Feature/custom exc handler: Support optional custom exception handling in Table.iter() + Table.read()

[hjoukl]

Support custom exception handling through optional Table.iter() / Table.read() argument.

Implements: #258

This pull requests aims to add an exc_handler option to Table.iter() + Table.read():

def iter(self, keyed=False, extended=False, cast=True,                   
         integrity=False, relations=False,
         foreign_keys_values=False, exc_handler=None):

Current behaviour will remain unchanged if this exc_handler option is not used.

The pull request contains this change plus some "support infrastructure":

• add this option to Table.iter() + Table.read() and Schema.cast_row()
• optional row_number argument to Schema.cast_row for better error messages
• specialized Exception classes for Foreign key and unique key errors
• test data files for checking proper stream closing

I'm happy to discuss any changes that would be necessary to make this acceptable :-).

Some notes:

• iter() should really use a context manager to open the stream; I've deliberately not included this in the PR to not create a huge diff caused by the indentation change
• at some places I wasn't so sure about formatting; while your contribution guidelines indicate 79 characters per line this isn't always the case in the current code. So I apologize if I might have changed one or the other source line to 79 chars max that isn't directly related to this PR.
• the exception specializations for FK and unique key errors come in handy for nicer error reporting and better distinguishable errors imho
• FailedCast is a means to actually yield rows/values but still be able to realize something's gone wrong with a data cell and that the data couldn't be successfully casted.

---

Please preserve this line to notify @roll (lead of this repository)

[roll]

@hjoukl I'm a little bit busy at the moment but I will have reviewed it in a week or so.

Thanks for your input!

[hjoukl]

I'm a little bit busy at the moment but I will have reviewed it in a week or so.

Great! I haven't yet added any documentation - which I'd of course do if this looks like an acceptable addition, along with anything else that needs fixing in the PR.

[roll]

iter() should really use a context manager to open the stream; I've deliberately not included this in the PR to not create a huge diff caused by the indentation change

Probably it's because somewhere we had to use it outside a context manager, like returning to the client. If it's possible we're, of course, better to wrap it into one

[hjoukl]

iter() should really use a context manager to open the stream; I've deliberately not included this in the PR to not create a huge diff caused by the indentation change

Probably it's because somewhere we had to use it outside a context manager, like returning to the client. If it's possible we're, of course, better to wrap it into one

I actually had it running and passing all unit tests with simply using the existing context manager support. I removed that to keep the diff smaller, so this could be changed later on.

[hjoukl]

So, I finally got round to adding documentation, sorry for the delay.

For custom exception handlers, I also slightly changed the schema.py error reporting for casting errors (error_data arg to custom exception handlers) to only include the uncastable fields instead of the full row, which is available anyway in the row_data arg. Tests reflect this change.

Some additional test_table.py tests to test the behavioural differences when using cast=True vs cast=False for iter()/read() have been added.

Please let me know if anything's missing or should be changed to further advance this!

Btw. not being a proper git hero I got pretty much stumped by your commit bringing some master changes into the feature/custom-exc-handler branch. I just couldn't push afterwards any more and ended up cherry-picking my latest changes into a fresh local clone. I suspect I would have needed some "git fetch" or "git pull --rebase" from my github server-side fork, for the branch? I did actually do git fetch upstream and git merge upstream master.

Greetings, Holger

[hjoukl]

Great :-) - I'll go ahead and close #258, then.

Thanks for providing us with tableschema-py!