path: root/src/tax_table.c

#include "tax_table.h"

#include <errno.h>
#include <limits.h>
#include <netdb.h>
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <strings.h>
#include <sys/socket.h>
#include <sys/time.h>
#include <time.h>
#include <unistd.h>

#include <openssl/err.h>
#include <openssl/ssl.h>
#include <openssl/x509.h>

#include "db.h"
#include "tls_ca.h"
#include "version.h"

#define TT_LINE_LEN 49
#define TT_TIMEOUT_SEC 30
#define TT_MAX_REDIRECTS 5
#define TT_TABLE_FILE "allmanna-tabeller-manad.txt"
#define TT_PAGE_URL                                                          \
    "https://www.skatteverket.se/foretag/arbetsgivare/"                      \
    "arbetsgivaravgifterochskatteavdrag/skattetabeller/"                     \
    "specialversionerforprogramforetagmfl.4.319dc1451507f2f99e86ee.html"

static void set_err(char **err, const char *fmt, ...)
    __attribute__((format(printf, 2, 3)));

static void set_err(char **err, const char *fmt, ...)
{
    if (!err || *err)
        return;
    char buf[512];
    va_list ap;
    va_start(ap, fmt);
    vsnprintf(buf, sizeof buf, fmt, ap);
    va_end(ap);
    *err = xstrdup(buf);
}

/* ------------------------------------------------------------------ */
/* parser                                                              */
/* ------------------------------------------------------------------ */

static int parse_int_field(const char *s, int len, int64_t *v, int *empty)
{
    int i = 0, j = len;
    while (i < j && s[i] == ' ')
        i++;
    while (j > i && s[j - 1] == ' ')
        j--;
    *empty = i == j;
    if (*empty) {
        *v = 0;
        return 0;
    }
    int64_t x = 0;
    for (int k = i; k < j; k++) {
        if (s[k] < '0' || s[k] > '9')
            return -1;
        x = x * 10 + (s[k] - '0');
    }
    *v = x;
    return 0;
}

int tax_table_parse(const unsigned char *data, size_t len,
                    struct tax_row **out, size_t *out_n, char **err)
{
    *out = NULL;
    *out_n = 0;
    if (len >= 3 && data[0] == 0xef && data[1] == 0xbb && data[2] == 0xbf) {
        data += 3;
        len -= 3;
    }
    struct tax_row *rows = NULL;
    size_t n = 0, cap = 0;
    size_t pos = 0;
    int line_no = 0;
    while (pos < len) {
        const char *line = (const char *)data + pos;
        size_t end = pos;
        while (end < len && data[end] != '\n')
            end++;
        size_t llen = end - pos;
        pos = end + 1;
        line_no++;
        while (llen > 0 && (line[llen - 1] == '\r' ||
                            line[llen - 1] == ' ' || line[llen - 1] == '\t'))
            llen--;
        if (llen == 0)
            continue;
        if (llen != TT_LINE_LEN) {
            set_err(err, "line %d: expected %d characters, got %zu", line_no,
                    TT_LINE_LEN, llen);
            goto bad;
        }
        if (line[0] != '3' || line[1] != '0' ||
            (line[2] != 'B' && line[2] != '%')) {
            set_err(err, "line %d: not a 30B/30%% record", line_no);
            goto bad;
        }
        int table_no = (line[3] - '0') * 10 + (line[4] - '0');
        if (table_no < 29 || table_no > 42) {
            set_err(err, "line %d: table %d is out of range", line_no,
                    table_no);
            goto bad;
        }
        int is_pct = line[2] == '%';
        int64_t from_kr = 0, to_kr = 0, col_kr = 0;
        int empty = 0;
        if (parse_int_field(line + 5, 7, &from_kr, &empty) != 0 || empty) {
            set_err(err, "line %d: bad income from", line_no);
            goto bad;
        }
        if (parse_int_field(line + 12, 7, &to_kr, &empty) != 0) {
            set_err(err, "line %d: bad income to", line_no);
            goto bad;
        }
        int has_to = !empty;
        if (!has_to && !is_pct) {
            set_err(err, "line %d: B record needs an income to", line_no);
            goto bad;
        }
        if (has_to && to_kr < from_kr) {
            set_err(err, "line %d: income range is reversed", line_no);
            goto bad;
        }
        for (int c = 0; c < 6; c++) {
            if (parse_int_field(line + 19 + 5 * c, 5, &col_kr, &empty) != 0 ||
                empty) {
                set_err(err, "line %d: bad column %d", line_no, c + 1);
                goto bad;
            }
            if (n == cap) {
                cap = cap ? cap * 2 : 1024;
                rows = xrealloc(rows, cap * sizeof *rows);
            }
            struct tax_row *row = &rows[n++];
            row->table_no = table_no;
            row->column_no = c + 1;
            row->income_from_ore = from_kr * 100;
            row->income_to_ore = has_to ? to_kr * 100 : -1;
            row->tax_ore = is_pct ? 0 : col_kr * 100;
            row->pct = is_pct ? col_kr * 100 : 0;
            row->is_pct = is_pct;
        }
    }
    if (n == 0) {
        set_err(err, "no table records found");
        goto bad;
    }
    *out = rows;
    *out_n = n;
    return 0;

