Psv

BUG FIXES


1. `delimiter` argument no longer not being passed in loading functions


API CHANGES

Addition: `safe_load` has been implemented, allow safe loading of more dirty csvs
Addition: ''opencsv' has been implemented, which decides to use safe_load or regular load depending on the data.
Behavior Change: `tabulate` library version bumped to 0.8.3
Addition: Added `index()` method to rows, equivalent to `dict(row["Column 1", "Column 2"])`
Addition: Added 'fast_find`, which allows much quicker finding of a single row. Uses .index()


Behavior Change: `threading`'s `RLock` have been added to: `addcolumn`, `delcolumn`, `non_hash_merge`, and selections processing.


Minor details
- Testing no longer uses hypothesis, was transferred over to custom written csv generator.
- Updated Copyright notice

BUG FIXES

1. The `non_hash_merge` method now correctly returns the results instead of just returning every row.
2. The `remove_duplicated` selection method now doesn't bug the `MainSelection` object if soft argument was `False`
2 `psv.column_names` now has proper error handeling for empty files.
3. `psv.loads` now correctly handles dictionaries, before it would bug the columns


API CHANGES

Removed: The `Selection.columns` setter has been removed, since it was actually not used internally. Outside access of the variable is not supported, use the addcolumn and delcolumn methods if you need to edit columns.
Removed: `Selection.fast_add` was just duplicate code to it's `__add__`, and has been removed.

Change: `BaseRow` renamed to `Row`. For backwards compability reasons a variable called BaseRow exists in the same namespace to allow code referencing it to still work.
Change: All psv csv loading functions no longer accept the `outputfile` argument, it was unused inside the library.

Addition: `addcolumn` method now supports passing a function that will give it the value to set the newly created column for that row.
Addition: `delcolumn` method added to allow deleting columns in a supported and safe way.
Addition: `update_values` added to `Row`. Built to allow a safe way to use .update() like method on `Row` instances.
Addition: Added a find family of methods that use the any() function generation code instead of the select method's

Behavior Change: `outputtedrows` and `nonoutputtedrows` now use a safe_select(), due to the nature of those functions being a current snapshot and not lazy-loadable.
Behavior Change: `Selection.output()` now accepts an opened file or io object instead of just a string location of the file. Note: Closes the file within the method.


Minor details
- Improvements to comments
- Updated Copyright notice
- .checkignore removed from the repo, the service using it was shutdown

First (and beta) Release of PSV to pypi distribution.

Further releases will include changes from last version