source journal's homepage_url from OpenAlex

[smierz]

This PR sources homepage_urls for journals from OpenAlex:

• querying OpenAlex stays close to the existing code in crossref with light changes
  • removed the email parameter from the requesting method as it is never used
  • always taking the value from config + loading the config only once at startup
  • because config is located in config file, I need pytest-mock to set value in tests
• simple auto-generated db migration, adding column "homepage_url" to the JournalBase object
• when querying OpenAlex, I'm only putting one of the ISSNs into the query parameters to avoid repetition
• copy-pasted the code to update a journal's homepage_url from crossref.store_papers because I'm not familiar with SQLAlchemy

[coveralls]

Pull Request Test Coverage Report for Build 7441611676

• 0 of 0 changed or added relevant lines in 0 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage decreased (-3.6%) to 71.343%

---

Totals
Change from base Build 7435789001:	-3.6%
Covered Lines:	371
Relevant Lines:	502

---

💛 - Coveralls

[smierz]

thanks for the quick review, I will look into it asap, but probably due to work not before the weekend :/

[smierz]

update:

• separated all db operations from journal_service into separate file journal_repository
• added tests for methods in journal_repository (test coverage up to 79%)
• added docstrings to all methods in openalex, journal_service and journal_repository

[smierz]

re:config

i do have a mild preference for treating the Config object as a singleton that gets instantiated when used or passed explicitly, rather than created as a module-level object - the goal is to have config be single-source and so we want to strongly discourage storing and modifying it.

hmm I think we were both going for a singleton approach, but the difference is the passing of parameters: I was going for imports (=import singleton config from config.py) while you would prefer dependency injection (pass parameters from config singleton in app.py)

Both are valid approaches for handling config propagation in applications and I'm ok with either. The only concern I have is with creating new Config objects around the application (like crossref.crossref_get does) because I do not want the config to be mutable from outside the application (= environment variables) during runtime (good luck finding this "bug" if user accidentally changes them).

Anyway, I kicked out the config variable from config.py and switched to dependency injection passing the email parameter through to openalex._openalex_get

[sneakers-the-rat]

Sry for the delay, will merge when I get to keyboard. I think ill deploy this today too

[sneakers-the-rat]

thanks for this! looks great :)

[sneakers-the-rat]

Released in v0.0.5 :) https://github.com/sneakers-the-rat/paper-feeds/releases/tag/v0.0.5 https://github.com/sneakers-the-rat/paper-feeds/tree/v0.0.5 https://pypi.org/project/paper-feeds/0.0.5/