/* common.c - Contains functions needed for a number of barcodes */ /* libzint - the open source barcode library Copyright (C) 2008-2022 Robin Stuart 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 #ifdef ZINT_TEST #include #endif #include "common.h" /* Converts a character 0-9, A-F to its equivalent integer value */ INTERNAL int ctoi(const char source) { if (z_isdigit(source)) return (source - '0'); if ((source >= 'A') && (source <= 'F')) return (source - 'A' + 10); if ((source >= 'a') && (source <= 'f')) return (source - 'a' + 10); return -1; } /* Converts an integer value to its hexadecimal character */ INTERNAL char itoc(const int source) { if ((source >= 0) && (source <= 9)) { return ('0' + source); } else { return ('A' + (source - 10)); } } /* Converts decimal string of length <= 9 to integer value. Returns -1 if not numeric */ INTERNAL int to_int(const unsigned char source[], const int length) { int val = 0; int i; for (i = 0; i < length; i++) { if (!z_isdigit(source[i])) { return -1; } val *= 10; val += source[i] - '0'; } return val; } /* Converts lower case characters to upper case in a string source[] */ INTERNAL void to_upper(unsigned char source[], const int length) { int i; for (i = 0; i < length; i++) { if (z_islower(source[i])) { source[i] = (source[i] - 'a') + 'A'; } } } /* Returns the number of times a character occurs in a string */ INTERNAL int chr_cnt(const unsigned char string[], const int length, const unsigned char c) { int count = 0; int i; for (i = 0; i < length; i++) { if (string[i] == c) { count++; } } return count; } /* Flag table for `is_chr()` and `is_sane()` */ #define IS_CLS_F (IS_CLI_F | IS_SIL_F) static const unsigned short flgs[256] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /*00-1F*/ IS_SPC_F, IS_C82_F, IS_C82_F, IS_HSH_F, /*20-23*/ /* !"# */ IS_CLS_F, IS_SIL_F | IS_C82_F, IS_C82_F, IS_C82_F, /*24-27*/ /* $%&' */ IS_C82_F, IS_C82_F, IS_AST_F, IS_PLS_F, /*28-2B*/ /* ()*+ */ IS_C82_F, IS_MNS_F, IS_CLS_F | IS_C82_F, IS_CLS_F | IS_C82_F, /*2B-2F*/ /* ,-./ */ IS_NUM_F, IS_NUM_F, IS_NUM_F, IS_NUM_F, /*30-33*/ /* 0123 */ IS_NUM_F, IS_NUM_F, IS_NUM_F, IS_NUM_F, /*34-37*/ /* 4567 */ IS_NUM_F, IS_NUM_F, IS_CLI_F | IS_C82_F, IS_C82_F, /*38-3B*/ /* 89:; */ IS_C82_F, IS_C82_F, IS_C82_F, IS_C82_F, /*3B-3F*/ /* <=>? */ 0, IS_UHX_F | IS_ARS_F, IS_UHX_F | IS_ARS_F, IS_UHX_F | IS_ARS_F, /*40-43*/ /* @ABC */ IS_UHX_F | IS_ARS_F, IS_UHX_F | IS_ARS_F, IS_UHX_F | IS_ARS_F, IS_UPO_F | IS_ARS_F, /*44-47*/ /* DEFG */ IS_UPO_F | IS_ARS_F, IS_UPO_F, IS_UPO_F | IS_ARS_F, IS_UPO_F | IS_ARS_F, /*48-4B*/ /* HIJK */ IS_UPO_F | IS_ARS_F, IS_UPO_F | IS_ARS_F, IS_UPO_F | IS_ARS_F, IS_UPO_F, /*4B-4F*/ /* LMNO */ IS_UPO_F | IS_ARS_F, IS_UPO_F, IS_UPO_F | IS_ARS_F, IS_UPO_F | IS_ARS_F, /*50-53*/ /* PQRS */ IS_UPO_F | IS_ARS_F, IS_UPO_F | IS_ARS_F, IS_UPO_F | IS_ARS_F, IS_UPO_F | IS_ARS_F, /*53-57*/ /* TUVW */ IS_UX__F | IS_ARS_F, IS_UPO_F | IS_ARS_F, IS_UPO_F | IS_ARS_F, 0, /*58-5B*/ /* XYZ[ */ 0, 0, 0, IS_C82_F, /*5B-5F*/ /* \]^_ */ 0, IS_LHX_F, IS_LHX_F, IS_LHX_F, /*60-63*/ /* `abc */ IS_LHX_F, IS_LHX_F, IS_LHX_F, IS_LWO_F, /*64-67*/ /* defg */ IS_LWO_F, IS_LWO_F, IS_LWO_F, IS_LWO_F, /*68-6B*/ /* hijk */ IS_LWO_F, IS_LWO_F, IS_LWO_F, IS_LWO_F, /*6B-6F*/ /* lmno */ IS_LWO_F, IS_LWO_F, IS_LWO_F, IS_LWO_F, /*70-73*/ /* pqrs */ IS_LWO_F, IS_LWO_F, IS_LWO_F, IS_LWO_F, /*74-77*/ /* tuvw */ IS_LX__F, IS_LWO_F, IS_LWO_F, 0, /*78-7B*/ /* xyz{ */ 0, 0, 0, 0, /*7B-7F*/ /* |}~D */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /*80-9F*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /*A0-BF*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /*C0-DF*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /*E0-FF*/ }; /* Whether a character matches `flg` */ INTERNAL int is_chr(const unsigned int flg, const unsigned int c) { return c < 0x80 && (flgs[c] & flg) != 0; } /* Verifies that a string only uses valid characters */ INTERNAL int is_sane(const unsigned int flg, const unsigned char source[], const int length) { int i; for (i = 0; i < length; i++) { if (!(flgs[source[i]] & flg)) { return 0; } } return 1; } /* Replaces huge switch statements for looking up in tables */ /* Verifies that a string only uses valid characters, and returns `test_string` position of each in `posns` array */ INTERNAL int is_sane_lookup(const char test_string[], const int test_length, const unsigned char source[], const int length, int *posns) { int i, j; for (i = 0; i < length; i++) { posns[i] = -1; for (j = 0; j < test_length; j++) { if (source[i] == test_string[j]) { posns[i] = j; break; } } if (posns[i] == -1) { return 0; } } return 1; } /* Returns the position of data in set_string */ INTERNAL int posn(const char set_string[], const char data) { const char *s; for (s = set_string; *s; s++) { if (data == *s) { return s - set_string; } } return -1; } /* Convert an integer value to a string representing its binary equivalent and place at a given position */ INTERNAL int bin_append_posn(const int arg, const int length, char *binary, const int bin_posn) { int i; int start; start = 0x01 << (length - 1); for (i = 0; i < length; i++) { if (arg & (start >> i)) { binary[bin_posn + i] = '1'; } else { binary[bin_posn + i] = '0'; } } return bin_posn + length; } #ifndef Z_COMMON_INLINE /* Return true (1) if a module is dark/black, otherwise false (0) */ INTERNAL int module_is_set(const struct zint_symbol *symbol, const int y_coord, const int x_coord) { return (symbol->encoded_data[y_coord][x_coord >> 3] >> (x_coord & 0x07)) & 1; } /* Set a module to dark/black */ INTERNAL void set_module(struct zint_symbol *symbol, const int y_coord, const int x_coord) { symbol->encoded_data[y_coord][x_coord >> 3] |= 1 << (x_coord & 0x07); } /* Return true (1-8) if a module is colour, otherwise false (0) */ INTERNAL int module_colour_is_set(const struct zint_symbol *symbol, const int y_coord, const int x_coord) { return symbol->encoded_data[y_coord][x_coord]; } /* Set a module to a colour */ INTERNAL void set_module_colour(struct zint_symbol *symbol, const int y_coord, const int x_coord, const int colour) { symbol->encoded_data[y_coord][x_coord] = colour; } #endif /* Set a dark/black module to white (i.e. unset) */ INTERNAL void unset_module(struct zint_symbol *symbol, const int y_coord, const int x_coord) { symbol->encoded_data[y_coord][x_coord >> 3] &= ~(1 << (x_coord & 0x07)); } /* Expands from a width pattern to a bit pattern */ INTERNAL void expand(struct zint_symbol *symbol, const char data[], const int length) { int reader; int writer, i; int latch, num; const int row = symbol->rows; symbol->rows++; writer = 0; latch = 1; for (reader = 0; reader < length; reader++) { num = ctoi(data[reader]); assert(num >= 0); for (i = 0; i < num; i++) { if (latch) { set_module(symbol, row, writer); } writer++; } latch = !latch; } if (writer > symbol->width) { symbol->width = writer; } } /* Indicates which symbologies can have row binding */ INTERNAL int is_stackable(const int symbology) { if (symbology < BARCODE_PHARMA_TWO && symbology != BARCODE_POSTNET) { return 1; } switch (symbology) { case BARCODE_CODE128B: case BARCODE_ISBNX: case BARCODE_EAN14: case BARCODE_NVE18: case BARCODE_KOREAPOST: case BARCODE_PLESSEY: case BARCODE_TELEPEN_NUM: case BARCODE_ITF14: case BARCODE_CODE32: case BARCODE_CODABLOCKF: case BARCODE_HIBC_BLOCKF: return 1; break; } return 0; } /* Indicates which symbols can have addon (EAN-2 and EAN-5) */ INTERNAL int is_extendable(const int symbology) { switch (symbology) { case BARCODE_EANX: case BARCODE_EANX_CHK: case BARCODE_UPCA: case BARCODE_UPCA_CHK: case BARCODE_UPCE: case BARCODE_UPCE_CHK: case BARCODE_ISBNX: case BARCODE_EANX_CC: case BARCODE_UPCA_CC: case BARCODE_UPCE_CC: return 1; break; } return 0; } /* Indicates which symbols can have composite 2D component data */ INTERNAL int is_composite(const int symbology) { return symbology >= BARCODE_EANX_CC && symbology <= BARCODE_DBAR_EXPSTK_CC; } /* Returns 1 if symbology is a matrix design renderable as dots */ INTERNAL int is_dotty(const int symbology) { switch (symbology) { /* Note MAXICODE and ULTRA absent */ case BARCODE_QRCODE: case BARCODE_DATAMATRIX: case BARCODE_MICROQR: case BARCODE_HIBC_DM: case BARCODE_AZTEC: case BARCODE_HIBC_QR: case BARCODE_HIBC_AZTEC: case BARCODE_AZRUNE: case BARCODE_CODEONE: case BARCODE_GRIDMATRIX: case BARCODE_HANXIN: case BARCODE_DOTCODE: case BARCODE_UPNQR: case BARCODE_RMQR: return 1; break; } return 0; } /* Returns 1 if symbology has fixed aspect ratio (matrix design) */ INTERNAL int is_fixed_ratio(const int symbology) { if (is_dotty(symbology)) { return 1; } switch (symbology) { case BARCODE_MAXICODE: case BARCODE_ULTRA: return 1; break; } return 0; } /* Whether next two characters are digits */ INTERNAL int is_twodigits(const unsigned char source[], const int length, const int position) { if ((position + 1 < length) && z_isdigit(source[position]) && z_isdigit(source[position + 1])) { return 1; } return 0; } /* State machine to decode UTF-8 to Unicode codepoints (state 0 means done, state 12 means error) */ INTERNAL unsigned int decode_utf8(unsigned int *state, unsigned int *codep, const unsigned char byte) { /* Copyright (c) 2008-2009 Bjoern Hoehrmann Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. See https://bjoern.hoehrmann.de/utf-8/decoder/dfa/ for details. */ static const unsigned char utf8d[] = { /* The first part of the table maps bytes to character classes that * reduce the size of the transition table and create bitmasks. */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9, 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, 8,8,2,2,2,2,2,2,2,2,2,2,2,2,2,2, 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, 10,3,3,3,3,3,3,3,3,3,3,3,3,4,3,3, 11,6,6,6,5,8,8,8,8,8,8,8,8,8,8,8, /* The second part is a transition table that maps a combination * of a state of the automaton and a character class to a state. */ 0,12,24,36,60,96,84,12,12,12,48,72, 12,12,12,12,12,12,12,12,12,12,12,12, 12, 0,12,12,12,12,12, 0,12, 0,12,12, 