public static void main(String[] args) {
QPTreeTransformer transformer = new QPTreeTransformer();
Treebank tb = new MemoryTreebank();
Properties props = StringUtils.argsToProperties(args);
String treeFileName = props.getProperty("treeFile");
if (treeFileName != null) {
try {
TreeReader tr =
new PennTreeReader(
new BufferedReader(new InputStreamReader(new FileInputStream(treeFileName))),
new LabeledScoredTreeFactory());
Tree t;
while ((t = tr.readTree()) != null) {
tb.add(t);
}
} catch (IOException e) {
throw new RuntimeException("File problem: " + e);
}
}
for (Tree t : tb) {
System.out.println("Original tree");
t.pennPrint();
System.out.println();
System.out.println("Tree transformed");
Tree tree = transformer.transformTree(t);
tree.pennPrint();
System.out.println();
System.out.println("----------------------------");
}
}
/**
* Load a collection of parse trees from a Reader. Each tree may optionally be encased in parens
* to allow for Penn Treebank style trees.
*
* @param r The reader to read trees from. (If you want it buffered, you should already have
* buffered it!)
* @param id An ID for where these files come from (arbitrary, but something like a filename. Can
* be <code>null</code> for none.
*/
public void load(Reader r, String id) {
try {
// could throw an IO exception?
TreeReader tr = treeReaderFactory().newTreeReader(r);
int sentIndex = 0;
for (Tree pt; (pt = tr.readTree()) != null; ) {
if (pt.label() instanceof HasIndex) { // so we can trace where this tree came from
HasIndex hi = (HasIndex) pt.label();
if (id != null) {
hi.setDocID(id);
}
hi.setSentIndex(sentIndex);
}
parseTrees.add(pt);
sentIndex++;
}
} catch (IOException e) {
System.err.println("load IO Exception: " + e);
}
}
/**
* Load a collection of parse trees from the file of given name. Each tree may optionally be
* encased in parens to allow for Penn Treebank style trees. This methods implements the <code>
* FileProcessor</code> interface.
*
* @param file file to load a tree from
*/
public void processFile(File file) {
TreeReader tr = null;
// SRL stuff
CollectionValuedMap<Integer, String> srlMap = null;
if (this.srlMap != null) {
// there must be a better way ...
String filename = file.getAbsolutePath();
for (String suffix : this.srlMap.keySet()) {
if (filename.endsWith(suffix)) {
srlMap = this.srlMap.get(suffix);
break;
}
}
if (srlMap == null) {
System.err.println("could not find SRL entries for file: " + file);
}
}
try {
// maybe print file name to stdout to get some feedback
if (PRINT_FILENAMES) {
System.err.println(file);
}
// could throw an IO exception if can't open for reading
tr =
treeReaderFactory()
.newTreeReader(
new BufferedReader(new InputStreamReader(new FileInputStream(file), encoding())));
int sentIndex = 0;
Tree pt;
while ((pt = tr.readTree()) != null) {
if (pt.label() instanceof HasIndex) { // so we can trace where this tree came from
HasIndex hi = (HasIndex) pt.label();
hi.setDocID(file.getName());
hi.setSentIndex(sentIndex);
}
if (srlMap == null) {
parseTrees.add(pt);
} else {
Collection<String> srls = srlMap.get(sentIndex);
// pt.pennPrint();
// System.err.println(srls);
parseTrees.add(pt);
if (srls.isEmpty()) {
// parseTrees.add(pt);
} else {
for (String srl : srls) {
// Tree t = pt.deepCopy();
String[] bits = srl.split("\\s+");
int verbIndex = Integer.parseInt(bits[0]);
String lemma = bits[2].split("\\.")[0];
// Tree verb = Trees.getTerminal(t, verbIndex);
Tree verb = Trees.getTerminal(pt, verbIndex);
// ((CoreLabel)verb.label()).set(SRLIDAnnotation.class, SRL_ID.REL);
((CoreLabel) verb.label()).set(CoreAnnotations.CoNLLPredicateAnnotation.class, true);
for (int i = 4; i < bits.length; i++) {
String arg = bits[i];
String[] bits1;
if (arg.indexOf("ARGM") >= 0) {
bits1 = arg.split("-");
} else {
bits1 = arg.split("-");
}
String locs = bits1[0];
String argType = bits1[1];
if (argType.equals("rel")) {
continue;
}
for (String loc : locs.split("[*,]")) {
bits1 = loc.split(":");
int term = Integer.parseInt(bits1[0]);
int height = Integer.parseInt(bits1[1]);
// Tree t1 = Trees.getPreTerminal(t, term);
Tree t1 = Trees.getPreTerminal(pt, term);
for (int j = 0; j < height; j++) {
// t1 = t1.parent(t);
t1 = t1.parent(pt);
}
Map<Integer, String> roleMap =
((CoreLabel) t1.label()).get(CoreAnnotations.CoNLLSRLAnnotation.class);
if (roleMap == null) {
roleMap = new HashMap<Integer, String>();
