All Downloads are FREE. Search and download functionalities are using the official Maven repository.

com.google.zxing.datamatrix.encoder.MinimalEncoder Maven / Gradle / Ivy

There is a newer version: 3.5.3_1
Show newest version
/*
 * Copyright 2021 ZXing authors
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *      http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

package com.google.zxing.datamatrix.encoder;

import java.nio.charset.Charset;
import java.nio.charset.StandardCharsets;
import java.util.ArrayList;
import java.util.List;

import com.google.zxing.common.MinimalECIInput;

/**
 * Encoder that encodes minimally
 *
 * Algorithm:
 *
 * Uses Dijkstra to produce mathematically minimal encodings that are in some cases smaller than the results produced
 * by the algorithm described in annex S in the specification ISO/IEC 16022:200(E). The biggest improvment of this
 * algorithm over that one is the case when the algorithm enters the most inefficient mode, the B256 mode. The 
 * algorithm from the specification algorithm will exit this mode only if it encounters digits so that arbitrarily
 * inefficient results can be produced if the postfix contains no digits.
 *
 * Multi ECI support and ECI switching:
 *
 * For multi language content the algorithm selects the most compact representation using ECI modes. Note that unlike
 * the compaction algorithm used for QR-Codes, this implementation operates in two stages and therfore is not
 * mathematically optimal. In the first stage, the input string is encoded minimally as a stream of ECI character set
 * selectors and bytes encoded in the selected encoding. In this stage the algorithm might for example decide to
 * encode ocurrences of the characters "\u0150\u015C" (O-double-acute, S-circumflex) in UTF-8 by a single ECI or
 * alternatively by multiple ECIs that switch between IS0-8859-2 and ISO-8859-3 (e.g. in the case that the input
 * contains many * characters from ISO-8859-2 (Latin 2) and few from ISO-8859-3 (Latin 3)).
 * In a second stage this stream of ECIs and bytes is minimally encoded using the various Data Matrix encoding modes.
 * While both stages encode mathematically minimally it is not ensured that the result is mathematically minimal since
 * the size growth for inserting an ECI in the first stage can only be approximated as the first stage does not know 
 * in which mode the ECI will occur in the second stage (may, or may not require an extra latch to ASCII depending on
 * the current mode). The reason for this shortcoming are difficulties in implementing it in a straightforward and
 * readable manner.
 *
 * GS1 support
 *
 * FNC1 delimiters can be encoded in the input string by using the FNC1 character specified in the encoding function.
 * When a FNC1 character is specified then a leading FNC1 will be encoded and all ocurrences of delimiter characters
 * while result in FNC1 codewords in the symbol.
 *
 * @author Alex Geller
 */
public final class MinimalEncoder {

  enum Mode {
    ASCII,
    C40,
    TEXT,
    X12,
    EDF,
    B256
  }
  static final char[] C40_SHIFT2_CHARS = {'!', '"', '#', '$', '%', '&', '\'', '(', ')', '*',  '+', ',', '-', '.', '/',
                                          ':', ';', '<', '=', '>', '?',  '@', '[', '\\', ']', '^', '_' };


  private MinimalEncoder() {
  }

  static boolean isExtendedASCII(char ch, int fnc1) {
    return ch != fnc1 && ch >= 128 && ch <= 255;
  }

  private static boolean isInC40Shift1Set(char ch) {
    return ch <= 31;
  }

  private static boolean isInC40Shift2Set(char ch, int fnc1) {
    for (char c40Shift2Char : C40_SHIFT2_CHARS) {
      if (c40Shift2Char == ch) {
        return true;
      }
    }
    return ch == fnc1;
  }

  private static boolean isInTextShift1Set(char ch) {
    return isInC40Shift1Set(ch);
  }

  private static boolean isInTextShift2Set(char ch, int fnc1) {
    return isInC40Shift2Set(ch, fnc1);
  }

  /**
   * Performs message encoding of a DataMatrix message
   *
   * @param msg the message
   * @return the encoded message (the char values range from 0 to 255)
   */
  public static String encodeHighLevel(String msg) {
    return encodeHighLevel(msg, null, -1, SymbolShapeHint.FORCE_NONE);
  }

  /**
   * Performs message encoding of a DataMatrix message
   *
   * @param msg the message
   * @param priorityCharset The preferred {@link Charset}. When the value of the argument is null, the algorithm
   *   chooses charsets that leads to a minimal representation. Otherwise the algorithm will use the priority
   *   charset to encode any character in the input that can be encoded by it if the charset is among the
   *   supported charsets.
   * @param fnc1 denotes the character in the input that represents the FNC1 character or -1 if this is not a GS1
   *   bar code. If the value is not -1 then a FNC1 is also prepended.
   * @param shape requested shape.
   * @return the encoded message (the char values range from 0 to 255)
   */
  public static String encodeHighLevel(String msg, Charset priorityCharset, int fnc1, SymbolShapeHint shape) {
    int macroId = 0;
    if (msg.startsWith(HighLevelEncoder.MACRO_05_HEADER) && msg.endsWith(HighLevelEncoder.MACRO_TRAILER)) {
      macroId = 5;
      msg = msg.substring(HighLevelEncoder.MACRO_05_HEADER.length(), msg.length() - 2);
    } else if (msg.startsWith(HighLevelEncoder.MACRO_06_HEADER) && msg.endsWith(HighLevelEncoder.MACRO_TRAILER)) {
      macroId = 6;
      msg = msg.substring(HighLevelEncoder.MACRO_06_HEADER.length(), msg.length() - 2);
    }
    return new String(encode(msg, priorityCharset, fnc1, shape, macroId), StandardCharsets.ISO_8859_1);
  }