bad:
    free(rows);
    return -1;
}

/* ------------------------------------------------------------------ */
/* storage                                                             */
/* ------------------------------------------------------------------ */

int tax_table_store(sqlite3 *db, int year, const struct tax_row *rows,
                    size_t n, const char *source_url,
                    const unsigned char sha256[32], const char *fetched_at,
                    char **err)
{
    char *del = sqlite3_mprintf("DELETE FROM tax_tables WHERE in_year=%d",
                                year);
    if (!del) {
        set_err(err, "out of memory");
        return -1;
    }
    int drc = db_exec(db, del, err);
    sqlite3_free(del);
    if (drc != 0)
        return -1;
    sqlite3_stmt *st = NULL;
    if (sqlite3_prepare_v2(
            db,
            "INSERT INTO tax_tables(in_year,table_no,column_no,income_from_ore,"
            "income_to_ore,tax_ore,pct)"
            " VALUES(?1,?2,?3,?4,?5,?6,?7)",
            -1, &st, NULL) != SQLITE_OK) {
        set_err(err, "database error: %s", sqlite3_errmsg(db));
        return -1;
    }
    for (size_t i = 0; i < n; i++) {
        sqlite3_bind_int(st, 1, year);
        sqlite3_bind_int(st, 2, rows[i].table_no);
        sqlite3_bind_int(st, 3, rows[i].column_no);
        sqlite3_bind_int64(st, 4, rows[i].income_from_ore);
        if (rows[i].income_to_ore < 0)
            sqlite3_bind_null(st, 5);
        else
            sqlite3_bind_int64(st, 5, rows[i].income_to_ore);
        sqlite3_bind_int64(st, 6, rows[i].tax_ore);
        if (rows[i].is_pct)
            sqlite3_bind_int64(st, 7, rows[i].pct);
        else
            sqlite3_bind_null(st, 7);
        int rc = sqlite3_step(st);
        sqlite3_reset(st);
        sqlite3_clear_bindings(st);
        if (rc != SQLITE_DONE) {
            set_err(err, "could not store row %zu: %s", i + 1,
                    sqlite3_errmsg(db));
            sqlite3_finalize(st);
            return -1;
        }
    }
    sqlite3_finalize(st);
    if (sqlite3_prepare_v2(
            db,
            "INSERT INTO tax_table_meta(in_year,source_url,sha256,fetched_at)"
            " VALUES(?1,?2,?3,?4)"
            " ON CONFLICT(in_year) DO UPDATE SET"
            " source_url=excluded.source_url, sha256=excluded.sha256,"
            " fetched_at=excluded.fetched_at",
            -1, &st, NULL) != SQLITE_OK) {
        set_err(err, "database error: %s", sqlite3_errmsg(db));
        return -1;
    }
    sqlite3_bind_int(st, 1, year);
    sqlite3_bind_text(st, 2, source_url, -1, SQLITE_TRANSIENT);
    sqlite3_bind_blob(st, 3, sha256, 32, SQLITE_TRANSIENT);
    sqlite3_bind_text(st, 4, fetched_at, -1, SQLITE_TRANSIENT);
    int rc = sqlite3_step(st);
    sqlite3_finalize(st);
    if (rc != SQLITE_DONE) {
        set_err(err, "could not store metadata: %s", sqlite3_errmsg(db));
        return -1;
    }
    return 0;
}

