Examples with CoNLLReader org.grupolys.nlputils.parser.CoNLLReader used on opensource projects

Example 1 with CoNLLReader

use of org.grupolys.nlputils.parser.CoNLLReader in project uuusa by aghie.

the class Samulan method analyse.

// Obtains a list of SentimentDependencyGraph given a CoNLL file
// private static List<SentimentDependencyGraph> getGraphs(String path, String encoding){
// CoNLLReader conllReader = new CoNLLReader();
// List<DependencyGraph> graphs =conllReader.read(path, encoding);
// List<SentimentDependencyGraph> sgraphs = new ArrayList<SentimentDependencyGraph>();
// 
// for (DependencyGraph dg: graphs){
// 
// HashMap<Short, DependencyNode> nodes = dg.getNodes();
// HashMap<Short, DependencyNode> snodes = new HashMap<Short, DependencyNode>();
// for (short address: nodes.keySet()){
// snodes.put(address,new SentimentDependencyNode(nodes.get(address), null));
// }
// 
// sgraphs.add(new SentimentDependencyGraph(snodes));
// }
// return sgraphs;
// }
// Obtains the sentiment classification for each graph in a CoNLL file
private static void analyse(String conllFile, String encoding, RuleBasedAnalyser rba, String pathOutput, String scale, boolean verbose) {
    BufferedReader br = null;
    String line, conll = null, textID = null, previousTextID = null;
    boolean newFileGraphs = false;
    boolean first = true;
    CoNLLReader conllReader = new CoNLLReader();
    List<SentimentDependencyGraph> sdgs = new ArrayList<SentimentDependencyGraph>();
    double totalAnalyseTime = 0, textAnalyseTime = 0;
    try {
        br = new BufferedReader(new FileReader(conllFile));
    } catch (FileNotFoundException e1) {
        System.err.println("File or directory: " + conllFile + " not found");
        e1.printStackTrace();
    }
    Writer writer = null;
    try {
        if (pathOutput != null)
            writer = new PrintWriter(pathOutput, encoding);
        else
            writer = new BufferedWriter(new OutputStreamWriter(System.out));
    } catch (FileNotFoundException | UnsupportedEncodingException e1) {
        // TODO Auto-generated catch block
        e1.printStackTrace();
    }
    try {
        line = br.readLine();
        while (line != null) {
            String[] ls = line.split("\t");
            if (newFileGraphs && (previousTextID != null)) {
                long initAnalyseTextTime = System.nanoTime();
                List<SentimentInformation> sis = sdgs.stream().map((SentimentDependencyGraph dg) -> (rba.analyse(dg, (short) 0))).collect(Collectors.toList());
                long stopAnalyseTextTime = System.nanoTime();
                String text = String.join(" ", sdgs.stream().map((SentimentDependencyGraph dg) -> dg.subgraphToString((short) 0)).collect(Collectors.toList()));
                SentimentInformation siFinal = rba.merge(sis);
                try {
                    textAnalyseTime = (stopAnalyseTextTime - initAnalyseTextTime) / 1000000000.0;
                    totalAnalyseTime += textAnalyseTime;
                    writer.write(printOutputScaled(siFinal, scale, rba.getAc().isBinaryNeutralAsNegative()) + "\t" + "\t" + text + "\t" + " [The analysis took: " + textAnalyseTime + " seg.] [Accumulated time is: " + totalAnalyseTime + "]\n");
                    // writer.write(previousTextID+"\t"+printOutputScaled(siFinal,scale, rba.getAc().isBinaryNeutralAsNegative())+"\t"+"\t"+text+"\t"+" [The analysis took: "+textAnalyseTime+" seg.] [Accumulated time is: "+totalAnalyseTime+"]\n");
                    writer.flush();
                    if (verbose) {
                        sdgs.stream().forEach(sdg -> sdg.printLandscapeGraph((short) 0));
                    }
                } catch (IOException e) {
                    // TODO Auto-generated catch block
                    e.printStackTrace();
                }
                sdgs.clear();
                previousTextID = null;
                newFileGraphs = false;
            }
            // We process the line
            if (line.startsWith(CONLL_IDENTIFIER_SYMBOL)) {
                String lcleaned = line.replace(CONLL_IDENTIFIER_SYMBOL, "").replace(" ", "").replace("\n", "");
                if (!first) {
                    previousTextID = textID;
                }
                first = false;
                textID = lcleaned.split("\t")[1];
                conll = "";
                newFileGraphs = true;
            } else // We are still reading conll graphs from the same text
            {
                // We are reading a new conll graph, but from the same text
                if (line.equals("")) {
                    HashMap<Short, DependencyNode> nodes = conllReader.read(conll).getNodes();
                    HashMap<Short, DependencyNode> snodes = new HashMap<Short, DependencyNode>();
                    for (short address : nodes.keySet()) {
                        snodes.put(address, new SentimentDependencyNode(nodes.get(address), null));
                    }
                    sdgs.add(new SentimentDependencyGraph(snodes));
                    conll = "";
                } else {
                    conll = conll.concat(line + "\n");
                }
            }
            line = br.readLine();
        }
        // Last graph
        if (!sdgs.isEmpty()) {
            List<SentimentInformation> sis = sdgs.stream().map((SentimentDependencyGraph dg) -> (rba.analyse(dg, (short) 0))).collect(Collectors.toList());
            ;
            String text = String.join(" ", sdgs.stream().map((SentimentDependencyGraph dg) -> dg.subgraphToString((short) 0)).collect(Collectors.toList()));
            SentimentInformation siFinal = rba.merge(sis);
            try {
                writer.write(printOutputScaled(siFinal, scale, rba.getAc().isBinaryNeutralAsNegative()) + "\t" + "\t" + text + "\t" + " [The analysis took: " + textAnalyseTime + " seg.] [Accumulated time is: " + totalAnalyseTime + "]\n");
                // writer.write(textID+"\t"+printOutputScaled(siFinal,scale,rba.getAc().isBinaryNeutralAsNegative())+"\t"+"\t"+text+"\t"+" [The analysis took: "+textAnalyseTime+" seg.] [Accumulated time is: "+totalAnalyseTime+"]\n");
                writer.flush();
                if (verbose) {
                    sdgs.stream().forEach(sdg -> sdg.printLandscapeGraph((short) 0));
                }
            } catch (IOException e) {
                // TODO Auto-generated catch block
                e.printStackTrace();
            }
            sdgs.clear();
            textID = null;
            newFileGraphs = false;
        }
        br.close();
        writer.close();
    } catch (IOException e) {
        e.printStackTrace();
    }
}