Skip to content

Commit

Permalink
Reduce code duplication.
Browse files Browse the repository at this point in the history
  • Loading branch information
@kno10
kno10 committed Aug 1, 2017
1 parent 1073eac commit f001d81
Showing 1 changed file with 56 additions and 61 deletions.
117 changes: 56 additions & 61 deletions src/de/unihd/dbs/uima/annotator/heideltime/utilities/ContextAnalyzer.java
View file
Original file line number Diff line number Diff line change
@@ -1,8 +1,8 @@
package de.unihd.dbs.uima.annotator.heideltime.utilities;

import java.util.ArrayList;
import java.util.Comparator;
import java.util.List;
import java.util.Map;
import java.util.TreeMap;
import java.util.function.Function;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
Expand All @@ -21,11 +21,9 @@
import de.unihd.dbs.uima.types.heideltime.Token;

/**
*
* This class contains methods that work with the dependence of a subject with its surrounding data; namely via the jcas element or a subset list.
*
* @author jannik stroetgen
*
*/
public class ContextAnalyzer {
/** Class logger */
Expand Down Expand Up @@ -331,30 +329,12 @@ public static Tense getClosestTense(Timex3 timex, JCas jcas, Language language)
int lastid = 0, nextid = 0;
int tid = 0;

// Get the sentence
AnnotationIndex<Sentence> sentences = jcas.getAnnotationIndex(Sentence.type);
Sentence s = null;
for (FSIterator<Sentence> iterSentence = sentences.iterator(); iterSentence.hasNext();) {
s = iterSentence.next();
if (s.getBegin() <= timex.getBegin() && s.getEnd() >= timex.getEnd()) {
break;
}
}

// Get the tokens
TreeMap<Integer, Token> tmToken = new TreeMap<Integer, Token>();
AnnotationIndex<Token> tokens = jcas.getAnnotationIndex(Token.type);
FSIterator<Token> iter = (s != null) ? tokens.subiterator(s) : tokens.iterator();
while (iter.hasNext()) {
Token token = iter.next();
tmToken.put(token.getEnd(), token);
}
ArrayList<Token> tmToken = getCloseTokens(timex, jcas);

// Get the last VERB token
for (Map.Entry<Integer, Token> ent : tmToken.entrySet()) {
for (Token token : tmToken) {
tokenCounter++;
if (ent.getKey() < timex.getBegin()) {
Token token = ent.getValue();
if (token.getEnd() < timex.getBegin()) {
String pos = token.getPos();
if (pos == null)
continue; // POS not available?
Expand Down Expand Up @@ -385,11 +365,10 @@ public static Tense getClosestTense(Timex3 timex, JCas jcas, Language language)
}
}
tokenCounter = 0;
for (Map.Entry<Integer, Token> ent : tmToken.entrySet()) {
for (Token token : tmToken) {
tokenCounter++;
if (nextTense == null) {
if (ent.getKey() > timex.getEnd()) {
Token token = ent.getValue();
if (token.getEnd() > timex.getEnd()) {
String pos = token.getPos();
if (pos == null)
continue; // No POS available?
Expand Down Expand Up @@ -453,27 +432,11 @@ public static Tense getLastTense(Timex3 timex, JCas jcas, Language language) {
Tense lastTense = null;

// Get the sentence
AnnotationIndex<Sentence> sentences = jcas.getAnnotationIndex(Sentence.type);
Sentence s = null;
for (FSIterator<Sentence> iterSentence = sentences.iterator(); iterSentence.hasNext();) {
s = iterSentence.next();
if (s.getBegin() <= timex.getBegin() && s.getEnd() >= timex.getEnd())
break;
}

// Get the tokens
TreeMap<Integer, Token> tmToken = new TreeMap<Integer, Token>();
AnnotationIndex<Token> tokens = jcas.getAnnotationIndex(Token.type);
FSIterator<Token> iter = (s != null) ? tokens.subiterator(s) : tokens.iterator();
while (iter.hasNext()) {
Token token = iter.next();
tmToken.put(token.getEnd(), token);
}
ArrayList<Token> tmToken = getCloseTokens(timex, jcas);

