Restructure Dockerfile for better caching

Dockerfile

@@ -4,11 +4,13 @@ FROM python:3.10.13-slim-bookworm
 LABEL authors="Patrick Jentsch <p.jentsch@uni-bielefeld.de>"
 
 
+# Set environment variables
 ENV LANG="C.UTF-8"
 ENV PYTHONDONTWRITEBYTECODE="1"
 ENV PYTHONUNBUFFERED="1"
 
 
+# Install system dependencies
 RUN apt-get update \
  && apt-get install --no-install-recommends --yes \
       build-essential \
@@ -17,37 +19,42 @@ RUN apt-get update \
  && rm --recursive /var/lib/apt/lists/*
 
 
+# Create a non-root user
 RUN useradd --create-home --no-log-init nopaque \
  && groupadd docker \
  && usermod --append --groups docker nopaque
 
-
 USER nopaque
 WORKDIR /home/nopaque
 
 
+# Create a Python virtual environment
 ENV NOPAQUE_PYTHON3_VENV_PATH="/home/nopaque/.venv"
 RUN python3 -m venv "${NOPAQUE_PYTHON3_VENV_PATH}"
 ENV PATH="${NOPAQUE_PYTHON3_VENV_PATH}/bin:${PATH}"
 
 
+# Install Python dependencies
+COPY --chown=nopaque:nopaque requirements.txt requirements.txt
+RUN python3 -m pip install --requirement requirements.txt \
+ && rm requirements.txt
+
+
+# Install the application
+COPY docker-nopaque-entrypoint.sh /usr/local/bin/
+
 COPY --chown=nopaque:nopaque app app
 COPY --chown=nopaque:nopaque migrations migrations
 COPY --chown=nopaque:nopaque tests tests
 COPY --chown=nopaque:nopaque .flaskenv boot.sh config.py nopaque.py requirements.txt ./
 
-
-RUN python3 -m pip install --requirement requirements.txt \
- && mkdir logs
-
-
-USER root
-
-
-COPY docker-nopaque-entrypoint.sh /usr/local/bin/
+RUN mkdir logs
 
 
 EXPOSE 5000
 
 
+USER root
+
+
 ENTRYPOINT ["docker-nopaque-entrypoint.sh"]

app/converters/sandpaper.py

@@ -2,32 +2,34 @@ from flask import current_app
 from app import db
 from app.models import User, Corpus, CorpusFile
 from datetime import datetime
+from pathlib import Path
+from typing import Dict, List
 import json
-import os
 import shutil
 
 
 class SandpaperConverter:
-    def __init__(self, json_db_file, data_dir):
+    def __init__(self, json_db_file: Path, data_dir: Path):
         self.json_db_file = json_db_file
         self.data_dir = data_dir
 
     def run(self):
-        with open(self.json_db_file, 'r') as f:
-            json_db = json.loads(f.read())
+        with self.json_db_file.open('r') as f:
+            json_db: List[Dict] = json.load(f)
 
         for json_user in json_db:
             if not json_user['confirmed']:
                 current_app.logger.info(f'Skip unconfirmed user {json_user["username"]}')
                 continue
-            user_dir = os.path.join(self.data_dir, str(json_user['id']))
+            user_dir = self.data_dir / f'{json_user["id"]}'
             self.convert_user(json_user, user_dir)
             db.session.commit()
 
 
-    def convert_user(self, json_user, user_dir):
+    def convert_user(self, json_user: Dict, user_dir: Path):
         current_app.logger.info(f'Create User {json_user["username"]}...')
-        user = User(
+        try:
+            user = User.create(
                 confirmed=json_user['confirmed'],
                 email=json_user['email'],
                 last_seen=datetime.fromtimestamp(json_user['last_seen']),
@@ -35,47 +37,34 @@ class SandpaperConverter:
                 password_hash=json_user['password_hash'],  # TODO: Needs to be added manually
                 username=json_user['username']
             )
-        db.session.add(user)
-        db.session.flush(objects=[user])
-        db.session.refresh(user)
-        try:
-            user.makedirs()
-        except OSError as e:
-            current_app.logger.error(e)
-            db.session.rollback()
+        except OSError:
             raise Exception('Internal Server Error')
         for json_corpus in json_user['corpora'].values():
             if not json_corpus['files'].values():
                 current_app.logger.info(f'Skip empty corpus {json_corpus["title"]}')
                 continue
-            corpus_dir = os.path.join(user_dir, 'corpora', str(json_corpus['id']))
+            corpus_dir = user_dir / 'corpora' / f'{json_corpus["id"]}'
             self.convert_corpus(json_corpus, user, corpus_dir)
         current_app.logger.info('Done')
 
