com.articulate.sigma.nlg.NLGUtils Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of sigma-component Show documentation
Show all versions of sigma-component Show documentation
Sigma knowledge engineering system is an system for developing, viewing and debugging theories in first
order logic. It works with Knowledge Interchange Format (KIF) and is optimized for the Suggested Upper Merged
Ontology (SUMO) www.ontologyportal.org.
The newest version!
package com.articulate.sigma.nlg;
import com.articulate.sigma.Formula;
import com.articulate.sigma.KB;
import com.articulate.sigma.KBmanager;
import com.articulate.sigma.StringUtil;
import com.google.common.collect.Maps;
import java.io.BufferedReader;
import java.io.File;
import java.io.FileInputStream;
import java.io.InputStreamReader;
import java.util.*;
/**
* Utilities and variables used by LanguageFormatter and other NLG classes.
*/
public class NLGUtils {
private static final String SIGMA_HOME = System.getenv("SIGMA_HOME");
//private static final String KB_PATH = (new File(SIGMA_HOME, "KBs")).getAbsolutePath() + File.separator + "WordNetMappings";
private static final String PHRASES_FILENAME = "Translations/language.txt";
static HashMap> keywordMap;
/** *************************************************************
*/
static String prettyPrint(String term) {
if (term.endsWith("Fn"))
term = term.substring(0,term.length()-2);
StringBuilder result = new StringBuilder();
for (int i = 0; i < term.length(); i++) {
if (Character.isLowerCase(term.charAt(i)) || !Character.isLetter(term.charAt(i)))
result.append(term.charAt(i));
else {
if (i + 1 < term.length() && Character.isUpperCase(term.charAt(i+1)))
result.append(term.charAt(i));
else {
if (i != 0)
result.append(" ");
result.append(Character.toLowerCase(term.charAt(i)));
}
}
}
return result.toString();
}
/** ***************************************************************
* Resolve the "format specifiers" in the given printf type of statement.
* @param template
* @param href
* @return
*/
public static String resolveFormatSpecifiers(String template, String href) {
String anchorStart = ("= sblen)
break;
di = sb.indexOf(ditok, tj);
if (di == -1)
break;
String termName = sb.substring(tj, di);
dj = (di + ditoklen);
if (dj >= sblen)
break;
dk = sb.indexOf(dktok, dj);
if (dk == -1)
break;
String displayName = sb.substring(dj, dk);
if (StringUtil.emptyString(displayName))
displayName = termName;
StringBuilder rsb = new StringBuilder();
rsb.append(anchorStart);
rsb.append(termName);
rsb.append("\">");
rsb.append(displayName);
rsb.append("");
dl = (dk + dktoklen);
if (dl > sblen)
break;
int rsblen = rsb.length();
sb = sb.replace(ti, dl, rsb.toString());
sblen = sb.length();
ti = (ti + rsblen);
if (ti >= sblen)
break;
tj = -1;
di = -1;
dj = -1;
dk = -1;
dl = -1;
}
return sb.toString();
}
/** ***************************************************************
* Format a list of variables which are not enclosed by parens.
* Formatting includes inserting the appropriate separator between the elements (usually a comma), as well as
* inserting the conjunction ("and" or its equivalent in another language) if the conjunction doesn't already exist.
* @param strseq
* the list of variables
* @param language
* the target language (used for the conjunction "and")
* @return
* the formatted string
*/
public static String formatList(String strseq, String language) {
if (language == null || language.isEmpty()) {
throw new IllegalArgumentException("Parameter language is empty or null.");
}
StringBuilder result = new StringBuilder();
String comma = NLGUtils.getKeyword(",", language);
String space = " ";
String[] arr = strseq.split(space);
int lastIdx = (arr.length - 1);
for (int i = 0; i < arr.length; i++) {
String val = arr[i];
if (i > 0) {
if(val.equals(NLGUtils.getKeyword("and", language))) {
// Make behavior for lists that include "and" the same as for those that don't.
continue;
}
if (i == lastIdx) {
result.append(space);
result.append(NLGUtils.getKeyword("and", language));
}
else {
result.append(comma);
}
result.append(space);
}
result.append(val);
}
return result.toString();
}
/** ***************************************************************
*/
static boolean logicalOperator(String word) {
String logops = "if,then,=>,and,or,<=>,not,forall,exists,holds";
return logops.contains(word);
}
/** ***************************************************************
* Read a set of standard words and phrases in several languages.
