mirror of
https://github.com/zint/zint
synced 2024-11-16 20:57:25 +13:00
fb3b3001aa
changing internal FNC1 marker '[' -> '\x1D' (GS), ticket #319, props Moli Sojet; also fix non-AI square brackets -> round brackets in GS1_128 HRT - BWIPP: update to latest
1334 lines
49 KiB
C
1334 lines
49 KiB
C
/* code128.c - Handles Code 128 and derivatives */
|
|
/*
|
|
libzint - the open source barcode library
|
|
Copyright (C) 2008-2024 Robin Stuart <rstuart114@gmail.com>
|
|
Bugfixes thanks to Christian Sakowski and BogDan Vatra
|
|
|
|
Redistribution and use in source and binary forms, with or without
|
|
modification, are permitted provided that the following conditions
|
|
are met:
|
|
|
|
1. Redistributions of source code must retain the above copyright
|
|
notice, this list of conditions and the following disclaimer.
|
|
2. Redistributions in binary form must reproduce the above copyright
|
|
notice, this list of conditions and the following disclaimer in the
|
|
documentation and/or other materials provided with the distribution.
|
|
3. Neither the name of the project nor the names of its contributors
|
|
may be used to endorse or promote products derived from this software
|
|
without specific prior written permission.
|
|
|
|
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
|
|
ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
|
|
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
|
DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
|
OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
|
HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
|
LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
|
OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
|
SUCH DAMAGE.
|
|
*/
|
|
/* SPDX-License-Identifier: BSD-3-Clause */
|
|
|
|
#include <assert.h>
|
|
#include <stdio.h>
|
|
#include "common.h"
|
|
#include "code128.h"
|
|
#include "gs1.h"
|
|
|
|
#define C128_SYMBOL_MAX 99
|
|
|
|
static const char KRSET[] = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ";
|
|
#define KRSET_F (IS_NUM_F | IS_UPR_F)
|
|
|
|
/* Code 128 tables checked against ISO/IEC 15417:2007 */
|
|
|
|
INTERNAL_DATA const char C128Table[107][6] = { /* Used by CODABLOCKF and CODE16K also */
|
|
/* Code 128 character encodation - Table 1 (with final CODE16K-only character in place of Stop character) */
|
|
{'2','1','2','2','2','2'}, {'2','2','2','1','2','2'}, {'2','2','2','2','2','1'}, {'1','2','1','2','2','3'},
|
|
{'1','2','1','3','2','2'}, {'1','3','1','2','2','2'}, {'1','2','2','2','1','3'}, {'1','2','2','3','1','2'},
|
|
{'1','3','2','2','1','2'}, {'2','2','1','2','1','3'}, {'2','2','1','3','1','2'}, {'2','3','1','2','1','2'},
|
|
{'1','1','2','2','3','2'}, {'1','2','2','1','3','2'}, {'1','2','2','2','3','1'}, {'1','1','3','2','2','2'},
|
|
{'1','2','3','1','2','2'}, {'1','2','3','2','2','1'}, {'2','2','3','2','1','1'}, {'2','2','1','1','3','2'},
|
|
{'2','2','1','2','3','1'}, {'2','1','3','2','1','2'}, {'2','2','3','1','1','2'}, {'3','1','2','1','3','1'},
|
|
{'3','1','1','2','2','2'}, {'3','2','1','1','2','2'}, {'3','2','1','2','2','1'}, {'3','1','2','2','1','2'},
|
|
{'3','2','2','1','1','2'}, {'3','2','2','2','1','1'}, {'2','1','2','1','2','3'}, {'2','1','2','3','2','1'},
|
|
{'2','3','2','1','2','1'}, {'1','1','1','3','2','3'}, {'1','3','1','1','2','3'}, {'1','3','1','3','2','1'},
|
|
{'1','1','2','3','1','3'}, {'1','3','2','1','1','3'}, {'1','3','2','3','1','1'}, {'2','1','1','3','1','3'},
|
|
{'2','3','1','1','1','3'}, {'2','3','1','3','1','1'}, {'1','1','2','1','3','3'}, {'1','1','2','3','3','1'},
|
|
{'1','3','2','1','3','1'}, {'1','1','3','1','2','3'}, {'1','1','3','3','2','1'}, {'1','3','3','1','2','1'},
|
|
{'3','1','3','1','2','1'}, {'2','1','1','3','3','1'}, {'2','3','1','1','3','1'}, {'2','1','3','1','1','3'},
|
|
{'2','1','3','3','1','1'}, {'2','1','3','1','3','1'}, {'3','1','1','1','2','3'}, {'3','1','1','3','2','1'},
|
|
{'3','3','1','1','2','1'}, {'3','1','2','1','1','3'}, {'3','1','2','3','1','1'}, {'3','3','2','1','1','1'},
|
|
{'3','1','4','1','1','1'}, {'2','2','1','4','1','1'}, {'4','3','1','1','1','1'}, {'1','1','1','2','2','4'},
|
|
{'1','1','1','4','2','2'}, {'1','2','1','1','2','4'}, {'1','2','1','4','2','1'}, {'1','4','1','1','2','2'},
|
|
{'1','4','1','2','2','1'}, {'1','1','2','2','1','4'}, {'1','1','2','4','1','2'}, {'1','2','2','1','1','4'},
|
|
{'1','2','2','4','1','1'}, {'1','4','2','1','1','2'}, {'1','4','2','2','1','1'}, {'2','4','1','2','1','1'},
|
|
{'2','2','1','1','1','4'}, {'4','1','3','1','1','1'}, {'2','4','1','1','1','2'}, {'1','3','4','1','1','1'},
|
|
{'1','1','1','2','4','2'}, {'1','2','1','1','4','2'}, {'1','2','1','2','4','1'}, {'1','1','4','2','1','2'},
|
|
{'1','2','4','1','1','2'}, {'1','2','4','2','1','1'}, {'4','1','1','2','1','2'}, {'4','2','1','1','1','2'},
|
|
{'4','2','1','2','1','1'}, {'2','1','2','1','4','1'}, {'2','1','4','1','2','1'}, {'4','1','2','1','2','1'},
|
|
{'1','1','1','1','4','3'}, {'1','1','1','3','4','1'}, {'1','3','1','1','4','1'}, {'1','1','4','1','1','3'},
|
|
{'1','1','4','3','1','1'}, {'4','1','1','1','1','3'}, {'4','1','1','3','1','1'}, {'1','1','3','1','4','1'},
|
|
{'1','1','4','1','3','1'}, {'3','1','1','1','4','1'}, {'4','1','1','1','3','1'}, {'2','1','1','4','1','2'},
|
|
{'2','1','1','2','1','4'}, {'2','1','1','2','3','2'}, {/* Only used by CODE16K */ '2','1','1','1','3','3'}
|
|
};
|
|
|
|
/* Determine appropriate mode for a given character */
|
|
INTERNAL int c128_parunmodd(const unsigned char llyth) {
