tap2sna.py

SYNOPSIS

tap2sna.py [options] INPUT [INPUT] [OUTFILE]
tap2sna.py @FILE [args]

DESCRIPTION

tap2sna.py converts one or two PZX, TAP or TZX files (which may be inside a zip archive) into an SZX or Z80 snapshot. INPUT may be the full URL to a remote zip archive or tape file, or the path to a local file. If two INPUTs are given, they must both be (local or remote) tape files. Arguments may be read from FILE instead of (or as well as) being given on the command line.

OPTIONS

-c, –sim-load-config name=value

Set the value of a simulated LOAD configuration parameter. Do -c help for more information, or -c help-name for help on a specific parameter. Also see the section on SIMULATED LOAD below. This option may be used multiple times.

-d, –output-dir DIR

Write the snapshot file in this directory.

-I, –ini param=value

Set the value of a configuration parameter (see CONFIGURATION), overriding any value found in skoolkit.ini. This option may be used multiple times.

–press N:KEYS

Pause the tape at block number N and press KEYS before resuming. KEYS must be a space-separated list of key identifiers (see USER INPUT). This option may be used multiple times.

-p, –stack STACK

Set the stack pointer. This option is equivalent to --reg sp=STACK. STACK must be a decimal number, or a hexadecimal number prefixed by ‘0x’.

--ram OPERATION

Perform a load operation or otherwise modify the memory snapshot being built. Do --ram help for more information, or see the sections on the CALL, LOAD, MOVE, PATCH, POKE and SYSVARS operations below. This option may be used multiple times.

–reg name=value

Set the value of a register. Do --reg help for more information, or see the section on REGISTERS below. This option may be used multiple times.

--show-config

Show configuration parameter values.

-s, –start START

Set the start address to JP to. This option is equivalent to --reg pc=START. START must be a decimal number, or a hexadecimal number prefixed by ‘0x’.

–state name=value

Set a hardware state attribute. Do --state help for more information, or see the section on HARDWARE STATE below. This option may be used multiple times.

--tape-analysis

Show an analysis of the tape’s tones, pulse sequences and data blocks.

--tape-name NAME

Specify the name of a tape file in a zip archive. By default, the first tape file found in the zip archive is selected. Use this option twice when loading two tape files.

–tape-skip A[-B]

Skip block numbers A-B on the tape.

--tape-start BLOCK

Start the tape at this block number. In a tape file, the first block is number 1, the second is 2, etc.

--tape-stop BLOCK

Stop the tape at this block number. In a tape file, the first block is number 1, the second is 2, etc.

--tape-sum MD5SUM

Specify the MD5 checksum of the tape file. tap2sna.py will abort if there is a checksum mismatch. This option may be used twice if loading two tape files.

-u, –user-agent AGENT

Set the User-Agent header used in an HTTP(S) request.

-V, --version

Show the SkoolKit version number and exit.

TZX SUPPORT

tap2sna.py cannot read data from TZX block types 0x18 (CSW recording) or 0x19 (generalized data block).

SIMULATED LOAD

By default, tap2sna.py simulates a freshly booted 48K ZX Spectrum running LOAD “” (or LOAD “”CODE, if the first block on the tape is a ‘Bytes’ header). Whenever the Spectrum ROM’s load routine at $0556 is called, a shortcut is taken by “fast loading” the next block on the tape. All other code (including any custom loader) is fully simulated. Simulation continues until the program counter hits the start address given by the --start option, or 15 minutes of simulated Z80 CPU time has elapsed, or the end of the tape is reached and one of the following conditions is satisfied:

  • a custom loader was detected

  • the program counter hits an address outside the ROM

  • more than one second of simulated Z80 CPU time has elapsed since the end of the tape was reached

A simulated LOAD can also be aborted by pressing Ctrl-C. When a simulated LOAD has completed or been aborted, the values of the registers (including the program counter) in the simulator are used to populate the snapshot.