int tax_table_lookup(sqlite3 *db, int year, int table_no, int column_no,
                     int64_t gross_ore, int64_t *tax_ore)
{
    sqlite3_stmt *st = NULL;
    if (sqlite3_prepare_v2(
            db,
            "SELECT MAX(income_to_ore) FROM tax_tables"
            " WHERE in_year=?1 AND table_no=?2 AND column_no=?3"
            " AND pct IS NULL",
            -1, &st, NULL) != SQLITE_OK)
        return -1;
    sqlite3_bind_int(st, 1, year);
    sqlite3_bind_int(st, 2, table_no);
    sqlite3_bind_int(st, 3, column_no);
    int step = sqlite3_step(st);
    if (step != SQLITE_ROW ||
        sqlite3_column_type(st, 0) == SQLITE_NULL) {
        sqlite3_finalize(st);
        return 2;
    }
    int64_t top = sqlite3_column_int64(st, 0);
    sqlite3_finalize(st);
    if (gross_ore > top)
        return 1;
    if (sqlite3_prepare_v2(
            db,
            "SELECT tax_ore FROM tax_tables"
            " WHERE in_year=?1 AND table_no=?2 AND column_no=?3"
            " AND pct IS NULL AND income_from_ore<=?4 AND income_to_ore>=?4",
            -1, &st, NULL) != SQLITE_OK)
        return -1;
    sqlite3_bind_int(st, 1, year);
    sqlite3_bind_int(st, 2, table_no);
    sqlite3_bind_int(st, 3, column_no);
    sqlite3_bind_int64(st, 4, gross_ore);
    step = sqlite3_step(st);
    if (step != SQLITE_ROW) {
        sqlite3_finalize(st);
        return 2;
    }
    *tax_ore = sqlite3_column_int64(st, 0);
    sqlite3_finalize(st);
    return 0;
}

/* ------------------------------------------------------------------ */
/* HTTPS GET                                                           */
/* ------------------------------------------------------------------ */

struct tt_url {
    char host[256];
    char path[2048];
    int port;
};

static int tt_url_parse(const char *url, struct tt_url *u, char **err)
{
    if (strncmp(url, "https://", 8) != 0) {
        set_err(err, "not an https URL: %.120s", url);
        return -1;
    }
    const char *p = url + 8;
    const char *slash = strchr(p, '/');
    const char *host_end = slash ? slash : p + strlen(p);
    const char *colon = memchr(p, ':', (size_t)(host_end - p));
    size_t hl = colon ? (size_t)(colon - p) : (size_t)(host_end - p);
    if (hl == 0 || hl >= sizeof u->host) {
        set_err(err, "bad host in URL: %.120s", url);
        return -1;
    }
    memcpy(u->host, p, hl);
    u->host[hl] = '\0';
    u->port = 443;
    if (colon) {
        long port = strtol(colon + 1, NULL, 10);
        if (port < 1 || port > 65535) {
            set_err(err, "bad port in URL: %.120s", url);
            return -1;
        }
        u->port = (int)port;
    }
    snprintf(u->path, sizeof u->path, "%s", slash ? slash : "/");
    return 0;
}

static int tt_tcp_connect(const struct tt_url *u, char **err)
{
    char portstr[16];
    snprintf(portstr, sizeof portstr, "%d", u->port);
    struct addrinfo hints;
    memset(&hints, 0, sizeof hints);
    hints.ai_family = AF_UNSPEC;
    hints.ai_socktype = SOCK_STREAM;
    struct addrinfo *res = NULL;
    int gai = getaddrinfo(u->host, portstr, &hints, &res);
    if (gai != 0) {
        set_err(err, "cannot resolve %s: %s", u->host, gai_strerror(gai));
        return -1;
    }
    int last = 0;
    for (struct addrinfo *ai = res; ai; ai = ai->ai_next) {
        int fd = socket(ai->ai_family, ai->ai_socktype | SOCK_CLOEXEC,
                        ai->ai_protocol);
        if (fd < 0) {
            last = errno;
            continue;
        }
        struct timeval tv;
        tv.tv_sec = TT_TIMEOUT_SEC;
        tv.tv_usec = 0;
        (void)setsockopt(fd, SOL_SOCKET, SO_RCVTIMEO, &tv, sizeof tv);
        (void)setsockopt(fd, SOL_SOCKET, SO_SNDTIMEO, &tv, sizeof tv);
        if (connect(fd, ai->ai_addr, ai->ai_addrlen) == 0) {
            freeaddrinfo(res);
            return fd;
        }
        last = errno;
        close(fd);
    }
    freeaddrinfo(res);
    set_err(err, "cannot connect to %s:%d: %s", u->host, u->port,
            last ? strerror(last) : "unknown error");
    return -1;
}