|
|
int modd;
|
|
|
|
if (llyth <= 31) {
|
|
modd = C128_SHIFTA;
|
|
} else if ((llyth >= 48) && (llyth <= 57)) {
|
|
modd = C128_ABORC;
|
|
} else if (llyth <= 95) {
|
|
modd = C128_AORB;
|
|
} else if (llyth <= 127) {
|
|
modd = C128_SHIFTB;
|
|
} else if (llyth <= 159) {
|
|
modd = C128_SHIFTA;
|
|
} else if (llyth <= 223) {
|
|
modd = C128_AORB;
|
|
} else {
|
|
modd = C128_SHIFTB;
|
|
}
|
|
|
|
return modd;
|
|
}
|
|
|
|
/**
|
|
* Bring together same type blocks
|
|
*/
|
|
static void c128_grwp(int list[2][C128_MAX], int *p_indexliste) {
|
|
|
|
/* bring together same type blocks */
|
|
if (*p_indexliste > 1) {
|
|
int i = 1;
|
|
while (i < *p_indexliste) {
|
|
if (list[1][i - 1] == list[1][i]) {
|
|
int j;
|
|
/* bring together */
|
|
list[0][i - 1] = list[0][i - 1] + list[0][i];
|
|
j = i + 1;
|
|
|
|
/* decrease the list */
|
|
while (j < *p_indexliste) {
|
|
list[0][j - 1] = list[0][j];
|
|
list[1][j - 1] = list[1][j];
|
|
j++;
|
|
}
|
|
*p_indexliste = *p_indexliste - 1;
|
|
i--;
|
|
}
|
|
i++;
|
|
}
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Implements rules from ISO 15417 Annex E
|
|
*/
|
|
INTERNAL void c128_dxsmooth(int list[2][C128_MAX], int *p_indexliste, const char *manual_set) {
|
|
int i, j, nextshift = 0 /*Suppresses gcc -Wmaybe-uninitialized false positive*/, nextshift_i = 0;
|
|
const int indexliste = *p_indexliste;
|
|
|
|
for (i = 0; i < indexliste; i++) {
|
|
int current = list[1][i]; /* Either C128_ABORC, C128_AORB, C128_SHIFTA or C128_SHIFTB */
|
|
int length = list[0][i];
|
|
if (i == nextshift_i) {
|
|
nextshift = 0;
|
|
/* Set next shift to aid deciding between latching to A or B - taken from Okapi, props Daniel Gredler */
|
|
for (j = i + 1; j < indexliste; j++) {
|
|
if (list[1][j] == C128_SHIFTA || list[1][j] == C128_SHIFTB) {
|
|
nextshift = list[1][j];
|
|
nextshift_i = j;
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
if (i == 0) { /* first block */
|
|
if (current == C128_ABORC) {
|
|
if (manual_set && manual_set[i]) {
|
|
list[1][i] = manual_set[i];
|
|
current = manual_set[i];
|
|
} else if ((indexliste == 1) && (length == 2)) {
|
|
/* Rule 1a */
|
|
list[1][i] = C128_LATCHC;
|
|
current = C128_LATCHC;
|
|
} else if (length >= 4) {
|
|
/* Rule 1b */
|
|
list[1][i] = C128_LATCHC;
|
|
current = C128_LATCHC;
|
|
} else {
|
|
current = C128_AORB; /* Determine below */
|
|
}
|
|
}
|
|
if (current == C128_AORB) {
|
|
if (manual_set && (manual_set[i] == 'A' || manual_set[i] == 'B')) {
|
|
list[1][i] = manual_set[i];
|
|
} else if (nextshift == C128_SHIFTA) {
|
|
/* Rule 1c */
|
|
list[1][i] = C128_LATCHA;
|
|
} else {
|
|
/* Rule 1d */
|
|
list[1][i] = C128_LATCHB;
|
|
}
|
|
} else if (current == C128_SHIFTA) {
|
|
/* Rule 1c */
|
|
list[1][i] = C128_LATCHA;
|
|
} else if (current == C128_SHIFTB) { /* Unless C128_LATCHX set above, can only be C128_SHIFTB */
|
|
/* Rule 1d */
|
|
list[1][i] = C128_LATCHB;
|
|
}
|
|
} else {
|
|
int last = list[1][i - 1];
|
|
if (current == C128_ABORC) {
|
|
if (manual_set && manual_set[i]) {
|
|
list[1][i] = manual_set[i];
|
|
current = manual_set[i];
|
|
} else if (length >= 4) {
|
|
/* Rule 3 - note Rule 3b (odd C blocks) dealt with later */
|
|
list[1][i] = C128_LATCHC;
|
|
current = C128_LATCHC;
|
|
} else {
|
|
current = C128_AORB; /* Determine below */
|
|
}
|
|
}
|
|
if (current == C128_AORB) {
|
|
if (manual_set && (manual_set[i] == 'A' || manual_set[i] == 'B')) {
|
|
list[1][i] = manual_set[i];
|
|
} else if (last == C128_LATCHA || last == C128_SHIFTB) { /* Maintain state */
|
|
list[1][i] = C128_LATCHA;
|
|
} else if (last == C128_LATCHB || last == C128_SHIFTA) { /* Maintain state */
|
|
list[1][i] = C128_LATCHB;
|
|
} else if (nextshift == C128_SHIFTA) {
|
|
list[1][i] = C128_LATCHA;
|
|
} else {
|
|
list[1][i] = C128_LATCHB;
|
|
}
|
|
} else if (current == C128_SHIFTA) {
|
|
if (manual_set && manual_set[i] == 'A') {
|
|
list[1][i] = C128_LATCHA;
|
|
} else if (length > 1) {
|
|
/* Rule 4 */
|
|
list[1][i] = C128_LATCHA;
|
|
} else if (last == C128_LATCHA || last == C128_SHIFTB) { /* Maintain state */
|
|
list[1][i] = C128_LATCHA;
|
|
} else if (last == C128_LATCHC) {
|
|
list[1][i] = C128_LATCHA;
|
|
}
|
|
} else if (current == C128_SHIFTB) { /* Unless C128_LATCHX set above, can only be C128_SHIFTB */
|
|
if (manual_set && manual_set[i] == 'B') {
|
|
list[1][i] = C128_LATCHB;
|
|
} else if (length > 1) {
|
|
/* Rule 5 */
|
|
list[1][i] = C128_LATCHB;
|
|
} else if (last == C128_LATCHB || last == C128_SHIFTA) { /* Maintain state */
|
|
list[1][i] = C128_LATCHB;
|
|
} else if (last == C128_LATCHC) {
|
|
list[1][i] = C128_LATCHB;
|
|
}
|
|
}
|
|
} /* Rule 2 is implemented elsewhere, Rule 6 is implied */
|
|
}
|
|
|
|
c128_grwp(list, p_indexliste);
|
|
}
|
|
|
|
/**
|
|
* Translate Code 128 Set A characters into barcodes.
|
|
* This set handles all control characters NUL to US.
|
|
*/
|
|
INTERNAL void c128_set_a(const unsigned char source, int values[], int *bar_chars) {
|
|
|
|
if (source > 127) {
|
|
if (source < 160) {
|
|
values[(*bar_chars)] = (source - 128) + 64;
|
|
} else {
|
|
values[(*bar_chars)] = (source - 128) - 32;
|
|
}
|
|
} else {
|
|
if (source < 32) {
|
|
values[(*bar_chars)] = source + 64;
|
|
} else {
|
|
values[(*bar_chars)] = source - 32;
|
|
}
|
|
}
|
|
(*bar_chars)++;
|
|
}
|
|
|
|
/**
|
|
* Translate Code 128 Set B characters into barcodes.
|
|
* This set handles all characters which are not part of long numbers and not
|
|
* control characters.