A simulated LOAD can be configured via parameters that are set by the by the --sim-load-config (or -c) option. The recognised configuration parameters are:

  • accelerate-dec-a - enable acceleration of ‘DEC A: JR NZ,$-1’ delay loops only (1), or ‘DEC A: JP NZ,$-1’ delay loops only (2), or both (3, the default) or neither (0)

  • accelerator - a comma-separated list of tape-sampling loop accelerators to use (see the ACCELERATORS section below)

  • cmio - enable simulation of memory contention and I/O contention delays (1), or disable it (0); this is disabled by default to improve performance, but some loaders may require it; when this is enabled, all acceleration is disabled

  • fast-load - enable fast loading whenever the ROM loader is called (1, the default), or disable it (0); fast loading (also known as “flash loading”) significantly reduces the load time for many tapes, but can also cause some loaders to fail

  • finish-tape - run the tape to the end before stopping the simulation at the address specified by the --start option (1), or stop the simulation as soon as that address is reached, regardless of whether the tape has finished (0, the default)

  • first-edge - the time (in T-states) from the start of the tape at which to place the leading edge of the first pulse (default: 0)

  • in-flags - various flags specifying how to handle ‘IN’ instructions (see below)

  • load - a space-separated list of keys to press to build an alternative command line to load the tape (see the LOAD COMMAND section below)

  • machine - the type of machine to simulate: a 48K Spectrum (48, the default), or a 128K Spectrum (128)

  • pause - pause the tape between blocks and resume playback when port 254 is read (1, the default), or run the tape continuously (0); pausing can help with tapes that require (but do not actually contain) long pauses between blocks, but can cause some loaders to fail

  • polarity - the EAR bit reading produced by the first pulse on the tape: 0 (the default) or 1; subsequent pulses give readings that alternate between 0 and 1

  • python - whether to use the pure Python Z80 simulator (1), or the much faster C version if available (0, the default)

  • timeout - the number of seconds of Z80 CPU time after which to abort the simulated LOAD if it’s still in progress (default: 900)

  • trace - the file to which to log all instructions executed during the simulated LOAD (default: none)

The in-flags parameter is the sum of the following values, chosen according to the desired behaviour:

  • 1 - interpret ‘IN A,($FE)’ instructions in the address range $4000-$7FFF as reading the tape (by default they are ignored)

  • 2 - ignore ‘IN’ instructions in the address range $4000-$FFFF (i.e. in RAM) that read port $FE

  • 4 - yield a simulated port reading when executing an ‘IN r,(C)’ instruction (by default such an instruction always yields the value $FF)

By default, the EAR bit reading produced by a pulse is 0 if the 0-based index of the pulse is even (i.e. first, third, fifth pulses etc.), or 1 otherwise. This can be reversed by setting polarity=1. Run tap2sna.py with the --tape-analysis option to see the timings and EAR bit readings of the pulses on a tape.

ACCELERATORS

The accelerator simulated LOAD configuration parameter must be either a comma-separated list of specific accelerator names or one of the following special values:

  • auto - select accelerators automatically (this is the default)

  • list - list the accelerators used during a simulated LOAD, along with the hit/miss counts generated by the tape-sampling loop detector

  • none - disable acceleration; the loading time for a game with a custom loader that uses an unrecognised tape-sampling loop may be reduced by specifying this value

A tape-sampling loop accelerator works by effectively fast-forwarding the tape (and the state of the loop itself) to the next edge whenever the loop is entered. This technique is known as “edge loading”.

