feat!: add new nlp subcommand, require it for any NLP tasks

Right now, there isn't much difference between the "etl" and "nlp"
subcommands, except that:
- all FHIR resource type tasks must use "etl"
- all NLP tasks must use "nlp"
- completion tracking isn't used for "nlp"

In the future, we'll add more NLP-specific options to the "nlp"
subcommand, like document selection args and eventually, some sort
of dynamic NLP config options. But for right now, this is just a
refactor and splitting out those two subcommands.

Author: mikix

cumulus_etl/cli.py

@@ -10,7 +10,7 @@
 import rich.logging
 
 from cumulus_etl import common, etl, export, inliner, upload_notes
-from cumulus_etl.etl import convert, init
+from cumulus_etl.etl import convert, init, nlp
 
 
 class Command(enum.Enum):
@@ -21,6 +21,7 @@ class Command(enum.Enum):
     EXPORT = "export"
     INIT = "init"
     INLINE = "inline"
+    NLP = "nlp"
     UPLOAD_NOTES = "upload-notes"
 
     # Why isn't this part of Enum directly...?
@@ -72,13 +73,15 @@ async def main(argv: list[str]) -> None:
         run_method = init.run_init
     elif subcommand == Command.INLINE.value:
         run_method = inliner.run_inline
+    elif subcommand == Command.NLP.value:
+        run_method = nlp.run_nlp
     else:
         parser.description = "Extract, transform, and load FHIR data."
         if not subcommand:
             # Add a note about other subcommands we offer, and tell argparse not to wrap our formatting
             parser.formatter_class = argparse.RawDescriptionHelpFormatter
             parser.description += "\n\nother commands available:\n"
-            parser.description += "  convert\n  export\n  init\n  inline\n  upload-notes"
+            parser.description += "  convert\n  export\n  init\n  inline\n  nlp\n  upload-notes"
         run_method = etl.run_etl
 
     with tempfile.TemporaryDirectory() as tempdir:

cumulus_etl/cli_utils.py

@@ -117,7 +117,7 @@ def add_task_selection(parser: argparse.ArgumentParser):
     task.add_argument(
         "--task-filter",
         action="append",
-        choices=["covid_symptom", "cpu", "gpu"],
+        choices=["covid_symptom", "irae", "cpu", "gpu"],
         help="restrict tasks to only the given sets (comma separated)",
     )
 

cumulus_etl/errors.py

@@ -39,6 +39,8 @@
 INLINE_TASK_FAILED = 39
 INLINE_WITHOUT_FOLDER = 40
 WRONG_PHI_FOLDER = 41
+TASK_NOT_PROVIDED = 42
+TASK_MISMATCH = 43
 
 
 class FatalError(Exception):