  /**
   * Encodes input minimally and returns an array of the codewords
   *
   * @param input The string to encode
   * @param priorityCharset The preferred {@link Charset}. When the value of the argument is null, the algorithm
   *   chooses charsets that leads to a minimal representation. Otherwise the algorithm will use the priority
   *   charset to encode any character in the input that can be encoded by it if the charset is among the
   *   supported charsets.
   * @param fnc1 denotes the character in the input that represents the FNC1 character or -1 if this is not a GS1
   *   bar code. If the value is not -1 then a FNC1 is also prepended.
   * @param shape requested shape.
   * @param macroId Prepends the specified macro function in case that a value of 5 or 6 is specified.
   * @return An array of bytes representing the codewords of a minimal encoding.
   */
  static byte[] encode(String input, Charset priorityCharset, int fnc1, SymbolShapeHint shape, int macroId) {
    return encodeMinimally(new Input(input, priorityCharset, fnc1, shape, macroId)).getBytes();
  }

  static void addEdge(Edge[][] edges, Edge edge) {
    int vertexIndex = edge.fromPosition + edge.characterLength;
    if (edges[vertexIndex][edge.getEndMode().ordinal()] == null ||
        edges[vertexIndex][edge.getEndMode().ordinal()].cachedTotalSize > edge.cachedTotalSize) {
      edges[vertexIndex][edge.getEndMode().ordinal()] = edge;
    }
  }

  /** @return the number of words in which the string starting at from can be encoded in c40 or text mode.
   *  The number of characters encoded is returned in characterLength.
   *  The number of characters encoded is also minimal in the sense that the algorithm stops as soon
   *  as a character encoding fills a C40 word competely (three C40 values). An exception is at the
   *  end of the string where two C40 values are allowed (according to the spec the third c40 value 
   *  is filled  with 0 (Shift 1) in this case).
   */
  static int getNumberOfC40Words(Input input, int from, boolean c40,int[] characterLength) {
    int thirdsCount = 0;
    for (int i = from; i < input.length(); i++) {
      if (input.isECI(i)) {
        characterLength[0] = 0;
        return 0;
      }
      char ci = input.charAt(i);
      if (c40 && HighLevelEncoder.isNativeC40(ci) || !c40 && HighLevelEncoder.isNativeText(ci)) {
        thirdsCount++; //native
      } else if (!isExtendedASCII(ci, input.getFNC1Character())) {
        thirdsCount += 2; //shift
      } else {
        int asciiValue = ci & 0xff;
        if (asciiValue >= 128 && (c40 && HighLevelEncoder.isNativeC40((char) (asciiValue - 128)) ||
                                  !c40 && HighLevelEncoder.isNativeText((char) (asciiValue - 128)))) {
          thirdsCount += 3; // shift, Upper shift
        } else {
          thirdsCount += 4; // shift, Upper shift, shift
        }
      }

      if (thirdsCount % 3 == 0 || ((thirdsCount - 2) % 3 == 0 && i + 1 == input.length())) {
        characterLength[0] = i - from + 1;
        return (int) Math.ceil(((double) thirdsCount) / 3.0);
      }
    }
    characterLength[0] = 0;
    return 0;
  }

  static void addEdges(Input input, Edge[][] edges, int from, Edge previous) {

    if (input.isECI(from)) {
      addEdge(edges, new Edge(input, Mode.ASCII, from, 1, previous));
      return;
    }

    char ch = input.charAt(from);
    if (previous == null || previous.getEndMode() != Mode.EDF) { //not possible to unlatch a full EDF edge to something
                                                                 //else
      if (HighLevelEncoder.isDigit(ch) && input.haveNCharacters(from, 2) &&
          HighLevelEncoder.isDigit(input.charAt(from + 1))) {
        // two digits ASCII encoded
        addEdge(edges, new Edge(input, Mode.ASCII, from, 2, previous));
      } else {
        // one ASCII encoded character or an extended character via Upper Shift
        addEdge(edges, new Edge(input, Mode.ASCII, from, 1, previous));
      }
  
      Mode[] modes = {Mode.C40, Mode.TEXT};
      for (Mode mode : modes) {
        int[] characterLength = new int[1];
        if (getNumberOfC40Words(input, from, mode == Mode.C40, characterLength) > 0) {
          addEdge(edges, new Edge(input, mode, from, characterLength[0], previous));
        }
      }
  
      if (input.haveNCharacters(from,3) &&
          HighLevelEncoder.isNativeX12(input.charAt(from)) &&
          HighLevelEncoder.isNativeX12(input.charAt(from + 1)) &&
          HighLevelEncoder.isNativeX12(input.charAt(from + 2))) {
        addEdge(edges, new Edge(input, Mode.X12, from, 3, previous));
      }

      addEdge(edges, new Edge(input, Mode.B256, from, 1, previous));
    }

    //We create 4 EDF edges,  with 1, 2 3 or 4 characters length. The fourth normally doesn't have a latch to ASCII
    //unless it is 2 characters away from the end of the input.
    int i;
    for (i = 0; i < 3; i++) {
      int pos = from + i;
      if (input.haveNCharacters(pos,1) && HighLevelEncoder.isNativeEDIFACT(input.charAt(pos))) {
        addEdge(edges, new Edge(input, Mode.EDF, from, i + 1, previous));
      } else {
        break;
      }
    }
    if (i == 3 && input.haveNCharacters(from, 4) && HighLevelEncoder.isNativeEDIFACT(input.charAt(from + 3))) {
      addEdge(edges, new Edge(input, Mode.EDF, from, 4, previous));
    }
  }
  static Result encodeMinimally(Input input) {

