josephaxisa / henry Goto Github PK

This is a great example of bloat in a project that can be improved.
While there are lists of unused joins in explores, I don't currently see views that are not used in any explores.

Given a particular explore, find other explores in the instance that are very similar (in terms of joins and fields exposed).

Python 2.7.15, pip 18.0, mac os 10.13.4

MacBook-Pro-2:~ looker$ pip install henry
Collecting henry
  Could not find a version that satisfies the requirement henry (from versions: )
No matching distribution found for henry

When 'henry pulse' is run, I get the following error message. To me it seems that the results list empty, although when I check the client id and client secret manually ( https://:19999/api-docs/index.html) , the connection works fine. I hope you have any suggestions.

RUNNING (1/5) Testing Connections: 0% | |[00:00<?]
Traceback (most recent call last):
File "/usr/local/bin/henry", line 11, in
load_entry_point('henry==0.0.1b2', 'console_scripts', 'henry')()
File "/usr/local/lib/python3.7/site-packages/henry-0.0.1b2-py3.7.egg/henry/cli.py", line 273, in main
result = pulse.run_all()
File "/usr/local/lib/python3.7/site-packages/henry-0.0.1b2-py3.7.egg/henry/commands/pulse.py", line 28, in run_all
result = self.check_connections()
File "/usr/local/lib/python3.7/site-packages/henry-0.0.1b2-py3.7.egg/henry/commands/pulse.py", line 102, in check_connections
for i in results:
TypeError: 'NoneType' object is not iterable

-- output switch allows export to csv rather thanb just txt

Currently view_count is the number of view files in a particular project. If a view file contains multiple view definitions then the view_count becomes inaccurate.

According to the documentation, the config.yml file should be as below.

hosts:
host_alias:
access_token: ''
host: foo.bar.companyname.com
id: AbCdEfGhIjKlMnOp
secret: QrStUvWxYz1234567890

However, auth.py, references the config file using the name 'kwargs', where the value of host is assigned the value 'looker' in cli.py, as such, throwing an error.

host = params['hosts'][kwargs['host']]['host']

I'm surprised no one else has encountered this issue so thought I'd bring it up.

I fixed the issue by finding and replacing all instances of '[kwargs['host']]' with '[host_alias]'.

Second issue I found was maybe because I'm running on a windows machine. In cli.py, line 29 was giving me issues because of file path backslashes, so had to use encode, decode for logging.config to be able to read this correctly.

LOGGING_LOG_PATH = os.path.join(LOGGING_LOG_PATH,'henry.log').encode('unicode_escape').decode()

Thanks,

Longfei

As users of a large looker instance with millions of queries a month, Henry's default API timeout of 60 seconds means no diagnostic query (besides connection tests) return.

Allowing users to manually fiddle with the timeout window (outside of a development install) would be very useful.

Big fan of this project - thanks.

The default behaviour in henry is to create the config.yml file in the current working directory. This could be confusing for some users. It would be great to create a config.yml file in the default install folder and then allow users to override this either globally or on a per-use basis.