The output produced by accelerator=list looks something like this:

Accelerators: microsphere: 5500; rom: 793036; misses: 0; dec-a: 800708/0/23

This means that:

  • the microsphere and rom tape-sampling loops were detected, 5500 times and 793036 times respectively

  • no instances of an ‘IN A,($FE)’ instruction outside a recognised tape-sampling loop were executed (0 misses)

  • 800708 ‘DEC A: JR NZ,$-1’ delay loops were entered, no ‘DEC A: JP NZ,$-1’ delay loops were entered, and 23 instances of ‘DEC A’ outside such delay loops were executed

To show the names of the available tape-sampling loop accelerators:


tap2sna.py -c help-accelerator

LOAD COMMAND

The load simulated LOAD configuration parameter may be used to specify an alternative command line to load the tape in cases where neither ‘LOAD “”’ nor ‘LOAD “”CODE’ works. Its value is a space-separated list of ‘words’ (a ‘word’ being a sequence of any characters other than space), each of which is broken down into a sequence of one or more keypresses. If a word contains the ‘+’ symbol, the tokens it separates are converted into keypresses made simultaneously. If a word matches a BASIC token, the corresponding sequence of keypresses to produce that token are substituted. Otherwise, each character in the word is converted individually into the appropriate keypresses.

The following special tokens are also recognised:


CS - CAPS SHIFT
SS - SYMBOL SHIFT
SPACE - SPACE
ENTER - ENTER
DOWN - Cursor down (‘CS+6’)
GOTO - GO TO (‘g’)
GOSUB - GO SUB (‘h’)
DEFFN - DEF FN (‘CS+SS SS+1’)
OPEN# - OPEN # (‘CS+SS SS+4’)
CLOSE# - CLOSE # (‘CS+SS SS+5’)
PC=address - Stop the keyboard input simulation at this address

The PC=address token, if present, must appear last. The default address is either 0x0605 (when a 48K Spectrum is being simulated) or 0x13BE (on a 128K Spectrum). The simulated LOAD begins at this address.

ENTER is automatically appended to the command line if not already present.

For example, the load parameter may be set to:


CLEAR 34999: LOAD “” CODE : RANDOMIZE USR 35000

Note that the spaces around CLEAR, LOAD, CODE, RANDOMIZE and USR are required in order for them to be recognised as BASIC tokens.

USER INPUT

Some tapes require the user to pause the tape before loading has finished, press one or more keys, and then start the tape again to resume loading. Such tapes can be handled by using the --press option. Its single argument takes the form:

N:KEYS

where:

  • N is the block number at which to pause the tape

  • KEYS is a space-separated list of key identifiers

For example:


tap2sna.py --press 5:ENTER game.tzx

This will load blocks 1-4 of game.tzx, pause the tape, simulate pressing the ENTER key until the appropriate key row has been read, and then start the tape (whereupon block 5 and any other remaining blocks will be loaded).

The KEYS list recognises the digits 0-9 and lower case letters a-z as valid key identifiers, along with the following special tokens:


CS - CAPS SHIFT
SS - SYMBOL SHIFT
SPACE - SPACE
ENTER - ENTER
NONE - no key

Sometimes a single keypress may need to be read more than once before the game’s loader responds to it. In such cases the * notation is useful:


tap2sna.py --press 6:s*3 game.tzx

Equivalent to 6:s s s, this will wait until the ‘s’ keypress has been read three times before resuming the tape at block number 6.

CALL OPERATIONS

The --ram option can be used to call a Python function to perform arbitrary modification of the memory snapshot.


--ram call=[/path/to/moduledir:]module.function

The function is called with the memory snapshot (a list of 65536 byte values) as the sole positional argument. The function must modify the snapshot in place. The path to the module’s location may be omitted if the module is already in the module search path.