 
-    def convert_corpus(self, json_corpus, user, corpus_dir):
+    def convert_corpus(self, json_corpus: Dict, user: User, corpus_dir: Path):
         current_app.logger.info(f'Create Corpus {json_corpus["title"]}...')
-        corpus = Corpus(
+        try:
+            corpus = Corpus.create(
                 user=user,
                 creation_date=datetime.fromtimestamp(json_corpus['creation_date']),
                 description=json_corpus['description'],
                 title=json_corpus['title']
             )
-        db.session.add(corpus)
-        db.session.flush(objects=[corpus])
-        db.session.refresh(corpus)
-        try:
-            corpus.makedirs()
-        except OSError as e:
-            current_app.logger.error(e)
-            db.session.rollback()
+        except OSError:
             raise Exception('Internal Server Error')
         for json_corpus_file in json_corpus['files'].values():
             self.convert_corpus_file(json_corpus_file, corpus, corpus_dir)
         current_app.logger.info('Done')
 
 
-    def convert_corpus_file(self, json_corpus_file, corpus, corpus_dir):
+    def convert_corpus_file(self, json_corpus_file: Dict, corpus: Corpus, corpus_dir: Path):
         current_app.logger.info(f'Create CorpusFile {json_corpus_file["title"]}...')
         corpus_file = CorpusFile(
             corpus=corpus,
@@ -99,13 +88,13 @@ class SandpaperConverter:
         db.session.refresh(corpus_file)
         try:
             shutil.copy2(
-                os.path.join(corpus_dir, json_corpus_file['filename']),
+                corpus_dir / json_corpus_file['filename'],
                 corpus_file.path
             )
         except:
             current_app.logger.warning(
                 'Can not convert corpus file: '
-                f'{os.path.join(corpus_dir, json_corpus_file["filename"])}'
+                f'{corpus_dir / json_corpus_file["filename"]}'
                 ' -> '
                 f'{corpus_file.path}'
             )

app/corpora/cli.py

@@ -1,7 +1,7 @@
-from app.models import Corpus, CorpusStatus
-import os
+from flask import current_app
 import shutil
 from app import db
+from app.models import Corpus, CorpusStatus
 from . import bp
 
 
@@ -18,10 +18,17 @@ def reset():
     ]
     for corpus in [x for x in Corpus.query.all() if x.status in status]:
         print(f'Resetting corpus {corpus}')
-        shutil.rmtree(os.path.join(corpus.path, 'cwb'), ignore_errors=True)
-        os.mkdir(os.path.join(corpus.path, 'cwb'))
-        os.mkdir(os.path.join(corpus.path, 'cwb', 'data'))
-        os.mkdir(os.path.join(corpus.path, 'cwb', 'registry'))
+        corpus_cwb_dir = corpus.path / 'cwb'
+        corpus_cwb_data_dir = corpus_cwb_dir / 'data'
+        corpus_cwb_registry_dir = corpus_cwb_dir / 'registry'
+        try:
+            shutil.rmtree(corpus.path / 'cwb', ignore_errors=True)
+            corpus_cwb_dir.mkdir()
+            corpus_cwb_data_dir.mkdir()
+            corpus_cwb_registry_dir.mkdir()
+        except OSError as e:
+            current_app.logger.error(e)
+            raise
         corpus.status = CorpusStatus.UNPREPARED
         corpus.num_analysis_sessions = 0
     db.session.commit()

app/corpora/cqi_over_sio/extensions.py

@@ -12,7 +12,6 @@ from typing import Dict, List
 import gzip
 import json
 import math
-import os
 from app import db
 from app.models import Corpus
 from .utils import lookups_by_cpos, partial_export_subcorpus, export_subcorpus
@@ -42,9 +41,9 @@ def ext_corpus_static_data(corpus: str) -> Dict:
     db_corpus_id: int = session['cqi_over_sio']['db_corpus_id']
     db_corpus: Corpus = Corpus.query.get(db_corpus_id)
 
