replace epsilon and arrow right by constants

SamyaDaleh · Jan 7, 2024 · 17bab7b · 17bab7b
1 parent b082006
commit 17bab7b
Show file tree

Hide file tree

Showing 33 changed files with 148 additions and 81 deletions.
diff --git a/src/main/java/com/github/samyadaleh/cltoolbox/chartparsing/Deduction.java b/src/main/java/com/github/samyadaleh/cltoolbox/chartparsing/Deduction.java
@@ -13,6 +13,9 @@
 import java.util.*;
 import java.util.stream.Collectors;
 
+import static com.github.samyadaleh.cltoolbox.common.Constants.ARROW_RIGHT;
+import static com.github.samyadaleh.cltoolbox.common.Constants.EPSILON;
+
 /**
  * A deduction system that derives consequences from antecendence items and
  * tries to generate a goal item. Based on the slides from Laura Kallmeyer about
@@ -194,11 +197,11 @@ public void printTraceLatex(String[][] trace) {
     for (String[] line : trace) {
       String id = line[0];
       String item = line[1].replace("$", "\\$").replace("•", "\\textbullet{}")
-          .replace("ε", "$\\epsilon$").replace("->", "$\\rightarrow$");
+          .replace(EPSILON, "$\\epsilon$").replace(ARROW_RIGHT, "$\\rightarrow$");
       String rules =
-          line[2].replace("ε", "$\\epsilon$").replace("->", "$\\rightarrow$");
+          line[2].replace(EPSILON, "$\\epsilon$").replace(ARROW_RIGHT, "$\\rightarrow$");
       String backPointers = line[3].replace("{", "\\{").replace("}", "\\}");
-      String trees = line[4].replace("ε", "$\\epsilon$");
+      String trees = line[4].replace(EPSILON, "$\\epsilon$");
       log.info(
           id + " & " + item + " & " + rules + " & " + backPointers + " & " + trees + " \\\\");
     }

diff --git a/...va/com/github/samyadaleh/cltoolbox/chartparsing/converter/cfg/CfgToCykRulesConverter.java b/...va/com/github/samyadaleh/cltoolbox/chartparsing/converter/cfg/CfgToCykRulesConverter.java
@@ -16,8 +16,7 @@
 import java.util.ArrayList;
 import java.util.List;
 
-import static com.github.samyadaleh.cltoolbox.common.Constants.DEDUCTION_RULE_CFG_CYK_AXIOM;
-import static com.github.samyadaleh.cltoolbox.common.Constants.DEDUCTION_RULE_CFG_CYK_AXIOM_EPSILON;
+import static com.github.samyadaleh.cltoolbox.common.Constants.*;
 
 public class CfgToCykRulesConverter {
 
@@ -30,7 +29,7 @@ public static ParsingSchema cfgToCykRules(Cfg cfg, String w)
       throw new ParseException("Grammar has to be in Chomsky Normal Form.", 1);
     }
     String[] wSplit = w.split(" ");
-    int wLength = wSplit[0].equals("") || wSplit[0].equals("ε")
+    int wLength = wSplit[0].equals("") || wSplit[0].equals(EPSILON)
         ? 0 : wSplit.length;
     ParsingSchema schema = new ParsingSchema();
 
@@ -84,7 +83,7 @@ private static void addCykScanRules(String[] wSplit, ParsingSchema schema,
     for (int i = 0; i < wSplit.length; i++) {
       if (wSplit[i].equals(rule.getRhs()[0])) {
         StaticDeductionRule scan = new StaticDeductionRule();
-        String itemLength = wSplit[i].equals("") || wSplit[i].equals("ε")
+        String itemLength = wSplit[i].equals("") || wSplit[i].equals(EPSILON)
             ? "0" : "1";
         ChartItemInterface consequence =
             new DeductionChartItem(rule.getLhs(), String.valueOf(i),

diff --git a/src/main/java/com/github/samyadaleh/cltoolbox/chartparsing/converter/lag/LagChartItem.java b/src/main/java/com/github/samyadaleh/cltoolbox/chartparsing/converter/lag/LagChartItem.java
@@ -3,6 +3,8 @@
 import com.github.samyadaleh.cltoolbox.chartparsing.item.AbstractChartItem;
 import com.github.samyadaleh.cltoolbox.chartparsing.item.ChartItemInterface;
 
