
+sdr = MASKSET('SDR')
+
+
+expecting_eof = True
+
+
+# one of the commands that take the shiftParts after the length, the parse
+# template for all of these commands is identical
+shiftOps = ('SDR', 'SIR', 'LSDR', 'HDR', 'HIR', 'TDR', 'TIR')
+
+# the order must correspond to shiftOps, this holds the MASKSETS.  'LSDR' shares sdr with 'SDR'
+shiftSets = (sdr, sir, sdr, hdr, hir, tdr, tir )
+
+# what to expect as parameters to a shiftOp, i.e. after a SDR length or SIR length
+shiftParts = ('TDI', 'TDO', 'MASK', 'SMASK')
+
+# the set of legal states which can trail the RUNTEST command
+run_state_allowed = ('IRPAUSE', 'DRPAUSE', 'RESET', 'IDLE')
+
+enddr_state_allowed = ('DRPAUSE', 'IDLE')
+endir_state_allowed = ('IRPAUSE', 'IDLE')
+
+trst_mode_allowed = ('ON', 'OFF', 'Z', 'ABSENT')
+
+enddr_state = IDLE
+endir_state = IDLE
+
+frequency =    1.00e+006 # HZ;
+
+# change detection for xsdrsize and xtdomask
+xsdrsize = -1           # the last one sent, send only on change
+xtdomask = bytearray()  # the last one sent, send only on change
+
+
+# we use a number of single byte writes for the XSVF command below
+cmdbuf = bytearray(1)
+
+
+# Save the XREPEAT setting into the file as first thing.
+obuf = bytearray(2)
+obuf[0] = XREPEAT
+obuf[1] = xrepeat
+output.write( obuf )
+
+
+try:
+    while 1:
+        expecting_eof = True
+        nextTok()
+        expecting_eof = False
+        # print( tokType, tokVal, tokLn )
+
+        if tokVal in shiftOps:
+            shiftOp_linenum = tokLn
+            shiftOp = tokVal
+
+            set = shiftSets[shiftOps.index(shiftOp)]
+
+            # set flags false, if we see one later, set that one true later
+            sawTDI = sawTDO = sawMASK = sawSMASK = False
+
+            nextTok()
+            if tokType != 'int':
+                raise ParseError( tokLn, tokVal, "Expecting 'int' giving %s length, got '%s'" % (shiftOp, tokType) )
+            length = tokVal
+
+            nextTok()
+
+            while tokVal != ';':
+                if tokVal not in shiftParts:
+                    raise ParseError( tokLn, tokVal, "Expecting TDI, TDO, MASK, SMASK, or ';'")
+                shiftPart = tokVal
+
+                nextTok()
+
+                if tokType != 'hex':
+                    raise ParseError( tokLn, tokVal, "Expecting hex bits" )
+                bits = makeBitArray( tokVal, length )
+
+                if shiftPart == 'TDI':
+                    sawTDI = True
+                    set.tdi = bits
+
+                elif shiftPart == 'TDO':
+                    sawTDO = True
+                    set.tdo = bits
+
+                elif shiftPart == 'MASK':
+                    sawMASK = True
+                    set.mask = bits
+
+                elif shiftPart == 'SMASK':
+                    sawSMASK = True
+                    set.smask = bits
+
+                nextTok()
+
+            set.syncLengths( sawTDI, sawTDO, sawMASK, sawSMASK, length )
+
+            # process all the gathered parameters and generate outputs here
+            if shiftOp == 'SIR':
+                if doCOMMENTs:
+                    writeComment( output, shiftOp_linenum, 'SIR' )
+
+                tdi = combineBitVectors( tir.tdi, sir.tdi, hir.tdi )
+                if len(tdi) > 255:
+                    obuf = bytearray(3)
+                    obuf[0] = XSIR2
+                    struct.pack_into( ">h", obuf, 1, len(tdi) )
+                else:
+                    obuf = bytearray(2)
+                    obuf[0] = XSIR
+                    obuf[1] = len(tdi)
+                output.write( obuf )
+                obuf = makeXSVFbytes( tdi )
+                output.write( obuf )
+
+            elif shiftOp == 'SDR':
+                if doCOMMENTs:
+                    writeComment( output, shiftOp_linenum, shiftOp )
+