static int tt_tls_start(SSL_CTX *ctx, SSL **ssl, int fd, const char *host,
                        char **err)
{
    *ssl = SSL_new(ctx);
    if (!*ssl) {
        set_err(err, "cannot create TLS connection");
        return -1;
    }
    if (SSL_set_fd(*ssl, fd) != 1 ||
        SSL_set_tlsext_host_name(*ssl, host) != 1 ||
        SSL_set1_host(*ssl, host) != 1) {
        set_err(err, "cannot set up TLS connection");
        return -1;
    }
    if (SSL_connect(*ssl) != 1) {
        long vr = SSL_get_verify_result(*ssl);
        unsigned long ec = ERR_get_error();
        if (vr != X509_V_OK)
            set_err(err, "TLS certificate verification failed: %s",
                    X509_verify_cert_error_string(vr));
        else if (ec)
            set_err(err, "TLS handshake failed: %s",
                    ERR_error_string(ec, NULL));
        else
            set_err(err, "TLS handshake failed");
        return -1;
    }
    if (SSL_get_verify_result(*ssl) != X509_V_OK) {
        set_err(err, "TLS certificate verification failed: %s",
                X509_verify_cert_error_string(SSL_get_verify_result(*ssl)));
        return -1;
    }
    return 0;
}

static int tt_write(SSL *ssl, const void *data, size_t n, char **err)
{
    const unsigned char *p = data;
    while (n > 0) {
        int chunk = n > (size_t)INT_MAX ? INT_MAX : (int)n;
        int w = SSL_write(ssl, p, chunk);
        if (w <= 0) {
            int e = SSL_get_error(ssl, w);
            if (e == SSL_ERROR_WANT_READ || e == SSL_ERROR_WANT_WRITE)
                continue;
            set_err(err, "TLS write failed");
            return -1;
        }
        p += w;
        n -= (size_t)w;
    }
    return 0;
}

static int tt_read_all(SSL *ssl, struct buf *out, char **err)
{
    for (;;) {
        unsigned char chunk[65536];
        int r = SSL_read(ssl, chunk, (int)sizeof chunk);
        if (r > 0) {
            buf_append(out, chunk, (size_t)r);
            continue;
        }
        int e = SSL_get_error(ssl, r);
        if (e == SSL_ERROR_ZERO_RETURN)
            return 0;
        if (e == SSL_ERROR_WANT_READ || e == SSL_ERROR_WANT_WRITE)
            continue;
        if (e == SSL_ERROR_SYSCALL && r == 0)
            return 0;
#ifdef SSL_R_UNEXPECTED_EOF_WHILE_READING
        if (e == SSL_ERROR_SSL) {
            unsigned long ec = ERR_peek_last_error();
            if (ec &&
                ERR_GET_REASON(ec) == SSL_R_UNEXPECTED_EOF_WHILE_READING)
                return 0;
        }
#endif
        if (e == SSL_ERROR_SYSCALL &&
            (errno == EAGAIN || errno == EWOULDBLOCK)) {
            set_err(err, "receive timed out");
            return -1;
        }
        unsigned long ec = ERR_get_error();
        if (ec)
            set_err(err, "TLS read failed: %s", ERR_error_string(ec, NULL));
        else
            set_err(err, "TLS read failed");
        return -1;
    }
}