-    static_data_file_path: str = os.path.join(db_corpus.path, 'cwb', 'static.json.gz')
-    if os.path.exists(static_data_file_path):
-        with open(static_data_file_path, 'rb') as f:
+    static_data_file_path = db_corpus.path / 'cwb' / 'static.json.gz'
+    if static_data_file_path.exists():
+        with static_data_file_path.open('rb') as f:
             return f.read()
 
     cqi_client: CQiClient = session['cqi_over_sio']['cqi_client']

app/corpora/files/routes.py

@@ -7,7 +7,6 @@ from flask import (
     url_for
 )
 from flask_breadcrumbs import register_breadcrumb
-import os
 from app import db
 from app.models import Corpus, CorpusFile, CorpusStatus
 from ..decorators import corpus_follower_permission_required
@@ -92,8 +91,8 @@ def corpus_file(corpus_id, corpus_file_id):
 def download_corpus_file(corpus_id, corpus_file_id):
     corpus_file = CorpusFile.query.filter_by(corpus_id=corpus_id, id=corpus_file_id).first_or_404()
     return send_from_directory(
-        os.path.dirname(corpus_file.path),
-        os.path.basename(corpus_file.path),
+        corpus_file.path.parent,
+        corpus_file.path.name,
         as_attachment=True,
         attachment_filename=corpus_file.filename,
         mimetype=corpus_file.mimetype

app/jobs/json_routes.py

@@ -1,7 +1,6 @@
 from flask import abort, current_app
 from flask_login import current_user
 from threading import Thread
-import os
 from app import db
 from app.decorators import admin_required, content_negotiation
 from app.models import Job, JobStatus
@@ -39,7 +38,7 @@ def job_log(job_id):
     if job.status not in [JobStatus.COMPLETED, JobStatus.FAILED]:
         response = {'errors': {'message': 'Job status is not completed or failed'}}
         return response, 409
-    with open(os.path.join(job.path, 'pipeline_data', 'logs', 'pyflow_log.txt')) as log_file:
+    with open(job.path / 'pipeline_data' / 'logs' / 'pyflow_log.txt') as log_file:
         log = log_file.read()
     response_data = {
         'jobLog': log

app/jobs/routes.py

@@ -7,7 +7,6 @@ from flask import (
 )
 from flask_breadcrumbs import register_breadcrumb
 from flask_login import current_user
-import os
 from app.models import Job, JobInput, JobResult
 from . import bp
 from .utils import job_dynamic_list_constructor as job_dlc
@@ -38,8 +37,8 @@ def download_job_input(job_id, job_input_id):
     if not (job_input.job.user == current_user or current_user.is_administrator()):
         abort(403)
     return send_from_directory(
-        os.path.dirname(job_input.path),
-        os.path.basename(job_input.path),
+        job_input.path.parent,
+        job_input.path.name,
         as_attachment=True,
         attachment_filename=job_input.filename,
         mimetype=job_input.mimetype
@@ -52,8 +51,8 @@ def download_job_result(job_id, job_result_id):
     if not (job_result.job.user == current_user or current_user.is_administrator()):
         abort(403)
     return send_from_directory(
-        os.path.dirname(job_result.path),
-        os.path.basename(job_result.path),
+        job_result.path.parent,
+        job_result.path.name,
         as_attachment=True,
         attachment_filename=job_result.filename,
         mimetype=job_result.mimetype