+import static com.github.samyadaleh.cltoolbox.common.Constants.EPSILON;
+
 /**
  * Chart Item for Left associative grammar that uses the usual itemForm for
  * display, but stores its own data structures for further use.
@@ -31,7 +33,7 @@ public LagChartItem(String[][] categories, String[] rulePackage) {
         categorieRepr.append(cat);
       }
       if (categorieRepr.charAt(categorieRepr.length()-1) == '(') {
-        categorieRepr.append("ε");
+        categorieRepr.append(EPSILON);
       }
       categorieRepr.append(")");
     }

diff --git a/...om/github/samyadaleh/cltoolbox/chartparsing/converter/lcfrs/LcfrsToCykRulesConverter.java b/...om/github/samyadaleh/cltoolbox/chartparsing/converter/lcfrs/LcfrsToCykRulesConverter.java
@@ -19,6 +19,7 @@
 import java.util.List;
 
 import static com.github.samyadaleh.cltoolbox.common.Constants.DEDUCTION_RULE_LCFRS_CYK_AXIOM;
+import static com.github.samyadaleh.cltoolbox.common.Constants.EPSILON;
 
 public class LcfrsToCykRulesConverter {
 
@@ -34,7 +35,7 @@ private static void addSrcgCykScanRules(String[] wsplit, ParsingSchema schema,
         String[] arg = clause.getLhs().getArgumentByIndex(i + 1);
         for (int j = 0; j < arg.length; j++) {
           if (arg[j].length() == 0) {
-            treeString.append("ε");
+            treeString.append(EPSILON);
           } else {
             treeString.append(clause.getLhs().getSymAt(i + 1, j));
           }

diff --git a/src/main/java/com/github/samyadaleh/cltoolbox/chartparsing/tag/cyk/TagCykAdjoin.java b/src/main/java/com/github/samyadaleh/cltoolbox/chartparsing/tag/cyk/TagCykAdjoin.java
@@ -11,6 +11,7 @@
 import com.github.samyadaleh.cltoolbox.common.tag.Tree;
 
 import static com.github.samyadaleh.cltoolbox.common.Constants.DEDUCTION_RULE_TAG_ADJOIN;
+import static com.github.samyadaleh.cltoolbox.common.Constants.EPSILON;
 
 /**
  * Adjoin an auxiliary tree into an appropriate node in any other tree.
@@ -64,7 +65,7 @@ protected void calculateConsequences(String[] itemForm1, String[] itemForm2) {
         }
       }
       String node2Name =
-          (node2.length() > 1) ? node2.substring(0, node2.length() - 1) : "ε";
+          (node2.length() > 1) ? node2.substring(0, node2.length() - 1) : EPSILON;
       this.name =
           DEDUCTION_RULE_TAG_ADJOIN + " " + treeName2 + "[" + node2Name + ","
               + treeName1 + "]";

diff --git a/src/main/java/com/github/samyadaleh/cltoolbox/chartparsing/tag/earley/TagEarleyAdjoin.java b/src/main/java/com/github/samyadaleh/cltoolbox/chartparsing/tag/earley/TagEarleyAdjoin.java
@@ -11,6 +11,7 @@
 import com.github.samyadaleh.cltoolbox.common.tag.Tree;
 
 import static com.github.samyadaleh.cltoolbox.common.Constants.DEDUCTION_RULE_TAG_EARLEY_ADJOIN;
+import static com.github.samyadaleh.cltoolbox.common.Constants.EPSILON;
 
