test(codes): compare Data Matrix and PDF417 with references and decode them, also from screenshots

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

1 files changed, 146 insertions, 0 deletions

diff --git a/tools/gen_code_vectors.py b/tools/gen_code_vectors.py
index a87b0c9..5ce2611 100644
--- a/tools/gen_code_vectors.py
+++ b/tools/gen_code_vectors.py
@@ -21,7 +21,10 @@ from pathlib import Path
 from typing import Any, cast
 
 import barcode_ref as ref
+import ppf.datamatrix.codec_ascii  # noqa: F401  (registers the datamatrix.ascii codec)
 import segno
+from pdf417gen import encode as pdf417_encode
+from ppf.datamatrix import DataMatrix
 
 CHECK_ADD, CHECK_FIX, CHECK_OFF = 0, 1, 2
 TEXT_ON = 0x10
@@ -183,6 +186,126 @@ def qr_reference(payload: str, level: int, version: int, mask: int) -> tuple[int
     return int(qr.designator.split("-")[0]), qr.mask, packed.hex()
 
 
+def dm_cases(rng: random.Random, count: int) -> list[str]:
+    """ASCII payloads for Data Matrix (ppf.datamatrix has no Base256 or
+    extended ASCII; those are covered by decoding). Lengths cover small,
+    multi-region and the largest (144 x 144, multi-block) sizes."""
+    fixed = ["a", "123456", "Hello, World!", "0" * 44, "ABC-0042/fake member", "9" * 300, "x" * 500, "7" * 3116]
+    printable = string.ascii_letters + string.digits + " .,:;-/+()"
+    out = list(fixed)
+    for _ in range(count):
+        n = rng.choice([rng.randint(1, 40), rng.randint(40, 400)])
+        out.append("".join(rng.choice(printable) for _ in range(n)))
+    return out
+
+
+# Data codewords per square Data Matrix size (ISO/IEC 16022 table 7).
+DM_CAPACITY = {
+    10: 3,
+    12: 5,
+    14: 8,
+    16: 12,
+    18: 18,
+    20: 22,
+    22: 30,
+    24: 36,
+    26: 44,
+    32: 62,
+    36: 86,
+    40: 114,
+    44: 144,
+    48: 174,
+    52: 204,
+    64: 280,
+    72: 368,
+    80: 456,
+    88: 576,
+    96: 696,
+    104: 816,
+    120: 1050,
+    132: 1304,
+    144: 1558,
+}
+
+
+def ppf_pad_differs(payload: str, size: int) -> bool:
+    """True when ppf.datamatrix would write a pad codeword of 0 instead of 254.
+
+    ppf computes the 253-state pad as (R + 130) % 254, which turns 254 into 0;
+    ISO/IEC 16022 5.2.3 only subtracts 254 above 254. The watch follows the
+    standard, so such payloads make no exact reference.
+    """
+    used = len(payload.encode("datamatrix.ascii"))
+    capacity = DM_CAPACITY[size]
+    return any((149 * pos) % 253 == 124 for pos in range(used + 2, capacity + 1))
+
+
+def dm_reference(payload: str) -> tuple[int, str]:
+    """ppf.datamatrix's (size, matrix as hex of row-major bits), ASCII only."""
+    matrix = DataMatrix(payload, codecs=["ascii"]).matrix
+    bits = "".join(str(b) for row in matrix for b in row)
+    bits += "0" * (-len(bits) % 8)
+    packed = bytes(int(bits[i : i + 8], 2) for i in range(0, len(bits), 8))
+    return len(matrix), packed.hex()
+
+
+def pdf417_cases(rng: random.Random, count: int) -> list[tuple[str, int, int]]:
+    """(payload, columns, level): text in all submodes, digit runs below and
+    above 13, UTF-8 bytes, exact 6-byte groups, levels 0-6, 1-12 columns."""
+    fixed = [
+        ("Hello, World!", 3, 2),
+        ("abcABC123;:@#", 4, 1),
+        ("Member 0042 - fake card", 2, 2),
+        ("12345678901234567890123456789012345678901234567890", 3, 3),
+        ("Zürich ✓ fake data", 4, 2),
+        ("\x01\x02\x03\x04\x05\x06", 2, 0),
+        ("tab\there\nnew line\r end", 5, 4),
+        ("x" * 200, 8, 5),
+        ("0" * 13 + "abc" + "1" * 12, 3, 2),
+        ("https://example.com/member/fake?id=0042&level=gold", 6, 6),
+    ]
+    out = list(fixed)
