Palace Manager

This is a The Palace Project maintained fork of the NYPL Library Simplified Circulation Manager.

Installation

Docker images created from this code are available at:

• circ-webapp
• circ-scripts
• circ-exec

Docker images are the preferred way to deploy this code in a production environment.

Git Branch Workflow

Branch	Python Version
main	Python 3
python2	Python 2

The default branch is main and that's the working branch that should be used when branching off for bug fixes or new features.

Python 2 stopped being supported after January 1st, 2020 but there is still a python2 branch which can be used. As of August 2021, development will be done in the main branch and the python2 branch will not be updated unless absolutely necessary.

Set Up

Docker Compose

In order to help quickly set up a development environment, we include a docker-compose.yml file. This docker-compose file, will build the webapp and scripts containers from your local repository, and start those containers as well as all the necessary service containers.

You can give this a try by running the following command:

docker-compose up --build

Python Set Up

Homebrew (OSX)

If you do not have Python 3 installed, you can use Homebrew to install it by running the command brew install python3.

If you do not yet have Homebrew, you can install it by running the following:

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

While you're at it, go ahead and install the following required dependencies:

brew install pkg-config libffi
brew install libjpeg

The version of libxmlsec1 (v1.3) that ships with brew is currently not compatible with the python xmlsec package that we use (for more details see this xmlsec issue). In order to install a compatible version of libxmlsec1, you can use the following command:

brew tap jonathangreen/xmlsec
brew install [email protected]

Linux

Most distributions will offer Python packages. On Arch Linux, the following command is sufficient:

pacman -S python

pyenv

pyenv pyenv lets you easily switch between multiple versions of Python. It can be installed using the command curl https://pyenv.run | bash. You can then install the version of Python you want to work with.

It is recommended that pyenv-virtualenv be used to allow pyenv to manage virtual environments in a manner that can be used by the poetry tool. The pyenv-virtualenv plugin can be installed by cloning the relevant repository into the plugins subdirectory of your $PYENV_ROOT:

mkdir -p $PYENV_ROOT/plugins
cd $PYENV_ROOT/plugins
git clone https://github.com/pyenv/pyenv-virtualenv

After cloning the repository, pyenv now has a new virtualenv command:

$ pyenv virtualenv
pyenv-virtualenv: no virtualenv name given.

Poetry

You will need to set up a local virtual environment to install packages and run the project. This project uses poetry for dependency management.

Poetry can be installed using the command curl -sSL https://install.python-poetry.org | python3 -.

More information about installation options can be found in the poetry documentation.

Opensearch

Palace now supports Opensearch: please use it instead of Elasticsearch. Elasticsearch is no longer supported.

Docker

We recommend that you run Opensearch with docker using the following docker commands:

docker run --name opensearch -d --rm -p 9200:9200 -e "discovery.type=single-node" -e "plugins.security.disabled=true" "opensearchproject/opensearch:1"
docker exec opensearch opensearch-plugin -s install analysis-icu
docker restart opensearch

Database

Docker

docker run -d --name pg -e POSTGRES_USER=palace -e POSTGRES_PASSWORD=test -e POSTGRES_DB=circ -p 5432:5432 postgres:12

You can run psql in the container using the command

docker exec -it pg psql -U palace circ

Local

1. Download and install Postgres if you don't have it already.
2. Use the command psql to access the Postgresql client.
3. Within the session, run the following commands:

CREATE DATABASE circ;
CREATE USER palace with password 'test';
grant all privileges on database circ to palace;

Environment variables

Database

To let the application know which database to use, set the SIMPLIFIED_PRODUCTION_DATABASE environment variable.

export SIMPLIFIED_PRODUCTION_DATABASE="postgresql://palace:test@localhost:5432/circ"

Opensearch

To let the application know which Opensearch instance to use, you can set the following environment variables:

• PALACE_SEARCH_URL: The url of the Opensearch instance (required).
• PALACE_SEARCH_INDEX_PREFIX: The prefix to use for the Opensearch indices. The default is circulation-works. This is useful if you want to use the same Opensearch instance for multiple CM (optional).
• PALACE_SEARCH_TIMEOUT: The timeout in seconds to use when connecting to the Opensearch instance. The default is 20 (optional).
• PALACE_SEARCH_MAXSIZE: The maximum size of the connection pool to use when connecting to the Opensearch instance. (optional).

export PALACE_SEARCH_URL="http://localhost:9200"

General

• PALACE_BASE_URL: The base URL of the application. Used to create absolute links. (optional)
• PALACE_PATRON_WEB_HOSTNAMES: Only web applications from these hosts can access this circulation manager. This can be a single hostname (http://catalog.library.org) or a pipe-separated list of hostnames (http://catalog.library.org|https://beta.library.org). You can also set this to * to allow access from any host, but you must not do this in a production environment -- only during development. (optional)
• PALACE_AUTHENTICATION_DOCUMENT_CACHE_TIME: Cache time for authentication documents (in seconds). The default is 3600 (optional).