* Each phrase must appear on a new line with alternatives separated by '|'.
* The first entry should be a set of two letter language identifiers.
*
* @return a HashMap of HashMaps where the first HashMap has a key of the
* English phrase, and the interior HashMap has a key of the two letter
* language identifier.
*/
public static HashMap> readKeywordMap(String dir) {
System.out.println("INFO in NLGUtils.readKeywordMap(" + dir + "/" +
PHRASES_FILENAME + ")");
if(dir == null || dir.isEmpty()) {
throw new IllegalArgumentException("Parameter dir is null or empty.");
}
File dirFile = new File(dir);
if(! dirFile.exists()) {
throw new IllegalArgumentException("Parameter dir points to non-existent path: " + dir);
}
if (keywordMap == null)
keywordMap = new HashMap<>();
int lc = 0;
BufferedReader br = null;
File phrasesFile = null;
try {
if (keywordMap.isEmpty()) {
System.out.println("Filling keywordMap");
phrasesFile = new File(dirFile, PHRASES_FILENAME);
if (!phrasesFile.canRead())
throw new Exception("Cannot read \"" + phrasesFile.getCanonicalPath() + "\"");
br = new BufferedReader(new InputStreamReader(new FileInputStream(phrasesFile),"UTF-8"));
HashMap phrasesByLang = null;
List phraseList = null;
List languageKeys = null;
String delim = "|";
String key = null;
String line = null;
while ((line = br.readLine()) != null) {
lc++;
line = line.trim();
if (line.startsWith(";") || line.equals("")) {
continue;
}
if (line.contains(delim)) {
if (line.startsWith("EnglishLanguage|")) // The language key line.
languageKeys = Arrays.asList(line.split("\\" + delim));
else {
phraseList = Arrays.asList(line.split("\\" + delim));
phrasesByLang = new HashMap<>();
key = phraseList.get(0);
int plLen = phraseList.size();
for (int i = 0; i < plLen; i++)
phrasesByLang.put(languageKeys.get(i), phraseList.get(i));
keywordMap.put(key.intern(), phrasesByLang);
}
}
else {
System.out.println("WARNING in NLGUtils.readKeywordMap(): "
+ "Unrecognized line");
System.out.println(lc + ": \"" + line + "\"");
}
}
}
}
catch (Exception ex) {
System.out.println("ERROR loading " + PHRASES_FILENAME + " at line " + lc + ":");
System.out.println(ex.getMessage());
ex.printStackTrace();
}
finally {
try {
if (br != null) { br.close(); }
}
catch (Exception ex2) {
ex2.printStackTrace();
}
System.out.println("EXIT NLGUtils.readKeywordMap(" + dir + ")");
//return keywordMap;
}
return keywordMap;
}
/** **************************************************************
* Generate a linguistic article appropriate to how many times in a
* paraphrase a particular type has already occurred.
