SchOS/tools/qrtest.c

// Host harness for kernel/core/qr.c.
//
// The QR encoder cannot be unit-tested against itself in any meaningful way: a
// symbol with a mistyped table entry is still a plausible-looking grid of black
// squares. The only real test is whether a decoder reads it back, so this builds
// the same encoder on the host and writes a PNG to scan.
//
//   make qrtest                       a sample crash report
//   build/tools/qrtest out.png "text" arbitrary payload
//
// It also self-checks the parts that *can* be checked mechanically -- the
// version table's internal consistency, and the Reed-Solomon remainder against
// a known vector -- so a transcription error is caught before anyone reaches
// for a phone.

#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#define QR_HOST
#include "../kernel/core/qr.c"

// --- Minimal PNG writer ----------------------------------------------------
//
// Uncompressed (stored) deflate blocks, so there is no zlib dependency for a
// tool whose entire job is to produce one small image. A stored block costs
// five bytes of header per 65535 bytes of data, which for a 41-module symbol is
// irrelevant.

static unsigned long crc_table[256];

static void
crc_init(void)
{
    for (int n = 0; n < 256; n++) {
        unsigned long c = (unsigned long)n;
        for (int k = 0; k < 8; k++)
            c = (c & 1) ? 0xEDB88320UL ^ (c >> 1) : c >> 1;
        crc_table[n] = c;
    }
}

static unsigned long
crc(const unsigned char *buf, int len)
{
    unsigned long c = 0xFFFFFFFFUL;
    for (int n = 0; n < len; n++)
        c = crc_table[(c ^ buf[n]) & 0xFF] ^ (c >> 8);
    return c ^ 0xFFFFFFFFUL;
}

static void
be32(unsigned char *p, unsigned long v)
{
    p[0] = (unsigned char)(v >> 24); p[1] = (unsigned char)(v >> 16);
    p[2] = (unsigned char)(v >> 8);  p[3] = (unsigned char)v;
}

static void
put_chunk(FILE *f, const char *type, const unsigned char *data, int len)
{
    unsigned char hdr[4];
    be32(hdr, (unsigned long)len);
    fwrite(hdr, 1, 4, f);

    unsigned char *buf = malloc((size_t)len + 4);
    memcpy(buf, type, 4);
    if (len)
        memcpy(buf + 4, data, (size_t)len);
    fwrite(buf, 1, (size_t)len + 4, f);

    be32(hdr, crc(buf, len + 4));
    fwrite(hdr, 1, 4, f);
    free(buf);
}

// Write the symbol as a PNG, `scale` pixels per module, with the 4-module quiet
// zone the standard requires. The quiet zone is not decoration: without it many
// decoders will not find the symbol at all.
static int
write_png(const char *path, const struct qr *q, int scale, int quiet)
{
    int mods = q->size + 2 * quiet;
    int w = mods * scale, h = w;

    // Greyscale, 8-bit, one filter byte per row.
    int stride = w + 1;
    unsigned char *raw = malloc((size_t)stride * h);
    for (int y = 0; y < h; y++) {
        unsigned char *row = raw + (size_t)y * stride;
        row[0] = 0;                            // filter: none
        int my = y / scale - quiet;
        for (int x = 0; x < w; x++) {
            int mx = x / scale - quiet;
            int dark = (mx >= 0 && my >= 0 && mx < q->size && my < q->size)
                       ? q->mod[my][mx] : 0;
            row[1 + x] = dark ? 0x00 : 0xFF;
        }
    }
    int rawlen = stride * h;

    // zlib stream: header, stored deflate blocks, adler32.
    int cap = rawlen + (rawlen / 65535 + 1) * 5 + 6;
    unsigned char *z = malloc((size_t)cap);
    int zn = 0;
    z[zn++] = 0x78; z[zn++] = 0x01;
    int off = 0;
    while (off < rawlen) {
        int n = rawlen - off;
        if (n > 65535) n = 65535;
        z[zn++] = (off + n >= rawlen) ? 1 : 0;
        z[zn++] = (unsigned char)(n & 0xFF);
        z[zn++] = (unsigned char)(n >> 8);
        z[zn++] = (unsigned char)(~n & 0xFF);
        z[zn++] = (unsigned char)((~n >> 8) & 0xFF);
        memcpy(z + zn, raw + off, (size_t)n);
        zn += n;
        off += n;
    }
    unsigned long a = 1, b = 0;
    for (int i = 0; i < rawlen; i++) {
        a = (a + raw[i]) % 65521;
        b = (b + a) % 65521;
    }
    be32(z + zn, (b << 16) | a);
    zn += 4;

