make_sample.py

#!/usr/bin/env python3
"""
Generate a small synthetic X9.100-187 file so x9view can be exercised without
production data. Writes EBCDIC, variable-length records, with two hand-built
TIFF images per item (front and back).

    python3 make_sample.py sample.x937
"""

import struct
import sys

EBCDIC = "cp037"


def fld(value, length, numeric=False):
    """Pad a field to its fixed width: numerics right-justified zero-filled,
    alphanumerics left-justified blank-filled, as X9 requires."""
    text = str(value)
    if numeric:
        return text[-length:].rjust(length, "0")
    return text[:length].ljust(length)


def rec(*parts):
    """Encode a text record to EBCDIC and prefix it with its 4-byte length."""
    body = "".join(parts).encode(EBCDIC)
    return struct.pack(">I", len(body)) + body


def binary_rec(prefix_text, image):
    body = prefix_text.encode(EBCDIC) + image
    return struct.pack(">I", len(body)) + body


def tiny_tiff(width=96, height=40, stripe=8):
    """A 1-bit uncompressed TIFF with horizontal stripes.

    Real X9 files carry CCITT Group 4 images; uncompressed keeps this generator
    dependency-free while still producing a file that `sips` and Preview read.
    """
    row_bytes = (width + 7) // 8
    pixels = bytearray()
    for y in range(height):
        # Photometric is WhiteIsZero, so 0xFF is a black run.
        fill = 0xFF if (y // stripe) % 2 else 0x00
        pixels.extend(bytes([fill]) * row_bytes)

    header_size = 8
    entries = [
        (256, 3, 1, width),          # ImageWidth
        (257, 3, 1, height),         # ImageLength
        (258, 3, 1, 1),              # BitsPerSample
        (259, 3, 1, 1),              # Compression: none
        (262, 3, 1, 0),              # PhotometricInterpretation: WhiteIsZero
        (273, 4, 1, None),           # StripOffsets       (patched below)
        (277, 3, 1, 1),              # SamplesPerPixel
        (278, 4, 1, height),         # RowsPerStrip
        (279, 4, 1, len(pixels)),    # StripByteCounts
        (282, 5, 1, None),           # XResolution        (patched below)
        (283, 5, 1, None),           # YResolution        (patched below)
        (296, 3, 1, 2),              # ResolutionUnit: inch
    ]
    ifd_size = 2 + 12 * len(entries) + 4
    xres_off = header_size + ifd_size
    yres_off = xres_off + 8
    data_off = yres_off + 8

    patched = []
    for tag, typ, count, value in entries:
        if tag == 273:
            value = data_off
        elif tag == 282:
            value = xres_off
        elif tag == 283:
            value = yres_off
        patched.append((tag, typ, count, value))

    out = bytearray(b"II" + struct.pack("<HI", 42, header_size))
    out += struct.pack("<H", len(patched))
    for tag, typ, count, value in patched:
        out += struct.pack("<HHI", tag, typ, count)
        if typ == 3:  # SHORT occupies the first two bytes of the value field
            out += struct.pack("<HH", value, 0)
        else:
            out += struct.pack("<I", value)
    out += struct.pack("<I", 0)          # no next IFD
    out += struct.pack("<II", 200, 1)    # XResolution 200 dpi
    out += struct.pack("<II", 200, 1)    # YResolution 200 dpi
    out += pixels
    return bytes(out)


def image_view_detail(side):
    return rec(
        "50",
        fld("1", 1),                      # image indicator: image present
        fld("061000227", 9),              # image creator routing number
        fld("20260805", 8),               # image creator date
        fld("00", 2),                     # format indicator: TIFF
        fld("00", 2),                     # compression: G4
        fld(0, 7, numeric=True),          # image view data size (0 = not used)
        fld(side, 1),                     # 0 front, 1 back
        fld("00", 2),                     # view descriptor: full view
        fld("0", 1),                      # digital signature indicator: none
        fld("00", 2),
        fld(0, 5, numeric=True),
        fld(0, 7, numeric=True),
        fld(0, 7, numeric=True),
        fld("0", 1),
        fld("", 8),
        fld("", 15),
    )


def image_view_data(seq, image):
    prefix = "".join([
        "52",
        fld("061000227", 9),
        fld("20260805", 8),
        fld("01", 2),
        fld(seq, 15, numeric=True),
        fld("", 16),                      # security originator name
        fld("", 16),                      # security authenticator name
        fld("", 16),                      # security key name
        fld("0", 1),                      # clipping origin: no clipping
        fld(0, 4, numeric=True),
        fld(0, 4, numeric=True),
        fld(0, 4, numeric=True),
        fld(0, 4, numeric=True),
        fld(0, 4, numeric=True),          # length of image reference key: 0
        fld(0, 5, numeric=True),          # length of digital signature: 0
        fld(len(image), 7, numeric=True), # length of image data
    ])
    return binary_rec(prefix, image)