* @param occurrence is the number of times a variables of a
* given type have appeared
* @param count is the number of times a given variable has
* appeared
*/
public static String getArticle(String s, int count, int occurrence, String language) {
String ordinal = "";
switch (occurrence) {
case 3: ordinal = NLGUtils.getKeyword("third", language); break;
case 4: ordinal = NLGUtils.getKeyword("fourth", language); break;
case 5: ordinal = NLGUtils.getKeyword("fifth", language); break;
case 6: ordinal = NLGUtils.getKeyword("sixth", language); break;
case 7: ordinal = NLGUtils.getKeyword("seventh", language); break;
case 8: ordinal = NLGUtils.getKeyword("eighth", language); break;
case 9: ordinal = NLGUtils.getKeyword("ninth", language); break;
case 10: ordinal = NLGUtils.getKeyword("tenth", language); break;
case 11: ordinal = NLGUtils.getKeyword("eleventh", language); break;
case 12: ordinal = NLGUtils.getKeyword("twelfth", language); break;
}
boolean isArabic = (language.matches(".*(?i)arabic.*")
|| language.equalsIgnoreCase("ar"));
if (count == 1 && occurrence == 2)
return NLGUtils.getKeyword("another", language);
if (count > 1) {
if (occurrence == 1) {
if (isArabic)
return ordinal;
else
return (NLGUtils.getKeyword("the", language));
}
else if (occurrence > 2) {
if (isArabic)
return ordinal;
else
return (NLGUtils.getKeyword("the", language) + " " + ordinal);
}
else {
if (isArabic)
return (NLGUtils.getKeyword("the", language) + " " + NLGUtils.getKeyword("other", language));
else
return (NLGUtils.getKeyword("the", language) + " " + NLGUtils.getKeyword("other", language));
}
}
// count = 1 (first occurrence of a type)
if (language.equalsIgnoreCase("EnglishLanguage")) {
String vowels = "AEIOUaeiou";
if ((vowels.indexOf(s.charAt(0)) != -1) && (occurrence == 1))
return "an";
else
return "a " + ordinal;
}
else if (isArabic) {
String defArt = NLGUtils.getKeyword("the", language);
if (ordinal.startsWith(defArt)) {
// remove the definite article
ordinal = ordinal.substring(defArt.length());
// remove shadda
ordinal = ordinal.replaceFirst("\\&\\#\\x651\\;","");
}
return ordinal;
}
else
return ordinal;
}
/** **************************************************************
* Collect all the variables occurring in a formula in order. Return
* an ArrayList of Strings.
*/
static ArrayList collectOrderedVariables(String form) {
boolean inString = false;
boolean inVar = false;
String var = "";
ArrayList result = new ArrayList<>();
for (int i = 0; i < form.length(); i++) {
char ch = form.charAt(i);
switch (ch) {
case '"': inString = !inString; break;
case '?': if (!inString) inVar = true; break;
case '@': if (!inString) inVar = true; break;
}
if (inVar && !Character.isLetterOrDigit(ch) && ch != '?' && ch != '@') {
if (!result.contains(var))
result.add(var);
inVar = false;
var = "";
}
if (inVar)
var = var + ch;
}
// Add input-final var if it exists.
if (! var.trim().isEmpty()) {
result.add(var);
}
return result;
}
/** **************************************************************
* Return a defensive copy of the keyword map.
*/
public static HashMap> getKeywordMap() {
if (keywordMap == null) {
return null;
}
return Maps.newHashMap(keywordMap);
}
/** ***************************************************************
*/
static String getKeyword(String englishWord, String language) {
String ans = "";
if (keywordMap == null) {
System.out.println("Error in NLGUtils.getKeyword(): keyword map is null");
return ans;
}
HashMap hm = keywordMap.get(englishWord);
if (hm != null) {
String tmp = hm.get(language);
if (tmp != null)
ans = tmp;
}
return ans;
}
/** **************************************************************
* Hyperlink terms in a natural language format string. This assumes that
* terms to be hyperlinked are in the form &%termName$termString , where
* termName is the name of the term to be browsed in the knowledge base and
* termString is the text that should be displayed hyperlinked.
*
* @param href the anchor string up to the term= parameter, which this method
* will fill in.
* @param stmt the KIF statement that will be passed to paraphraseStatement for formatting.
* @param phraseMap the set of NL formatting statements that will be passed to paraphraseStatement.
* @param termMap the set of NL statements for terms that will be passed to paraphraseStatement.
* @param language the natural language in which the paraphrase should be generated.
