added a script for depccg parsing

Author: kmineshima

ja/rte.py

@@ -0,0 +1,49 @@
+
+from __future__ import print_function, unicode_literals
+from lxml import etree
+from depccg import PyJaAStarParser
+import argparse
+
+def depccg2xml(tree, sid):
+    def traverse(node, spid=0):
+        id = "s{}_sp{}".format(sid, spid)
+        xml_node = etree.SubElement(res, "span")
+        xml_node.set("category", str(node.cat))
+        xml_node.set("begin", str(node.start_of_span))
+        xml_node.set("end", str(node.start_of_span+len(node)))
+        xml_node.set("id", id)
+        if node.is_leaf:
+            xml_node.set("terminal",
+                    "s{}_{}".format(sid, node.head_id))
+        else:
+            spid, childid = traverse(node.left_child, spid+1)
+            if not node.is_unary:
+                spid, tmp = traverse(node.right_child, spid+1)
+                childid += " " + tmp
+            xml_node.set("child", childid)
+            xml_node.set("rule", node.op_string)
+        return spid, id
+    res = etree.Element("ccg")
+    res.set("id", "s{}_ccg0".format(sid))
+    _, id = traverse(tree)
+    res.set("root", str(id))
+    return res
+
+
+parser = argparse.ArgumentParser("A* CCG parser")
+parser.add_argument("model", help="model directory")
+parser.add_argument("input", help="input xml")
+args = parser.parse_args()
+
+xml_root = etree.parse(args.input).getroot()
+sents = xml_root[0][0]
+
+parser = PyJaAStarParser(args.model)
+res = parser.parse_doc(
+        [[t.get("surf") for t in sent[0]] for sent in sents])
+
+for i, (sent, tree) in enumerate(zip(sents, res)):
+    sent.append(depccg2xml(tree, i))
+
+print(etree.tostring(xml_root,
+    pretty_print=True).decode("utf-8"))