+                if not sawTDO:
+                    # pass a zero filled bit vector for the sdr.mask
+                    mask = combineBitVectors( tdr.mask, bytearray(sdr.size), hdr.mask )
+                    tdi  = combineBitVectors( tdr.tdi,  sdr.tdi,  hdr.tdi )
+
+                    if xsdrsize != len(tdi):
+                        xsdrsize = len(tdi)
+                        cmdbuf[0] = XSDRSIZE
+                        output.write( cmdbuf )
+                        obuf = bytearray(4)
+                        struct.pack_into( ">i", obuf, 0, xsdrsize )  # big endian 4 byte int to obuf
+                        output.write( obuf )
+
+                    if xtdomask != mask:
+                        xtdomask = mask
+                        cmdbuf[0] = XTDOMASK
+                        output.write( cmdbuf )
+                        obuf = makeXSVFbytes( mask )
+                        output.write( obuf )
+
+                    cmdbuf[0] = XSDR
+                    output.write( cmdbuf )
+                    obuf = makeXSVFbytes( tdi )
+                    output.write( obuf )
+
+                else:
+                    mask = combineBitVectors( tdr.mask, sdr.mask, hdr.mask )
+                    tdi  = combineBitVectors( tdr.tdi,  sdr.tdi,  hdr.tdi )
+                    tdo  = combineBitVectors( tdr.tdo,  sdr.tdo,  hdr.tdo )
+
+                    if xsdrsize != len(tdi):
+                        xsdrsize = len(tdi)
+                        cmdbuf[0] = XSDRSIZE
+                        output.write( cmdbuf )
+                        obuf = bytearray(4)
+                        struct.pack_into(">i", obuf, 0, xsdrsize )  # big endian 4 byte int to obuf
+                        output.write( obuf )
+
+                    if xtdomask != mask:
+                        xtdomask = mask
+                        cmdbuf[0] = XTDOMASK
+                        output.write( cmdbuf )
+                        obuf = makeXSVFbytes( mask )
+                        output.write( obuf )
+
+                    cmdbuf[0] = XSDRTDO
+                    output.write( cmdbuf )
+                    obuf = makeXSVFbytes( tdi )
+                    output.write( obuf )
+                    obuf = makeXSVFbytes( tdo )
+                    output.write( obuf )
+                    #print( "len(tdo)=", len(tdo), "len(tdr.tdo)=", len(tdr.tdo), "len(sdr.tdo)=", len(sdr.tdo), "len(hdr.tdo)=", len(hdr.tdo) )
+
+            elif shiftOp == 'LSDR':
+                if doCOMMENTs:
+                    writeComment( output, shiftOp_linenum, shiftOp )
+
+                mask = combineBitVectors( tdr.mask, sdr.mask, hdr.mask )
+                tdi  = combineBitVectors( tdr.tdi,  sdr.tdi,  hdr.tdi )
+                tdo  = combineBitVectors( tdr.tdo,  sdr.tdo,  hdr.tdo )
+
+                if xsdrsize != len(tdi):
+                    xsdrsize = len(tdi)
+                    cmdbuf[0] = XSDRSIZE
+                    output.write( cmdbuf )
+                    obuf = bytearray(4)
+                    struct.pack_into(">i", obuf, 0, xsdrsize )  # big endian 4 byte int to obuf
+                    output.write( obuf )
+
+                if xtdomask != mask:
+                    xtdomask = mask
+                    cmdbuf[0] = XTDOMASK
+                    output.write( cmdbuf )
+                    obuf = makeXSVFbytes( mask )
+                    output.write( obuf )
+
+                cmdbuf[0] = LSDR
+                output.write( cmdbuf )
+                obuf = makeXSVFbytes( tdi )
+                output.write( obuf )
+                obuf = makeXSVFbytes( tdo )
+                output.write( obuf )
+                #print( "len(tdo)=", len(tdo), "len(tdr.tdo)=", len(tdr.tdo), "len(sdr.tdo)=", len(sdr.tdo), "len(hdr.tdo)=", len(hdr.tdo) )
+
+        elif tokVal == 'RUNTEST' or tokVal == 'LDELAY':