((CoreLabel) t1.label()).set(CoreAnnotations.CoNLLSRLAnnotation.class, roleMap);
}
roleMap.put(verbIndex, argType);
// ((CoreLabel)t1.label()).set(SRLIDAnnotation.class,
// SRL_ID.ARG);
}
}
// for (Tree t1 : t) {
// if (t1.isLeaf()) { continue; }
// CoreLabel fl = (CoreLabel)t1.label();
// if (fl.value() == null) { continue; }
// if (!fl.has(SRLIDAnnotation.class)) {
// boolean allNone = true;
// for (Tree t2 : t1) {
// SRL_ID s = ((CoreLabel)t2.label()).get(SRLIDAnnotation.class);
// if (s == SRL_ID.ARG || s == SRL_ID.REL) {
// allNone = false;
// break;
// }
// }
// if (allNone) {
// fl.set(SRLIDAnnotation.class, SRL_ID.ALL_NO);
// } else {
// fl.set(SRLIDAnnotation.class, SRL_ID.NO);
// }
// }
// }
// parseTrees.add(t);
}
}
}
sentIndex++;
}
} catch (IOException e) {
System.err.println("loadTree IO Exception: " + e + " in file " + file);
} finally {
try {
if (tr != null) {
tr.close(); // important: closes file even if error!
}
} catch (IOException e) {
// do nothin'
}
}
}
/**
* Prints out all matches of a tree pattern on each tree in the path. Usage: <br>
* <br>
* <code>
* java edu.stanford.nlp.trees.tregex.TregexPattern [[-TCwfosnu] [-filter] [-h <node-name>]]* pattern
* filepath </code>
*
* <p>Arguments:<br>
*
* <ul>
* <li><code>pattern</code>: the tree pattern which optionally names some set of nodes (i.e.,
* gives it the "handle") <code>=name</code> (for some arbitrary string "name")
* <li><code>filepath</code>: the path to files with trees. If this is a directory, there will
* be recursive descent and the pattern will be run on all files beneath the specified
* directory.
* </ul>
*
* <p>Options:<br>
* <li><code>-C</code> suppresses printing of matches, so only the number of matches is printed.
* <li><code>-w</code> causes the whole of a tree that matches to be printed.
* <li><code>-f</code> causes the filename to be printed.
* <li><code>-i <filename></code> causes the pattern to be matched to be read from <code>
* <filename></code> rather than the command line. Don't specify a pattern when this
* option is used.
* <li><code>-o</code> Specifies that each tree node can be reported only once as the root of a
* match (by default a node will be printed once for every <em>way</em> the pattern matches).
* <li><code>-s</code> causes trees to be printed all on one line (by default they are pretty
* printed).
* <li><code>-n</code> causes the number of the tree in which the match was found to be printed
* before every match.
* <li><code>-u</code> causes only the label of each matching node to be printed, not complete
* subtrees.
* <li><code>-t</code> causes only the yield (terminal words) of the selected node to be printed
* (or the yield of the whole tree, if the <code>-w</code> option is used).
* <li><code>-encoding <charset_encoding></code> option allows specification of character
* encoding of trees..
* <li><code>-h <node-handle></code> If a <code>-h</code> option is given, the root tree
* node will not be printed. Instead, for each <code>node-handle</code> specified, the node
* matched and given that handle will be printed. Multiple nodes can be printed by using the
* <code>-h</code> option multiple times on a single command line.
* <li><code>-hf <headfinder-class-name></code> use the specified {@link HeadFinder} class
* to determine headship relations.
* <li><code>-hfArg <string></code> pass a string argument in to the {@link HeadFinder}
* class's constructor. <code>-hfArg</code> can be used multiple times to pass in multiple
* arguments.
* <li><code>-trf <TreeReaderFactory-class-name></code> use the specified {@link
* TreeReaderFactory} class to read trees from files.
* <li><code>-v</code> print every tree that contains no matches of the specified pattern, but
* print no matches to the pattern.
* <li><code>-x</code> Instead of the matched subtree, print the matched subtree's identifying
* number as defined in <tt>tgrep2</tt>:a unique identifier for the subtree and is in the form
* s:n, where s is an integer specifying the sentence number in the corpus (starting with 1),
* and n is an integer giving the order in which the node is encountered in a depth-first
* search starting with 1 at top node in the sentence tree.
* <li><code>-extract <code> <tree-file></code> extracts the subtree s:n specified by
* <tt>code</tt> from the specified <tt>tree-file</tt>. Overrides all other behavior of
* tregex. Can't specify multiple encodings etc. yet.