    @SuppressWarnings("checkstyle:lineLength")
    /* The minimal encoding is computed by Dijkstra. The acyclic graph is modeled as follows:
     * A vertex represents a combination of a position in the input and an encoding mode where position 0
     * denotes the position left of the first character, 1 the position left of the second character and so on.
     * Likewise the end vertices are located after the last character at position input.length().
     * For any position there might be up to six vertices, one for each of the encoding types ASCII, C40, TEXT, X12,
     * EDF and B256.
     * 
     * As an example consider the input string "ABC123" then at position 0 there is only one vertex with the default
     * ASCII encodation. At position 3 there might be vertices for the types ASCII, C40, X12, EDF and B256.
     *
     * An edge leading to such a vertex encodes one or more of the characters left of the position that the vertex
     * represents. It encodes the characters in the encoding mode of the vertex that it ends on. In other words,
     * all edges leading to a particular vertex encode the same characters (the length of the suffix can vary) using the same 
     * encoding mode.
     * As an example consider the input string "ABC123" and the vertex (4,EDF). Possible edges leading to this vertex
     * are: 
     *   (0,ASCII)  --EDF(ABC1)--> (4,EDF)
     *   (1,ASCII)  --EDF(BC1)-->  (4,EDF)
     *   (1,B256)   --EDF(BC1)-->  (4,EDF)
     *   (1,EDF)    --EDF(BC1)-->  (4,EDF)
     *   (2,ASCII)  --EDF(C1)-->   (4,EDF)
     *   (2,B256)   --EDF(C1)-->   (4,EDF)
     *   (2,EDF)    --EDF(C1)-->   (4,EDF)
     *   (3,ASCII)  --EDF(1)-->    (4,EDF)
     *   (3,B256)   --EDF(1)-->    (4,EDF)
     *   (3,EDF)    --EDF(1)-->    (4,EDF)
     *   (3,C40)    --EDF(1)-->    (4,EDF)
     *   (3,X12)    --EDF(1)-->    (4,EDF)
     *
     * The edges leading to a vertex are stored in such a way that there is a fast way to enumerate the edges ending
     * on a particular vertex.
     *
     * The algorithm processes the vertices in order of their position thereby performing the following:
     *
     * For every vertex at position i the algorithm enumerates the edges ending on the vertex and removes all but the
     * shortest from that list.
     * Then it processes the vertices for the position i+1. If i+1 == input.length() then the algorithm ends
     * and chooses the the edge with the smallest size from any of the edges leading to vertices at this position.
     * Otherwise the algorithm computes all possible outgoing edges for the vertices at the position i+1
     *
     * Examples:
     * The process is illustrated by showing the graph (edges) after each iteration from left to right over the input:
     * An edge is drawn as follows "(" + fromVertex + ") -- " + encodingMode + "(" + encodedInput + ") (" +
     * accumulatedSize + ") --> (" + toVertex + ")"
     *
     * Example 1 encoding the string "ABCDEFG":
     *
     *
     * Situation after adding edges to the start vertex (0,ASCII)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII)
     * (0,ASCII) B256(A) (3) --> (1,B256)
     * (0,ASCII) EDF(AB) (4) --> (2,EDF)
     * (0,ASCII) C40(ABC) (3) --> (3,C40)
     * (0,ASCII) TEXT(ABC) (5) --> (3,TEXT)
     * (0,ASCII) X12(ABC) (3) --> (3,X12)
     * (0,ASCII) EDF(ABC) (4) --> (3,EDF)
     * (0,ASCII) EDF(ABCD) (4) --> (4,EDF)
     *
     * Situation after adding edges to vertices at position 1
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII)
     * (0,ASCII) B256(A) (3) --> (1,B256)
     * (0,ASCII) EDF(AB) (4) --> (2,EDF)
     * (0,ASCII) C40(ABC) (3) --> (3,C40)
     * (0,ASCII) TEXT(ABC) (5) --> (3,TEXT)
     * (0,ASCII) X12(ABC) (3) --> (3,X12)
     * (0,ASCII) EDF(ABC) (4) --> (3,EDF)
     * (0,ASCII) EDF(ABCD) (4) --> (4,EDF)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) B256(B) (4) --> (2,B256)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) EDF(BC) (5) --> (3,EDF)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) C40(BCD) (4) --> (4,C40)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) TEXT(BCD) (6) --> (4,TEXT)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) X12(BCD) (4) --> (4,X12)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) EDF(BCD) (5) --> (4,EDF)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) EDF(BCDE) (5) --> (5,EDF)
     * (0,ASCII) B256(A) (3) --> (1,B256) ASCII(B) (4) --> (2,ASCII)
     * (0,ASCII) B256(A) (3) --> (1,B256) B256(B) (3) --> (2,B256)
     * (0,ASCII) B256(A) (3) --> (1,B256) EDF(BC) (6) --> (3,EDF)
     * (0,ASCII) B256(A) (3) --> (1,B256) C40(BCD) (5) --> (4,C40)
     * (0,ASCII) B256(A) (3) --> (1,B256) TEXT(BCD) (7) --> (4,TEXT)
     * (0,ASCII) B256(A) (3) --> (1,B256) X12(BCD) (5) --> (4,X12)
     * (0,ASCII) B256(A) (3) --> (1,B256) EDF(BCD) (6) --> (4,EDF)
     * (0,ASCII) B256(A) (3) --> (1,B256) EDF(BCDE) (6) --> (5,EDF)
     *
     * Edge "(1,ASCII) ASCII(B) (2) --> (2,ASCII)" is minimal for the vertex (2,ASCII) so that edge "(1,B256) ASCII(B) (4) --> (2,ASCII)" is removed.
     * Edge "(1,B256) B256(B) (3) --> (2,B256)" is minimal for the vertext (2,B256) so that the edge "(1,ASCII) B256(B) (4) --> (2,B256)" is removed.
     *
     * Situation after adding edges to vertices at position 2
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII)
     * (0,ASCII) B256(A) (3) --> (1,B256)
     * (0,ASCII) EDF(AB) (4) --> (2,EDF)
     * (0,ASCII) C40(ABC) (3) --> (3,C40)
     * (0,ASCII) TEXT(ABC) (5) --> (3,TEXT)
     * (0,ASCII) X12(ABC) (3) --> (3,X12)
     * (0,ASCII) EDF(ABC) (4) --> (3,EDF)
     * (0,ASCII) EDF(ABCD) (4) --> (4,EDF)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) EDF(BC) (5) --> (3,EDF)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) C40(BCD) (4) --> (4,C40)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) TEXT(BCD) (6) --> (4,TEXT)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) X12(BCD) (4) --> (4,X12)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) EDF(BCD) (5) --> (4,EDF)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) EDF(BCDE) (5) --> (5,EDF)
     * (0,ASCII) B256(A) (3) --> (1,B256) B256(B) (3) --> (2,B256)
     * (0,ASCII) B256(A) (3) --> (1,B256) EDF(BC) (6) --> (3,EDF)