 /**
  * Combines an auxiliary tree with another tree to get a new item in which has
@@ -67,7 +68,7 @@ protected void calculateConsequences(String[] itemForm1, String[] itemForm2) {
           }
         }
         consequence.setTrees(derivedTrees);
-        String node2name = node2.length() == 0 ? "ε" : node2;
+        String node2name = node2.length() == 0 ? EPSILON : node2;
         this.name =
             DEDUCTION_RULE_TAG_EARLEY_ADJOIN + " " + treeName2 + "[" + node2name
                 + "," + treeName1 + "]";

diff --git a/...ain/java/com/github/samyadaleh/cltoolbox/chartparsing/tag/earley/TagEarleySubstitute.java b/...ain/java/com/github/samyadaleh/cltoolbox/chartparsing/tag/earley/TagEarleySubstitute.java
@@ -11,6 +11,7 @@
 import com.github.samyadaleh.cltoolbox.common.tag.Tree;
 
 import static com.github.samyadaleh.cltoolbox.common.Constants.DEDUCTION_RULE_TAG_EARLEY_SUBSTITUTE;
+import static com.github.samyadaleh.cltoolbox.common.Constants.EPSILON;
 
 /**
  * If a potential initial tree is complete, substitute it if possible.
@@ -63,7 +64,7 @@ public TagEarleySubstitute(String outTreeName, String outNode, Tag tag) {
         }
         // imagine a tree with 1 node where you would substitute into the root
         // ...
-        String outNodeName = outNode.length() == 0 ? "ε" : outNode;
+        String outNodeName = outNode.length() == 0 ? EPSILON : outNode;
         this.name =
             "substitute " + outTreeName + "[" + outNodeName + "," + treeName
                 + "]";

diff --git a/...leh/cltoolbox/chartparsing/tag/earleyprefixvalid/TagEarleyPrefixValidPredictAdjoined.java b/...leh/cltoolbox/chartparsing/tag/earleyprefixvalid/TagEarleyPrefixValidPredictAdjoined.java
@@ -10,6 +10,7 @@
 import java.util.List;
 
 import static com.github.samyadaleh.cltoolbox.common.Constants.DEDUCTION_RULE_TAG_EARLEY_PREFIXVALID_PREDICTADJOINED;
+import static com.github.samyadaleh.cltoolbox.common.Constants.EPSILON;
 
 public class TagEarleyPrefixValidPredictAdjoined
     extends AbstractDynamicDecutionRuleTwoAntecedences {
@@ -42,7 +43,7 @@ protected void calculateConsequences(String[] itemForm1, String[] itemForm2) {
             new DeductionChartItem(treeName2, node2, "lb", iGamma2, m, "-", "-",
                 m, "0");
         List<Tree> derivedTrees = generateDerivatedTrees(itemForm1);
-        String node2name = node2.length() == 0 ? "ε" : node2;
+        String node2name = node2.length() == 0 ? EPSILON : node2;
         this.name =
             DEDUCTION_RULE_TAG_EARLEY_PREFIXVALID_PREDICTADJOINED + " " + treeName2 + "["
                 + node2name + "," + treeName1 + "]";

diff --git a/src/main/java/com/github/samyadaleh/cltoolbox/common/ArrayUtils.java b/src/main/java/com/github/samyadaleh/cltoolbox/common/ArrayUtils.java
@@ -6,6 +6,8 @@
 import java.util.List;
 import java.util.stream.Stream;
 
+import static com.github.samyadaleh.cltoolbox.common.Constants.EPSILON;
+
 /**
  * Open collection of functions to work with arrays.
  */
@@ -85,7 +87,7 @@ public static String toString(String[] item) {
         representation.append(",");
       }
       if (element.equals("")) {
-        representation.append("ε");
+        representation.append(EPSILON);
       } else {
         representation.append(element);
       }