Storage Service

The application optionally uses a s3 compatible storage service to store files. To configure the application to use a storage service, you can set the following environment variables:

• PALACE_STORAGE_PUBLIC_ACCESS_BUCKET: Required if you want to use the storage service to serve files directly to users. This is the name of the bucket that will be used to serve files. This bucket should be configured to allow public access to the files.
• PALACE_STORAGE_ANALYTICS_BUCKET: Required if you want to use the storage service to store analytics data.
• PALACE_STORAGE_ACCESS_KEY: The access key (optional).
  • If this key is set it will be passed to boto3 when connecting to the storage service.
  • If it is not set boto3 will attempt to find credentials as outlined in their documentation.
• PALACE_STORAGE_SECRET_KEY: The secret key (optional).
• PALACE_STORAGE_REGION: The AWS region of the storage service (optional).
• PALACE_STORAGE_ENDPOINT_URL: The endpoint of the storage service (optional). This is used if you are using a s3 compatible storage service like minio.
• PALACE_STORAGE_URL_TEMPLATE: The url template to use when generating urls for files stored in the storage service (optional).
  • The default value is https://{bucket}.s3.{region}.amazonaws.com/{key}.
  • The following variables can be used in the template:
    • {bucket}: The name of the bucket.
    • {key}: The key of the file.
    • {region}: The region of the storage service.

Reporting

• PALACE_REPORTING_NAME: (Optional) A name used to identify the CM instance associated with generated reports.
• SIMPLIFIED_REPORTING_EMAIL: (Required) Email address of recipient of reports.

Logging

The application uses the Python logging module for logging. Optionally logs can be configured to be sent to AWS CloudWatch logs. The following environment variables can be used to configure the logging:

• PALACE_LOG_LEVEL: The log level to use for the application. The default is INFO.
• PALACE_LOG_VERBOSE_LEVEL: The log level to use for particularly verbose loggers. Keeping these loggers at a higher log level by default makes it easier to troubleshoot issues. The default is WARNING.
• PALACE_LOG_CLOUDWATCH_ENABLED: Enable / disable sending logs to CloudWatch. The default is false.
• PALACE_LOG_CLOUDWATCH_REGION: The AWS region of the CloudWatch logs. This must be set if using CloudWatch logs.
• PALACE_LOG_CLOUDWATCH_GROUP: The name of the CloudWatch log group to send logs to. Default is palace.
• PALACE_LOG_CLOUDWATCH_STREAM: The name of the CloudWatch log stream to send logs to. Default is {machine_name}/{program_name}/{logger_name}/{process_id}. See watchtower docs for details.
• PALACE_LOG_CLOUDWATCH_INTERVAL: The interval in seconds to send logs to CloudWatch. Default is 60.
• PALACE_LOG_CLOUDWATCH_CREATE_GROUP: Whether to create the log group if it does not exist. Default is true.
• PALACE_LOG_CLOUDWATCH_ACCESS_KEY: The access key to use when sending logs to CloudWatch. This is optional.
  • If this key is set it will be passed to boto3 when connecting to CloudWatch.
  • If it is not set boto3 will attempt to find credentials as outlined in their documentation.
• PALACE_LOG_CLOUDWATCH_SECRET_KEY: The secret key to use when sending logs to CloudWatch. This is optional.

Patron Basic Token authentication

Enables/disables patron "basic token" authentication through setting the designated environment variable to any (case-insensitive) value of "true"/"yes"/"on"/"1" or "false"/"no"/"off"/"0", respectively. If the value is the empty string or the variable is not present in the environment, it is disabled by default.

• SIMPLIFIED_ENABLE_BASIC_TOKEN_AUTH

export SIMPLIFIED_ENABLE_BASIC_TOKEN_AUTH=true

Firebase Cloud Messaging

For Firebase Cloud Messaging (FCM) support (e.g., for notifications), one (and only one) of the following should be set:

• SIMPLIFIED_FCM_CREDENTIALS_JSON - the JSON-format Google Cloud Platform (GCP) service account key or
• SIMPLIFIED_FCM_CREDENTIALS_FILE - the name of the file containing that key.

export SIMPLIFIED_FCM_CREDENTIALS_JSON='{"type":"service_account","project_id":"<id>", "private_key_id":"f8...d1", ...}'

...or...

export SIMPLIFIED_FCM_CREDENTIALS_FILE="/opt/credentials/fcm_credentials.json"

The FCM credentials can be downloaded once a Google Service account has been created. More details in the FCM documentation

Quicksight Dashboards

For generating quicksight dashboard links the following environment variable is required QUICKSIGHT_AUTHORIZED_ARNS - A dictionary of the format "<dashboard name>": ["arn:aws:quicksight:...",...] where each quicksight dashboard gets treated with an arbitrary "name", and a list of "authorized arns". The first the "authorized arns" is always considered as the InitialDashboardID when creating an embed URL for the respective "dashboard name".