     * (0,ASCII) B256(A) (3) --> (1,B256) C40(BCD) (5) --> (4,C40)
     * (0,ASCII) B256(A) (3) --> (1,B256) TEXT(BCD) (7) --> (4,TEXT)
     * (0,ASCII) B256(A) (3) --> (1,B256) X12(BCD) (5) --> (4,X12)
     * (0,ASCII) B256(A) (3) --> (1,B256) EDF(BCD) (6) --> (4,EDF)
     * (0,ASCII) B256(A) (3) --> (1,B256) EDF(BCDE) (6) --> (5,EDF)
     * (0,ASCII) EDF(AB) (4) --> (2,EDF) ASCII(C) (5) --> (3,ASCII)
     * (0,ASCII) EDF(AB) (4) --> (2,EDF) B256(C) (6) --> (3,B256)
     * (0,ASCII) EDF(AB) (4) --> (2,EDF) EDF(CD) (7) --> (4,EDF)
     * (0,ASCII) EDF(AB) (4) --> (2,EDF) C40(CDE) (6) --> (5,C40)
     * (0,ASCII) EDF(AB) (4) --> (2,EDF) TEXT(CDE) (8) --> (5,TEXT)
     * (0,ASCII) EDF(AB) (4) --> (2,EDF) X12(CDE) (6) --> (5,X12)
     * (0,ASCII) EDF(AB) (4) --> (2,EDF) EDF(CDE) (7) --> (5,EDF)
     * (0,ASCII) EDF(AB) (4) --> (2,EDF) EDF(CDEF) (7) --> (6,EDF)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) ASCII(C) (3) --> (3,ASCII)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) B256(C) (5) --> (3,B256)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) EDF(CD) (6) --> (4,EDF)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) C40(CDE) (5) --> (5,C40)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) TEXT(CDE) (7) --> (5,TEXT)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) X12(CDE) (5) --> (5,X12)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) EDF(CDE) (6) --> (5,EDF)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) EDF(CDEF) (6) --> (6,EDF)
     * (0,ASCII) B256(A) (3) --> (1,B256) B256(B) (3) --> (2,B256) ASCII(C) (4) --> (3,ASCII)
     * (0,ASCII) B256(A) (3) --> (1,B256) B256(B) (3) --> (2,B256) B256(C) (4) --> (3,B256)
     * (0,ASCII) B256(A) (3) --> (1,B256) B256(B) (3) --> (2,B256) EDF(CD) (6) --> (4,EDF)
     * (0,ASCII) B256(A) (3) --> (1,B256) B256(B) (3) --> (2,B256) C40(CDE) (5) --> (5,C40)
     * (0,ASCII) B256(A) (3) --> (1,B256) B256(B) (3) --> (2,B256) TEXT(CDE) (7) --> (5,TEXT)
     * (0,ASCII) B256(A) (3) --> (1,B256) B256(B) (3) --> (2,B256) X12(CDE) (5) --> (5,X12)
     * (0,ASCII) B256(A) (3) --> (1,B256) B256(B) (3) --> (2,B256) EDF(CDE) (6) --> (5,EDF)
     * (0,ASCII) B256(A) (3) --> (1,B256) B256(B) (3) --> (2,B256) EDF(CDEF) (6) --> (6,EDF)
     *
     * Edge "(2,ASCII) ASCII(C) (3) --> (3,ASCII)" is minimal for the vertex (3,ASCII) so that edges "(2,EDF) ASCII(C) (5) --> (3,ASCII)" 
     * and "(2,B256) ASCII(C) (4) --> (3,ASCII)" can be removed.
     * Edge "(0,ASCII) EDF(ABC) (4) --> (3,EDF)" is minimal for the vertex (3,EDF) so that edges "(1,ASCII) EDF(BC) (5) --> (3,EDF)" 
     * and "(1,B256) EDF(BC) (6) --> (3,EDF)" can be removed.
     * Edge "(2,B256) B256(C) (4) --> (3,B256)" is minimal for the vertex (3,B256) so that edges "(2,ASCII) B256(C) (5) --> (3,B256)" 
     * and "(2,EDF) B256(C) (6) --> (3,B256)" can be removed.
     *
     * This continues for vertices 3 thru 7
     *
     * Situation after adding edges to vertices at position 7
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII)
     * (0,ASCII) B256(A) (3) --> (1,B256)
     * (0,ASCII) EDF(AB) (4) --> (2,EDF)
     * (0,ASCII) C40(ABC) (3) --> (3,C40)
     * (0,ASCII) TEXT(ABC) (5) --> (3,TEXT)
     * (0,ASCII) X12(ABC) (3) --> (3,X12)
     * (0,ASCII) EDF(ABC) (4) --> (3,EDF)
     * (0,ASCII) EDF(ABCD) (4) --> (4,EDF)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) C40(BCD) (4) --> (4,C40)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) TEXT(BCD) (6) --> (4,TEXT)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) X12(BCD) (4) --> (4,X12)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) EDF(BCDE) (5) --> (5,EDF)
     * (0,ASCII) B256(A) (3) --> (1,B256) B256(B) (3) --> (2,B256)
     * (0,ASCII) C40(ABC) (3) --> (3,C40) C40(DEF) (5) --> (6,C40)
     * (0,ASCII) X12(ABC) (3) --> (3,X12) X12(DEF) (5) --> (6,X12)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) ASCII(C) (3) --> (3,ASCII)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) C40(CDE) (5) --> (5,C40)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) TEXT(CDE) (7) --> (5,TEXT)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) X12(CDE) (5) --> (5,X12)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) EDF(CDEF) (6) --> (6,EDF)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) C40(BCD) (4) --> (4,C40) C40(EFG) (6) --> (7,C40)    //Solution 1
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) X12(BCD) (4) --> (4,X12) X12(EFG) (6) --> (7,X12)    //Solution 2
     * (0,ASCII) B256(A) (3) --> (1,B256) B256(B) (3) --> (2,B256) B256(C) (4) --> (3,B256)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) ASCII(C) (3) --> (3,ASCII) ASCII(D) (4) --> (4,ASCII)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) ASCII(C) (3) --> (3,ASCII) TEXT(DEF) (8) --> (6,TEXT)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) ASCII(C) (3) --> (3,ASCII) EDF(DEFG) (7) --> (7,EDF)
     * (0,ASCII) B256(A) (3) --> (1,B256) B256(B) (3) --> (2,B256) B256(C) (4) --> (3,B256) B256(D) (5) --> (4,B256)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) ASCII(C) (3) --> (3,ASCII) ASCII(D) (4) --> (4,ASCII) ASCII(E) (5) --> (5,ASCII)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) ASCII(C) (3) --> (3,ASCII) ASCII(D) (4) --> (4,ASCII) TEXT(EFG) (9) --> (7,TEXT)
     * (0,ASCII) B256(A) (3) --> (1,B256) B256(B) (3) --> (2,B256) B256(C) (4) --> (3,B256) B256(D) (5) --> (4,B256) B256(E) (6) --> (5,B256)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) ASCII(C) (3) --> (3,ASCII) ASCII(D) (4) --> (4,ASCII) ASCII(E) (5) --> (5,ASCII) ASCII(F) (6) --> (6,ASCII)
     * (0,ASCII) B256(A) (3) --> (1,B256) B256(B) (3) --> (2,B256) B256(C) (4) --> (3,B256) B256(D) (5) --> (4,B256) B256(E) (6) --> (5,B256) B256(F) (7) --> (6,B256)
     * (0,ASCII) ASCII(A) (1) --> (1,ASCII) ASCII(B) (2) --> (2,ASCII) ASCII(C) (3) --> (3,ASCII) ASCII(D) (4) --> (4,ASCII) ASCII(E) (5) --> (5,ASCII) ASCII(F) (6) --> (6,ASCII) ASCII(G) (7) --> (7,ASCII)
     * (0,ASCII) B256(A) (3) --> (1,B256) B256(B) (3) --> (2,B256) B256(C) (4) --> (3,B256) B256(D) (5) --> (4,B256) B256(E) (6) --> (5,B256) B256(F) (7) --> (6,B256) B256(G) (8) --> (7,B256)
     *
     * Hence a minimal encoding of "ABCDEFG" is either ASCII(A),C40(BCDEFG) or ASCII(A), X12(BCDEFG) with a size of 5 bytes.
     */