    FILE *f = fopen(path, "wb");
    if (!f) { perror(path); return -1; }
    fwrite("\x89PNG\r\n\x1a\n", 1, 8, f);

    unsigned char ihdr[13];
    be32(ihdr, (unsigned long)w);
    be32(ihdr + 4, (unsigned long)h);
    ihdr[8] = 8; ihdr[9] = 0; ihdr[10] = 0; ihdr[11] = 0; ihdr[12] = 0;
    put_chunk(f, "IHDR", ihdr, 13);
    put_chunk(f, "IDAT", z, zn);
    put_chunk(f, "IEND", NULL, 0);
    fclose(f);

    free(raw);
    free(z);
    return 0;
}

// --- Self-checks -----------------------------------------------------------

static int failures;

static void
check(int ok, const char *what)
{
    printf("  [%s] %s\n", ok ? "ok" : "FAIL", what);
    if (!ok) failures++;
}

// The version table is the one part of qr.c that is transcription rather than
// reasoning, so check what can be checked: block sizes must divide evenly, and
// the symbol must have room for exactly the codewords claimed.
static void
check_tables(void)
{
    printf("version table\n");
    for (int v = 1; v <= QR_MAX_VERSION; v++) {
        int total = vinfo[v].total;
        int data  = total - vinfo[v].ecc * vinfo[v].blocks;
        char msg[128];

        snprintf(msg, sizeof msg, "v%d: %d data codewords divide into %d blocks",
                 v, data, vinfo[v].blocks);
        check(data > 0 && data % vinfo[v].blocks == 0, msg);

        // Count the modules a symbol of this version actually leaves for data,
        // by building one and asking. If the table's codeword count disagrees
        // with the geometry, one of the two is wrong.
        struct qr q;
        memset(&q, 0, sizeof q);
        q.size = 17 + 4 * v;
        draw_function_patterns(&q);
        int free_modules = 0;
        for (int y = 0; y < q.size; y++)
            for (int x = 0; x < q.size; x++)
                if (!q.fn[y][x]) free_modules++;

        snprintf(msg, sizeof msg,
                 "v%d: %d free modules holds %d codewords (%d spare bits)",
                 v, free_modules, total, free_modules - total * 8);
        check(free_modules >= total * 8 && free_modules - total * 8 < 8, msg);
    }
}

// A known Reed-Solomon vector. "HELLO WORLD" as a version 1-M byte-mode symbol
// has data codewords 0x40 0xB3 0x42 ... ; rather than depend on remembering the
// whole thing, check the standard's own worked example: the generator
// polynomial for degree 10 begins with the coefficients below.
static void
check_rs(void)
{
    printf("reed-solomon\n");
    unsigned char g[30];

    rs_divisor(10, g);
    // Generator for 10 EC codewords, highest power first, leading 1 implicit.
    static const unsigned char want10[10] = {
        216, 194, 159, 111, 199, 94, 95, 113, 157, 193
    };
    check(memcmp(g, want10, 10) == 0, "degree-10 generator polynomial");

    rs_divisor(7, g);
    static const unsigned char want7[7] = { 127, 122, 154, 164, 11, 68, 117 };
    check(memcmp(g, want7, 7) == 0, "degree-7 generator polynomial");

    // Multiplication is the primitive everything else rests on.
    check(gmul(0, 5) == 0 && gmul(1, 5) == 5, "gf(256) identity and zero");
    check(gmul(0x80, 0x02) == 0x1D, "gf(256) reduction by 0x11D");
}

static void
check_encode(void)
{
    printf("encode\n");
    struct qr q;

    check(qr_encode(&q, (const unsigned char *)"HELLO", 5) == 0, "short payload encodes");
    check(q.size == 21, "and picks version 1");

    unsigned char big[QR_MAX_BYTES];
    memset(big, 'A', sizeof big);
    check(qr_encode(&q, big, QR_MAX_BYTES) == 0, "maximum payload encodes");
    check(q.size == 41, "and picks version 6");
    check(qr_encode(&q, big, QR_MAX_BYTES + 1) == -1, "one byte over is refused");

