Update callbacks

autophrasex/autophrase.py

@@ -81,12 +81,14 @@ def mine(self,
         callback = CallbackWrapper(callbacks=callbacks)
         callback.begin()
 
+        callback.on_read_corpus_begin()
         self.corpus_reader.read(
             corpus_files=corpus_files,
             extractor=self.extractor_wrapper,
             N=N,
             verbose=kwargs.get('verbose', True),
             logsteps=kwargs.get('logsteps', 1000))
+        callback.on_read_corpus_end()
 
         callback.on_build_quality_phrases_begin(quality_phrase_files)
         quality_phrases = load_quality_phrase_files(quality_phrase_files)

autophrasex/callbacks.py

@@ -7,6 +7,12 @@ class Callback(abc.ABC):
     def begin(self):
         pass
 
+    def on_read_corpus_begin(self):
+        pass
+
+    def on_read_corpus_end(self):
+        pass
+
     def on_build_quality_phrases_begin(self, quality_file):
         pass
 
@@ -63,6 +69,14 @@ def begin(self):
         for cb in self.callbacks:
             cb.begin()
 
+    def on_read_corpus_begin(self):
+        for cb in self.callbacks:
+            cb.on_read_corpus_begin()
+
+    def on_read_corpus_end(self):
+        for cb in self.callbacks:
+            cb.on_read_corpus_end()
+
     def on_build_quality_phrases_begin(self, quality_file):
         for cb in self.callbacks:
             cb.on_build_quality_phrases_begin(quality_file)
@@ -129,6 +143,12 @@ class LoggingCallback(Callback):
     def begin(self):
         logging.info('Starting to mining phrases...')
 
+    def on_read_corpus_begin(self):
+        logging.info('Starting to read corpus...')
+
+    def on_read_corpus_end(self):
+        logging.info('Finished to read corpus.')
+
     def on_build_quality_phrases_begin(self, quality_file):
         pass