    int inputLength = input.length();

    // Array that represents vertices. There is a vertex for every character and mode.
    // The last dimension in the array below encodes the 6 modes ASCII, C40, TEXT, X12, EDF and B256
    Edge[][] edges = new Edge[inputLength + 1][6];
    addEdges(input, edges, 0, null);

    for (int i = 1; i <= inputLength; i++) {
      for (int j = 0; j < 6; j++) {
        if (edges[i][j] != null && i < inputLength) {
          addEdges(input, edges, i, edges[i][j]);
        }
      }
      //optimize memory by removing edges that have been passed.
      for (int j = 0; j < 6; j++) {
        edges[i - 1][j] = null;
      }
    }

    int minimalJ = -1;
    int minimalSize = Integer.MAX_VALUE;
    for (int j = 0; j < 6; j++) {
      if (edges[inputLength][j] != null) {
        Edge edge = edges[inputLength][j];
        int size = j >= 1 && j <= 3 ? edge.cachedTotalSize + 1 : edge.cachedTotalSize; //C40, TEXT and X12 need an
                                                                                       // extra unlatch at the end
        if (size < minimalSize) {
          minimalSize = size;
          minimalJ = j;
        }
      }
    }

    if (minimalJ < 0) {
      throw new RuntimeException("Internal error: failed to encode \"" + input + "\"");
    }
    return new Result(edges[inputLength][minimalJ]);
  }

  private static final class Edge {
    private static final int[] allCodewordCapacities = {3, 5, 8, 10, 12, 16, 18, 22, 30, 32, 36, 44, 49, 62, 86, 114,
                                                        144, 174, 204, 280, 368, 456, 576, 696, 816, 1050, 1304, 1558};
    private static final int[] squareCodewordCapacities = {3, 5, 8, 12, 18, 22, 30, 36, 44, 62, 86, 114, 144, 174, 204,
                                                           280, 368, 456, 576, 696, 816, 1050, 1304, 1558};
    private static final int[] rectangularCodewordCapacities = {5, 10, 16, 33, 32, 49};
    private final Input input;
    private final Mode mode; //the mode at the start of this edge.
    private final int fromPosition;
    private final int characterLength;
    private final Edge previous;
    private final int cachedTotalSize;

    private Edge(Input input, Mode mode, int fromPosition, int characterLength, Edge previous) {
      this.input = input;
      this.mode = mode;
      this.fromPosition = fromPosition;
      this.characterLength = characterLength;
      this.previous = previous;
      assert fromPosition + characterLength <= input.length();

      int size = previous != null ? previous.cachedTotalSize : 0;

      Mode previousMode = getPreviousMode();

     /*
      * Switching modes
      * ASCII -> C40: latch 230
      * ASCII -> TEXT: latch 239
      * ASCII -> X12: latch 238
      * ASCII -> EDF: latch 240
      * ASCII -> B256: latch 231
      * C40 -> ASCII: word(c1,c2,c3), 254
      * TEXT -> ASCII: word(c1,c2,c3), 254
      * X12 -> ASCII: word(c1,c2,c3), 254
      * EDIFACT -> ASCII: Unlatch character,0,0,0 or c1,Unlatch character,0,0 or c1,c2,Unlatch character,0 or 
      * c1,c2,c3,Unlatch character
      * B256 -> ASCII: without latch after n bytes
      */
      switch (mode) {
        case ASCII:
          size++;
          if (input.isECI(fromPosition) || isExtendedASCII(input.charAt(fromPosition), input.getFNC1Character())) {
            size++;
          }
          if (previousMode == Mode.C40 ||
              previousMode == Mode.TEXT ||
              previousMode == Mode.X12) {
            size++; // unlatch 254 to ASCII
          }
          break;
        case B256:
          size++;
          if (previousMode != Mode.B256) {
            size++; //byte count
          } else if (getB256Size() == 250) {
            size++; //extra byte count
          }
          if (previousMode == Mode.ASCII) {
            size++; //latch to B256
          } else if (previousMode == Mode.C40 ||
                     previousMode == Mode.TEXT ||
                     previousMode == Mode.X12) {
            size += 2; //unlatch to ASCII, latch to B256
          }
          break;
        case C40:
        case TEXT:
        case X12:
          if (mode == Mode.X12) {
            size += 2;
          } else {
            int[] charLen = new int[1];
            size += getNumberOfC40Words(input, fromPosition, mode == Mode.C40, charLen) * 2;
          }