diff --git a/src/main/java/com/github/samyadaleh/cltoolbox/common/Constants.java b/src/main/java/com/github/samyadaleh/cltoolbox/common/Constants.java
@@ -4,6 +4,10 @@
  * Class holding constant strings used all around the toolbox.
  */
 public class Constants {
+  /* Symbols */
+  public static final String EPSILON = "ε";
+  public static final String ARROW_RIGHT = "->";
+
   /* Static rules*/
   public final static String DEDUCTION_RULE_CCG_DEDUCTION_AXIOM = "axiom";
   public final static String DEDUCTION_RULE_CFG_CYK_AXIOM = "scan";

diff --git a/src/main/java/com/github/samyadaleh/cltoolbox/common/TreeUtils.java b/src/main/java/com/github/samyadaleh/cltoolbox/common/TreeUtils.java
@@ -10,11 +10,14 @@
 import java.util.ArrayList;
 import java.util.List;
 
+import static com.github.samyadaleh.cltoolbox.common.Constants.ARROW_RIGHT;
+import static com.github.samyadaleh.cltoolbox.common.Constants.EPSILON;
+
 public class TreeUtils {
 
   public static Tree getTreeOfSrcgClause(Clause clause, List<Integer> vector) {
     StringBuilder extractedRule = new StringBuilder();
-    extractedRule.append(clause.getLhs().getNonterminal()).append(" ->");
+    extractedRule.append(clause.getLhs().getNonterminal()).append(" ").append(ARROW_RIGHT);
     int terminalsInLhs = 0;
     for (String symbol : clause.getLhs().getSymbolsAsPlainArray()) {
       if (!TreeUtils.symbolIsVariable(clause, symbol)) {
@@ -55,7 +58,7 @@ public static Tree getTreeOfSrcgClause(Clause clause) {
     if (clause.getRhs().size() == 0) {
       try {
         return new Tree(new CfgProductionRule(
-            clause.getLhs().getNonterminal() + " -> " + ArrayUtils
+            clause.getLhs().getNonterminal() + " " + ARROW_RIGHT + " " + ArrayUtils
                 .getSubSequenceAsString(
                     clause.getLhs().getSymbolsAsPlainArray(), 0,
                     clause.getLhs().getSymbolsAsPlainArray().length)));
@@ -65,7 +68,7 @@ public static Tree getTreeOfSrcgClause(Clause clause) {
     }
     StringBuilder cfgRuleString =
         new StringBuilder(clause.getLhs().getNonterminal());
-    cfgRuleString.append(" ->");
+    cfgRuleString.append(" ").append(ARROW_RIGHT);
     for (Predicate rhsPred : clause.getRhs()) {
       cfgRuleString.append(" ").append(rhsPred.getNonterminal());
     }
@@ -222,7 +225,7 @@ private static List<Tree> collectSubTrees(Tree tree1, Vertex node1,
   private static String collectSubtreeAsString(Tree tree2, Vertex node2) {
     StringBuilder newTree = new StringBuilder("(");
     if (node2.getLabel().equals("")) {
-      newTree.append("ε");
+      newTree.append(EPSILON);
     } else {
       newTree.append(node2.getLabel());
     }

diff --git a/src/main/java/com/github/samyadaleh/cltoolbox/common/cfg/Cfg.java b/src/main/java/com/github/samyadaleh/cltoolbox/common/cfg/Cfg.java
@@ -10,6 +10,8 @@
 import java.text.ParseException;
 import java.util.*;
 
+import static com.github.samyadaleh.cltoolbox.common.Constants.ARROW_RIGHT;
+import static com.github.samyadaleh.cltoolbox.common.Constants.EPSILON;
 import static com.github.samyadaleh.cltoolbox.common.cfg.util.EmptyProductions.isEmptyProductionRule;
 import static com.github.samyadaleh.cltoolbox.common.cfg.util.EmptyProductions.nonterminalOccursInAnyRhs;
 