// Get the last VERB token
for (Map.Entry<Integer, Token> ent : tmToken.entrySet()) {
if (ent.getKey() < timex.getBegin()) {
Token token = ent.getValue();
for (Token token : tmToken) {
if (token.getEnd() < timex.getBegin()) {
String coveredText = token.getCoveredText();
String pos = token.getPos();
if (pos == null)
Expand All @@ -500,8 +463,7 @@ public static Tense getLastTense(Timex3 timex, JCas jcas, Language language) {
lastTense = Tense.PAST;
}
}
if (lastTense == null && ent.getKey() > timex.getEnd()) {
Token token = ent.getValue();
if (lastTense == null && token.getEnd() > timex.getEnd()) {
String pos = token.getPos();

if (LOG.isTraceEnabled()) {
Expand All @@ -524,16 +486,15 @@ public static Tense getLastTense(Timex3 timex, JCas jcas, Language language) {
}
}
if (lastTense != null)
LOG.trace("this tense: {} {}", ent.getValue().getCoveredText(), lastTense);
LOG.trace("this tense: {} {}", token.getCoveredText(), lastTense);
}
// check for double POS Constraints (not included in the rule language, yet) TODO
// VHZ VNN and VHZ VNN and VHP VNN and VBP VVN
String prevPos = "";
Tense longTense = null;
if (lastTense == Tense.PRESENTFUTURE) {
for (Map.Entry<Integer, Token> ent : tmToken.entrySet()) {
if (ent.getKey() < timex.getBegin()) {
Token token = ent.getValue();
for (Token token : tmToken) {
if (token.getEnd() < timex.getBegin()) {
String pos = token.getPos();
if ("VHZ".equals(prevPos) || "VBZ".equals(prevPos) || "VHP".equals(prevPos) || "VBP".equals(prevPos) || prevPos.equals("VER:pres")) {
if ("VVN".equals(pos) || "VER:pper".equals(pos)) {
Expand All @@ -546,8 +507,7 @@ public static Tense getLastTense(Timex3 timex, JCas jcas, Language language) {
}
prevPos = pos;
}
if (longTense == null && ent.getKey() > timex.getEnd()) {
Token token = ent.getValue();
if (longTense == null && token.getEnd() > timex.getEnd()) {
if ("VHZ".equals(prevPos) || "VBZ".equals(prevPos) || "VHP".equals(prevPos) || "VBP".equals(prevPos) || "VER:pres".equals(prevPos)) {
if ("VVN".equals(token.getPos()) || "VER:pper".equals(token.getPos())) {
String covered = token.getCoveredText();
Expand All @@ -563,9 +523,8 @@ public static Tense getLastTense(Timex3 timex, JCas jcas, Language language) {
}
// French: VER:pres VER:pper
if (lastTense == Tense.PAST) {
for (Map.Entry<Integer, Token> ent : tmToken.entrySet()) {
if (ent.getKey() < timex.getBegin()) {
Token token = ent.getValue();
for (Token token : tmToken) {
if (token.getEnd() < timex.getBegin()) {
String pos = token.getPos();
if ("VER:pres".equals(prevPos) && "VER:pper".equals(pos)) {
if (PREVUE_ENVISAGEE.matcher(token.getCoveredText()).matches()) {
Expand All @@ -576,8 +535,7 @@ public static Tense getLastTense(Timex3 timex, JCas jcas, Language language) {
prevPos = pos;
}
if (longTense == null) {
if (ent.getKey() > timex.getEnd()) {
Token token = ent.getValue();
if (token.getEnd() > timex.getEnd()) {
String pos = token.getPos();
if ("VER:pres".equals(prevPos) && "VER:pper".equals(pos)) {
if (PREVUE_ENVISAGEE.matcher(token.getCoveredText()).matches()) {
Expand All @@ -591,7 +549,44 @@ public static Tense getLastTense(Timex3 timex, JCas jcas, Language language) {
}
}
LOG.trace("TENSE: {}", lastTense);

return lastTense;
}

/**
* Get the tokens close to the given timex (i.e. the same sentence).
*
* @param timex
* Timex
* @param jcas
* Cas
* @return Tokens, sorted by end.
*/
private static ArrayList<Token> getCloseTokens(Timex3 timex, JCas jcas) {
// Get the sentence
AnnotationIndex<Sentence> sentences = jcas.getAnnotationIndex(Sentence.type);
Sentence s = null;
for (FSIterator<Sentence> iterSentence = sentences.iterator(); iterSentence.hasNext();) {
s = iterSentence.next();
if (s.getBegin() <= timex.getBegin() && s.getEnd() >= timex.getEnd())
break;
}

// Get the tokens
AnnotationIndex<Token> tokens = jcas.getAnnotationIndex(Token.type);
FSIterator<Token> iter = (s != null) ? tokens.subiterator(s) : tokens.iterator();
ArrayList<Token> tmToken = new ArrayList<Token>();
while (iter.hasNext())
tmToken.add(iter.next());
tmToken.sort(SORT_TOKENS);
return tmToken;
}

/**
* Sort tokens by the token end.
*/
private static final Comparator<Token> SORT_TOKENS = new Comparator<Token>() {
public int compare(Token o1, Token o2) {
return Integer.compare(o1.getEnd(), o2.getEnd());
}
};
}

0 comments on commit f001d81

Please sign in to comment.