|
|
*/
|
|
INTERNAL int c128_set_b(const unsigned char source, int values[], int *bar_chars) {
|
|
if (source >= 128 + 32) {
|
|
values[(*bar_chars)] = source - 32 - 128;
|
|
} else if (source >= 128) { /* Should never happen */
|
|
return 0; /* Not reached */
|
|
} else if (source >= 32) {
|
|
values[(*bar_chars)] = source - 32;
|
|
} else { /* Should never happen */
|
|
return 0; /* Not reached */
|
|
}
|
|
(*bar_chars)++;
|
|
return 1;
|
|
}
|
|
|
|
/* Translate Code 128 Set C characters into barcodes
|
|
* This set handles numbers in a compressed form
|
|
*/
|
|
INTERNAL void c128_set_c(const unsigned char source_a, const unsigned char source_b, int values[], int *bar_chars) {
|
|
int weight;
|
|
|
|
weight = (10 * ctoi(source_a)) + ctoi(source_b);
|
|
values[(*bar_chars)] = weight;
|
|
(*bar_chars)++;
|
|
}
|
|
|
|
/* Put set data into set[]. If source given (GS1_MODE) then resolves odd C blocks */
|
|
INTERNAL void c128_put_in_set(int list[2][C128_MAX], const int indexliste, char set[C128_MAX],
|
|
const unsigned char *source) {
|
|
int read = 0;
|
|
int i, j;
|
|
|
|
for (i = 0; i < indexliste; i++) {
|
|
for (j = 0; j < list[0][i]; j++) {
|
|
set[read++] = list[1][i];
|
|
}
|
|
}
|
|
if (source) {
|
|
/* Watch out for odd-length Mode C blocks */
|
|
int c_count = 0;
|
|
for (i = 0; i < read; i++) {
|
|
if (set[i] == 'C') {
|
|
if (source[i] == '\x1D') {
|
|
if (c_count & 1) {
|
|
if ((i - c_count) != 0) {
|
|
set[i - c_count] = 'B';
|
|
} else {
|
|
set[i - 1] = 'B';
|
|
}
|
|
}
|
|
c_count = 0;
|
|
} else {
|
|
c_count++;
|
|
}
|
|
} else {
|
|
if (c_count & 1) {
|
|
if ((i - c_count) != 0) {
|
|
set[i - c_count] = 'B';
|
|
} else {
|
|
set[i - 1] = 'B';
|
|
}
|
|
}
|
|
c_count = 0;
|
|
}
|
|
}
|
|
if (c_count & 1) {
|
|
if ((i - c_count) != 0) {
|
|
set[i - c_count] = 'B';
|
|
} else {
|
|
set[i - 1] = 'B';
|
|
}
|
|
}
|
|
for (i = 1; i < read - 1; i++) {
|
|
if ((set[i] == 'C') && ((set[i - 1] == 'B') && (set[i + 1] == 'B'))) {
|
|
set[i] = 'B';
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
/* Handle Code 128, 128B and HIBC 128 */
|
|
INTERNAL int code128(struct zint_symbol *symbol, unsigned char source[], int length) {
|
|
int i, j, k, values[C128_MAX] = {0}, bar_characters = 0, read, total_sum;
|
|
int error_number, indexchaine, indexliste, f_state = 0;
|
|
unsigned char src_buf[C128_MAX + 1];
|
|
unsigned char *src = source;
|
|
char manual_set[C128_MAX] = {0};
|
|
int list[2][C128_MAX] = {{0}};
|
|
char set[C128_MAX] = {0}, fset[C128_MAX], mode, last_set, current_set = ' ';
|
|
int glyph_count = 0; /* Codeword estimate times 2 */
|
|
char dest[1000];
|
|
char *d = dest;
|
|
|
|
/* Suppresses clang-analyzer-core.UndefinedBinaryOperatorResult warning on fset which is fully set */
|
|
assert(length > 0);
|
|
|
|
if (length > C128_MAX) {
|
|
/* This only blocks ridiculously long input - the actual length of the
|
|
resulting barcode depends on the type of data, so this is trapped later */
|
|
sprintf(symbol->errtxt, "340: Input too long (%d character maximum)", C128_MAX);
|
|
return ZINT_ERROR_TOO_LONG;
|
|
}
|
|
|
|
/* Detect special Code Set escapes for Code 128 in extra escape mode only */
|
|
if ((symbol->input_mode & EXTRA_ESCAPE_MODE) && symbol->symbology == BARCODE_CODE128) {
|
|
char manual_ch = '\0';
|
|
j = 0;
|
|
for (i = 0; i < length; i++) {
|
|
if (source[i] == '\\' && i + 2 < length && source[i + 1] == '^'
|
|
&& ((source[i + 2] >= 'A' && source[i + 2] <= 'C') || source[i + 2] == '^')) {
|
|
if (source[i + 2] != '^') {
|
|
i += 2;
|
|
manual_ch = source[i];
|
|
} else { /* Escape sequence '\^^' */
|
|
manual_set[j] = manual_ch;
|
|
src_buf[j++] = source[i++];
|
|
manual_set[j] = manual_ch;
|
|
src_buf[j++] = source[i++];
|
|
/* Drop second '^' */
|
|
}
|
|
} else {
|
|
manual_set[j] = manual_ch;
|
|
src_buf[j++] = source[i];
|
|
}
|
|
}
|
|
if (j != length) {
|
|
length = j;
|
|
if (length == 0) {
|
|
strcpy(symbol->errtxt, "842: No input data");
|
|
return ZINT_ERROR_INVALID_DATA;
|
|
}
|
|
src = src_buf;
|
|
src[length] = '\0';
|
|
if (symbol->debug & ZINT_DEBUG_PRINT) {
|
|
fputs("MSet: ", stdout);
|
|
for (i = 0; i < length; i++) printf("%c", manual_set[i] ? manual_set[i] : '.');
|
|
fputc('\n', stdout);
|
|
}
|
|
}
|
|
}
|
|
|
|
/* Detect extended ASCII characters */
|
|
for (i = 0; i < length; i++) {
|
|
fset[i] = src[i] >= 128 ? 'f' : ' ';
|
|
}
|
|
|
|
/* Decide when to latch to extended mode - Annex E note 3 */
|
|
j = 0;
|
|
for (i = 0; i < length; i++) {
|
|
if (fset[i] == 'f') {
|
|
j++;
|
|
} else {
|
|
j = 0;
|
|
}
|
|
|
|
if (j >= 5) {
|
|
for (k = i; k > (i - 5); k--) {
|
|
fset[k] = 'F';
|
|
}
|
|
}
|
|
|
|
if ((j >= 3) && (i == (length - 1))) {
|
|
for (k = i; k > (i - 3); k--) {
|
|
fset[k] = 'F';
|
|
}
|
|
}
|
|
}
|
|
|
|
/* Decide if it is worth reverting to 646 encodation for a few characters as described in 4.3.4.2 (d) */
|
|
for (i = 1; i < length; i++) {
|
|
if ((fset[i - 1] == 'F') && (fset[i] == ' ')) {
|
|
/* Detected a change from 8859-1 to 646 - count how long for */
|
|
for (j = 0; ((i + j) < length) && (fset[i + j] == ' '); j++);
|
|
/* Count how many 8859-1 beyond */
|
|
k = 0;
|
|
if (i + j < length) {
|
|
for (k = 1; ((i + j + k) < length) && (fset[i + j + k] != ' '); k++);
|
|
}
|
|
if (j < 3 || (j < 5 && k > 2)) {
|
|
/* Change to shifting back rather than latching back */
|
|
/* Inverts the same figures recommended by Annex E note 3 */
|
|
for (k = 0; k < j; k++) {
|
|
fset[i + k] = 'n';
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