12,24,12,12,12,12,12,24,12,24,12,12, 12,12,12,12,12,12,12,24,12,12,12,12, 12,24,12,12,12,12,12,12,12,24,12,12, 12,12,12,12,12,12,12,36,12,36,12,12, 12,36,12,12,12,12,12,36,12,36,12,12, 12,36,12,12,12,12,12,12,12,12,12,12, }; const unsigned int type = utf8d[byte]; *codep = *state != 0 ? (byte & 0x3fu) | (*codep << 6) : (0xff >> type) & byte; *state = utf8d[256 + *state + type]; return *state; } /* Is string valid UTF-8? */ INTERNAL int is_valid_utf8(const unsigned char source[], const int length) { int i; unsigned int codepoint, state = 0; for (i = 0; i < length; i++) { if (decode_utf8(&state, &codepoint, source[i]) == 12) { return 0; } } return state == 0; } /* Convert UTF-8 to Unicode. If `disallow_4byte` unset, allow all values (UTF-32). If `disallow_4byte` set, * only allow codepoints <= U+FFFF (ie four-byte sequences not allowed) (UTF-16, no surrogates) */ INTERNAL int utf8_to_unicode(struct zint_symbol *symbol, const unsigned char source[], unsigned int vals[], int *length, const int disallow_4byte) { int bpos; int jpos; unsigned int codepoint, state = 0; bpos = 0; jpos = 0; while (bpos < *length) { do { decode_utf8(&state, &codepoint, source[bpos++]); } while (bpos < *length && state != 0 && state != 12); if (state != 0) { strcpy(symbol->errtxt, "240: Corrupt Unicode data"); return ZINT_ERROR_INVALID_DATA; } if (disallow_4byte && codepoint > 0xffff) { strcpy(symbol->errtxt, "242: Unicode sequences of more than 3 bytes not supported"); return ZINT_ERROR_INVALID_DATA; } vals[jpos] = codepoint; jpos++; } *length = jpos; return 0; } /* Set symbol height, returning a warning if not within minimum and/or maximum if given. `default_height` does not include height of fixed-height rows (i.e. separators/composite data) */ INTERNAL int set_height(struct zint_symbol *symbol, const float min_row_height, const float default_height, const float max_height, const int no_errtxt) { int error_number = 0; float fixed_height = 0.0f; int zero_count = 0; float row_height; int i; const int rows = symbol->rows ? symbol->rows : 1; /* Sometimes called before expand() */ for (i = 0; i < rows; i++) { if (symbol->row_height[i]) { fixed_height += symbol->row_height[i]; } else { zero_count++; } } if (zero_count) { if (symbol->height) { if (symbol->input_mode & HEIGHTPERROW_MODE) { row_height = stripf(symbol->height); } else { row_height = stripf((symbol->height - fixed_height) / zero_count); } } else if (default_height) { row_height = stripf(default_height / zero_count); } else { row_height = stripf(min_row_height); } if (row_height < 0.5f) { /* Absolute minimum */ row_height = 0.5f; } if (min_row_height) { if (stripf(row_height) < stripf(min_row_height)) { error_number = ZINT_WARN_NONCOMPLIANT; if (!no_errtxt) { strcpy(symbol->errtxt, "247: Height not compliant with standards"); } } } symbol->height = stripf(row_height * zero_count + fixed_height); } else { symbol->height = stripf(fixed_height); /* Ignore any given height */ } if (max_height) { if (stripf(symbol->height) > stripf(max_height)) { error_number = ZINT_WARN_NONCOMPLIANT; if (!no_errtxt) { strcpy(symbol->errtxt, "248: Height not compliant with standards"); } } } return error_number; } /* Prevent inlining of `stripf()` which can optimize away its effect */ #if defined(__GNUC__) || defined(__clang__) __attribute__((__noinline__)) #endif #if defined(_MSC_VER) && _MSC_VER >= 1310 /* MSVC 2003 (VC++ 7.1) */ __declspec(noinline) #endif /* Removes excess precision from floats - see https://stackoverflow.com/q/503436 */ INTERNAL float stripf(const float arg) { return *((volatile const float *) &arg); } /* Returns total length of segments */ INTERNAL int segs_length(const struct zint_seg segs[], const int seg_count) { int total_len = 0; int i; for (i = 0; i < seg_count; i++) { total_len += segs[i].length == -1 ? (int) ustrlen(segs[i].source) : segs[i].length; } return total_len; } /* Shallow copy segments, adjusting default ECIs */ INTERNAL void segs_cpy(const struct zint_symbol *symbol, const struct zint_seg segs[], const int seg_count, struct zint_seg local_segs[]) { const int default_eci = symbol->symbology == BARCODE_GRIDMATRIX ? 29 : symbol->symbology == BARCODE_UPNQR ? 4 : 3; int i; local_segs[0] = segs[0]; for (i = 1; i < seg_count; i++) { local_segs[i] = segs[i]; /* Ensure default ECI set if follows non-default ECI */ if (local_segs[i].eci == 0 && local_segs[i - 1].eci != 0 && local_segs[i - 1].eci != default_eci) { local_segs[i].eci = default_eci; } } } /* Returns red component if any of ultra colour indexing "0CBMRYGKW" */ INTERNAL int colour_to_red(const int colour) { int return_val = 0; switch (colour) { case 8: /* White */ case 3: /* Magenta */ case 4: /* Red */ case 5: /* Yellow */ return_val = 255; break; } return return_val; } /* Returns green component if any of ultra colour indexing "0CBMRYGKW" */ INTERNAL int colour_to_green(const int colour) { int return_val = 0; switch (colour) { case 8: /* White */ case 1: /* Cyan */ case 5: /* Yellow */ case 6: /* Green */ return_val = 255; break; } return return_val; } /* Returns blue component if any of ultra colour indexing "0CBMRYGKW" */ INTERNAL int colour_to_blue(const int colour) { int return_val = 0; switch (colour) { case 8: /* White */ case 1: /* Cyan */ case 2: /* Blue */ case 3: /* Magenta */ return_val = 255; break; } return return_val; } #ifdef ZINT_TEST /* Dumps hex-formatted codewords in symbol->errtxt (for use in testing) */ void debug_test_codeword_dump(struct zint_symbol *symbol, const unsigned char *codewords, const int length) { int i, max = length, cnt_len = 0; if (length > 30) { /* 30*3 < errtxt 92 (100 - "Warning ") chars */ sprintf(symbol->errtxt, "(%d) ", length); /* Place the number of codewords at the front */ cnt_len = (int) strlen(symbol->errtxt); max = 30 - (cnt_len + 2) / 3; } for (i = 0; i < max; i++) { sprintf(symbol->errtxt + cnt_len + i * 3, "%02X ", codewords[i]); } symbol->errtxt[strlen(symbol->errtxt) - 1] = '\0'; /* Zap last space */ } /* Dumps decimal-formatted codewords in symbol->errtxt (for use in testing) */ void debug_test_codeword_dump_int(struct zint_symbol *symbol, const int *codewords, const int length) { int i, max = 0, cnt_len, errtxt_len; char temp[20]; errtxt_len = sprintf(symbol->errtxt, "(%d) ", length); /* Place the number of codewords at the front */ for (i = 0, cnt_len = errtxt_len; i < length; i++) { cnt_len += sprintf(temp, "%d ", codewords[i]); if (cnt_len > 92) { break; } max++; } for (i = 0; i < max; i++) { errtxt_len += sprintf(symbol->errtxt + errtxt_len, "%d ", codewords[i]); } symbol->errtxt[strlen(symbol->errtxt) - 1] = '\0'; /* Zap last space */ } #endif /* vim: set ts=4 sw=4 et : */