+            # e.g. from lattice tools:
+            # "RUNTEST IDLE    5 TCK   1.00E-003 SEC;"
+            saveTok = tokVal
+            nextTok()
+            min_time = 0
+            run_count = 0
+            max_time = 600  # ten minutes
+            if tokVal in run_state_allowed:
+                run_state = StateTxt.index(tokVal)
+                end_state = run_state  # bottom of page 17 of SVF spec
+                nextTok()
+            if tokType != 'int' and tokType != 'float':
+                raise ParseError( tokLn, tokVal, "Expecting 'int' or 'float' after RUNTEST [run_state]")
+            timeval = tokVal;
+            nextTok()
+            if tokVal != 'TCK' and tokVal != 'SEC' and tokVal != 'SCK':
+                raise ParseError( tokLn, tokVal, "Expecting 'TCK' or 'SEC' or 'SCK' after RUNTEST [run_state] (run_count|min_time)")
+            if tokVal == 'TCK' or tokVal == 'SCK':
+                run_count = int( timeval )
+            else:
+                min_time = timeval
+            nextTok()
+            if tokType == 'int' or tokType == 'float':
+                min_time = tokVal
+                nextTok()
+                if tokVal != 'SEC':
+                    raise ParseError( tokLn, tokVal, "Expecting 'SEC' after RUNTEST [run_state] run_count min_time")
+                nextTok()
+            if tokVal == 'MAXIMUM':
+                nextTok()
+                if tokType != 'int' and tokType != 'float':
+                    raise ParseError( tokLn, tokVal, "Expecting 'max_time' after RUNTEST [run_state] min_time SEC MAXIMUM")
+                max_time = tokVal
+                nextTok()
+                if tokVal != 'SEC':
+                    raise ParseError( tokLn, tokVal, "Expecting 'max_time' after RUNTEST [run_state] min_time SEC MAXIMUM max_time")
+                nextTok()
+            if tokVal == 'ENDSTATE':
+                nextTok()
+                if tokVal not in run_state_allowed:
+                    raise ParseError( tokLn, tokVal, "Expecting 'run_state' after RUNTEST .... ENDSTATE")
+                end_state = StateTxt.index(tokVal)
+                nextTok()
+            if tokVal != ';':
+                raise ParseError( tokLn, tokVal, "Expecting ';' after RUNTEST ....")
+            # print( "run_count=", run_count, "min_time=", min_time,
+                # "max_time=", max_time, "run_state=", State[run_state], "end_state=", State[end_state] )
+            writeRUNTEST( output, run_state, end_state, run_count, min_time, saveTok )
+
+        elif tokVal == 'LCOUNT':
+            nextTok()
+            if tokType != 'int':
+                raise ParseError( tokLn, tokVal, "Expecting integer 'count' after LCOUNT")
+            loopCount = tokVal
+            nextTok()
+            if tokVal != ';':
+                raise ParseError( tokLn, tokVal, "Expecting ';' after LCOUNT count")
+            if doCOMMENTs:
+                writeComment( output, tokLn, 'LCOUNT' )
+            obuf = bytearray(5)
+            obuf[0] = LCOUNT
+            struct.pack_into(">i", obuf, 1, loopCount )  # big endian 4 byte int to obuf
+            output.write( obuf )
+
+        elif tokVal == 'ENDDR':
+            nextTok()
+            if tokVal not in enddr_state_allowed:
+                raise ParseError( tokLn, tokVal, "Expecting 'stable_state' after ENDDR. (one of: DRPAUSE, IDLE)")
+            enddr_state = StateTxt.index(tokVal)
+            nextTok()
+            if tokVal != ';':
+                raise ParseError( tokLn, tokVal, "Expecting ';' after ENDDR stable_state")
+            if doCOMMENTs:
+                writeComment( output, tokLn, 'ENDDR' )
+            obuf = bytearray(2)
+            obuf[0] = XENDDR
+            # Page 10 of the March 1999 SVF spec shows that RESET is also allowed here.