@@ -52,7 +54,7 @@ public Cfg(NondeterministicFiniteAutomaton nfa) {
       for (String transitionState : entry.getValue()) {
         try {
           productionRules.add(new CfgProductionRule(
-              entry.getKey()[0] + " -> " + entry.getKey()[1] + " "
+              entry.getKey()[0] + " " + ARROW_RIGHT + " " + entry.getKey()[1] + " "
                   + transitionState));
         } catch (ParseException e) {
           throw new RuntimeException(e);
@@ -61,7 +63,7 @@ public Cfg(NondeterministicFiniteAutomaton nfa) {
     }
     for (String finalState : nfa.getFinalStates()) {
       try {
-        productionRules.add(new CfgProductionRule(finalState + " -> ε"));
+        productionRules.add(new CfgProductionRule(finalState + " " + ARROW_RIGHT + " " + EPSILON));
       } catch (ParseException e) {
         throw new RuntimeException(e);
       }

diff --git a/src/main/java/com/github/samyadaleh/cltoolbox/common/cfg/CfgProductionRule.java b/src/main/java/com/github/samyadaleh/cltoolbox/common/cfg/CfgProductionRule.java
@@ -2,6 +2,9 @@
 
 import java.text.ParseException;
 
+import static com.github.samyadaleh.cltoolbox.common.Constants.ARROW_RIGHT;
+import static com.github.samyadaleh.cltoolbox.common.Constants.EPSILON;
+
 /**
  * Representation of a CFG production rule where the lhs consists of one
  * nonterminal and the rhs can be any length.
@@ -14,7 +17,7 @@ public class CfgProductionRule {
   public CfgProductionRule(String[] rule) {
     this.lhs = rule[0];
     String[] ruleSplit = rule[1].split(" ");
-    if ((ruleSplit.length == 1 && ruleSplit[0].equals("ε"))) {
+    if ((ruleSplit.length == 1 && ruleSplit[0].equals(EPSILON))) {
       this.rhs = new String[] {""};
     } else {
       this.rhs = ruleSplit;
@@ -27,7 +30,7 @@ public CfgProductionRule(String[] rule) {
    */
   public CfgProductionRule(String lhs, String[] rhs) {
     this.lhs = lhs;
-    if (rhs.length == 1 && rhs[0].equals("ε")) {
+    if (rhs.length == 1 && rhs[0].equals(EPSILON)) {
       this.rhs = new String[] {""};
     } else {
       this.rhs = rhs;
@@ -38,12 +41,12 @@ public CfgProductionRule(String lhs, String[] rhs) {
    * Creates a rule from a String representation like S -> A B
    */
   public CfgProductionRule(String ruleString) throws ParseException {
-    if (!ruleString.contains("->")) {
-      throw new ParseException("Separator -> missing in rule " + ruleString, 0);
+    if (!ruleString.contains(ARROW_RIGHT)) {
+      throw new ParseException("Separator " + ARROW_RIGHT + " missing in rule " + ruleString, 0);
     }
-    String[] ruleSplit = ruleString.split("->", 2);
+    String[] ruleSplit = ruleString.split(ARROW_RIGHT, 2);
     this.lhs = ruleSplit[0].trim();
-    if (ruleSplit[1].trim().equals("") || ruleSplit[1].trim().equals("ε")) {
+    if (ruleSplit[1].trim().equals("") || ruleSplit[1].trim().equals(EPSILON)) {
       this.rhs = new String[] {""};
     } else {
       this.rhs = ruleSplit[1].trim().split(" ");
@@ -60,9 +63,9 @@ public String[] getRhs() {
 
   @Override public String toString() {
     if (rhs.length == 0 || rhs[0].equals("")) {
-      return lhs + " -> ε";
+      return lhs + " " + ARROW_RIGHT + " " + EPSILON;
     } else {
-      return lhs + " -> " + String.join(" ", rhs);
+      return lhs + " " + ARROW_RIGHT + " " + String.join(" ", rhs);
     }
   }
 

diff --git a/src/main/java/com/github/samyadaleh/cltoolbox/common/cfg/PcfgProductionRule.java b/src/main/java/com/github/samyadaleh/cltoolbox/common/cfg/PcfgProductionRule.java
@@ -2,6 +2,9 @@
 
 import java.text.ParseException;
 
