Error attempting to use Stanza engine

[kmatt]

Attempting to test the Stanza engine using as a guide: https://microsoft.github.io/presidio/analyzer/languages/

Creating the engine fails with a Wrong NLP engine configuration error. Using spaCy succeeds with the same approach (2nd code example below).

Is the Stanza config below incorrect?

In [1]: import stanza
   ...: stanza.download("en")
Downloading https://raw.githubusercontent.com/stanfordnlp/stanza-resources/master/resources_1.2.0.json: 128kB [00:00, 29.2MB/s]
2021-06-02 12:36:48 INFO: Downloading default packages for language: en (English)...
2021-06-02 12:36:55 INFO: Finished downloading models and saved to /home/kmatt/stanza_resources.

In [2]: from presidio_analyzer import AnalyzerEngine
   ...: from presidio_anonymizer import AnonymizerEngine
   ...: from presidio_analyzer.nlp_engine import NlpEngineProvider

In [3]:     configuration = {
   ...:         "nlp_engine_name": "stanza",
   ...:         "models": [{"lang_code": "en", "model_name": "en"}]
   ...:     }
   ...:

In [4]:     provider = NlpEngineProvider(nlp_configuration=configuration)

In [5]:     nlp_engine = provider.create_engine()
---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
~/.local/lib/python3.9/site-packages/presidio_analyzer/nlp_engine/nlp_engine_provider.py in create_engine(self)
     75             nlp_engine_name = self.nlp_configuration["nlp_engine_name"]
---> 76             nlp_engine_class = self.nlp_engines[nlp_engine_name]
     77             nlp_engine_opts = {

KeyError: 'stanza'

In [7]: configuration = {
   ...:     "nlp_engine_name": "spacy",
   ...:     "models": [{"lang_code": "en", "model_name": "en_core_web_lg"}]
   ...:     }

In [8]: provider = NlpEngineProvider(nlp_configuration=configuration)

In [9]: nlp_engine = provider.create_engine()

[SharonHart]

Hi @kmatt!
From your code snippet, it seems like you've installed stanza manually, have you done the same with spacy_stanza? pipenv install spacy_stanza --skip-lock

[kmatt]

I did, using pip instead of pipenv as this is a development VM. Ran pip install -U spacy_stanza to double check and same error persists as above.

[kmatt]

Is my configuration{} above correct for Stanza?

[SharonHart]

It is correct, I've reproduced the issue using your code snippet and resolved it by installing stanza and spacy_stanza. Try to install and run the script from pipenv or virtualenv environments to eliminate all dependency issues.
Additionally, try to see which line of code is raising a 'KeyError' on nlp_engine_provider.py:88

[kmatt]

Success in a clean venv: #726 (comment)

[SharonHart]

Great! 🥳

[kmatt]

With a clean virtualenv, and installing only:

1. presidio_analyzer
2. presidio_anonymizer
3. stanza
4. spacy_stanza

import stanza
stanza.download("en")

from presidio_analyzer import AnalyzerEngine
from presidio_anonymizer import AnonymizerEngine
from presidio_analyzer.nlp_engine import NlpEngineProvider

configuration = {
    "nlp_engine_name": "stanza",
    "models": [{"lang_code": "en", "model_name": "en"}]
}

provider = NlpEngineProvider(nlp_configuration=configuration)
nlp_engine = provider.create_engine()

print(nlp_engine)

Downloading https://raw.githubusercontent.com/stanfordnlp/stanza-resources/main/resources_1.2.1.json: 138kB [00:00, 29.4MB/s]
2021-06-09 13:57:30 INFO: Downloading default packages for language: en (English)...
2021-06-09 13:57:31 INFO: File exists: /home/keranem1/stanza_resources/en/default.zip.
2021-06-09 13:57:33 INFO: Finished downloading models and saved to /home/keranem1/stanza_resources.
2021-06-09 13:57:34 INFO: Loading these models for language: en (English):
=========================
| Processor | Package   |
-------------------------
| tokenize  | combined  |
| pos       | combined  |
| lemma     | combined  |
| ner       | ontonotes |
=========================

2021-06-09 13:57:34 INFO: Use device: cpu
2021-06-09 13:57:34 INFO: Loading: tokenize
2021-06-09 13:57:34 INFO: Loading: pos
2021-06-09 13:57:34 INFO: Loading: lemma
2021-06-09 13:57:34 INFO: Loading: ner
2021-06-09 13:57:34 INFO: Done loading processors!
<presidio_analyzer.nlp_engine.stanza_nlp_engine.StanzaNlpEngine object at 0x7f695f9fb220>

[omri374]

Great to see this was sorted out. We updated the spacy stanza code on this PR #711 but it was only released to PyPI today.

[Momilijaz96]

totally works!