Added new ValueError in place of assertion error for no model data provided in lsi model

gensim/models/lsimodel.py

@@ -62,6 +62,7 @@
 import logging
 import sys
 import time
+import types
 
 import numpy as np
 import scipy.linalg
@@ -482,14 +483,29 @@ def add_documents(self, corpus, chunksize=None, decay=None):
         If the distributed mode is on, each chunk is sent to a different worker/computer.
 
         """
+        def is_empty(corpus):
+            """Is the corpus (an iterable or a scipy.sparse array) empty?"""
+            if scipy.sparse.issparse(corpus):
+                return corpus.shape[1] == 0  # by convention, scipy.sparse documents are columns
+            if isinstance(corpus, types.GeneratorType):
+                return False  # don't try to guess emptiness of generators, may lose elements irretrievably
+            try:
+                first_doc = next(iter(corpus))  # list, numpy array etc
+                return False # first document exists => not empty
+            except StopIteration:
+                return True
+            except Exception:
+                return False
+
         logger.info("updating model with new documents")
 
         # get computation parameters; if not specified, use the ones from constructor
         if chunksize is None:
             chunksize = self.chunksize
         if decay is None:
             decay = self.decay
-
+        if is_empty(corpus):
+            logger.warning('LsiModel.add_documents() called but no documents provided, is this intended?')
         if not scipy.sparse.issparse(corpus):
             if not self.onepass:
                 # we are allowed multiple passes over the input => use a faster, randomized two-pass algo
@@ -590,7 +606,8 @@ def __getitem__(self, bow, scaled=False, chunksize=512):
             Latent representation of corpus in BoW format if `bow` is corpus.
 
         """
-        assert self.projection.u is not None, "decomposition not initialized yet"
+        if self.projection.u is None:
+            raise ValueError('No training data provided - LSI model not initialized yet')
 
         # if the input vector is in fact a corpus, return a transformed corpus as a result
         is_corpus, bow = utils.is_corpus(bow)