          if (previousMode == Mode.ASCII || previousMode == Mode.B256) {
            size++; //additional byte for latch from ASCII to this mode
          } else if (previousMode != mode && (previousMode == Mode.C40 ||
                                             previousMode == Mode.TEXT ||
                                             previousMode == Mode.X12)) {
            size += 2; //unlatch 254 to ASCII followed by latch to this mode
          }
          break;
        case EDF:
          size += 3;
          if (previousMode == Mode.ASCII || previousMode == Mode.B256) {
            size++; //additional byte for latch from ASCII to this mode
          } else if (previousMode == Mode.C40 ||
                    previousMode == Mode.TEXT ||
                    previousMode == Mode.X12) {
            size += 2; //unlatch 254 to ASCII followed by latch to this mode
          }
          break;
      }
      cachedTotalSize = size;
    }

    // does not count beyond 250
    int getB256Size() {
      int cnt = 0;
      Edge current = this;
      while (current != null && current.mode == Mode.B256 && cnt <= 250) {
        cnt++;
        current = current.previous;
      }
      return cnt;
    }

    Mode getPreviousStartMode() {
      return  previous == null ? Mode.ASCII : previous.mode;
    }

    Mode getPreviousMode() {
      return  previous == null ? Mode.ASCII : previous.getEndMode();
    }

    /** Returns Mode.ASCII in case that:
     *  - Mode is EDIFACT and characterLength is less than 4 or the remaining characters can be encoded in at most 2
     *    ASCII bytes.
     *  - Mode is C40, TEXT or X12 and the remaining characters can be encoded in at most 1 ASCII byte.
     *  Returns mode in all other cases.
     * */
    Mode getEndMode() {
      if (mode == Mode.EDF) {
        if (characterLength < 4) {
          return Mode.ASCII;
        }
        int lastASCII = getLastASCII(); // see 5.2.8.2 EDIFACT encodation Rules
        if (lastASCII > 0 && getCodewordsRemaining(cachedTotalSize + lastASCII) <= 2 - lastASCII) {
          return Mode.ASCII;
        }
      }
      if (mode == Mode.C40 ||
          mode == Mode.TEXT ||
          mode == Mode.X12) {

        // see 5.2.5.2 C40 encodation rules and 5.2.7.2 ANSI X12 encodation rules
        if (fromPosition + characterLength >= input.length() && getCodewordsRemaining(cachedTotalSize) == 0) {
          return Mode.ASCII; 
        }
        int lastASCII = getLastASCII();
        if (lastASCII == 1 && getCodewordsRemaining(cachedTotalSize + 1) == 0) {
          return Mode.ASCII;
        }
      }
      return mode;
    }

    Mode getMode() {
      return mode;
    }

    /** Peeks ahead and returns 1 if the postfix consists of exactly two digits, 2 if the postfix consists of exactly
     *  two consecutive digits and a non extended character or of 4 digits. 
     *  Returns 0 in any other case
     **/
    int getLastASCII() {
      int length = input.length();
      int from = fromPosition + characterLength;
      if (length - from > 4 || from >= length) {
        return 0;
      }
      if (length - from == 1) {
        if (isExtendedASCII(input.charAt(from), input.getFNC1Character())) {
          return 0;
        }
        return 1;
      }
      if (length - from == 2) {
        if (isExtendedASCII(input.charAt(from), input.getFNC1Character()) || isExtendedASCII(input.charAt(from + 1),
            input.getFNC1Character())) {
          return 0;
        }
        if (HighLevelEncoder.isDigit(input.charAt(from)) && HighLevelEncoder.isDigit(input.charAt(from + 1))) {
          return 1;
        }
        return 2;
      }
      if (length - from == 3) {
        if (HighLevelEncoder.isDigit(input.charAt(from)) && HighLevelEncoder.isDigit(input.charAt(from + 1))
            && !isExtendedASCII(input.charAt(from + 2), input.getFNC1Character())) {
          return 2;
        }
        if (HighLevelEncoder.isDigit(input.charAt(from + 1)) && HighLevelEncoder.isDigit(input.charAt(from + 2))
            && !isExtendedASCII(input.charAt(from), input.getFNC1Character())) {
          return 2;
        }
        return 0;
      }
      if (HighLevelEncoder.isDigit(input.charAt(from)) && HighLevelEncoder.isDigit(input.charAt(from + 1))
          && HighLevelEncoder.isDigit(input.charAt(from + 2)) && HighLevelEncoder.isDigit(input.charAt(from + 3))) {
        return 2;
      }
      return 0;
    }

    /** Returns the capacity in codewords of the smallest symbol that has enough capacity to fit the given minimal
     * number of codewords.
     **/
    int getMinSymbolSize(int minimum) {
      switch (input.getShapeHint()) {
        case FORCE_SQUARE:
          for (int capacity : squareCodewordCapacities) {
            if (capacity >= minimum) {
              return capacity;
            }
          }
          break;
        case FORCE_RECTANGLE:
          for (int capacity : rectangularCodewordCapacities) {
            if (capacity >= minimum) {
              return capacity;
            }
          }
          break;
      }
      for (int capacity : allCodewordCapacities) {
        if (capacity >= minimum) {
          return capacity;
        }
      }
      return allCodewordCapacities[allCodewordCapacities.length - 1];
    }