/* One GET with Connection: close; returns status, Location (or NULL) and
   the response body. */
static int tt_get_once(const struct tt_url *u, struct buf *body, int *status,
                       char **location, char **err)
{
    int fd = tt_tcp_connect(u, err);
    if (fd < 0)
        return -1;
    SSL_CTX *ctx = SSL_CTX_new(TLS_client_method());
    SSL *ssl = NULL;
    int rc = -1;
    if (!ctx) {
        set_err(err, "cannot create TLS context");
        goto done;
    }
    SSL_CTX_set_min_proto_version(ctx, TLS1_2_VERSION);
    SSL_CTX_set_options(ctx, SSL_OP_NO_COMPRESSION | SSL_OP_NO_RENEGOTIATION);
    SSL_CTX_set_verify(ctx, SSL_VERIFY_PEER, NULL);
    if (tls_load_default_cas(ctx) != 1) {
        set_err(err, "cannot load system CA certificates");
        goto done;
    }
    if (tt_tls_start(ctx, &ssl, fd, u->host, err) != 0)
        goto done;

    struct buf req;
    buf_init(&req);
    static const char head_start[] = "GET ";
    static const char head_mid[] = " HTTP/1.1\r\nHost: ";
    static const char head_end[] = "\r\nUser-Agent: bokf/" BOKF_VERSION
                                   "\r\nAccept: */*\r\nConnection: close\r\n\r\n";
    buf_append(&req, head_start, sizeof head_start - 1);
    buf_append(&req, u->path, strlen(u->path));
    buf_append(&req, head_mid, sizeof head_mid - 1);
    buf_append(&req, u->host, strlen(u->host));
    buf_append(&req, head_end, sizeof head_end - 1);
    int ok = tt_write(ssl, req.p, req.len, err);
    buf_free(&req);
    if (ok != 0)
        goto done;

    struct buf raw;
    buf_init(&raw);
    if (tt_read_all(ssl, &raw, err) != 0) {
        buf_free(&raw);
        goto done;
    }
    size_t hdr_end = 0;
    int found = 0;
    for (size_t i = 0; i + 3 < raw.len; i++) {
        if (raw.p[i] == '\r' && raw.p[i + 1] == '\n' &&
            raw.p[i + 2] == '\r' && raw.p[i + 3] == '\n') {
            hdr_end = i + 4;
            found = 1;
            break;
        }
    }
    if (!found) {
        set_err(err, "malformed HTTP response");
        buf_free(&raw);
        goto done;
    }
    *status = 0;
    if (raw.len >= 12 && memcmp(raw.p, "HTTP/", 5) == 0)
        *status = atoi((const char *)raw.p + 9);
    char line[1024];
    size_t p = 0;
    int first = 1;
    while (p < hdr_end) {
        size_t e = p;
        while (e < hdr_end && raw.p[e] != '\n')
            e++;
        size_t n = e - p;
        if (n > 0 && raw.p[e - 1] == '\r')
            n--;
        if (n >= sizeof line)
            n = sizeof line - 1;
        memcpy(line, raw.p + p, n);
        line[n] = '\0';
        if (!first && n >= 9 && strncasecmp(line, "location:", 9) == 0) {
            const char *v = line + 9;
            while (*v == ' ' || *v == '\t')
                v++;
            free(*location);
            *location = xstrdup(v);
        }
        first = 0;
        p = e + 1;
    }
    for (size_t i = hdr_end; i < raw.len; i++)
        buf_append(body, raw.p + i, 1);
    buf_free(&raw);
    rc = 0;

done:
    if (ssl) {
        SSL_shutdown(ssl);
        SSL_free(ssl);
    }
    if (ctx)
        SSL_CTX_free(ctx);
    close(fd);
    return rc;
}

static int tt_is_absolute(const char *u)
{
    return strncmp(u, "https://", 8) == 0 ||
           strncmp(u, "http://", 7) == 0;
}

static char *tt_resolve(const char *base, const char *location, char **err)
{
    if (tt_is_absolute(location))
        return xstrdup(location);
    struct tt_url u;
    if (tt_url_parse(base, &u, err) != 0)
        return NULL;
    char out[2560];
    if (location[0] == '/') {
        if (u.port == 443)
            snprintf(out, sizeof out, "https://%s%s", u.host, location);
        else
            snprintf(out, sizeof out, "https://%s:%d%s", u.host, u.port,
                     location);
        return xstrdup(out);
    }
    const char *last = strrchr(u.path, '/');
    size_t dir = last ? (size_t)(last - u.path + 1) : 0;
    char path[2048];
    if (dir >= sizeof path) {
        set_err(err, "redirect path is too long");
        return NULL;
    }
    memcpy(path, u.path, dir);
    snprintf(path + dir, sizeof path - dir, "%s", location);
    if (u.port == 443)
        snprintf(out, sizeof out, "https://%s%s", u.host, path);
    else
        snprintf(out, sizeof out, "https://%s:%d%s", u.host, u.port, path);
    return xstrdup(out);
}