/* Decide on mode using same system as PDF417 and rules of ISO 15417 Annex E */
|
|
indexliste = 0;
|
|
indexchaine = 0;
|
|
|
|
mode = c128_parunmodd(src[indexchaine]);
|
|
if (mode == C128_ABORC
|
|
&& (symbol->symbology == BARCODE_CODE128AB
|
|
|| (manual_set[indexchaine] == 'A' || manual_set[indexchaine] == 'B'))) {
|
|
mode = C128_AORB;
|
|
}
|
|
|
|
do {
|
|
list[1][indexliste] = mode;
|
|
while ((list[1][indexliste] == mode) && (indexchaine < length)) {
|
|
list[0][indexliste]++;
|
|
indexchaine++;
|
|
if (indexchaine == length) {
|
|
break;
|
|
}
|
|
mode = c128_parunmodd(src[indexchaine]);
|
|
if (mode == C128_ABORC
|
|
&& (symbol->symbology == BARCODE_CODE128AB
|
|
|| (manual_set[indexchaine] == 'A' || manual_set[indexchaine] == 'B'))) {
|
|
mode = C128_AORB;
|
|
}
|
|
if (manual_set[indexchaine] != manual_set[indexchaine - 1]) {
|
|
break;
|
|
}
|
|
}
|
|
indexliste++;
|
|
} while (indexchaine < length);
|
|
|
|
if (src == src_buf) {
|
|
/* Need to re-index `manual_set` to have sames indexes as `list` blocks for `c128_dxsmooth()` */
|
|
j = 0;
|
|
for (i = 1; i < indexliste; i++) {
|
|
j += list[0][i - 1];
|
|
manual_set[i] = manual_set[j];
|
|
}
|
|
}
|
|
c128_dxsmooth(list, &indexliste, src == src_buf ? manual_set : NULL);
|
|
|
|
/* Resolve odd length C128_LATCHC blocks */
|
|
if ((list[1][0] == C128_LATCHC) && (list[0][0] & 1)) {
|
|
/* Rule 2 */
|
|
list[0][1]++;
|
|
list[0][0]--;
|
|
if (indexliste == 1) {
|
|
list[0][1] = 1;
|
|
list[1][1] = C128_LATCHB;
|
|
indexliste = 2;
|
|
}
|
|
}
|
|
if (indexliste > 1) {
|
|
for (i = 1; i < indexliste; i++) {
|
|
if ((list[1][i] == C128_LATCHC) && (list[0][i] & 1)) {
|
|
/* Rule 3b */
|
|
list[0][i - 1]++;
|
|
list[0][i]--;
|
|
}
|
|
}
|
|
}
|
|
|
|
/* Put set data into set[]. Giving NULL as source as used to resolve odd C blocks which has been done above */
|
|
c128_put_in_set(list, indexliste, set, NULL /*source*/);
|
|
|
|
if (symbol->debug & ZINT_DEBUG_PRINT) {
|
|
printf("Data: %.*s (%d)\n", length, src, length);
|
|
printf(" Set: %.*s\n", length, set);
|
|
printf("FSet: %.*s\n", length, fset);
|
|
}
|
|
|
|
/* Now we can calculate how long the barcode is going to be - and stop it from
|
|
being too long */
|
|
last_set = set[0];
|
|
for (i = 0; i < length; i++) {
|
|
if ((set[i] == 'a') || (set[i] == 'b')) {
|
|
glyph_count += 2; /* 1 codeword */
|
|
}
|
|
if ((fset[i] == 'f') || (fset[i] == 'n')) {
|
|
glyph_count += 2; /* May be overestimate if in latch */
|
|
}
|
|
if (((set[i] == 'A') || (set[i] == 'B')) || (set[i] == 'C')) {
|
|
if (set[i] != last_set) {
|
|
last_set = set[i];
|
|
glyph_count += 2;
|
|
}
|
|
}
|
|
if (i == 0) {
|
|
if (fset[i] == 'F') {
|
|
glyph_count += 4; /* 2 codewords */
|
|
}
|
|
} else {
|
|
if ((fset[i] == 'F') && (fset[i - 1] != 'F')) {
|
|
glyph_count += 4;
|
|
}
|
|
if ((fset[i] != 'F') && (fset[i - 1] == 'F')) {
|
|
glyph_count += 4;
|
|
}
|
|
}
|
|
|
|
if (set[i] == 'C') {
|
|
glyph_count += 1; /* Half a codeword */
|
|
} else {
|
|
glyph_count += 2;
|
|
}
|
|
}
|
|
if (glyph_count > C128_SYMBOL_MAX * 2) {
|
|
sprintf(symbol->errtxt, "341: Input too long (%d symbol character maximum)", C128_SYMBOL_MAX);
|
|
return ZINT_ERROR_TOO_LONG;
|
|
}
|
|
|
|
/* So now we know what start character to use - we can get on with it! */
|
|
if (symbol->output_options & READER_INIT) {
|
|
/* Reader Initialisation mode */
|
|
switch (set[0]) {
|
|
case 'A': /* Start A */
|
|
values[bar_characters++] = 103;
|
|
current_set = 'A';
|
|
values[bar_characters++] = 96; /* FNC3 */
|
|
break;
|
|
case 'B': /* Start B */
|
|
values[bar_characters++] = 104;
|
|
current_set = 'B';
|
|
values[bar_characters++] = 96; /* FNC3 */
|
|
break;
|
|
case 'C': /* Start C */
|
|
values[bar_characters++] = 104; /* Start B */
|
|
values[bar_characters++] = 96; /* FNC3 */
|
|
values[bar_characters++] = 99; /* Code C */
|
|
current_set = 'C';
|
|
break;
|
|
}
|
|
} else {
|
|
/* Normal mode */
|
|
switch (set[0]) {
|
|
case 'A': /* Start A */
|
|
values[bar_characters++] = 103;
|
|
current_set = 'A';
|
|
break;
|
|
case 'B': /* Start B */
|
|
values[bar_characters++] = 104;
|
|
current_set = 'B';
|
|
break;
|
|
case 'C': /* Start C */
|
|
values[bar_characters++] = 105;
|
|
current_set = 'C';
|
|
break;
|
|
}
|
|
}
|
|
|
|
if (fset[0] == 'F') {
|
|
switch (current_set) {
|
|
case 'A':
|
|
values[bar_characters++] = 101;
|
|
values[bar_characters++] = 101;
|
|
f_state = 1;
|
|
break;
|
|
case 'B':
|
|
values[bar_characters++] = 100;
|
|
values[bar_characters++] = 100;
|
|
f_state = 1;
|
|
break;
|
|
}
|
|
}
|
|
|
|
/* Encode the data */
|
|
read = 0;
|
|
do {
|
|
|
|
if ((read != 0) && (set[read] != current_set)) {
|
|
/* Latch different code set */
|
|
switch (set[read]) {
|
|
case 'A':
|
|
values[bar_characters++] = 101;
|
|
current_set = 'A';
|
|
break;
|
|
case 'B':
|
|
values[bar_characters++] = 100;
|
|
current_set = 'B';
|
|
break;
|
|
case 'C':
|
|
values[bar_characters++] = 99;
|
|
current_set = 'C';
|
|
break;
|
|
}
|
|
}
|
|
|
|
if (read != 0) {
|
|
if ((fset[read] == 'F') && (f_state == 0)) {
|
|
/* Latch beginning of extended mode */
|
|
switch (current_set) {
|
|
case 'A':
|
|
values[bar_characters++] = 101;
|
|
values[bar_characters++] = 101;
|
|
f_state = 1;
|
|
break;
|
|
case 'B':
|
|
values[bar_characters++] = 100;
|
|
values[bar_characters++] = 100;
|
|
f_state = 1;
|
|
break;
|
|
}
|
|
}
|
|
if ((fset[read] == ' ') && (f_state == 1)) {
|
|
/* Latch end of extended mode */
|
|
switch (current_set) {
|
|
case 'A':
|
|
values[bar_characters++] = 101;
|
|