    /** Returns the remaining capacity in codewords of the smallest symbol that has enough capacity to fit the given
     * minimal number of codewords.
     **/
    int getCodewordsRemaining(int minimum) {
      return getMinSymbolSize(minimum) - minimum;
    }

    static byte[] getBytes(int c) {
      byte[] result = new byte[1];
      result[0] = (byte) c;
      return result;
    }

    static byte[] getBytes(int c1,int c2) {
      byte[] result = new byte[2];
      result[0] = (byte) c1;
      result[1] = (byte) c2;
      return result;
    }

    static void setC40Word(byte[] bytes, int offset, int c1, int c2, int c3) {
      int val16 = (1600 * (c1 & 0xff)) + (40 * (c2 & 0xff)) + (c3 & 0xff) + 1;
      bytes[offset] = (byte) (val16 / 256);
      bytes[offset + 1] = (byte) (val16 % 256);
    }

    private static int getX12Value(char c) {
      return c == 13 ? 0 :
             c == 42 ? 1 :
             c == 62 ? 2 :
             c == 32 ? 3 :
             c >= 48 && c <= 57 ? c - 44 :
             c >= 65 && c <= 90 ? c - 51 : c;
    }

    byte[] getX12Words() {
      assert characterLength % 3 == 0;
      byte[] result = new byte[characterLength / 3 * 2];
      for (int i = 0; i < result.length; i += 2) {
        setC40Word(result,i,getX12Value(input.charAt(fromPosition + i / 2 * 3)),
                              getX12Value(input.charAt(fromPosition + i / 2 * 3 + 1)),
                              getX12Value(input.charAt(fromPosition + i / 2 * 3 + 2)));
      }
      return result;
    }

    static int getShiftValue(char c, boolean c40, int fnc1) {
      return (c40 && isInC40Shift1Set(c) ||
             !c40 && isInTextShift1Set(c)) ? 0 :
             (c40 && isInC40Shift2Set(c, fnc1) ||
             !c40 && isInTextShift2Set(c, fnc1)) ? 1 : 2;
    }

    private static int getC40Value(boolean c40, int setIndex, char c, int fnc1) {
      if (c == fnc1) {
        assert setIndex ==  2;
        return  27;
      }
      if (c40) {
        return c <= 31 ? c :
               c == 32 ? 3 :
               c <= 47 ? c - 33 :
               c <= 57 ? c - 44 :
               c <= 64 ? c - 43 :
               c <= 90 ? c - 51 :
               c <= 95 ? c - 69 :
               c <= 127 ? c - 96 : c;
      } else {
        return c == 0 ? 0 :
               setIndex == 0 && c <= 3 ? c - 1 : //is this a bug in the spec?
               setIndex == 1 && c <= 31 ? c :
               c == 32 ? 3 :
               c >= 33 && c <= 47 ? c - 33 :
               c >= 48 && c <= 57 ? c - 44 :
               c >= 58 && c <= 64 ? c - 43 :
               c >= 65 && c <= 90 ? c - 64 :
               c >= 91 && c <= 95 ? c - 69 :
               c == 96 ? 0 :
               c >= 97 && c <= 122 ? c - 83 :
               c >= 123 && c <= 127 ? c - 96 : c;
      }
    }

    byte[] getC40Words(boolean c40, int fnc1) {
      List c40Values = new ArrayList<>();
      for (int i = 0; i < characterLength; i++) {
        char ci = input.charAt(fromPosition + i);
        if (c40 && HighLevelEncoder.isNativeC40(ci) || !c40 && HighLevelEncoder.isNativeText(ci)) {
          c40Values.add((byte) getC40Value(c40, 0, ci, fnc1));
        } else if (!isExtendedASCII(ci, fnc1)) {
          int shiftValue = getShiftValue(ci, c40, fnc1);
          c40Values.add((byte) shiftValue); //Shift[123]
          c40Values.add((byte) getC40Value(c40, shiftValue, ci, fnc1));
        } else {
          char asciiValue = (char) ((ci & 0xff) - 128);
          if (c40 && HighLevelEncoder.isNativeC40(asciiValue) ||
              !c40 && HighLevelEncoder.isNativeText(asciiValue)) {
            c40Values.add((byte) 1); //Shift 2
            c40Values.add((byte) 30); //Upper Shift
            c40Values.add((byte) getC40Value(c40, 0, asciiValue, fnc1));
          } else {
            c40Values.add((byte) 1); //Shift 2
            c40Values.add((byte) 30); //Upper Shift
            int shiftValue = getShiftValue(asciiValue, c40, fnc1);
            c40Values.add((byte) shiftValue); // Shift[123]
            c40Values.add((byte) getC40Value(c40, shiftValue, asciiValue, fnc1));
          }
        }
      }

      if ((c40Values.size() % 3) != 0) {
        assert (c40Values.size() - 2) % 3 == 0 && fromPosition + characterLength == input.length();
        c40Values.add((byte) 0); // pad with 0 (Shift 1)
      }

      byte[] result = new byte[c40Values.size() / 3 * 2];
      int byteIndex = 0;
      for (int i = 0; i < c40Values.size(); i += 3) {
        setC40Word(result,byteIndex, c40Values.get(i) & 0xff, c40Values.get(i + 1) & 0xff, c40Values.get(i + 2) & 0xff);
        byteIndex += 2;
      }
      return result;
    }

    byte[] getEDFBytes() {
      int numberOfThirds = (int) Math.ceil(characterLength / 4.0);
      byte[] result = new byte[numberOfThirds * 3];
      int pos = fromPosition;
      int endPos = Math.min(fromPosition + characterLength - 1 , input.length() - 1);
      for (int i = 0; i < numberOfThirds; i += 3) {
        int[] edfValues = new int[4];
        for (int j = 0; j < 4; j++) {
          if (pos <= endPos) {
            edfValues[j] = input.charAt(pos++) & 0x3f;
          } else {
            edfValues[j] = pos == endPos + 1 ? 0x1f : 0;
          }
        }
        int val24 = edfValues[0] << 18;
        val24 |= edfValues[1] << 12;
        val24 |= edfValues[2] << 6;
        val24 |= edfValues[3];
        result[i] = (byte) ((val24 >> 16) & 0xff);
        result[i + 1] = (byte) ((val24 >> 8) & 0xff);
        result[i + 2] = (byte) (val24 & 0xff);
      }
      return result;
    }