app/main/cli.py

@@ -1,6 +1,7 @@
 from flask import current_app
 from flask_migrate import upgrade
-import os
+from pathlib import Path
+from typing import List
 from app.models import (
     CorpusFollowerRole,
     Role,
@@ -17,16 +18,15 @@ def deploy():
     # Make default directories
     print('Make default directories')
     base_dir = current_app.config['NOPAQUE_DATA_DIR']
-    default_dirs = [
-        os.path.join(base_dir, 'tmp'),
-        os.path.join(base_dir, 'users')
+    default_dirs: List[Path] = [
+        base_dir / 'tmp',
+        base_dir / 'users'
     ]
-    for dir in default_dirs:
-        if os.path.exists(dir):
-            if not os.path.isdir(dir):
-                raise NotADirectoryError(f'{dir} is not a directory')
-        else:
-            os.mkdir(dir)
+    for default_dir in default_dirs:
+        if not default_dir.exists():
+            default_dir.mkdir()
+        if not default_dir.is_dir():
+            raise NotADirectoryError(f'{default_dir} is not a directory')
 
     # migrate database to latest revision
     print('Migrate database to latest revision')

app/models/__init__.py

@@ -11,6 +11,7 @@ from .spacy_nlp_pipeline_model import *
 from .tesseract_ocr_pipeline_model import *
 from .token import *
 from .user import *
+from app import login
 
 
 @login.user_loader

app/models/avatar.py

@@ -1,6 +1,6 @@
 from flask import current_app
 from flask_hashids import HashidMixin
-import os
+from pathlib import Path
 from app import db
 from .file_mixin import FileMixin
 
@@ -15,14 +15,16 @@ class Avatar(HashidMixin, FileMixin, db.Model):
     user = db.relationship('User', back_populates='avatar')
 
     @property
-    def path(self):
-        return os.path.join(self.user.path, 'avatar')
+    def path(self) -> Path:
+        return self.user.path / 'avatar'
+        # return os.path.join(self.user.path, 'avatar')
 
     def delete(self):
         try:
-            os.remove(self.path)
+            self.path.unlink(missing_ok=True)
         except OSError as e:
             current_app.logger.error(e)
+            raise
         db.session.delete(self)
 
     def to_json_serializeable(self, backrefs=False, relationships=False):

app/models/corpus.py

@@ -4,7 +4,7 @@ from flask import current_app, url_for
 from flask_hashids import HashidMixin
 from sqlalchemy.ext.associationproxy import association_proxy
 from typing import Union
-import os
+from pathlib import Path
 import shutil
 import xml.etree.ElementTree as ET
 from app import db
@@ -88,8 +88,8 @@ class Corpus(HashidMixin, db.Model):
         return f'{self.user.jsonpatch_path}/corpora/{self.hashid}'
 
     @property
-    def path(self):
-        return os.path.join(self.user.path, 'corpora', str(self.id))
+    def path(self) -> Path:
+        return self.user.path / 'corpora' / f'{self.id}'
 
     @property
     def url(self):
@@ -105,27 +105,39 @@ class Corpus(HashidMixin, db.Model):
         db.session.add(corpus)
         db.session.flush(objects=[corpus])
         db.session.refresh(corpus)
+        corpus_files_dir = corpus.path / 'files'
+        corpus_cwb_dir = corpus.path / 'cwb'
+        corpus_cwb_data_dir = corpus_cwb_dir / 'data'
+        corpus_cwb_registry_dir = corpus_cwb_dir / 'registry'
         try:
-            os.mkdir(corpus.path)
-            os.mkdir(os.path.join(corpus.path, 'files'))
-            os.mkdir(os.path.join(corpus.path, 'cwb'))
-            os.mkdir(os.path.join(corpus.path, 'cwb', 'data'))
-            os.mkdir(os.path.join(corpus.path, 'cwb', 'registry'))
+            corpus.path.mkdir()
+            corpus_files_dir.mkdir()
+            corpus_cwb_dir.mkdir()
+            corpus_cwb_data_dir.mkdir()
+            corpus_cwb_registry_dir.mkdir()
         except OSError as e:
+            # TODO: Potential leftover cleanup
             current_app.logger.error(e)
             db.session.rollback()
-            raise e
+            raise
         return corpus
 