*/
public static String htmlParaphrase(String href, String stmt, Map phraseMap,
Map termMap, KB kb, String language) {
LanguageFormatter languageFormatter = new LanguageFormatter(stmt, phraseMap, termMap, kb, language);
return languageFormatter.htmlParaphrase(href);
}
/** ***************************************************************
* This method expands all "star" (asterisk) directives in the input
* format string, and returns a new format string with individually
* numbered argument pointers.
*
* @param f The Formula being paraphrased.
*
* @param strFormat The format string that contains the patterns and
* directives for paraphrasing f.
*
* @param lang A two-character string indicating the language into
* which f should be paraphrased.
*
* @return A format string with all relevant argument pointers
* expanded.
*/
static String expandStar(Formula f, String strFormat, String lang) {
String result = strFormat;
ArrayList problems = new ArrayList<>();
try {
int flen = f.listLength();
if (StringUtil.isNonEmptyString(strFormat) && (flen > 1)) {
int p1 = 0;
int p2 = strFormat.indexOf("%*");
if (p2 != -1) {
int slen = strFormat.length();
String lb = null;
String rb = null;
int lbi = -1;
int rbi = -1;
String ss = null;
String range = null;
String[] rangeArr = null;
String[] rangeArr2 = null;
String lowStr = null;
String highStr = null;
int low = -1;
int high = -1;
String delim = " ";
boolean isRange = false;
boolean[] argsToPrint = new boolean[ flen ];
int nArgsSet = -1;
StringBuilder sb = new StringBuilder();
while ((p1 < slen) && (p2 >= 0) && (p2 < slen)) {
sb.append(strFormat.substring(p1, p2));
p1 = (p2 + 2);
for (int k = 0 ; k < argsToPrint.length ; k++) {
argsToPrint[k] = false;
}
lowStr = null;
highStr = null;
low = -1;
high = -1;
delim = " ";
nArgsSet = 0;
lb = null;
lbi = p1;
if (lbi < slen) { lb = strFormat.substring(lbi, (lbi + 1)); }
while ((lb != null) && (lb.equals("{") || lb.equals("["))) {
rb = "]";
if (lb.equals("{")) { rb = "}"; }
rbi = strFormat.indexOf(rb, lbi);
if (rbi == -1) {
problems.add("Error in format \"" + strFormat + "\": missing \"" + rb + "\"");
break;
}
p1 = (rbi + 1);
ss = strFormat.substring((lbi + 1), rbi);
if (lb.equals("{")) {
range = ss.trim();
rangeArr = range.split(",");
for (String aRangeArr : rangeArr) {
if (StringUtil.isNonEmptyString(aRangeArr)) {
isRange = (aRangeArr.contains("-"));
rangeArr2 = aRangeArr.split("-");
lowStr = rangeArr2[0].trim();
try {
low = Integer.parseInt(lowStr);
} catch (Exception e1) {
problems.add("Error in format \"" + strFormat + "\": bad value in \"" + ss + "\"");
low = 1;
}
high = low;
if (isRange) {
if (rangeArr2.length == 2) {
highStr = rangeArr2[1].trim();
try {
high = Integer.parseInt(highStr);
} catch (Exception e2) {
problems.add("Error in format \"" + strFormat + "\": bad value in \"" + ss + "\"");
high = (flen - 1);
}
} else
high = (flen - 1);
}
for (int j = low; (j <= high) && (j < argsToPrint.length); j++) {
argsToPrint[j] = true;
nArgsSet++;
}
}
}
}
else
delim = ss;
lb = null;
lbi = p1;
if (lbi < slen) { lb = strFormat.substring(lbi, (lbi + 1)); }
}
String AND = getKeyword("and", lang);
if (StringUtil.emptyString(AND))
AND = "+";
int nAdded = 0;
boolean addAll = (nArgsSet == 0);
int nToAdd = (addAll ? (argsToPrint.length - 1) : nArgsSet);
for (int i = 1 ; i < argsToPrint.length ; i++) {
if (addAll || argsToPrint[i]) {
if (nAdded >= 1) {
if (nToAdd == 2) {
sb.append(" ");