values[bar_characters++] = 101;
|
|
f_state = 0;
|
|
break;
|
|
case 'B':
|
|
values[bar_characters++] = 100;
|
|
values[bar_characters++] = 100;
|
|
f_state = 0;
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
if ((fset[read] == 'f' && f_state == 0) || (fset[read] == 'n' && f_state == 1)) {
|
|
/* Shift to or from extended mode */
|
|
switch (current_set) {
|
|
case 'A':
|
|
values[bar_characters++] = 101; /* FNC 4 */
|
|
break;
|
|
case 'B':
|
|
values[bar_characters++] = 100; /* FNC 4 */
|
|
break;
|
|
}
|
|
}
|
|
|
|
if ((set[read] == 'a') || (set[read] == 'b')) {
|
|
/* Insert shift character */
|
|
values[bar_characters++] = 98;
|
|
}
|
|
|
|
switch (set[read]) { /* Encode data characters */
|
|
case 'a':
|
|
case 'A': c128_set_a(src[read], values, &bar_characters);
|
|
read++;
|
|
break;
|
|
case 'b':
|
|
case 'B': (void) c128_set_b(src[read], values, &bar_characters);
|
|
read++;
|
|
break;
|
|
case 'C': c128_set_c(src[read], src[read + 1], values, &bar_characters);
|
|
read += 2;
|
|
break;
|
|
}
|
|
|
|
} while (read < length);
|
|
|
|
/* Destination setting and check digit calculation */
|
|
memcpy(d, C128Table[values[0]], 6);
|
|
d += 6;
|
|
total_sum = values[0];
|
|
|
|
for (i = 1; i < bar_characters; i++, d += 6) {
|
|
memcpy(d, C128Table[values[i]], 6);
|
|
total_sum += values[i] * i; /* Note can't overflow as 106 * C128_SYMBOL_MAX * C128_SYMBOL_MAX = 1038906 */
|
|
}
|
|
total_sum %= 103;
|
|
memcpy(d, C128Table[total_sum], 6);
|
|
d += 6;
|
|
values[bar_characters++] = total_sum;
|
|
|
|
/* Stop character */
|
|
memcpy(d, "2331112", 7);
|
|
d += 7;
|
|
values[bar_characters++] = 106;
|
|
|
|
if (symbol->debug & ZINT_DEBUG_PRINT) {
|
|
fputs("Codewords:", stdout);
|
|
for (i = 0; i < bar_characters; i++) {
|
|
printf(" %d", values[i]);
|
|
}
|
|
printf(" (%d)\n", bar_characters);
|
|
printf("Barspaces: %.*s\n", (int) (d - dest), dest);
|
|
printf("Checksum: %d\n", total_sum);
|
|
}
|
|
#ifdef ZINT_TEST
|
|
if (symbol->debug & ZINT_DEBUG_TEST) {
|
|
debug_test_codeword_dump_int(symbol, values, bar_characters);
|
|
}
|
|
#endif
|
|
|
|
expand(symbol, dest, d - dest);
|
|
|
|
/* ISO/IEC 15417:2007 leaves dimensions/height as application specification */
|
|
|
|
error_number = hrt_cpy_iso8859_1(symbol, src, length);
|
|
|
|
return error_number;
|
|
}
|
|
|
|
/* Handle EAN-128 (Now known as GS1-128), and composite version if `cc_mode` set */
|
|
INTERNAL int gs1_128_cc(struct zint_symbol *symbol, unsigned char source[], int length, const int cc_mode,
|
|
const int cc_rows) {
|
|
int i, values[C128_MAX] = {0}, bar_characters = 0, read, total_sum;
|
|
int error_number, indexchaine, indexliste;
|
|
int list[2][C128_MAX] = {{0}};
|
|
char set[C128_MAX] = {0}, mode, last_set;
|
|
int glyph_count = 0; /* Codeword estimate times 2 */
|
|
char dest[1000];
|
|
char *d = dest;
|
|
int separator_row = 0, linkage_flag = 0;
|
|
int reduced_length;
|
|
unsigned char *reduced = (unsigned char *) z_alloca(length + 1);
|
|
|
|
if (length > C128_MAX) {
|
|
/* This only blocks ridiculously long input - the actual length of the
|
|
resulting barcode depends on the type of data, so this is trapped later */
|
|
sprintf(symbol->errtxt, "342: Input too long (%d character maximum)", C128_MAX);
|
|
return ZINT_ERROR_TOO_LONG;
|
|
}
|
|
|
|
/* if part of a composite symbol make room for the separator pattern */
|
|
if (symbol->symbology == BARCODE_GS1_128_CC) {
|
|
separator_row = symbol->rows;
|
|
symbol->row_height[symbol->rows] = 1;
|
|
symbol->rows += 1;
|
|
}
|
|
|
|
error_number = gs1_verify(symbol, source, length, reduced);
|
|
if (error_number >= ZINT_ERROR) {
|
|
return error_number;
|
|
}
|
|
|
|
reduced_length = (int) ustrlen(reduced);
|
|
|
|
/* Decide on mode using same system as PDF417 and rules of ISO 15417 Annex E */
|
|
indexliste = 0;
|
|
indexchaine = 0;
|
|
|
|
mode = c128_parunmodd(reduced[indexchaine]);
|
|
|
|
do {
|
|
list[1][indexliste] = mode;
|
|
while ((list[1][indexliste] == mode) && (indexchaine < reduced_length)) {
|
|
list[0][indexliste]++;
|
|
indexchaine++;
|
|
if (indexchaine == reduced_length) {
|
|
break;
|
|
}
|
|
mode = c128_parunmodd(reduced[indexchaine]);
|
|
if (reduced[indexchaine] == '\x1D') {
|
|
mode = C128_ABORC;
|
|
}
|
|
}
|
|
indexliste++;
|
|
} while (indexchaine < reduced_length);
|
|
|
|
c128_dxsmooth(list, &indexliste, NULL /*manual_set*/);
|
|
|
|
/* Put set data into set[], resolving odd C blocks */
|
|
c128_put_in_set(list, indexliste, set, reduced);
|
|
|
|
if (symbol->debug & ZINT_DEBUG_PRINT) {
|
|
printf("Data: %s (%d)\n", reduced, reduced_length);
|
|
printf(" Set: %.*s\n", reduced_length, set);
|
|
}
|
|
|
|
/* Now we can calculate how long the barcode is going to be - and stop it from
|
|
being too long */
|
|
last_set = set[0];
|
|
for (i = 0; i < reduced_length; i++) {
|
|
if ((set[i] == 'A') || (set[i] == 'B') || (set[i] == 'C')) {
|
|
if (set[i] != last_set) {
|
|
last_set = set[i];
|
|
glyph_count += 2; /* 1 codeword */
|
|
}
|
|
} else if ((set[i] == 'a') || (set[i] == 'b')) {
|
|
glyph_count += 2; /* Not reached */
|
|
}
|
|
|
|
if ((set[i] == 'C') && (reduced[i] != '\x1D')) {
|
|
glyph_count += 1; /* Half a codeword */
|
|
} else {
|
|
glyph_count += 2;
|
|
}
|
|
}
|
|
if (glyph_count > C128_SYMBOL_MAX * 2) {
|
|
sprintf(symbol->errtxt, "344: Input too long (%d symbol character maximum)", C128_SYMBOL_MAX);
|
|
return ZINT_ERROR_TOO_LONG;
|
|
}
|
|
|
|
/* So now we know what start character to use - we can get on with it! */
|
|
switch (set[0]) {
|
|
case 'A': /* Start A */
|
|