int tax_table_https_get(const char *url, struct buf *out, char **err)
{
    char *current = xstrdup(url);
    for (int hop = 0; hop <= TT_MAX_REDIRECTS; hop++) {
        struct tt_url u;
        if (tt_url_parse(current, &u, err) != 0) {
            free(current);
            return -1;
        }
        struct buf body;
        buf_init(&body);
        int status = 0;
        char *location = NULL;
        int rc = tt_get_once(&u, &body, &status, &location, err);
        if (rc != 0) {
            free(location);
            buf_free(&body);
            free(current);
            return -1;
        }
        if (status >= 300 && status < 400 && location && *location) {
            char *next = tt_resolve(current, location, err);
            free(location);
            buf_free(&body);
            free(current);
            if (!next)
                return -1;
            current = next;
            continue;
        }
        if (status != 200) {
            set_err(err, "HTTP %d for %.160s", status, current);
            free(location);
            buf_free(&body);
            free(current);
            return -1;
        }
        free(location);
        free(current);
        *out = body;
        return 0;
    }
    set_err(err, "too many redirects for %.160s", url);
    free(current);
    return -1;
}

/* ------------------------------------------------------------------ */
/* Skatteverket page and download                                      */
/* ------------------------------------------------------------------ */

static char *find_link(const char *html, size_t len, const char *suffix,
                       int index)
{
    size_t slen = strlen(suffix);
    int seen = 0;
    size_t i = 0;
    while (i + 6 < len) {
        if (strncasecmp(html + i, "href=", 5) != 0) {
            i++;
            continue;
        }
        char quote = html[i + 5];
        if (quote != '"' && quote != '\'') {
            i += 5;
            continue;
        }
        size_t start = i + 6;
        size_t end = start;
        while (end < len && html[end] != quote)
            end++;
        size_t hlen = end - start;
        if (hlen >= slen &&
            strncmp(html + end - slen, suffix, slen) == 0) {
            if (seen == index) {
                char *href = xmalloc(hlen + 1);
                memcpy(href, html + start, hlen);
                href[hlen] = '\0';
                char *amp = href;
                while ((amp = strstr(amp, "&amp;")) != NULL) {
                    *amp = '&';
                    memmove(amp + 1, amp + 5, strlen(amp + 5) + 1);
                }
                return href;
            }
            seen++;
        }
        i = end + 1;
    }
    return NULL;
}

int tax_table_fetch_year(int year, unsigned char **data, size_t *len,
                         char **source_url, char **err)
{
    *data = NULL;
    *len = 0;
    *source_url = NULL;
    time_t now = time(NULL);
    struct tm tm;
    gmtime_r(&now, &tm);
    int current_year = tm.tm_year + 1900;
    if (year < 2000 || year > current_year) {
        set_err(err, "no published table for year %d", year);
        return -1;
    }

    struct buf page;
    buf_init(&page);
    if (tax_table_https_get(TT_PAGE_URL, &page, err) != 0) {
        buf_free(&page);
        return -1;
    }
    size_t page_len = page.len;
    buf_append(&page, "", 1);
    char *href = find_link((const char *)page.p, page_len, TT_TABLE_FILE,
                           current_year - year);
    buf_free(&page);
    if (!href) {
        set_err(err, "no %s link for %d on Skatteverket's page", TT_TABLE_FILE,
                year);
        return -1;
    }
    char *abs = tt_resolve(TT_PAGE_URL, href, err);
    free(href);
    if (!abs)
        return -1;

    struct buf file;
    buf_init(&file);
    if (tax_table_https_get(abs, &file, err) != 0) {
        buf_free(&file);
        free(abs);
        return -1;
    }
    *data = file.p;
    *len = file.len;
    *source_url = abs;
    return 0;
}