sb.append(AND);
sb.append(" ");
}
else {
sb.append(delim);
sb.append(" ");
}
if ((nToAdd > 2) && ((nAdded + 1) == nToAdd)) {
sb.append(AND);
sb.append(" ");
}
}
sb.append("%").append(i);
nAdded++;
}
}
if (p1 < slen) {
p2 = strFormat.indexOf("%*", p1);
if (p2 == -1) {
sb.append(strFormat.substring(p1, slen));
break;
}
}
}
if (sb.length() > 0) {
result = sb.toString();
}
}
}
}
catch (Exception ex) {
ex.printStackTrace();
}
if (! problems.isEmpty()) {
String errStr = KBmanager.getMgr().getError();
String str = null;
if (errStr == null) { errStr = ""; }
Iterator it = problems.iterator();
while (it.hasNext()) {
str = it.next();
System.out.println("Error in NLGUtils.expandStar(): ");
System.out.println(" " + str);
errStr += ("\n
" + str + "\n
");
}
KBmanager.getMgr().setError(errStr);
}
return result;
}
/** **************************************************************
* Capitalizes the first visible char of htmlParaphrase, if
* possible, and adds the full stop symbol for language at a
* workable place near the end of htmlParaphrase if addFullStop is
* true.
*
* @param htmlParaphrase Any String, but assumed to be a Formula
* paraphrase with HTML markup
*
* @param addFullStop If true, this method will try to add a full
* stop symbol to the result String.
*
* @param language The language of the paraphrase String.
*
* @return String
*/
public static String upcaseFirstVisibleChar(String htmlParaphrase,
boolean addFullStop,
String language) {
String ans = htmlParaphrase;
try {
if (StringUtil.isNonEmptyString(htmlParaphrase)) {
StringBuilder sb = new StringBuilder(htmlParaphrase.trim());
String termKey = "term=";
int sbLen = sb.length();
if (sbLen > 0) {
int codePoint = -1;
int termCodePoint = -1;
int termPos = -1;
String uc = null;
int i = 0;
while ((i > -1) && (i < sbLen)) {
// System.out.println("x");
codePoint = Character.codePointAt(sb, i);
if (Character.isLetter(codePoint)) {
if (Character.isLowerCase(codePoint)) {
boolean isKifTermCapitalized = true;
termPos = sb.indexOf(termKey);
if (termPos > -1) {
termPos = (termPos + termKey.length());
if (termPos < i) {
termCodePoint = Character.codePointAt(sb, termPos);
isKifTermCapitalized = Character.isUpperCase(termCodePoint);
}
}
if (isKifTermCapitalized) {
uc = sb.substring(i, i + 1).toUpperCase();
sb = sb.replace(i, i + 1, uc);
}
}
break;
}
i = sb.indexOf(">", i);
if (i > -1) {
i++;
while ((i < sbLen) && sb.substring(i, i + 1).matches("\\s"))
i++;
}
}
if (addFullStop) {
String fs = getKeyword(".", language);
if (StringUtil.isNonEmptyString(fs)) {
String ss = "";
sbLen = sb.length();
i = (sbLen - 1);
while ((i > 0) && (i < sbLen)) {
// System.out.println("m");
ss = sb.substring(i, i + 1);
if (ss.matches("\\s")) {
i--;
continue;
}
if (ss.matches("[\\w;]")) {
sb = sb.insert(i + 1, fs);
break;
}
while ((i > 0) && !ss.equals("<")) {
i--;
ss = sb.substring(i, i + 1);
}
i--;
}
}
}
ans = sb.toString();
}
}
}
catch (Exception ex) {
ex.printStackTrace();
}
return ans;
}
/**************************************************************************************************************
* Return true if the given list includes "Process", or if one of its elements is a subclass of Process.
* @param vals
* @param kb
* @return
*/
public static boolean containsProcess(Collection vals, KB kb) {
for (String val : vals) {
if (val.equals("Process") || kb.isSubclass(val, "Process")) {
return true;
}
}
return false;
}
}