     def build(self):
-        build_dir = os.path.join(self.path, 'cwb')
-        shutil.rmtree(build_dir, ignore_errors=True)
-        os.mkdir(build_dir)
-        os.mkdir(os.path.join(build_dir, 'data'))
-        os.mkdir(os.path.join(build_dir, 'registry'))
+        corpus_cwb_dir = self.path / 'cwb'
+        corpus_cwb_data_dir = corpus_cwb_dir / 'data'
+        corpus_cwb_registry_dir = corpus_cwb_dir / 'registry'
+        try:
+            shutil.rmtree(corpus_cwb_dir, ignore_errors=True)
+            corpus_cwb_dir.mkdir()
+            corpus_cwb_data_dir.mkdir()
+            corpus_cwb_registry_dir.mkdir()
+        except OSError as e:
+            current_app.logger.error(e)
+            self.status = CorpusStatus.FAILED
+            raise
         corpus_element = ET.fromstring('<corpus>\n</corpus>')
         for corpus_file in self.files:
-            normalized_vrt_path = os.path.join(build_dir, f'{corpus_file.id}.norm.vrt')
+            normalized_vrt_path = corpus_cwb_dir / f'{corpus_file.id}.norm.vrt'
             try:
                 normalize_vrt_file(corpus_file.path, normalized_vrt_path)
             except:
@@ -152,7 +164,7 @@ class Corpus(HashidMixin, db.Model):
             # corpus_element.insert(1, text_element)
             corpus_element.append(text_element)
         ET.ElementTree(corpus_element).write(
-            os.path.join(build_dir, 'corpus.vrt'),
+            corpus_cwb_dir / 'corpus.vrt',
             encoding='utf-8'
         )
         self.status = CorpusStatus.SUBMITTED

app/models/corpus_file.py

@@ -1,6 +1,6 @@
 from flask import current_app, url_for
 from flask_hashids import HashidMixin
-import os
+from pathlib import Path
 from app import db
 from .corpus import CorpusStatus
 from .file_mixin import FileMixin
@@ -45,8 +45,8 @@ class CorpusFile(FileMixin, HashidMixin, db.Model):
         return f'{self.corpus.jsonpatch_path}/files/{self.hashid}'
 
     @property
-    def path(self):
-        return os.path.join(self.corpus.path, 'files', str(self.id))
+    def path(self) -> Path:
+        return self.corpus.path / 'files' / f'{self.id}'
 
     @property
     def url(self):
@@ -66,9 +66,10 @@ class CorpusFile(FileMixin, HashidMixin, db.Model):
 
     def delete(self):
         try:
-            os.remove(self.path)
+            self.path.unlink(missing_ok=True)
         except OSError as e:
             current_app.logger.error(e)
+            raise
         db.session.delete(self)
         self.corpus.status = CorpusStatus.UNPREPARED
 

app/models/job.py

@@ -4,7 +4,7 @@ from flask import current_app, url_for
 from flask_hashids import HashidMixin
 from time import sleep
 from typing import Union
-import os
+from pathlib import Path
 import shutil
 from app import db
 from app.ext.flask_sqlalchemy import ContainerColumn, IntEnumColumn
@@ -79,8 +79,8 @@ class Job(HashidMixin, db.Model):
         return f'{self.user.jsonpatch_path}/jobs/{self.hashid}'
 
     @property
-    def path(self):
-        return os.path.join(self.user.path, 'jobs', str(self.id))
+    def path(self) -> Path:
+        return self.user.path / 'jobs' / f'{self.id}'
 
     @property
     def url(self):
@@ -96,15 +96,19 @@ class Job(HashidMixin, db.Model):
         db.session.add(job)
         db.session.flush(objects=[job])
         db.session.refresh(job)
+        job_inputs_dir = job.path / 'inputs'
+        job_pipeline_data_dir = job.path / 'pipeline_data'
+        job_results_dir = job.path / 'results'
         try:
-            os.mkdir(job.path)
-            os.mkdir(os.path.join(job.path, 'inputs'))
-            os.mkdir(os.path.join(job.path, 'pipeline_data'))
-            os.mkdir(os.path.join(job.path, 'results'))
+            job.path.mkdir()
+            job_inputs_dir.mkdir()
+            job_pipeline_data_dir.mkdir()
+            job_results_dir.mkdir()
         except OSError as e:
+            # TODO: Potential leftover cleanup
             current_app.logger.error(e)
             db.session.rollback()
-            raise e
+            raise
         return job
 