+import static com.github.samyadaleh.cltoolbox.common.Constants.ARROW_RIGHT;
+import static com.github.samyadaleh.cltoolbox.common.Constants.EPSILON;
+
 /**
  * Representation of a context-free rule where the lhs is only allowed to
  * contain one symbol and the rule has a probability.
@@ -28,8 +31,8 @@ public class PcfgProductionRule extends CfgProductionRule {
    */
   PcfgProductionRule(String ruleString) throws ParseException {
     super(ruleString.split(":", 2)[1]);
-    if (ruleString.indexOf("->") < ruleString.indexOf(':')) {
-      throw new ParseException(": has to be left of -> in rule " + ruleString,
+    if (ruleString.indexOf(ARROW_RIGHT) < ruleString.indexOf(':')) {
+      throw new ParseException(": has to be left of " + ARROW_RIGHT + " in rule " + ruleString,
         0);
     }
     String[] ruleSplit = ruleString.split(":", 2);
@@ -43,9 +46,9 @@ public Double getP() {
   @Override public String toString() {
     double roundedP = Math.round(p * 100.0) / 100.0;
     if (this.getRhs()[0].equals("")) {
-      return roundedP + " : " + this.getLhs() + " -> ε";
+      return roundedP + " : " + this.getLhs() + " " + ARROW_RIGHT + " " + EPSILON;
     } else {
-      return roundedP + " : " + this.getLhs() + " -> "
+      return roundedP + " : " + this.getLhs() + " " + ARROW_RIGHT + " "
           + String.join(" ", this.getRhs());
     }
   }

diff --git a/src/main/java/com/github/samyadaleh/cltoolbox/common/cfg/util/Doubling.java b/src/main/java/com/github/samyadaleh/cltoolbox/common/cfg/util/Doubling.java
@@ -9,6 +9,9 @@
 import java.util.Arrays;
 import java.util.List;
 
+import static com.github.samyadaleh.cltoolbox.common.Constants.ARROW_RIGHT;
+import static com.github.samyadaleh.cltoolbox.common.Constants.EPSILON;
+
 /**
  * Takes care that different nonterminals are used as left-corner and in other
  * places, so no symbol appears in both positions.
@@ -32,7 +35,7 @@ public static Cfg doubleSymbols(Cfg cfgOld) throws ParseException {
 
     for (int i = 0; i < newRules.size(); i++) {
       CfgProductionRule rule = newRules.get(i);
-      if (rule.getRhs().length > 0 && !"ε".equals(rule.getRhs()[0])) {
+      if (rule.getRhs().length > 0 && !EPSILON.equals(rule.getRhs()[0])) {
         handleLcOccuringInOtherRules(newRules, newNonterminals, rule);
       }
     }
@@ -52,7 +55,7 @@ private static void handleLcOccuringInOtherRules(
     //  if N occurs in any other production rule not as lc:
     for (int j = 0; j < newRules.size(); j++) {
       CfgProductionRule rule2 = newRules.get(j);
-      if (rule2.getRhs().length > 0 && !"ε".equals(rule2.getRhs()[0])) {
+      if (rule2.getRhs().length > 0 && !EPSILON.equals(rule2.getRhs()[0])) {
         for (int k = 1; k < rule2.getRhs().length; k++) {
           if (lc.equals(rule2.getRhs()[k])) {
             if (newNt == null) {
@@ -73,7 +76,7 @@ private static void duplicateRules(List<CfgProductionRule> newRules,
       CfgProductionRule rule3 = newRules.get(l);
       if (lc.equals(rule3.getLhs())) {
         newRules.add(new CfgProductionRule(
-            newNt + " -> " + rule3.toString().split("->")[1]));
+            newNt + " " + ARROW_RIGHT + " " + rule3.toString().split(ARROW_RIGHT)[1]));
       }
     }
   }