+            # Yet the XSVF spec has no provision for that, and uses a non-standard, i.e.
+            # boolean argument to XENDDR which only handles two of the 3 intended states.
+            obuf[1] = 1 if enddr_state == DRPAUSE else 0
+            output.write( obuf )
+
+        elif tokVal == 'ENDIR':
+            nextTok()
+            if tokVal not in endir_state_allowed:
+                raise ParseError( tokLn, tokVal, "Expecting 'stable_state' after ENDIR. (one of: IRPAUSE, IDLE)")
+            endir_state = StateTxt.index(tokVal)
+            nextTok()
+            if tokVal != ';':
+                raise ParseError( tokLn, tokVal, "Expecting ';' after ENDIR stable_state")
+            if doCOMMENTs:
+                writeComment( output, tokLn, 'ENDIR' )
+            obuf = bytearray(2)
+            obuf[0] = XENDIR
+            # Page 10 of the March 1999 SVF spec shows that RESET is also allowed here.
+            # Yet the XSVF spec has no provision for that, and uses a non-standard, i.e.
+            # boolean argument to XENDDR which only handles two of the 3 intended states.
+            obuf[1] = 1 if endir_state == IRPAUSE else 0
+            output.write( obuf )
+
+        elif tokVal == 'STATE':
+            nextTok()
+            ln = tokLn
+            while tokVal != ';':
+                if tokVal not in StateTxt:
+                    raise ParseError( tokLn, tokVal, "Expecting 'stable_state' after STATE")
+                stable_state = StateTxt.index( tokVal )
+
+                if doCOMMENTs and ln != -1:
+                    writeComment( output, ln, 'STATE' )
+                    ln = -1     # save comment only once
+
+                obuf = bytearray(2)
+                obuf[0] = XSTATE
+                obuf[1] = stable_state
+                output.write( obuf )
+                nextTok()
+
+        elif tokVal == 'FREQUENCY':
+            nextTok()
+            if tokVal != ';':
+                if tokType != 'int' and tokType != 'float':
+                    raise ParseError( tokLn, tokVal, "Expecting 'cycles HZ' after FREQUENCY")
+                frequency = tokVal
+                nextTok()
+                if tokVal != 'HZ':
+                    raise ParseError( tokLn, tokVal, "Expecting 'HZ' after FREQUENCY cycles")
+                nextTok()
+                if tokVal != ';':
+                    raise ParseError( tokLn, tokVal, "Expecting ';' after FREQUENCY cycles HZ")
+
+        elif tokVal == 'TRST':
+            nextTok()
+            if tokVal not in trst_mode_allowed:
+                raise ParseError( tokLn, tokVal, "Expecting 'ON|OFF|Z|ABSENT' after TRST")
+            trst_mode = tokVal
+            nextTok()
+            if tokVal != ';':
+                raise ParseError( tokLn, tokVal, "Expecting ';' after TRST trst_mode")
+            if doCOMMENTs:
+                writeComment( output, tokLn, 'TRST %s' % trst_mode )
+            obuf = bytearray( 2 )
+            obuf[0] = XTRST
+            obuf[1] = trst_mode_allowed.index( trst_mode )  # use the index as the binary argument to XTRST opcode
+            output.write( obuf )
+
+        else:
+            raise ParseError( tokLn, tokVal, "Unknown token '%s'" % tokVal)
+
+except StopIteration:
+    if not expecting_eof:
+        print( "Unexpected End of File at line ", tokLn )
+
+except ParseError as pe:
+    print( "\n", pe )
+
+finally:
+    # print( "closing file" )
+    cmdbuf[0] = XCOMPLETE
+    output.write( cmdbuf )
+    output.close()
+