values[bar_characters++] = 103; /* Not reached */
|
|
break;
|
|
case 'B': /* Start B */
|
|
values[bar_characters++] = 104;
|
|
break;
|
|
case 'C': /* Start C */
|
|
values[bar_characters++] = 105;
|
|
break;
|
|
}
|
|
|
|
values[bar_characters++] = 102;
|
|
|
|
/* Encode the data */
|
|
read = 0;
|
|
do {
|
|
|
|
if ((read != 0) && (set[read] != set[read - 1])) { /* Latch different code set */
|
|
switch (set[read]) {
|
|
case 'A':
|
|
values[bar_characters++] = 101; /* Not reached */
|
|
break;
|
|
case 'B':
|
|
values[bar_characters++] = 100;
|
|
break;
|
|
case 'C':
|
|
values[bar_characters++] = 99;
|
|
break;
|
|
}
|
|
}
|
|
|
|
if ((set[read] == 'a') || (set[read] == 'b')) {
|
|
/* Insert shift character */
|
|
values[bar_characters++] = 98; /* Not reached */
|
|
}
|
|
|
|
if (reduced[read] != '\x1D') {
|
|
switch (set[read]) { /* Encode data characters */
|
|
case 'A':
|
|
case 'a':
|
|
c128_set_a(reduced[read], values, &bar_characters); /* Not reached */
|
|
read++;
|
|
break;
|
|
case 'B':
|
|
case 'b':
|
|
(void) c128_set_b(reduced[read], values, &bar_characters);
|
|
read++;
|
|
break;
|
|
case 'C':
|
|
c128_set_c(reduced[read], reduced[read + 1], values, &bar_characters);
|
|
read += 2;
|
|
break;
|
|
}
|
|
} else {
|
|
values[bar_characters++] = 102;
|
|
read++;
|
|
}
|
|
} while (read < reduced_length);
|
|
|
|
/* "...note that the linkage flag is an extra code set character between
|
|
the last data character and the Symbol Check Character" (GS1 Specification) */
|
|
|
|
/* Linkage flags in GS1-128 are determined by ISO/IEC 24723 section 7.4 */
|
|
|
|
switch (cc_mode) {
|
|
case 1:
|
|
case 2:
|
|
/* CC-A or CC-B 2D component */
|
|
switch (set[reduced_length - 1]) {
|
|
case 'A': linkage_flag = 100; /* Not reached */
|
|
break;
|
|
case 'B': linkage_flag = 99;
|
|
break;
|
|
case 'C': linkage_flag = 101;
|
|
break;
|
|
}
|
|
break;
|
|
case 3:
|
|
/* CC-C 2D component */
|
|
switch (set[reduced_length - 1]) {
|
|
case 'A': linkage_flag = 99; /* Not reached */
|
|
break;
|
|
case 'B': linkage_flag = 101;
|
|
break;
|
|
case 'C': linkage_flag = 100;
|
|
break;
|
|
}
|
|
break;
|
|
}
|
|
|
|
if (linkage_flag != 0) {
|
|
values[bar_characters++] = linkage_flag;
|
|
}
|
|
|
|
/* Destination setting and check digit calculation */
|
|
memcpy(d, C128Table[values[0]], 6);
|
|
d += 6;
|
|
total_sum = values[0];
|
|
|
|
for (i = 1; i < bar_characters; i++, d += 6) {
|
|
memcpy(d, C128Table[values[i]], 6);
|
|
total_sum += values[i] * i; /* Note can't overflow as 106 * C128_SYMBOL_MAX * C128_SYMBOL_MAX = 1038906 */
|
|
}
|
|
total_sum %= 103;
|
|
memcpy(d, C128Table[total_sum], 6);
|
|
d += 6;
|
|
values[bar_characters++] = total_sum;
|
|
|
|
/* Stop character */
|
|
memcpy(d, "2331112", 7);
|
|
d += 7;
|
|
values[bar_characters++] = 106;
|
|
|
|
if (symbol->debug & ZINT_DEBUG_PRINT) {
|
|
fputs("Codewords:", stdout);
|
|
for (i = 0; i < bar_characters; i++) {
|
|
printf(" %d", values[i]);
|
|
}
|
|
printf(" (%d)\n", bar_characters);
|
|
printf("Barspaces: %.*s\n", (int) (d - dest), dest);
|
|
}
|
|
#ifdef ZINT_TEST
|
|
if (symbol->debug & ZINT_DEBUG_TEST) {
|
|
debug_test_codeword_dump_int(symbol, values, bar_characters);
|
|
}
|
|
#endif
|
|
|
|
expand(symbol, dest, d - dest);
|
|
|
|
/* Add the separator pattern for composite symbols */
|
|
if (symbol->symbology == BARCODE_GS1_128_CC) {
|
|
for (i = 0; i < symbol->width; i++) {
|
|
if (!(module_is_set(symbol, separator_row + 1, i))) {
|
|
set_module(symbol, separator_row, i);
|
|
}
|
|
}
|
|
}
|
|
|
|
if (reduced_length > 48) { /* GS1 General Specifications 5.4.4.3 */
|
|
if (error_number == 0) { /* Don't overwrite any `gs1_verify()` warning */
|
|
strcpy(symbol->errtxt, "843: GS1-128 input too long (48 character maximum)");
|
|
error_number = ZINT_WARN_NONCOMPLIANT;
|
|
}
|
|
}
|
|
|
|
if (symbol->output_options & COMPLIANT_HEIGHT) {
|
|
/* GS1 General Specifications 21.0.1 5.12.3.2 table 2, including footnote (**):
|
|
same as ITF-14: "in case of further space constraints" height 5.8mm / 1.016mm (X max) ~ 5.7;
|
|
default 31.75mm / 0.495mm ~ 64.14 */
|
|
const float min_height = 5.70866156f; /* 5.8 / 1.016 */
|
|
const float default_height = 64.1414108f; /* 31.75 / 0.495 */
|
|
if (symbol->symbology == BARCODE_GS1_128_CC) {
|
|
/* Pass back via temporary linear structure */
|
|
symbol->height = symbol->height ? min_height : default_height;
|
|
} else {
|
|
if (error_number == 0) { /* Don't overwrite any `gs1_verify()` warning */
|
|
error_number = set_height(symbol, min_height, default_height, 0.0f, 0 /*no_errtxt*/);
|
|
} else {
|
|
(void) set_height(symbol, min_height, default_height, 0.0f, 1 /*no_errtxt*/);
|
|
}
|
|
}
|
|
} else {
|
|
const float height = 50.0f;
|
|
if (symbol->symbology == BARCODE_GS1_128_CC) {
|
|
symbol->height = height - cc_rows * (cc_mode == 3 ? 3 : 2) - 1.0f;
|
|
} else {
|
|
(void) set_height(symbol, 0.0f, height, 0.0f, 1 /*no_errtxt*/);
|
|
}
|
|
}
|
|
|
|
if (symbol->input_mode & GS1PARENS_MODE) {
|
|
i = length < (int) sizeof(symbol->text) ? length : (int) sizeof(symbol->text);
|
|
memcpy(symbol->text, source, i);
|
|
} else {
|
|
int bracket_level = 0; /* Non-compliant closing square brackets may be in text */
|
|
for (i = 0; i < length && i < (int) sizeof(symbol->text); i++) {
|
|
if (source[i] == '[') {
|
|
symbol->text[i] = '(';
|
|
bracket_level++;
|
|
} else if (source[i] == ']' && bracket_level) {
|
|
symbol->text[i] = ')';
|
|
bracket_level--;
|
|
} else {
|
|
symbol->text[i] = source[i];
|
|
}
|
|
}
|
|
}
|
|