* <li><code>-extractFile <code-file> <tree-file></code> extracts every subtree
* specified by the subtree codes in <tt>code-file</tt>, which must appear exactly one per
* line, from the specified <tt>tree-file</tt>. Overrides all other behavior of tregex. Can't
* specify multiple encodings etc. yet.
* <li><code>-filter</code> causes this to act as a filter, reading tree input from stdin
* <li><code>-T</code> causes all trees to be printed as processed (for debugging purposes).
* Otherwise only matching nodes are printed.
* <li><code>-macros <filename></code> filename with macro substitutions to use. file with
* tab separated lines original-tab-replacement
* </ul>
*/
public static void main(String[] args) throws IOException {
Timing.startTime();
StringBuilder treePrintFormats = new StringBuilder();
String printNonMatchingTreesOption = "-v";
String subtreeCodeOption = "-x";
String extractSubtreesOption = "-extract";
String extractSubtreesFileOption = "-extractFile";
String inputFileOption = "-i";
String headFinderOption = "-hf";
String headFinderArgOption = "-hfArg";
String trfOption = "-trf";
String headFinderClassName = null;
String[] headFinderArgs = StringUtils.EMPTY_STRING_ARRAY;
String treeReaderFactoryClassName = null;
String printHandleOption = "-h";
String markHandleOption = "-k";
String encodingOption = "-encoding";
String encoding = "UTF-8";
String macroOption = "-macros";
String macroFilename = "";
String yieldOnly = "-t";
String printAllTrees = "-T";
String quietMode = "-C";
String wholeTreeMode = "-w";
String filenameOption = "-f";
String oneMatchPerRootNodeMode = "-o";
String reportTreeNumbers = "-n";
String rootLabelOnly = "-u";
String oneLine = "-s";
Map<String, Integer> flagMap = Generics.newHashMap();
flagMap.put(extractSubtreesOption, 2);
flagMap.put(extractSubtreesFileOption, 2);
flagMap.put(subtreeCodeOption, 0);
flagMap.put(printNonMatchingTreesOption, 0);
flagMap.put(encodingOption, 1);
flagMap.put(inputFileOption, 1);
flagMap.put(printHandleOption, 1);
flagMap.put(markHandleOption, 2);
flagMap.put(headFinderOption, 1);
flagMap.put(headFinderArgOption, 1);
flagMap.put(trfOption, 1);
flagMap.put(macroOption, 1);
flagMap.put(yieldOnly, 0);
flagMap.put(quietMode, 0);
flagMap.put(wholeTreeMode, 0);
flagMap.put(printAllTrees, 0);
flagMap.put(filenameOption, 0);
flagMap.put(oneMatchPerRootNodeMode, 0);
flagMap.put(reportTreeNumbers, 0);
flagMap.put(rootLabelOnly, 0);
flagMap.put(oneLine, 0);
Map<String, String[]> argsMap = StringUtils.argsToMap(args, flagMap);
args = argsMap.get(null);
if (argsMap.containsKey(encodingOption)) {
encoding = argsMap.get(encodingOption)[0];
System.err.println("Encoding set to " + encoding);
}
PrintWriter errPW = new PrintWriter(new OutputStreamWriter(System.err, encoding), true);
if (argsMap.containsKey(extractSubtreesOption)) {
List<String> subTreeStrings =
Collections.singletonList(argsMap.get(extractSubtreesOption)[0]);
extractSubtrees(subTreeStrings, argsMap.get(extractSubtreesOption)[1]);
return;
}
if (argsMap.containsKey(extractSubtreesFileOption)) {
List<String> subTreeStrings =
Arrays.asList(
IOUtils.slurpFile(argsMap.get(extractSubtreesFileOption)[0]).split("\n|\r|\n\r"));
extractSubtrees(subTreeStrings, argsMap.get(extractSubtreesFileOption)[0]);
return;
}
if (args.length < 1) {
errPW.println(
"Usage: java edu.stanford.nlp.trees.tregex.TregexPattern [-T] [-C] [-w] [-f] [-o] [-n] [-s] [-filter] [-hf class] [-trf class] [-h handle]* pattern [filepath]");
return;
}
String matchString = args[0];
if (argsMap.containsKey(macroOption)) {
macroFilename = argsMap.get(macroOption)[0];
}
if (argsMap.containsKey(headFinderOption)) {
headFinderClassName = argsMap.get(headFinderOption)[0];
errPW.println("Using head finder " + headFinderClassName + "...");
}
if (argsMap.containsKey(headFinderArgOption)) {
headFinderArgs = argsMap.get(headFinderArgOption);
}
if (argsMap.containsKey(trfOption)) {
treeReaderFactoryClassName = argsMap.get(trfOption)[0];