     def delete(self):
@@ -131,8 +135,8 @@ class Job(HashidMixin, db.Model):
         ''' Restart a job - only if the status is failed '''
         if self.status != JobStatus.FAILED:
             raise Exception('Job status is not "failed"')
-        shutil.rmtree(os.path.join(self.path, 'results'), ignore_errors=True)
-        shutil.rmtree(os.path.join(self.path, 'pyflow.data'), ignore_errors=True)
+        shutil.rmtree(self.path / 'results', ignore_errors=True)
+        shutil.rmtree(self.path / 'pyflow.data', ignore_errors=True)
         for result in self.results:
             db.session.delete(result)
         self.end_date = None

app/models/job_input.py

@@ -1,6 +1,6 @@
 from flask import url_for
 from flask_hashids import HashidMixin
-import os
+from pathlib import Path
 from app import db
 from .file_mixin import FileMixin
 
@@ -33,8 +33,8 @@ class JobInput(FileMixin, HashidMixin, db.Model):
         return f'{self.job.jsonpatch_path}/inputs/{self.hashid}'
 
     @property
-    def path(self):
-        return os.path.join(self.job.path, 'inputs', str(self.id))
+    def path(self) -> Path:
+        return self.job.path / 'inputs' / f'{self.id}'
 
     @property
     def url(self):

app/models/job_result.py

@@ -1,6 +1,6 @@
 from flask import url_for
 from flask_hashids import HashidMixin
-import os
+from pathlib import Path
 from app import db
 from .file_mixin import FileMixin
 
@@ -35,8 +35,8 @@ class JobResult(FileMixin, HashidMixin, db.Model):
         return f'{self.job.jsonpatch_path}/results/{self.hashid}'
 
     @property
-    def path(self):
-        return os.path.join(self.job.path, 'results', str(self.id))
+    def path(self) -> Path:
+        return self.job.path / 'results' / f'{self.id}'
 
     @property
     def url(self):

app/models/spacy_nlp_pipeline_model.py

@@ -1,8 +1,7 @@
-from flask import abort, current_app, url_for
+from flask import current_app, url_for
 from flask_hashids import HashidMixin
-from time import sleep
 from tqdm import tqdm
-import os
+from pathlib import Path
 import requests
 import yaml
 from app import db
@@ -32,12 +31,8 @@ class SpaCyNLPPipelineModel(FileMixin, HashidMixin, db.Model):
     user = db.relationship('User', back_populates='spacy_nlp_pipeline_models')
 
     @property
-    def path(self):
-        return os.path.join(
-            self.user.path,
-            'spacy_nlp_pipeline_models',
-            str(self.id)
-        )
+    def path(self) -> Path:
+        return self.user.path / 'spacy_nlp_pipeline_models' / f'{self.id}'
 
     @property
     def jsonpatch_path(self):
@@ -57,14 +52,10 @@ class SpaCyNLPPipelineModel(FileMixin, HashidMixin, db.Model):
     @staticmethod
     def insert_defaults(force_download=False):
         nopaque_user = User.query.filter_by(username='nopaque').first()
-        defaults_file = os.path.join(
-            os.path.dirname(os.path.abspath(__file__)),
-            'default_records',
-            'spacy_nlp_pipeline_model.yml'
-        )
-        with open(defaults_file, 'r') as f:
-            defaults = yaml.safe_load(f)
-        for m in defaults:
+        default_records_file = Path(__file__).parent / 'default_records' / 'spacy_nlp_pipeline_model.yml'
+        with default_records_file.open('r') as f:
+            default_records = yaml.safe_load(f)
+        for m in default_records:
             model = SpaCyNLPPipelineModel.query.filter_by(title=m['title'], version=m['version']).first()  # noqa
             if model is not None:
                 model.compatible_service_versions = m['compatible_service_versions']
@@ -96,7 +87,7 @@ class SpaCyNLPPipelineModel(FileMixin, HashidMixin, db.Model):
                 db.session.add(model)
                 db.session.flush(objects=[model])
                 db.session.refresh(model)
-            if not os.path.exists(model.path) or force_download:
+            if not model.path.exists() or force_download:
                 r = requests.get(m['url'], stream=True)
                 pbar = tqdm(
                     desc=f'{model.title} ({model.filename})',
@@ -116,9 +107,10 @@ class SpaCyNLPPipelineModel(FileMixin, HashidMixin, db.Model):
     
