fixes:

- dependency processor now only runs in dep_to_4lang - text_to_4lang.process also converts old-style deps to new ones
kornai · Oct 7, 2015 · 222530bf760b70d31b300ef147a5e1d1a72f60d4 · 222530b
1 parent cd6c569
commit 222530bf760b70d31b300ef147a5e1d1a72f60d4
Unified Split

Showing with 35 additions and 6 deletions.

+1 −0 4lang.primitive

+15 −0 conf/eksz_firsts.cfg

+15 −0 conf/nszt_firsts.cfg

+0 −6 src/fourlang/dict_to_4lang.py

+4 −0 src/fourlang/text_to_4lang.py
diff --git a/4lang.primitive b/4lang.primitive
@@ -0,0 +1 @@
+  
+have
diff --git a/conf/eksz_firsts.cfg b/conf/eksz_firsts.cfg
@@ -0,0 +1,15 @@
+  
+[dict]
+  
+input_type = eksz
+  
+input_file = input/eksz.xml
+  
+output_file = data/dict/eksz_firsts.json
+  
+
+  
+[machine]
+  
+ext_definitions = data/machines/eksz_firsts.pickle
+  
+graph_dir = data/graphs/eksz_firsts
+  
+
+  
+[deps]
+  
+lang = hu
+  
+dep_map = dep_to_4lang_hu.txt
+  
+
+  
+[magyarlanc]
+  
+path = magyarlanc/magyarlanc-2.0.jar
diff --git a/conf/nszt_firsts.cfg b/conf/nszt_firsts.cfg
@@ -0,0 +1,15 @@
+  
+[dict]
+  
+input_type = nszt
+  
+input_file = input/nsztb.xml
+  
+output_file = data/dict/nszt_firsts.json
+  
+
+  
+[machine]
+  
+ext_definitions = data/machines/nszt_firsts.pickle
+  
+graph_dir = data/graphs/nszt_firsts
+  
+
+  
+[deps]
+  
+lang = hu
+  
+dep_map = dep_to_4lang_hu.txt
+  
+
+  
+[magyarlanc]
+  
+path = magyarlanc/magyarlanc-2.0.jar
diff --git a/src/fourlang/dict_to_4lang.py b/src/fourlang/dict_to_4lang.py
@@ -9,7 +9,6 @@
 import traceback
 from dep_to_4lang import DepTo4lang
-  
-from dependency_processor import DependencyProcessor
 from entry_preprocessor import EntryPreprocessor
 from lexicon import Lexicon
 from longman_parser import LongmanParser
@@ -93,8 +92,6 @@ def process_entries(self, words):
        else:
            print 'incorrect lang'
-  
-        dependency_processor = DependencyProcessor(self.cfg)
-  
-
        for entry in entries:
            if entry['to_filter']:
                continue
@@ -103,9 +100,6 @@ def process_entries(self, words):
                definition = sense['definition']
                if definition is None:
                    continue
-  
-                # print 'printing deps' + str(definition['deps'])
-  
-                definition['deps'] = dependency_processor.process_dependencies(
-  
-                    definition['deps'])
            if word in self.dictionary:
                logging.warning(

diff --git a/src/fourlang/text_to_4lang.py b/src/fourlang/text_to_4lang.py
@@ -18,6 +18,7 @@ class TextTo4lang():
    def __init__(self, cfg):
        self.cfg = cfg
+  
+        self.lang = self.cfg.get("deps", "lang")
        self.deps_dir = self.cfg.get('data', 'deps_dir')
        ensure_dir(self.deps_dir)
        self.corenlp_wrapper = CoreNLPWrapper(self.cfg)
@@ -58,6 +59,9 @@ def process(self, text, dep_dir=None, fn=None):
        logging.getLogger().setLevel(__MACHINE_LOGLEVEL__)
        # logging.info("processing sentences...")
+  
+        if self.lang == 'en':
+  
+            parsed_sens = map(
+  
+                self.dep_to_4lang.convert_old_deps, parsed_sens)
        words_to_machines = self.dep_to_4lang.get_machines_from_deps_and_corefs(  # nopep8
            parsed_sens, corefs)