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

[mark-todd]

Added warning to lsi model for initialising a model with no data

Fixes #3270.

[codecov]

Codecov Report

Merging #3271 (9741a81) into develop (ac3bbcd) will decrease coverage by 0.05%.
The diff coverage is 55.55%.

@@             Coverage Diff             @@
##           develop    #3271      +/-   ##
===========================================
- Coverage    81.43%   81.38%   -0.06%     
===========================================
  Files          122      122              
  Lines        21052    21107      +55     
===========================================
+ Hits         17144    17177      +33     
- Misses        3908     3930      +22     

Impacted Files	Coverage Δ
gensim/utils.py	71.29% <53.84%> (-0.69%)	⬇️
gensim/models/lsimodel.py	63.88% <60.00%> (-0.16%)	⬇️
gensim/downloader.py	79.67% <0.00%> (-1.65%)	⬇️
gensim/matutils.py	77.07% <0.00%> (-1.07%)	⬇️
gensim/models/fasttext.py	89.73% <0.00%> (-0.23%)	⬇️
gensim/test/test_keyedvectors.py	99.29% <0.00%> (+0.02%)	⬆️
gensim/models/keyedvectors.py	82.43% <0.00%> (+0.19%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update ac3bbcd...9741a81. Read the comment docs.

[piskvorky]

@mpenkov could you please check why one of the builds is failing? Thank you!

[mark-todd]

I've now corrected the file according to your suggested modifications - hope this is all ok!

[mark-todd]

Not sure why it's still saying one change requested - the import line missing issue should now be resolved

[piskvorky]

We follow PEP8 & PEP257 in code style: no blank line after docstring; blank line after import group; more than one space before comment etc.

Plus types doesn't seem to be defined in utils.py. Does this code really work, did you try it?

[piskvorky]

Looks good, thanks!

@mpenkov what's the best way to shut up our failing flake8 CI test about this?

I'd prefer to have the variable there and named – makes the code much easier to read and understand. I don't really care that the variable is unused, so what.

Other than that LGTM.

[mpenkov]

Merging. Thank you for your contribution and your patience @mark-todd !