+    printable = string.ascii_letters + string.digits + " !\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~"
+    for _ in range(count):
+        n = rng.randint(1, 120)
+        payload = "".join(rng.choice(printable) if rng.random() < 0.8 else rng.choice("éßü€") for _ in range(n))
+        out.append((payload, rng.randint(1, 12), rng.randint(0, 5)))
+    return out
+
+
+def pdf417_reference(payload: str, columns: int, level: int) -> tuple[int, int, str] | None:
+    """pdf417gen's (rows, width, rows of modules as hex), or None when the
+    combination is outside PDF417's limits (fewer than 3 or more than 90 rows)."""
+    try:
+        codes = pdf417_encode(payload, columns=columns, security_level=level)
+    except ValueError:
+        return None
+    rows = ["".join(format(value, "b") for value in row) for row in codes]
+    bits = "".join(rows)
+    bits += "0" * (-len(bits) % 8)
+    packed = bytes(int(bits[i : i + 8], 2) for i in range(0, len(bits), 8))
+    return len(rows), len(rows[0]), packed.hex()
+
+
+def mc_string(text: str) -> str:
+    """A Monkey C string literal body for text (quotes, backslashes, controls escaped)."""
+    out = []
+    for ch in text:
+        if ch in '\\"':
+            out.append("\\" + ch)
+        elif ch == "\n":
+            out.append("\\n")
+        elif ch == "\t":
+            out.append("\\t")
+        elif ch == "\r":
+            out.append("\\r")
+        elif ord(ch) < 32:
+            out.append(f"\\u{ord(ch):04x}")
+        else:
+            out.append(ch)
+    return "".join(out)
+
+
 def render(rows: list[tuple[int, str, int, str]], seed: int) -> str:
     lines = []
     for symbology, payload, opts, text in rows:
@@ -202,6 +325,23 @@ def render(rows: list[tuple[int, str, int, str]], seed: int) -> str:
         escaped = payload.replace("\\", "\\\\").replace('"', '\\"')
         qr_lines.append(f'        ["{escaped}", {level}, {version}, {mask}, {ref_version}, {ref_mask}, "{hexbits}"]')
     qr_body = ",\n".join(qr_lines)
+    dm_lines = []
+    for dm_case in dm_cases(random.Random(seed + 2), 20):
+        payload = dm_case
+        size, hexbits = dm_reference(payload)
+        while ppf_pad_differs(payload, size):
+            payload += payload[-1]
+            size, hexbits = dm_reference(payload)
+        dm_lines.append(f'        ["{payload}", {size}, "{hexbits}"]')
+    dm_body = ",\n".join(dm_lines)
+    pdf_lines = []
+    for payload, columns, level in pdf417_cases(random.Random(seed + 3), 20):
+        reference = pdf417_reference(payload, columns, level)
+        if reference is None or any(ord(ch) < 32 and ch not in "\t\n\r" for ch in payload):
+            continue
+        n_rows, width, hexbits = reference
+        pdf_lines.append(f'        ["{mc_string(payload)}", {columns}, {level}, {n_rows}, {width}, "{hexbits}"]')
+    pdf_body = ",\n".join(pdf_lines)
     return (
         "// Author:  cirakg & claude ai\n"
         "// Summary: zxing-cpp reference module rows for the linear barcode encoders (generated; do not edit).\n"
@@ -212,6 +352,12 @@ def render(rows: list[tuple[int, str, int, str]], seed: int) -> str:
         + "\n    ];\n\n    // [payload, level, version (0 auto), mask (-1 auto), segno version, segno mask,\n"
         + "    //  segno matrix as hex of row-major bits]\n    const QR = [\n"
         + qr_body
+        + "\n    ];\n\n    // [payload, ppf.datamatrix size, its matrix as hex of row-major bits] (ASCII mode)\n"
+        + "    const DATAMATRIX = [\n"
+        + dm_body
+        + "\n    ];\n\n    // [payload, columns, level, pdf417gen rows, width, one module row per PDF417 row as hex]\n"
+        + "    const PDF417 = [\n"
+        + pdf_body
         + "\n    ];\n}\n"
     )