     def delete(self):
         try:
-            os.remove(self.path)
+            self.path.unlink(missing_ok=True)
         except OSError as e:
             current_app.logger.error(e)
+            raise
         db.session.delete(self)
 
     def to_json_serializeable(self, backrefs=False, relationships=False):

app/models/tesseract_ocr_pipeline_model.py

@@ -1,7 +1,7 @@
 from flask import current_app, url_for
 from flask_hashids import HashidMixin
 from tqdm import tqdm
-import os
+from pathlib import Path
 import requests
 import yaml
 from app import db
@@ -30,12 +30,8 @@ class TesseractOCRPipelineModel(FileMixin, HashidMixin, db.Model):
     user = db.relationship('User', back_populates='tesseract_ocr_pipeline_models')
 
     @property
-    def path(self):
-        return os.path.join(
-            self.user.path,
-            'tesseract_ocr_pipeline_models',
-            str(self.id)
-        )
+    def path(self) -> Path:
+        return self.user.path / 'tesseract_ocr_pipeline_models' / f'{self.id}'
 
     @property
     def jsonpatch_path(self):
@@ -55,14 +51,10 @@ class TesseractOCRPipelineModel(FileMixin, HashidMixin, db.Model):
     @staticmethod
     def insert_defaults(force_download=False):
         nopaque_user = User.query.filter_by(username='nopaque').first()
-        defaults_file = os.path.join(
-            os.path.dirname(os.path.abspath(__file__)),
-            'default_records',
-            'tesseract_ocr_pipeline_model.yml'
-        )
-        with open(defaults_file, 'r') as f:
-            defaults = yaml.safe_load(f)
-        for m in defaults:
+        default_records_file = Path(__file__).parent / 'default_records' / 'tesseract_ocr_pipeline_model.yml'
+        with default_records_file.open('r') as f:
+            default_records = yaml.safe_load(f)
+        for m in default_records:
             model = TesseractOCRPipelineModel.query.filter_by(title=m['title'], version=m['version']).first()  # noqa
             if model is not None:
                 model.compatible_service_versions = m['compatible_service_versions']
@@ -92,7 +84,7 @@ class TesseractOCRPipelineModel(FileMixin, HashidMixin, db.Model):
                 db.session.flush(objects=[model])
                 db.session.refresh(model)
                 model.filename = f'{model.id}.traineddata'
-            if not os.path.exists(model.path) or force_download:
+            if not model.path.exists() or force_download:
                 r = requests.get(m['url'], stream=True)
                 pbar = tqdm(
                     desc=f'{model.title} ({model.filename})',
@@ -112,9 +104,10 @@ class TesseractOCRPipelineModel(FileMixin, HashidMixin, db.Model):
 
     def delete(self):
         try:
-            os.remove(self.path)
+            self.path.unlink(missing_ok=True)
         except OSError as e:
             current_app.logger.error(e)
+            raise
         db.session.delete(self)
 
     def to_json_serializeable(self, backrefs=False, relationships=False):

app/models/user.py

@@ -4,14 +4,14 @@ from flask import current_app, url_for
 from flask_hashids import HashidMixin
 from flask_login import UserMixin
 from sqlalchemy.ext.associationproxy import association_proxy
+from pathlib import Path
 from typing import Union
 from werkzeug.security import generate_password_hash, check_password_hash
 import jwt
-import os
 import re
 import secrets
 import shutil
-from app import db, hashids, login
+from app import db, hashids
 from app.ext.flask_sqlalchemy import IntEnumColumn
 from .corpus import Corpus
 from .corpus_follower_association import CorpusFollowerAssociation
@@ -145,9 +145,8 @@ class User(HashidMixin, UserMixin, db.Model):
         self.password_hash = generate_password_hash(password)
 