For example:


--ram call=:ram.modify # Call modify(snapshot) in ./ram.py

LOAD OPERATIONS

By default, tap2sna.py attempts to load a tape exactly as a 48K Spectrum would (see the section on SIMULATED LOAD above). If that doesn’t work, the --ram option can be used to load bytes from specific tape blocks at the appropriate addresses. The syntax is:


--ram load=[+]block[+],start[,length,step,offset,inc]

where the parameters have the following meanings:

block

The tape block number; the first block is 1, the next is 2, etc. Attach a ‘+’ prefix to load the first byte of the block (which is usually the flag byte), and a ‘+’ suffix to load the last byte (which is usually the parity byte).

start

The destination address at which to start loading.

length

The number of bytes to load (optional; defaults to the number of bytes remaining in the block).

step

This number is added to the destination address after each byte is loaded (optional; default=1).

offset

This number is added to the destination address before a byte is loaded, and subtracted after the byte is loaded (optional; default=0). It is analogous to the offset d in the LD (IX+d),L operation that is commonly used in load routines to copy the byte just loaded from tape (L) into memory.

inc

After step is added to the destination address, this number is added too if the result overflowed past 65535 (optional; default=0).

A single tape block can be loaded in two or more stages; for example:


--ram load=2,32768,2048 # Load the first 2K at 32768
--ram load=2,0xC000 # Load the remainder at 49152

MOVE OPERATIONS

The --ram option can be used to copy a block of bytes from one location to another before saving the snapshot.


--ram move=[s:]src,N,[d:]dest

This copies a block of N bytes from src in RAM bank s to dest in RAM bank d. For example:


--ram move=32512,256,32768 # Copy 32512-32767 to 32768-33023
--ram move=3:0,8,4:0 # Copy the first 8 bytes of bank 3 to bank 4

PATCH OPERATIONS

The --ram option can be used to apply a binary patch file to the snapshot before saving it.


--ram patch=[p:]a,file

This applies the named binary patch file at address a in RAM bank p. For example:


--ram patch=32768,patch.bin # Apply patch.bin at 32768
--ram patch=1:256,patch.bin # Apply patch.bin at 256 in RAM bank 1

POKE OPERATIONS

The --ram option can be used to POKE values into the snapshot before saving it.


--ram poke=[P:]A[-B[-C]],[^+]V

This does POKE N,V in RAM bank P for N in {A, A+C, A+2C..., B}, where:

P is the RAM bank to POKE (0-7; 128K only)

A is the first address to POKE

B is the last address to POKE (optional; default is A)

C is the step (optional; default=1)

V is the value to POKE; prefix the value with ‘^’ to perform an XOR operation, or ‘+’ to perform an ADD operation

For example:


--ram poke=0x6000,0x10 # POKE 24576,16
--ram poke=30000-30002,^85 # Perform ‘XOR 85’ on addresses 30000-30002
--ram poke=40000-40004-2,1 # POKE 40000,1: POKE 40002,1: POKE 40004,1

SYSVARS OPERATION

The --ram option can be used to initialise the system variables at 23552-23754 (5C00-5CCA) with values suitable for a 48K ZX Spectrum.


--ram sysvars

REGISTERS

The --reg option sets the value of a register in the snapshot.


--reg name=value

For example:


--reg hl=32768
--reg b=0x1f

To set the value of an alternate (shadow) register, use the ‘^’ prefix:


--reg ^hl=10072

Recognised register names are:


^a, ^b, ^bc, ^c, ^d, ^de, ^e, ^f, ^h, ^hl, ^l,
a, b, bc, c, d, de, e, f, h, hl, l,
i, ix, iy, pc, r, sp

The default value for each register is 0, with the following exceptions:


i=63
iy=23610

HARDWARE STATE

The --state option sets a hardware state attribute.


--state name=value

Recognised attribute names and their default values are:


7ffd - last OUT to port 0x7ffd (128K only)
ay[N] - contents of AY register N (N=0-15; 128K only)
border - border colour (default=0)
fe - last OUT to port 0xfe (SZX only)
fffd - last OUT to port 0xfffd (128K only)
iff - interrupt flip-flop: 0=disabled, 1=enabled (default=1)
im - interrupt mode (default=1)
issue2 - issue 2 emulation: 0=disabled, 1=enabled (default=0)
tstates - T-states elapsed since start of frame (default=34943)

