Merge pull request #175 from xlate/174_dialect_specific_state_tables

Implement separate state tables for dialect search and each dialect

src/main/java/io/xlate/edi/internal/stream/StaEDIStreamWriter.java

@@ -315,7 +315,7 @@ private void write(int output) throws EDIStreamException {
             throw new EDIStreamException(String.format("Invalid character: 0x%04X", output), location);
         }
 
-        state = state.transition(clazz);
+        state = State.transition(state, dialect, clazz);
 
         switch (state) {
         case HEADER_X12_I: // I(SA)

src/main/java/io/xlate/edi/internal/stream/tokenization/Dialect.java

@@ -19,6 +19,9 @@
 
 public abstract class Dialect {
 
+    protected final String[] transactionVersion;
+    protected final int dialectStateCode;
+
     protected char segmentDelimiter;
     protected char segmentTagTerminator = '\0';
     protected char elementDelimiter;
@@ -28,16 +31,28 @@ public abstract class Dialect {
     protected char elementRepeater;
 
     protected boolean initialized;
-    protected boolean rejected;
+    protected String rejectionMessage;
 
     protected String transactionType;
-    protected final String[] transactionVersion;
     protected String transactionVersionString;
 
-    protected Dialect(String[] initialTransactionVersion) {
+    protected Dialect(int dialectStateCode, String[] initialTransactionVersion) {
+        this.dialectStateCode = dialectStateCode;
         this.transactionVersion = initialTransactionVersion;
     }
 
+    public static String getStandard(Dialect dialect) {
+        return dialect != null ? dialect.getStandard() : "UNKNOWN";
+    }
+
+    public int getDialectStateCode() {
+        return dialectStateCode;
+    }
+
+    public State getTagSearchState() {
+        return State.TAG_SEARCH;
+    }
+
     public char getComponentElementSeparator() {
         return componentDelimiter;
     }
@@ -75,7 +90,11 @@ public boolean isConfirmed() {
     }
 
     public boolean isRejected() {
-        return rejected;
+        return rejectionMessage != null;
+    }
+
+    public String getRejectionMessage() {
+        return rejectionMessage;
     }
 
     /**

src/main/java/io/xlate/edi/internal/stream/tokenization/EDIFACTDialect.java

@@ -46,7 +46,7 @@ public class EDIFACTDialect extends Dialect {
     private static final int TX_ASSIGNED_CODE = 3;
 
     EDIFACTDialect(String headerTag) {
-        super(new String[4]);
+        super(State.DialectCode.EDIFACT, new String[4]);
         componentDelimiter = DFLT_COMPONENT_ELEMENT_SEPARATOR;
         elementDelimiter = DFLT_DATA_ELEMENT_SEPARATOR;
         decimalMark = DFLT_DECIMAL_MARK;
@@ -100,6 +100,7 @@ boolean initialize(CharacterSet characters) {
             characters.setClass(segmentDelimiter, CharacterClass.SEGMENT_DELIMITER);
             initialized = true;
         } else {
+            rejectionMessage = "Unable to obtain version from EDIFACT header segment";
             initialized = false;
         }
 
@@ -138,6 +139,14 @@ public boolean isServiceAdviceSegment(CharSequence tag) {
         return UNA.contentEquals(tag);
     }
 
+    @Override
+    public State getTagSearchState() {
+        if (isServiceAdviceSegment(this.headerTag)) {
+            return State.HEADER_EDIFACT_UNB_SEARCH;
+        }
+        return State.TAG_SEARCH;
+    }
+
     @Override
     public String getStandard() {
         return Standards.EDIFACT;
@@ -181,7 +190,7 @@ boolean processInterchangeHeader(CharacterSet characters, char value) {
              */
             characters.setClass(elementDelimiter, CharacterClass.ELEMENT_DELIMITER);
         } else if (segmentDelimiter == value) {
-            rejected = !initialize(characters);
+            initialize(characters);
             return isConfirmed();
         }
 
@@ -224,7 +233,7 @@ boolean processServiceStringAdvice(CharacterSet characters, char value) {
                 header.deleteCharAt(index--);
             } else if (isIndexBeyondUNBFirstElement()) {
                 if (value == elementDelimiter || value == segmentDelimiter) {
-                    rejected = !initialize(characters);
+                    initialize(characters);
                     proceed = isConfirmed();
                 }
             } else if (value == 'B') {

src/main/java/io/xlate/edi/internal/stream/tokenization/Lexer.java

@@ -184,13 +184,13 @@ public void parse() throws IOException, EDIException {
 
             CharacterClass clazz = characters.getClass(input);
             previous = state;
-            state = state.transition(clazz);
-            LOGGER.finer(() -> "State " + previous + "(" + clazz + ") -> " + state);
+            state = State.transition(state, dialect, clazz);
+            LOGGER.finer(() -> String.format("%s + (%s, '%s', %s) -> %s", previous, Dialect.getStandard(dialect), (char) input, clazz, state));
 
             switch (state) {
             case INITIAL:
             case TAG_SEARCH:
-            case HEADER_TAG_SEARCH:
+            case HEADER_EDIFACT_UNB_SEARCH:
                 break;
             case HEADER_X12_I:
             case HEADER_X12_S:
@@ -219,11 +219,12 @@ public void parse() throws IOException, EDIException {
                     buffer.put((char) input);
                 }
                 break;
-            case HEADER_TAG_1: // U - When UNA is present
-            case HEADER_TAG_2: // N - When UNA is present
-            case HEADER_TAG_3: // B - When UNA is present
+            case HEADER_EDIFACT_UNB_1: // U - When UNA is present
+            case HEADER_EDIFACT_UNB_2: // N - When UNA is present
+            case HEADER_EDIFACT_UNB_3: // B - When UNA is present
                 handleStateHeaderTag(input);
                 break;
+            case HEADER_RELEASE:
             case DATA_RELEASE:
                 // Skip this character - next character will be literal value
                 break;
@@ -384,7 +385,7 @@ void handleStateHeaderData(int input) throws EDIException {
         switch (characters.getClass(input)) {
         case SEGMENT_DELIMITER:
             closeSegment();
-            state = State.HEADER_TAG_SEARCH;
+            state = dialect.getTagSearchState();
             break;
         case SEGMENT_TAG_DELIMITER:
         case ELEMENT_DELIMITER:
@@ -400,6 +401,15 @@ void handleStateHeaderData(int input) throws EDIException {
         }
     }
 
+    /**
+     * Determine if the input text has been confirmed by the dialect as being
+     * initially accepted. If so, transition to the state given by the
+     * <code>confirmed</code> parameter.
+     *
+     * @param confirmed the state to transition to if the dialect is confirmed.
+     * @return true if the dialect is confirmed, otherwise false.
+     * @throws EDIException when the input text has been rejected by the dialect.
+     */
     private boolean dialectConfirmed(State confirmed) throws EDIException {
         if (dialect.isConfirmed()) {
             state = confirmed;
@@ -408,9 +418,10 @@ private boolean dialectConfirmed(State confirmed) throws EDIException {
         } else if (dialect.isRejected()) {
             buffer.clear();
             clearQueues();
+            String rejectionMessage = dialect.getRejectionMessage();
             dialect = null;
             state = State.INITIAL;
-            throw error(EDIException.INVALID_STATE, "Invalid header segment");
+            throw error(EDIException.INVALID_STATE, rejectionMessage);
         }
 
         return false;