errPW.println("Using tree reader factory " + treeReaderFactoryClassName + "...");
}
if (argsMap.containsKey(printAllTrees)) {
TRegexTreeVisitor.printTree = true;
}
if (argsMap.containsKey(inputFileOption)) {
String inputFile = argsMap.get(inputFileOption)[0];
matchString = IOUtils.slurpFile(inputFile, encoding);
String[] newArgs = new String[args.length + 1];
System.arraycopy(args, 0, newArgs, 1, args.length);
args = newArgs;
}
if (argsMap.containsKey(quietMode)) {
TRegexTreeVisitor.printMatches = false;
TRegexTreeVisitor.printNumMatchesToStdOut = true;
}
if (argsMap.containsKey(printNonMatchingTreesOption)) {
TRegexTreeVisitor.printNonMatchingTrees = true;
}
if (argsMap.containsKey(subtreeCodeOption)) {
TRegexTreeVisitor.printSubtreeCode = true;
TRegexTreeVisitor.printMatches = false;
}
if (argsMap.containsKey(wholeTreeMode)) {
TRegexTreeVisitor.printWholeTree = true;
}
if (argsMap.containsKey(filenameOption)) {
TRegexTreeVisitor.printFilename = true;
}
if (argsMap.containsKey(oneMatchPerRootNodeMode)) TRegexTreeVisitor.oneMatchPerRootNode = true;
if (argsMap.containsKey(reportTreeNumbers)) TRegexTreeVisitor.reportTreeNumbers = true;
if (argsMap.containsKey(rootLabelOnly)) {
treePrintFormats.append(TreePrint.rootLabelOnlyFormat).append(',');
} else if (argsMap.containsKey(oneLine)) { // display short form
treePrintFormats.append("oneline,");
} else if (argsMap.containsKey(yieldOnly)) {
treePrintFormats.append("words,");
} else {
treePrintFormats.append("penn,");
}
HeadFinder hf = new CollinsHeadFinder();
if (headFinderClassName != null) {
Class[] hfArgClasses = new Class[headFinderArgs.length];
for (int i = 0; i < hfArgClasses.length; i++) hfArgClasses[i] = String.class;
try {
hf =
(HeadFinder)
Class.forName(headFinderClassName)
.getConstructor(hfArgClasses)
.newInstance(
(Object[])
headFinderArgs); // cast to Object[] necessary to avoid varargs-related
// warning.
} catch (Exception e) {
throw new RuntimeException("Error occurred while constructing HeadFinder: " + e);
}
}
TRegexTreeVisitor.tp =
new TreePrint(treePrintFormats.toString(), new PennTreebankLanguagePack());
try {
// TreePattern p = TreePattern.compile("/^S/ > S=dt $++ '' $-- ``");
TregexPatternCompiler tpc = new TregexPatternCompiler(hf);
Macros.addAllMacros(tpc, macroFilename, encoding);
TregexPattern p = tpc.compile(matchString);
errPW.println("Pattern string:\n" + p.pattern());
errPW.println("Parsed representation:");
p.prettyPrint(errPW);
String[] handles = argsMap.get(printHandleOption);
if (argsMap.containsKey("-filter")) {
TreeReaderFactory trf = getTreeReaderFactory(treeReaderFactoryClassName);
treebank =
new MemoryTreebank(
trf, encoding); // has to be in memory since we're not storing it on disk
// read from stdin
Reader reader = new BufferedReader(new InputStreamReader(System.in, encoding));
((MemoryTreebank) treebank).load(reader);
reader.close();
} else if (args.length == 1) {
errPW.println("using default tree");
TreeReader r =
new PennTreeReader(
new StringReader(
"(VP (VP (VBZ Try) (NP (NP (DT this) (NN wine)) (CC and) (NP (DT these) (NNS snails)))) (PUNCT .))"),
new LabeledScoredTreeFactory(new StringLabelFactory()));
Tree t = r.readTree();
treebank = new MemoryTreebank();
treebank.add(t);
} else {
int last = args.length - 1;
errPW.println("Reading trees from file(s) " + args[last]);
TreeReaderFactory trf = getTreeReaderFactory(treeReaderFactoryClassName);
treebank = new DiskTreebank(trf, encoding);
treebank.loadPath(args[last], null, true);
}
TRegexTreeVisitor vis = new TRegexTreeVisitor(p, handles, encoding);
treebank.apply(vis);
Timing.endTime();
if (TRegexTreeVisitor.printMatches) {
errPW.println("There were " + vis.numMatches() + " matches in total.");
}
if (TRegexTreeVisitor.printNumMatchesToStdOut) {
System.out.println(vis.numMatches());
}
} catch (IOException e) {
e.printStackTrace();
} catch (TregexParseException e) {
errPW.println("Error parsing expression: " + args[0]);
errPW.println("Parse exception: " + e.toString());
}
}