    byte[] getLatchBytes() {
      switch (getPreviousMode()) {
        case ASCII:
        case B256: //after B256 ends (via length) we are back to ASCII
          switch (mode) {
            case B256:
              return getBytes(231);
            case C40:
              return getBytes(230);
            case TEXT:
              return getBytes(239);
            case X12:
              return getBytes(238);
            case EDF:
              return getBytes(240);
          }
          break;
        case C40:
        case TEXT:
        case X12:
          if (mode != getPreviousMode()) {
            switch (mode) {
              case ASCII:
                return getBytes(254);
              case B256:
                return getBytes(254, 231);
              case C40:
                return getBytes(254, 230);
              case TEXT:
                return getBytes(254, 239);
              case X12:
                return getBytes(254, 238);
              case EDF:
                return getBytes(254, 240);
            }
          }
          break;
        case EDF:
          assert mode == Mode.EDF; //The rightmost EDIFACT edge always contains an unlatch character
          break;
      }
      return new byte[0];
    }

    // Important: The function does not return the length bytes (one or two) in case of B256 encoding
    byte[] getDataBytes() {
      switch (mode) {
        case ASCII:
          if (input.isECI(fromPosition)) {
            return getBytes(241,input.getECIValue(fromPosition) + 1);
          } else if (isExtendedASCII(input.charAt(fromPosition), input.getFNC1Character())) {
            return getBytes(235,input.charAt(fromPosition) - 127);
          } else if (characterLength == 2) {
            return getBytes((input.charAt(fromPosition) - '0') * 10 + input.charAt(fromPosition + 1) - '0' + 130);
          } else if (input.isFNC1(fromPosition)) {
            return getBytes(232);
          } else {
            return getBytes(input.charAt(fromPosition) + 1);
          }
        case B256:
          return getBytes(input.charAt(fromPosition));
        case C40:
          return getC40Words(true, input.getFNC1Character());
        case TEXT:
          return getC40Words(false, input.getFNC1Character());
        case X12:
          return getX12Words();
        case EDF:
          return getEDFBytes();
      }
      assert false;
      return new byte[0];
    }
  }

  private static final class Result {

    private final byte[] bytes;

    Result(Edge solution) {
      Input input = solution.input;
      int size = 0;
      List bytesAL = new ArrayList<>();
      List randomizePostfixLength = new ArrayList<>();
      List randomizeLengths = new ArrayList<>();
      if ((solution.mode == Mode.C40 ||
           solution.mode == Mode.TEXT ||
           solution.mode == Mode.X12) &&
           solution.getEndMode() != Mode.ASCII) {
        size += prepend(MinimalEncoder.Edge.getBytes(254),bytesAL);
      }
      Edge current = solution;
      while (current != null) {
        size += prepend(current.getDataBytes(),bytesAL);

        if (current.previous == null || current.getPreviousStartMode() != current.getMode()) {
          if (current.getMode() == Mode.B256) {
            if (size <= 249) {
              bytesAL.add(0, (byte) size);
              size++;
            } else {
              bytesAL.add(0, (byte) (size % 250));
              bytesAL.add(0, (byte) (size / 250 + 249));
              size += 2;
            }
            randomizePostfixLength.add(bytesAL.size());
            randomizeLengths.add(size);
          }
          prepend(current.getLatchBytes(), bytesAL);
          size = 0;
        }

        current = current.previous;
      }
      if (input.getMacroId() == 5) {
        size += prepend(MinimalEncoder.Edge.getBytes(236), bytesAL);
      } else if (input.getMacroId() == 6) {
        size += prepend(MinimalEncoder.Edge.getBytes(237), bytesAL);
      }
   
      if (input.getFNC1Character() > 0) {
        size += prepend(MinimalEncoder.Edge.getBytes(232), bytesAL);
      }
      for (int i = 0; i < randomizePostfixLength.size(); i++) {
        applyRandomPattern(bytesAL,bytesAL.size() - randomizePostfixLength.get(i), randomizeLengths.get(i));
      }
      //add padding
      int capacity = solution.getMinSymbolSize(bytesAL.size());
      if (bytesAL.size() < capacity) {
        bytesAL.add((byte) 129);
      }
      while (bytesAL.size() < capacity) {
        bytesAL.add((byte) randomize253State(bytesAL.size() + 1));
      }

      bytes = new byte[bytesAL.size()];
      for (int i = 0; i < bytes.length; i++) {
        bytes[i] = bytesAL.get(i);
      }
    }

    static int prepend(byte[] bytes, List into) {
      for (int i = bytes.length - 1; i >= 0; i--) {
        into.add(0, bytes[i]);
      }
      return bytes.length;
    }

    private static int randomize253State(int codewordPosition) {
      int pseudoRandom = ((149 * codewordPosition) % 253) + 1;
      int tempVariable = 129 + pseudoRandom;
      return tempVariable <= 254 ? tempVariable : tempVariable - 254;
    }

    static void applyRandomPattern(List bytesAL,int startPosition, int length) {
      for (int i = 0; i < length; i++) {
        //See "B.1 253-state algorithm
        int Pad_codeword_position = startPosition + i;
        int Pad_codeword_value = bytesAL.get(Pad_codeword_position) & 0xff;
        int pseudo_random_number = ((149 * (Pad_codeword_position + 1)) % 255) + 1;
        int temp_variable = Pad_codeword_value + pseudo_random_number;
        bytesAL.set(Pad_codeword_position, (byte) (temp_variable <= 255 ? temp_variable : temp_variable - 256));
      }
    }

    public byte[] getBytes() {
      return bytes;
    }

  }

  private static final class Input extends MinimalECIInput {

    private final SymbolShapeHint shape;
    private final int macroId;

    private Input(String stringToEncode, Charset priorityCharset, int fnc1, SymbolShapeHint shape, int macroId) {
      super(stringToEncode, priorityCharset, fnc1);
      this.shape = shape;
      this.macroId = macroId;
    }

    private int getMacroId() {
      return macroId;
    }

    private SymbolShapeHint getShapeHint() {
      return shape;
    }
  }
}




© 2015 - 2024 Weber Informatics LLC | Privacy Policy