160 is quite old, but i wanted to bring it up again because I just ran up against it.
I'm hacking around with a bibtex-to-sqlite converter to ingest mendeley- and zotero- managed bibtex files of collections of citations, and emit a database from it. Mendeley, in the great wisdom common to all the works of Elsevier, has injected fields in the bibtex entries whose names are things like mendeley-tags. This is guarded against by dataset at the moment, so we get:
File "/home/gvoysey/.cache/pypoetry/virtualenvs/bibtex-to-sqlite-uwnieG41-py3.8/lib/python3.8/site-packages/dataset/util.py", line 75, in normalize_column_name
raise ValueError("%r is not a valid column name." % name)
ValueError: 'mendeley-tags' is not a valid column name.
I'd like to be able to pass a no_i_really_mean_it_ruin_my_schema_give_me_the_footgun flag in this case to allow column IDs to be created that will subsequently be required to be quoted in SELECTs etc. I'm not sure where i'd want to pass it, but possibly in .create_table().
In my limited case, permitting weird column names is fine w/r/t dataset's table.find() method because the database i'm making will never see python again, but I understand the reluctance to expose the footgun given in #160. Still, for parity with the original data, I think it's worthwhile to have it passable. Right now i'm doing the extremely hacky thing of:
val = mendeley_entry.pop('mendeley-tags')
mendeley_entry['mendeleytags'] = val
160 is quite old, but i wanted to bring it up again because I just ran up against it.
I'm hacking around with a bibtex-to-sqlite converter to ingest mendeley- and zotero- managed bibtex files of collections of citations, and emit a database from it. Mendeley, in the great wisdom common to all the works of Elsevier, has injected fields in the bibtex entries whose names are things like
mendeley-tags. This is guarded against by dataset at the moment, so we get:I'd like to be able to pass a
no_i_really_mean_it_ruin_my_schema_give_me_the_footgunflag in this case to allow column IDs to be created that will subsequently be required to be quoted in SELECTs etc. I'm not sure where i'd want to pass it, but possibly in.create_table().In my limited case, permitting weird column names is fine w/r/t dataset's
table.find()method because the database i'm making will never see python again, but I understand the reluctance to expose the footgun given in #160. Still, for parity with the original data, I think it's worthwhile to have it passable. Right now i'm doing the extremely hacky thing of:and that's a little distasteful.