    // Every function module must still be a function module after masking --
    // i.e. masking must not have touched the finders.
    qr_encode(&q, (const unsigned char *)"test", 4);
    check(q.mod[0][0] && q.mod[6][6] && !q.mod[5][5], "finder pattern intact");
    check(q.mod[q.size - 8][8] == 1, "dark module set");
    // Row 6 alternates with the *even* coordinates dark, so the first timing
    // module clear of the separator (x=8) is dark and x=9 is light. Asserting
    // this the other way round was the one self-check that failed on the first
    // run, and the test was what was wrong -- worth keeping the orientation
    // written down.
    check(q.mod[6][8] == 1 && q.mod[6][9] == 0, "horizontal timing pattern");
    check(q.mod[8][6] == 1 && q.mod[9][6] == 0, "vertical timing pattern");
}

int
main(int argc, char **argv)
{
    crc_init();

    // -q suppresses the terminal rendering. It is on by default because a
    // phone can read the symbol straight off the terminal without anyone
    // opening a file, but it is 40 lines of escape codes and sometimes in the
    // way.
    int quiet = 0;
    if (argc > 1 && strcmp(argv[1], "-q") == 0) {
        quiet = 1;
        argc--; argv++;
    }

    // -m dumps the raw module matrix and nothing else, for comparison against
    // an independent encoder. This is the check that actually matters: a symbol
    // can pass every self-check above and still be unreadable if the zigzag,
    // the masking or the block interleave is subtly wrong, and none of those
    // have a plausible-looking failure mode -- they all just produce noise that
    // resembles a QR code. tools/qrcheck.py drives this.
    // -f dumps the function-pattern map alongside the modules, so a checker can
    // run the zigzag backwards and recover the codewords. Knowing *which*
    // codeword two encoders disagree about is the difference between a bug you
    // can see and one you can only stare at.
    if (argc > 2 && strcmp(argv[1], "-f") == 0) {
        struct qr q;
        int n = (int)strlen(argv[2]);
        if (n > QR_MAX_BYTES || qr_encode(&q, (const unsigned char *)argv[2], n) != 0)
            return 1;
        for (int y = 0; y < q.size; y++) {
            for (int x = 0; x < q.size; x++)
                putchar(q.fn[y][x] ? '1' : '0');
            putchar('\n');
        }
        return 0;
    }

    if (argc > 2 && strcmp(argv[1], "-m") == 0) {
        struct qr q;
        int n = (int)strlen(argv[2]);
        if (n > QR_MAX_BYTES || qr_encode(&q, (const unsigned char *)argv[2], n) != 0)
            return 1;
        for (int y = 0; y < q.size; y++) {
            for (int x = 0; x < q.size; x++)
                putchar(q.mod[y][x] ? '1' : '0');
            putchar('\n');
        }
        return 0;
    }

    const char *out = argc > 1 ? argv[1] : "build/qr.png";
    const char *text = argc > 2 ? argv[2] :
        "SCHOS PANIC\n"
        "build 8f2a91c4\n"
        "trap 14 err 4 cpu 0\n"
        "rip ffffffff80104e2a\n"
        "cr2 0000000000000018\n"
        "pid 7 schinit";

    check_tables();
    check_rs();
    check_encode();

    struct qr q;
    int len = (int)strlen(text);
    if (len > QR_MAX_BYTES) {
        fprintf(stderr, "qrtest: payload is %d bytes, max %d\n", len, QR_MAX_BYTES);
        return 1;
    }
    if (qr_encode(&q, (const unsigned char *)text, len) != 0) {
        fprintf(stderr, "qrtest: encode failed\n");
        return 1;
    }

    if (write_png(out, &q, 8, 4) != 0)
        return 1;

    printf("\n%d bytes -> version %d (%dx%d modules) -> %s\n",
           len, (q.size - 17) / 4, q.size, q.size, out);

    // Also dump it to the terminal, where a phone can read it directly off the
    // screen without opening the file. Two spaces per module because a terminal
    // cell is about twice as tall as it is wide, and a squashed symbol will not
    // scan.
    if (!quiet) {
    printf("\n");
    for (int i = 0; i < 2; i++) {
        printf("  ");
        for (int x = 0; x < q.size + 8; x++) printf("\033[107m  \033[0m");
        printf("\n");
    }
    for (int y = 0; y < q.size; y++) {
        printf("  \033[107m        \033[0m");
        for (int x = 0; x < q.size; x++)
            printf("%s  \033[0m", q.mod[y][x] ? "\033[40m" : "\033[107m");
        printf("\033[107m        \033[0m\n");
    }
    for (int i = 0; i < 2; i++) {
        printf("  ");
        for (int x = 0; x < q.size + 8; x++) printf("\033[107m  \033[0m");
        printf("\n");
    }
    }

    printf("\n%s\n", failures ? "SELF-CHECKS FAILED" : "self-checks passed");
    return failures ? 1 : 0;
}