READING ARGUMENTS FROM A FILE

For complex snapshots that require many options to build, it may be more convenient to store the arguments to tap2sna.py in a file. For example, if the file game.t2s has the following contents:


;
; tap2sna.py file for GAME
;
http://example.com/pub/games/GAME.zip
-c fast-load=0 # Disable fast loading
-c accelerator=none # Disable tape-sampling loop acceleration
–state issue2=1 # Enable issue 2 keyboard emulation
–start 34816 # Start at 34816

then:


tap2sna.py @game.t2s

will create game.z80 as if the arguments specified in game.t2s had been given on the command line. When tap2sna.py reads arguments from a file whose name ends with ‘.t2s’, the output snapshot filename defaults to the name of that arguments file with ‘.t2s’ replaced by either ‘.z80’ or ‘.szx’ (depending on the value of the DefaultSnapshotFormat configuration parameter).

CONFIGURATION

tap2sna.py will read configuration from a file named skoolkit.ini in the current working directory or in ~/.skoolkit, if present. The recognised configuration parameters are:

DefaultSnapshotFormat:

The format of the snapshot written when no output snapshot argument is specified. Valid values are z80 (the default) and szx.

TraceLine:

The format of each line in the trace log file for a simulated LOAD (default: ${pc:04X} {i}).

TraceOperand:

The prefix, byte format, and word format for the numeric operands of instructions in the trace log file for a simulated LOAD, separated by commas (default: $,02X,04X). The byte and word formats are standard Python format specifiers for numeric values, and default to empty strings if not supplied.

UserAgent:

The value of the User-Agent header in HTTP/HTTPS requests (default: empty string).

TraceLine is a standard Python format string that recognises the following replacement fields:


i - the current instruction
m[address] - the contents of a memory address
pc - the address of the current instruction (program counter)
r[X] - the X register (see below)
t - the current timestamp

When using the m (memory) replacement field, address must be either a decimal number, or a hexadecimal number prefixed by ‘$’ or ‘0x’.

The register name X in r[X] must be one of the following:

a b c d e f h l bc de hl
^a ^b ^c ^d ^e ^f ^h ^l ^bc ^de ^hl
ix ixh ixl iy iyh iyl
i r sp

The names that begin with ^ denote the shadow registers.

The current timestamp (t) is the number of T-states that have elapsed since the start of the simulation, according to the simulator’s internal clock. In order to maintain synchronisation with the tape being loaded, the simulator’s clock is adjusted to match the timestamp of the first pulse in each block (as shown by the --tape-analysis option) when that block is reached. (The simulator’s clock may at times become desynchronised with the tape because, by default, the tape is paused between blocks, and resumed when port 254 is read.)

Configuration parameters must appear in a [tap2sna] section. For example, to make tap2sna.py write instruction addresses and operands in a trace log file in decimal format by default, add the following section to skoolkit.ini:

[tap2sna]
TraceLine={pc:05} {i}
TraceOperand=

Configuration parameters may also be set on the command line by using the --ini option. Parameter values set this way will override any found in skoolkit.ini.

EXAMPLES

  1. Extract the tape file from a remote zip archive and convert it into a Z80 snapshot:


    tap2sna.py ftp://example.com/game.zip game.z80
  2. Extract the tape file from a zip archive, and convert it into an SZX snapshot with the program counter set to 32768:


    tap2sna.py --start 32768 game.zip game.szx
  3. Convert a TZX file into a Z80 snapshot by loading the third block on the tape at 25000:


    tap2sna.py --ram load=3,25000 game.tzx game.z80
  4. Convert a TZX file into an SZX snapshot using options read from the file game.t2s:


    tap2sna.py @game.t2s game.tzx game.szx
  5. Extract two tape files from a zip archive and convert them into a Z80 snapshot:


    tap2sna.py --tape-name side1.tzx --tape-name side2.tzx game.zip
  6. Convert two tape files into an SZX snapshot:


    tap2sna.py side1.tzx side2.tzx game.szx