Allow mirrors to be defined in the index

goodmami commented 2 years ago

Several wordnet projects have changes that will require some new features in the index.

The OMW English Wordnet was misnamed the Princeton WordNet previously. The old and new data should be renamed, along with the identifiers and versions (e.g., pwn:3.0 -> wn30:1.4+omw). When someone uses the old identifier, a warning should be issued as it redirects to the new resource.
It might also be useful to have a general mechanism for displaying messages when someone downloads a wordnet. One use for this is to recommend users of the upcoming OWN-PT wordnet to also get the OWN-EN wordnet to go along with it.
There have been some reliability issues with servers hosting wordnets (see #141), so it would be good to be able to specify a mirror in case one server times out.

These changes might look like this:

[pwn]
  warn = "This project has been renamed. See https://... for more information."
  [pwn.versions."3.0"]
    redirect = "wn30:1.4+omw"
  [pwn.versions."3.1"]
    redirect = "wn31:1.4+omw"

[wn30]
  label = "OMW English Wordnet 3.0"
  [wn30.versions."1.4+omw"]
    url = "https://..."

[wn31]
  label = "OMW English Wordnet 3.1"
  [wn31.versions."1.4+omw"]
    url = "https://..."

[ewn]
  label = "Estonian Wordnet"
  # ...
  [ewn.versions.2020]
    warn = "The Open English WordNet is now under the identifier 'oewn'. See https://... for more information."
    redirect = "oewn:2020"

[oewn]
  label = "Open English WordNet"
  language = "en"
  license = "https://creativecommons.org/licenses/by/4.0/"
  [oewn.versions.2020]
    urls = [
      "https://en-word.net/static/english-wordnet-2020.xml.gz",
      "https://...",
    ]
  [oewn.versions.2019]
    urls = [
      "https://en-word.net/static/english-wordnet-2019.xml.gz",
      "https://...",
    ]

goodmami commented 2 years ago

Trying to pin down the behavior, I think that having the redirect be automatic might actually be problematic, because if it happens silently, the user won't be aware that they should use a different identifier in order to use the lexicon. e.g:

>>> wn.download('pwn:3.0')  # really gets wn30:1.4
>>> pwn30 = wn.Wordnet('pwn:3.0')  # error

Some solutions:

Use the redirect to get the proper ID when creating a wn.Wordnet object. This will again issue any warning for the old project ID, but won't necessarily tell the user what the new name is. E.g.,
```
>>> pwn30 = wn.Wordnet('pwn:3.0')  # issues warning about pwn:3.0
>>> pwn30.lexicons()[0].specifier()  # user is unaware of the new ID unless they look for it
'wn30:1.4'
```
Raise an error for the redirect, instead, mentioning the new ID and version. This is initially disruptive, but then the user is made aware of what they should be doing. E.g.,
```
>>> pwn30 = wn.Wordnet('pwn:3.0')
Traceback (most recent call last):
...
wn.Error: lexicon specifier 'pwn:3.0' has been deprecated; use 'wn30:1.4' instead
```

goodmami commented 2 years ago

Another alternative, specify the error message in the index and don't bother with warnings or redirects. E.g.:

[pwn]
  error = "Instead of 'pwn:3.0' and 'pwn:3.1', please use 'wn30' and 'wn31'."

Then...


>>> wn.download('pwn:3.0')
Traceback (most recent call last):
  ...
wn.Error: Instead of 'pwn:3.0' and 'pwn:3.1', please use 'wn30' and 'wn31'.

fcbond commented 2 years ago

This looks reasonable.

On Fri, Oct 22, 2021 at 4:19 AM Michael Wayne Goodman < @.***> wrote:

Another alternative, specify the error message in the index and don't bother with warnings or redirects. E.g.:

[pwn] error = "Instead of 'pwn:3.0' and 'pwn:3.1', please use 'wn30' and 'wn31'."

Then...

wn.download('pwn:3.0') Traceback (most recent call last): ... wn.Error: Instead of 'pwn:3.0' and 'pwn:3.1', please use 'wn30' and 'wn31'.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/goodmami/wn/issues/142#issuecomment-948972456, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAIPZRVJ2CGJQLW3EZMP22LUIBYTVANCNFSM5FK22LOA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

-- Francis Bond http://www3.ntu.edu.sg/home/fcbond/ Division of Linguistics and Multilingual Studies Nanyang Technological University

goodmami commented 2 years ago

In that case #146 is for adding error and this issue is about adding mirrors.

I think I might just extend the current url key to accept a space-separated list of URLs:

[oewn]
  label = "Open English WordNet"
  language = "en"
  license = "https://creativecommons.org/licenses/by/4.0/"
  [oewn.versions.2020]
    url = """
      https://en-word.net/static/english-wordnet-2020.xml.gz
      https://...
    """

This way there's no change for those with only one URL.

fcbond commented 2 years ago

That seems like a very good idea.

On Sat, Oct 23, 2021 at 6:41 AM Michael Wayne Goodman < @.***> wrote:

In that case #146 https://github.com/goodmami/wn/issues/146 is for adding error and this issue is about adding mirrors.

I think I might just extend the current url key to accept a space-separated list of URLs:

[oewn] label = "Open English WordNet" language = "en" license = "https://creativecommons.org/licenses/by/4.0/" [oewn.versions.2020] url = """ https://en-word.net/static/english-wordnet-2020.xml.gz https://... """

This way there's no change for those with only one URL.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/goodmami/wn/issues/142#issuecomment-949994953, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAIPZRWREBAOMWAO3F5PU5TUIHQ5TANCNFSM5FK22LOA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

-- Francis Bond http://www3.ntu.edu.sg/home/fcbond/ Division of Linguistics and Multilingual Studies Nanyang Technological University

goodmami / wn

Allow mirrors to be defined in the index #142