if (i == sizeof(symbol->text)) {
|
|
/* Trumps all other warnings */
|
|
strcpy(symbol->errtxt, "844: Human Readable Text truncated");
|
|
error_number = ZINT_WARN_HRT_TRUNCATED;
|
|
i--;
|
|
}
|
|
symbol->text[i] = '\0';
|
|
|
|
return error_number;
|
|
}
|
|
|
|
/* Handle EAN-128 (Now known as GS1-128) */
|
|
INTERNAL int gs1_128(struct zint_symbol *symbol, unsigned char source[], int length) {
|
|
return gs1_128_cc(symbol, source, length, 0 /*cc_mode*/, 0 /*cc_rows*/);
|
|
}
|
|
|
|
/* Add check digit if encoding an NVE18 symbol */
|
|
INTERNAL int nve18(struct zint_symbol *symbol, unsigned char source[], int length) {
|
|
int error_number, zeroes;
|
|
unsigned char ean128_equiv[23];
|
|
|
|
if (length > 17) {
|
|
strcpy(symbol->errtxt, "345: Input too long (17 character maximum)");
|
|
return ZINT_ERROR_TOO_LONG;
|
|
}
|
|
|
|
if (!is_sane(NEON_F, source, length)) {
|
|
strcpy(symbol->errtxt, "346: Invalid character in data (digits only)");
|
|
return ZINT_ERROR_INVALID_DATA;
|
|
}
|
|
|
|
zeroes = 17 - length;
|
|
ustrcpy(ean128_equiv, symbol->input_mode & GS1PARENS_MODE ? "(00)" : "[00]");
|
|
memset(ean128_equiv + 4, '0', zeroes);
|
|
ustrcpy(ean128_equiv + 4 + zeroes, source);
|
|
|
|
ean128_equiv[21] = gs1_check_digit(ean128_equiv + 4, 17);
|
|
ean128_equiv[22] = '\0';
|
|
|
|
error_number = gs1_128(symbol, ean128_equiv, 22);
|
|
|
|
return error_number;
|
|
}
|
|
|
|
/* EAN-14 - A version of EAN-128 */
|
|
INTERNAL int ean14(struct zint_symbol *symbol, unsigned char source[], int length) {
|
|
int error_number, zeroes;
|
|
unsigned char ean128_equiv[19];
|
|
|
|
if (length > 13) {
|
|
strcpy(symbol->errtxt, "347: Input too long (13 character maximum)");
|
|
return ZINT_ERROR_TOO_LONG;
|
|
}
|
|
|
|
if (!is_sane(NEON_F, source, length)) {
|
|
strcpy(symbol->errtxt, "348: Invalid character in data (digits only)");
|
|
return ZINT_ERROR_INVALID_DATA;
|
|
}
|
|
|
|
zeroes = 13 - length;
|
|
ustrcpy(ean128_equiv, symbol->input_mode & GS1PARENS_MODE ? "(01)" : "[01]");
|
|
memset(ean128_equiv + 4, '0', zeroes);
|
|
ustrcpy(ean128_equiv + 4 + zeroes, source);
|
|
|
|
ean128_equiv[17] = gs1_check_digit(ean128_equiv + 4, 13);
|
|
ean128_equiv[18] = '\0';
|
|
|
|
error_number = gs1_128(symbol, ean128_equiv, 18);
|
|
|
|
return error_number;
|
|
}
|
|
|
|
/* DPD (Deutscher Paketdienst) Code */
|
|
/* Specification at https://esolutions.dpd.com/dokumente/DPD_Parcel_Label_Specification_2.4.1_EN.pdf
|
|
* and identification tag info (Barcode ID) at https://esolutions.dpd.com/dokumente/DPD_Routing_Database_1.3_EN.pdf */
|
|
INTERNAL int dpd(struct zint_symbol *symbol, unsigned char source[], int length) {
|
|
int error_number = 0;
|
|
int i, p;
|
|
unsigned char ident_tag;
|
|
unsigned char local_source_buf[29];
|
|
unsigned char *local_source;
|
|
const int mod = 36;
|
|
const int relabel = symbol->option_2 == 1; /* A "relabel" has no identification tag */
|
|
int cd; /* Check digit */
|
|
|
|
if ((length != 27 && length != 28) || (length == 28 && relabel)) {
|
|
if (relabel) {
|
|
strcpy(symbol->errtxt, "830: DPD relabel input wrong length (27 characters required)");
|
|
} else {
|
|
strcpy(symbol->errtxt, "349: DPD input wrong length (27 or 28 characters required)");
|
|
}
|
|
return ZINT_ERROR_TOO_LONG;
|
|
}
|
|
|
|
if (length == 27 && !relabel) {
|
|
local_source_buf[0] = '%';
|
|
ustrcpy(local_source_buf + 1, source);
|
|
local_source = local_source_buf;
|
|
length++;
|
|
} else {
|
|
local_source = source;
|
|
}
|
|
|
|
ident_tag = local_source[0];
|
|
|
|
to_upper(local_source + !relabel, length - !relabel);
|
|
if (!is_sane(KRSET_F, local_source + !relabel, length - !relabel)) {
|
|
if (local_source == local_source_buf || relabel) {
|
|
strcpy(symbol->errtxt, "300: Invalid character in data (alphanumerics only)");
|
|
} else {
|
|
strcpy(symbol->errtxt, "299: Invalid character in data (alphanumerics only after first character)");
|
|
}
|
|
return ZINT_ERROR_INVALID_DATA;
|
|
}
|
|
|
|
if ((ident_tag < 32) || (ident_tag > 127)) {
|
|
strcpy(symbol->errtxt, "343: Invalid DPD identification tag (first character), ASCII values 32 to 127 only");
|
|
return ZINT_ERROR_INVALID_DATA;
|
|
}
|
|
|
|
(void) code128(symbol, local_source, length); /* Only error returned is for large text which can't happen */
|
|
|
|
if (!(symbol->output_options & (BARCODE_BOX | BARCODE_BIND | BARCODE_BIND_TOP))) {
|
|
/* If no option has been selected then uses default bind top option */
|
|
symbol->output_options |= BARCODE_BIND_TOP; /* Note won't extend over quiet zones for DPD */
|
|
if (symbol->border_width == 0) { /* Allow override if non-zero */
|
|
symbol->border_width = 3; /* From examples, not mentioned in spec */
|
|
}
|
|
}
|
|
|
|
if (symbol->output_options & COMPLIANT_HEIGHT) {
|
|
/* DPD Parcel Label Specification Version 2.4.1 (19.01.2021) Section 4.6.1.2
|
|
25mm / 0.4mm (X max) = 62.5 min, 25mm / 0.375 (X) ~ 66.66 default */
|
|
if (relabel) { /* If relabel then half-size */
|
|
const float default_height = 33.3333321f; /* 12.5 / 0.375 */
|
|
error_number = set_height(symbol, 31.25f, default_height, 0.0f, 0 /*no_errtxt*/);
|
|
} else {
|
|
const float default_height = 66.6666641f; /* 25.0 / 0.375 */
|
|
error_number = set_height(symbol, 62.5f, default_height, 0.0f, 0 /*no_errtxt*/);
|
|
}
|
|
} else {
|
|
(void) set_height(symbol, 0.0f, relabel ? 25.0f : 50.0f, 0.0f, 1 /*no_errtxt*/);
|
|
}
|
|
|
|
cd = mod;
|
|
|
|
p = 0;
|
|
for (i = !relabel; i < length; i++) {
|
|
symbol->text[p] = local_source[i];
|
|
p++;
|
|
|
|
cd += posn(KRSET, local_source[i]);