def build(path):
    items = [
        ("000000000000001", 12550, "121000248", "1234567890         ", "ALEX MORGAN"),
        ("000000000000002", 899900, "021000021", "9876543210         ", "ACME SUPPLY CO"),
        ("000000000000003", 4500, "061000227", "5555000123         ", "JANE DOE"),
    ]
    total = sum(amount for _, amount, _, _, _ in items)

    out = [
        rec("01",
            fld("03", 2), fld("T", 1),                 # standard level, test file
            fld("061000227", 9), fld("121000248", 9),
            fld("20260805", 8), fld("1430", 4), fld("0", 1),
            fld("DEST BANK NA", 18), fld("ORIGIN BANK NA", 18),
            fld("A", 1), fld("US", 2), fld("", 4)),
        rec("10",
            fld("01", 2), fld("061000227", 9), fld("121000248", 9),
            fld("20260805", 8), fld("20260805", 8), fld("1430", 4),
            fld("I", 1),                                # record type indicator: images
            fld("G", 1),                                # documentation type: image
            fld("CL000001", 8), fld("OPS DESK", 14), fld("5551234567", 10),
            fld("", 1), fld("", 1), fld("", 2)),
        rec("20",
            fld("01", 2), fld("061000227", 9), fld("121000248", 9),
            fld("20260805", 8), fld("20260805", 8), fld("BUNDLE0001", 10),
            fld(1, 4, numeric=True), fld("01", 2), fld("", 9), fld("", 5),
            fld("", 12)),
    ]

    for seq, amount, routing, on_us, payee in items:
        out.append(rec("25",
                       fld("", 15),                     # auxiliary on-us
                       fld("", 1),                      # external processing code
                       fld(routing[:8], 8), fld(routing[8], 1),
                       fld(on_us, 20), fld(amount, 10, numeric=True),
                       fld(seq, 15, numeric=True),
                       fld("G", 1), fld("1", 1), fld("1", 1), fld("1", 1),
                       fld(1, 2, numeric=True), fld("0", 1), fld("A", 1)))
        out.append(rec("26",
                       fld("1", 1), fld("061000227", 9), fld("20260805", 8),
                       fld(seq, 15, numeric=True), fld("000123456789", 18),
                       fld("00042", 5), fld(payee, 15),
                       fld("Y", 1), fld("0", 1), fld("0", 1), fld("", 1), fld("", 3)))
        for side in ("0", "1"):
            image = tiny_tiff(stripe=6 if side == "0" else 10)
            out.append(image_view_detail(side))
            out.append(image_view_data(seq, image))

    # One returned item, so the type 31 layout gets exercised too.
    ret_seq, ret_amount = "000000000000004", 32100
    out.append(rec("31",
                   fld("12100024", 8), fld("8", 1),
                   fld("4444555566", 20), fld(ret_amount, 10, numeric=True),
                   fld("A", 1),                     # return reason: NSF
                   fld(0, 2, numeric=True), fld("G", 1), fld("20260805", 8),
                   fld(ret_seq, 15, numeric=True),
                   fld("", 1), fld("0", 1), fld("A", 1), fld("1", 1), fld("", 8)))
    for side in ("0", "1"):
        image = tiny_tiff(stripe=4 if side == "0" else 12)
        out.append(image_view_detail(side))
        out.append(image_view_data(ret_seq, image))
    total += ret_amount

    out.append(rec("70",
                   fld(len(items) + 1, 4, numeric=True),
                   fld(total, 12, numeric=True),
                   fld(total, 12, numeric=True),
                   fld((len(items) + 1) * 2, 5, numeric=True),
                   fld("", 20), fld("", 25)))
    out.append(rec("90",
                   fld(1, 6, numeric=True),
                   fld(len(items) + 1, 8, numeric=True),
                   fld(total, 14, numeric=True),
                   fld((len(items) + 1) * 2, 9, numeric=True),
                   fld("ORIGIN BANK NA", 18), fld("20260806", 8),
                   fld("0", 1), fld("", 14)))
    body = b"".join(out)
    record_count = len(out) + 1
    body += rec("99",
                fld(1, 6, numeric=True),
                fld(record_count, 8, numeric=True),
                fld(len(items) + 1, 8, numeric=True),
                fld(total, 16, numeric=True),
                fld("OPS DESK", 14), fld("5551234567", 10),
                fld("0", 1), fld("", 15))

    with open(path, "wb") as fh:
        fh.write(body)
    print("wrote %s (%d bytes, %d records)" % (path, len(body), record_count))


if __name__ == "__main__":
    build(sys.argv[1] if len(sys.argv) > 1 else "sample.x937")