     @property
-    def path(self):
-        return os.path.join(
-            current_app.config.get('NOPAQUE_DATA_DIR'), 'users', str(self.id))
+    def path(self) -> Path:
+        return current_app.config.get('NOPAQUE_DATA_DIR') / 'users' / f'{self.id}'
 
     @staticmethod
     def create(**kwargs):
@@ -155,16 +154,21 @@ class User(HashidMixin, UserMixin, db.Model):
         db.session.add(user)
         db.session.flush(objects=[user])
         db.session.refresh(user)
+        user_spacy_nlp_pipeline_models_dir = user.path / 'spacy_nlp_pipeline_models'
+        user_tesseract_ocr_pipeline_models_dir = user.path / 'tesseract_ocr_pipeline_models'
+        user_corpora_dir = user.path / 'corpora'
+        user_jobs_dir = user.path / 'jobs'
         try:
-            os.mkdir(user.path)
-            os.mkdir(os.path.join(user.path, 'spacy_nlp_pipeline_models'))
-            os.mkdir(os.path.join(user.path, 'tesseract_ocr_pipeline_models'))
-            os.mkdir(os.path.join(user.path, 'corpora'))
-            os.mkdir(os.path.join(user.path, 'jobs'))
+            user.path.mkdir()
+            user_spacy_nlp_pipeline_models_dir.mkdir()
+            user_tesseract_ocr_pipeline_models_dir.mkdir()
+            user_corpora_dir.mkdir()
+            user_jobs_dir.mkdir()
         except OSError as e:
+            # TODO: Potential leftover cleanup
             current_app.logger.error(e)
             db.session.rollback()
-            raise e
+            raise
         return user
 
     @staticmethod

app/services/__init__.py

@@ -1,12 +1,11 @@
 from flask import Blueprint
 from flask_login import login_required
-import os
+from pathlib import Path
 import yaml
 
 
-services_file = \
-    os.path.join(os.path.dirname(os.path.abspath(__file__)), 'services.yml')
-with open(services_file, 'r') as f:
+services_file = Path(__file__).parent / 'services.yml'
+with services_file.open('r') as f:
     SERVICES = yaml.safe_load(f)
 
 bp = Blueprint('services', __name__)

app/users/cli.py

@@ -1,6 +1,4 @@
 from app.models import User
-import os
-import shutil
 from app import db
 from . import bp
 

app/users/routes.py

@@ -7,7 +7,6 @@ from flask import (
 )
 from flask_breadcrumbs import register_breadcrumb
 from flask_login import current_user
-import os
 from app.models import User
 from . import bp
 from .utils import user_dynamic_list_constructor as user_dlc
@@ -40,8 +39,8 @@ def user_avatar(user_id):
     if user.avatar is None:
         return redirect(url_for('static', filename='images/user_avatar.png'))
     return send_from_directory(
-        os.path.dirname(user.avatar.path),
-        os.path.basename(user.avatar.path),
+        user.avatar.path.parent,
+        user.avatar.path.name,
         as_attachment=True,
         attachment_filename=user.avatar.filename,
         mimetype=user.avatar.mimetype

config.py

@@ -1,6 +1,7 @@
 from dotenv import load_dotenv
 from flask import Flask
 from logging.handlers import RotatingFileHandler
+from pathlib import Path
 from werkzeug.middleware.proxy_fix import ProxyFix
 import logging
 import os
@@ -57,8 +58,7 @@ class Config:
 
     ''' # nopaque # '''
     NOPAQUE_ADMIN = os.environ.get('NOPAQUE_ADMIN')
-    NOPAQUE_DATA_DIR = \
-        os.path.abspath(os.environ.get('NOPAQUE_DATA_PATH', '/mnt/nopaque'))
+    NOPAQUE_DATA_DIR = Path(os.environ.get('NOPAQUE_DATA_PATH', '/mnt/nopaque'))
     NOPAQUE_IS_PRIMARY_INSTANCE = \
         os.environ.get('NOPAQUE_IS_PRIMARY_INSTANCE', 'true').lower() == 'true'
     NOPAQUE_MAIL_SUBJECT_PREFIX = '[nopaque]'