|
|
if (cd > mod) cd -= mod;
|
|
cd *= 2;
|
|
if (cd >= (mod + 1)) cd -= mod + 1;
|
|
|
|
switch (i + relabel) {
|
|
case 4:
|
|
case 7:
|
|
case 11:
|
|
case 15:
|
|
case 19:
|
|
case 21:
|
|
case 24:
|
|
case 27:
|
|
symbol->text[p] = ' ';
|
|
p++;
|
|
break;
|
|
}
|
|
}
|
|
|
|
cd = mod + 1 - cd;
|
|
if (cd == mod) cd = 0;
|
|
|
|
if (cd < 10) {
|
|
symbol->text[p] = cd + '0';
|
|
} else {
|
|
symbol->text[p] = (cd - 10) + 'A';
|
|
}
|
|
p++;
|
|
|
|
symbol->text[p] = '\0';
|
|
|
|
/* Some compliance checks */
|
|
if (!is_sane(NEON_F, local_source + length - 16, 16)) {
|
|
if (!is_sane(NEON_F, local_source + length - 3, 3)) { /* 3-digit Country Code (ISO 3166-1) */
|
|
strcpy(symbol->errtxt, "831: Destination Country Code (last 3 characters) should be numeric");
|
|
} else if (!is_sane(NEON_F, local_source + length - 6, 3)) { /* 3-digit Service Code */
|
|
strcpy(symbol->errtxt, "832: Service Code (characters 6-4 from end) should be numeric");
|
|
} else { /* Last 10 characters of Tracking No. */
|
|
strcpy(symbol->errtxt,
|
|
"833: Last 10 characters of Tracking Number (characters 16-7 from end) should be numeric");
|
|
}
|
|
error_number = ZINT_WARN_NONCOMPLIANT;
|
|
}
|
|
|
|
return error_number;
|
|
}
|
|
|
|
/* Universal Postal Union S10 */
|
|
/* https://www.upu.int/UPU/media/upu/files/postalSolutions/programmesAndServices/standards/S10-12.pdf */
|
|
INTERNAL int upu_s10(struct zint_symbol *symbol, unsigned char source[], int length) {
|
|
int i, j;
|
|
unsigned char local_source[13 + 1];
|
|
unsigned char have_check_digit = '\0';
|
|
int check_digit;
|
|
static const char weights[8] = { 8, 6, 4, 2, 3, 5, 9, 7 };
|
|
int error_number = 0;
|
|
|
|
if (length != 12 && length != 13) {
|
|
strcpy(symbol->errtxt, "834: Input must be 12 or 13 characters long");
|
|
return ZINT_ERROR_TOO_LONG;
|
|
}
|
|
if (length == 13) { /* Includes check digit - remove for now */
|
|
have_check_digit = source[10];
|
|
memcpy(local_source, source, 10);
|
|
ustrcpy(local_source + 10, source + 11);
|
|
} else {
|
|
ustrcpy(local_source, source);
|
|
}
|
|
to_upper(local_source, length);
|
|
|
|
if (!z_isupper(local_source[0]) || !z_isupper(local_source[1])) {
|
|
strcpy(symbol->errtxt, "835: Invalid character in Service Indictor (first 2 characters) (alphabetic only)");
|
|
return ZINT_ERROR_INVALID_DATA;
|
|
}
|
|
if (!is_sane(NEON_F, local_source + 2, 12 - 4) || (have_check_digit && !z_isdigit(have_check_digit))) {
|
|
sprintf(symbol->errtxt, "836: Invalid character in Serial Number (middle %d characters) (digits only)",
|
|
have_check_digit ? 9 : 8);
|
|
return ZINT_ERROR_INVALID_DATA;
|
|
}
|
|
if (!z_isupper(local_source[10]) || !z_isupper(local_source[11])) {
|
|
strcpy(symbol->errtxt, "837: Invalid character in Country Code (last 2 characters) (alphabetic only)");
|
|
return ZINT_ERROR_INVALID_DATA;
|
|
}
|
|
|
|
check_digit = 0;
|
|
for (i = 2; i < 10; i++) { /* Serial Number only */
|
|
check_digit += ctoi(local_source[i]) * weights[i - 2];
|
|
}
|
|
check_digit %= 11;
|
|
check_digit = 11 - check_digit;
|
|
if (check_digit == 10) {
|
|
check_digit = 0;
|
|
} else if (check_digit == 11) {
|
|
check_digit = 5;
|
|
}
|
|
if (have_check_digit && ctoi(have_check_digit) != check_digit) {
|
|
sprintf(symbol->errtxt, "838: Invalid check digit '%c', expecting '%c'", have_check_digit, itoc(check_digit));
|
|
return ZINT_ERROR_INVALID_CHECK;
|
|
}
|
|
/* Add in (back) check digit */
|
|
local_source[12] = local_source[11];
|
|
local_source[11] = local_source[10];
|
|
local_source[10] = itoc(check_digit);
|
|
local_source[13] = '\0';
|
|
|
|
/* Do some checks on the Service Indicator (first char only) and Country Code */
|
|
if (strchr("JKSTW", local_source[0]) != NULL) { /* These are reserved & cannot be assigned */
|
|
strcpy(symbol->errtxt, "839: Invalid Service Indicator (first character should not be any of \"JKSTW\")");
|
|
error_number = ZINT_WARN_NONCOMPLIANT;
|
|
} else if (strchr("FHIOXY", local_source[0]) != NULL) { /* These aren't allocated as of spec Oct 2017 */
|
|
strcpy(symbol->errtxt, "840: Non-standard Service Indicator (first 2 characters)");
|
|
error_number = ZINT_WARN_NONCOMPLIANT;
|
|
} else if (!gs1_iso3166_alpha2(local_source + 11)) {
|
|
strcpy(symbol->errtxt, "841: Country code (last two characters) is not ISO 3166-1");
|
|
error_number = ZINT_WARN_NONCOMPLIANT;
|
|
}
|
|
|
|
(void) code128(symbol, local_source, 13); /* Only error returned is for large text which can't happen */
|
|
|
|
j = 0;
|
|
for (i = 0; i < 13; i++) {
|
|
if (i == 2 || i == 5 || i == 8 || i == 11) {
|
|
symbol->text[j++] = ' ';
|
|
}
|
|
symbol->text[j++] = local_source[i];
|
|
}
|
|
symbol->text[j] = '\0';
|
|
|
|
if (symbol->output_options & COMPLIANT_HEIGHT) {
|
|
/* Universal Postal Union S10 Section 8, using max X 0.51mm & minimum height 12.5mm or 15% of width */
|
|
const float min_height_min = 24.5098038f; /* 12.5 / 0.51 */
|
|
float min_height = stripf(symbol->width * 0.15f);
|
|
if (min_height < min_height_min) {
|
|
min_height = min_height_min;
|
|
}
|
|
/* Using 50 as default as none recommended */
|
|
if (error_number == 0) {
|
|
error_number = set_height(symbol, min_height, min_height > 50.0f ? min_height : 50.0f, 0.0f,
|
|
0 /*no_errtxt*/);
|
|
} else {
|
|
(void) set_height(symbol, min_height, min_height > 50.0f ? min_height : 50.0f, 0.0f, 1 /*no_errtxt*/);
|
|
}
|
|
} else {
|
|
(void) set_height(symbol, 0.0f, 50.0f, 0.0f, 1 /*no_errtxt*/);
|
|
}
|
|
|
|
return error_number;
|
|
}
|
|
|
|
/* vim: set ts=4 sw=4 et : */
|