fixed type errors and broken reset button

Author: charlottekostelic

overload_web/application/commands/file_io.py

@@ -89,7 +89,7 @@ class LoadAllWorkflowFiles:
     def execute(
         workflow_id: str, storage: ports.FileStorage, repo: ports.SqlRepositoryProtocol
     ) -> list[files.VendorFile]:
-        file_list = repo.list(workflow_id)
+        file_list = repo.list_by_id(workflow_id)
         vendor_files = [
             files.VendorFile(
                 file_name=i["filename"], content=storage.load(i["reference"])

overload_web/application/ports.py

@@ -3,15 +3,7 @@
 from __future__ import annotations
 
 import logging
-from typing import (
-    Any,
-    BinaryIO,
-    Iterator,
-    Protocol,
-    Sequence,
-    TypeVar,
-    runtime_checkable,
-)
+from typing import Any, Iterator, Protocol, Sequence, TypeVar, runtime_checkable
 
 logger = logging.getLogger(__name__)
 
@@ -52,11 +44,34 @@ def get_bibs_by_id(
 
 @runtime_checkable
 class FileStorage(Protocol):
-    def load(self, reference: str) -> BinaryIO: ...  # pragma: no branch
+    def load(self, reference: str) -> bytes: ...  # pragma: no branch
+
+    """
+    Load a file.
+
+    Args:
+        reference: the path to the file
+
+    Returns:
+        the content of the specified file as a `bytes` object
+    """
+
     def save(
         self, id: str, filename: str, content: bytes
     ) -> str: ...  # pragma: no branch
 
+    """
+    Save a file to a location on storage.
+
+    Args:
+        id: the workflow_id for the file.
+        filename: the name of the file.
+        content: the content of the file as a bytes object.
+
+    Returns:
+        the path where the file was saved as a string
+    """
+
 
 @runtime_checkable
 class FileLoader(Protocol):
@@ -191,7 +206,13 @@ def list(
         self, offset: int | None = 0, limit: int | None = 0
     ) -> Sequence[dict[str, Any]]: ...  # pragma: no branch
 
-    """List objects in a database."""
+    """List all objects in a database."""
+
+    def list_by_id(
+        self, id: str | int
+    ) -> Sequence[dict[str, Any]]: ...  # pragma: no branch
+
+    """List all objects in a database filtering by a specific id."""
 
     def save(self, obj: T) -> dict[str, Any]: ...  # pragma: no branch
 

overload_web/application/services/bib_processing.py

@@ -7,7 +7,7 @@
 from collections import Counter, defaultdict
 from typing import Any
 
-from overload_web.domain.models import bibs
+from overload_web.domain.models import bibs, reporting
 
 logger = logging.getLogger(__name__)
 
@@ -16,7 +16,7 @@ def create_full_records_report(
     analysis: list[bibs.MatchAnalysis],
     missing_barcodes: list[str],
     file_names: list[str],
-) -> dict[str, list[Any]]:
+) -> reporting.ProcessingStatistics:
     """Generate statistics from a batch of processed full-level records"""
     stats = defaultdict(list)
     stats["file_names"].extend(file_names)
@@ -27,12 +27,12 @@ def create_full_records_report(
     out: dict[str, Any] = dict(stats)
     out["total_records"] = len(analysis)
     out["total_files"] = len(stats["file_names"])
-    return out
+    return reporting.ProcessingStatistics(**out)
 
 
 def create_order_records_report(
     analysis: list[bibs.MatchAnalysis], file_names: list[str]
-) -> dict[str, list[Any]]:
+) -> reporting.ProcessingStatistics:
     """Generate statistics from a batch of processed order-level records"""
     stats = defaultdict(list)
     for rec in analysis:
@@ -42,7 +42,7 @@ def create_order_records_report(
     out["total_records"] = len(analysis)
     out["total_files"] = len(file_names)
     out["file_names"] = file_names
-    return out
+    return reporting.ProcessingStatistics(**out)
 
 
 def extract_barcodes(records: list[bibs.DomainBib]) -> list[str]:

overload_web/infrastructure/batch_db.py

@@ -133,7 +133,7 @@ def save(self, obj: PVFBatch) -> dict[str, Any]:
             ProcessedFileModel.model_validate(i, from_attributes=True)
             for i in obj.files
         ]
-        valid_stats = PVFReportModel(**obj.report)
+        valid_stats = PVFReportModel.model_validate(obj.report, from_attributes=True)
         valid_batch = PVFBatch(files=valid_files, report=valid_stats)
         self.session.add(valid_batch)
         self.session.commit()

overload_web/infrastructure/file_io.py

@@ -14,7 +14,7 @@
 import logging
 import os
 from pathlib import Path
-from typing import Any, BinaryIO, Sequence
+from typing import Any, Sequence
 
 from file_retriever import Client, File
 from sqlmodel import Field, Session, SQLModel, select
@@ -36,8 +36,10 @@ def save(self, id: str, filename: str, content: bytes) -> str:
 
         return str(path)
 
-    def load(self, reference: str) -> BinaryIO:
-        return open(reference, "rb")
+    def load(self, reference: str) -> bytes:
+        with open(reference, "rb") as fh:
+            file = fh.read()
+        return file
 
 
 class LocalFileLoader:
@@ -181,21 +183,20 @@ def get(self, id: str | int) -> dict[str, Any] | None:
         file = self.session.get(IncomingFileModel, id)
         return file.model_dump() if file else None
 
-    def list(self, workflow_id: str | int) -> Sequence[dict[str, Any]]:
+    def list_by_id(self, id: str | int) -> Sequence[dict[str, Any]]:
         """
         Retrieve all `IncomingFileModel` objects in the database.
 
         Args:
-            workflow_id: the `workflow_id` whose files to retrieve.
+            id: the `workflow_id` whose files to retrieve.
 
         Returns:
             a sequence of `IncomingFileModel` objects.
         """
-        statement = select(IncomingFileModel).where(
-            IncomingFileModel.workflow_id == workflow_id
-        )
-        results = self.session.exec(statement).all()
-        return [i.model_dump() for i in results]
+        statement = select(IncomingFileModel).where(IncomingFileModel.workflow_id == id)
+        results = self.session.exec(statement)
+        all_files = results.all()
+        return [i.model_dump() for i in all_files]
 
     def save(self, obj: IncomingFileModel) -> dict[str, Any]:
         """

overload_web/presentation/routers/files.py

@@ -62,7 +62,7 @@ async def select_ftp_file(
     UploadFileToWorkflow(storage=storage, repo=repository).execute(
         workflow_id=workflow_id, filename=remote_file, content=file_content.content
     )
-    selected = repository.list(workflow_id)
+    selected = repository.list_by_id(workflow_id)
     return request.app.state.templates.TemplateResponse(
         name="pvf_partials/selected_files.html",
         request=request,
@@ -82,7 +82,7 @@ async def upload_file(
     UploadFileToWorkflow(storage=storage, repo=repository).execute(
         workflow_id=workflow_id, filename=str(file.filename), content=file.file.read()
     )
-    selected = repository.list(workflow_id)
+    selected = repository.list_by_id(workflow_id)
     logger.info(f"Current file list: {selected}")
     return request.app.state.templates.TemplateResponse(
         name="pvf_partials/selected_files.html",
@@ -99,7 +99,7 @@ async def remove_file(
     workflow_id: str = Form(),
 ):
     DeleteFileFromWorkflow.execute(id=file_id, repo=repository)
-    selected = repository.list(workflow_id)
+    selected = repository.list_by_id(workflow_id)
     logger.info(f"Current file list: {selected}")
     return request.app.state.templates.TemplateResponse(
         "pvf_partials/selected_files.html", {"request": request, "files": selected}

overload_web/templates/pvf_partials/pvf_results.html

@@ -37,10 +37,10 @@
 <div class="pvf-button-section">
   <div class="row p-3">
     <div class="col">
-      <button id="pvf-button"
-              class="btn btn-nypl"
-              href='{{ url_for("vendor_file_page") }}'
-              type="submit">Reset form</button>
+      <a id="pvf-button"
+         class="btn btn-nypl"
+         href='{{ url_for("vendor_file_page") }}'
+         type="submit">Reset form</a>
     </div>
   </div>
 </div>