Analytics

Local analytics are enabled by default. S3 analytics can be enabled via the following environment variable:

• PALACE_S3_ANALYTICS_ENABLED: A boolean value to disable or enable s3 analytics. The default is false.

Email

To use the features that require sending emails, for example to reset the password for logged-out users, you will need to have a working SMTP server and set some environment variables:

• PALACE_MAIL_SERVER - The SMTP server to use. Required if you want to send emails.
• PALACE_MAIL_PORT - The port of the SMTP server. Default: 25. (optional)
• PALACE_MAIL_USERNAME - The username to use when connecting to the SMTP server. (optional)
• PALACE_MAIL_PASSWORD - The password to use when connecting to the SMTP server. (optional)
• PALACE_MAIL_SENDER - The email address to use as the sender of the emails. (optional)

Running the Application

As mentioned in the pyenv section, the poetry tool should be executed under a virtual environment in order to guarantee that it will use the Python version you expect. To use a particular Python version, you should create a local virtual environment in the cloned circulation repository directory. Assuming that you want to use, for example, Python 3.11.1:

pyenv virtualenv 3.11.1 circ

This will create a new local virtual environment called circ that uses Python 3.11.1. Switch to that environment:

pyenv local circ

On most systems, using pyenv will adjust your shell prompt to indicate which virtual environment you are now in. For example, the version of Python installed in your operating system might be 3.10.1, but using a virtual environment can substitute, for example, 3.11.1:

$ python --version
Python 3.10.1

$ pyenv local circ
(circ) $ python --version
Python 3.11.1

For brevity, these instructions assume that all shell commands will be executed within a virtual environment.

Install the dependencies (including dev and CI):

poetry install

Install only the production dependencies:

poetry install --only main,pg

Run the application with:

poetry run python app.py

Check that there is now a web server listening on port 6500:

curl http://localhost:6500/

The Admin Interface

Access

By default, the application is configured to provide a built-in version of the admin web interface. The admin interface can be accessed by visiting the /admin endpoint:

# On Linux
xdg-open http://localhost:6500/admin/

# On MacOS
open http://localhost:6500/admin/

If no existing users are configured (which will be the case if this is a fresh instance of the application), the admin interface will prompt you to specify an email address and password that will be used for subsequent logins. Extra users can be configured later.

Creating A Library

Navigate to System Configuration → Libraries and click Create new library. You will be prompted to enter various details such as the name of the library, a URL, and more. For example, the configuration for a hypothetical library, Hazelnut Peak, might look like this:

Note that the Patron support email address will appear in OPDS feeds served by the application, so make sure that it is an email address you are happy to make public.

At this point, the library exists but does not contain any collections and therefore won't be of much use to anyone.

Adding Collections

Navigate to System Configuration → Collections and click Create new collection. You will prompted to enter details that will be used to source the data for the collection. A good starting point, for testing purposes, is to use an open access OPDS feed as a data source. The Open Bookshelf is a good example of such a feed. Enter the following details:

Note that we associate the collection with the Hazelnut Peak library by selecting it in the Libraries drop-down. A collection can be associated with any number of libraries.

Importing

At this point, we have a library named Hazelnut Peak configured to use the Palace Bookshelf collection we created. It's now necessary to tell the application to start importing books from the OPDS feed. When the application is running inside a Docker image, the image is typically configured to execute various import operations on a regular schedule using cron. Because we're running the application from the command-line for development purposes, we need to execute these operations ourselves manually. In this particular case, we need to execute the opds_import_monitor:

(circ) $ ./bin/opds_import_monitor
{"host": "hazelnut",
 "app": "simplified",
 "name": "OPDS Import Monitor",
 "level": "INFO",
 "filename": "opds_import.py",
 "message": "[Palace Bookshelf] Following next link: http://openbookshelf.dp.la/lists/Open%20Bookshelf/crawlable",
 "timestamp": "2022-01-17T11:52:35.839978+00:00"}
...

The command will cause the application to crawl the configured OPDS feed and import every book in it. At the time of writing, this command will take around an hour to run the first time it is executed, but subsequent executions complete in seconds. Please wait for the import to complete before continuing!

When the import has completed, the books are imported but no OPDS feeds will have been generated, and no search service has been configured.

Configuring Search

Navigate to System Configuration → Search and add a new search configuration. The required URL is the URL of the Opensearch instance configured earlier:

Generating Search Indices

As with the collection configured earlier, the application depends upon various operations being executed on a regular schedule to generate search indices. Because we're running the application from the local command-line, we need to execute those operations manually:

./bin/search_index_clear
./bin/search_index_refresh

Neither of the commands will produce any output if the operations succeed.

Generating OPDS Feeds

When the collection has finished importing, we are required to generate OPDS feeds. Again, this operation is configured to execute on a regular schedule in the Docker image, but we'll need to execute it manually here:

./bin/opds_entry_coverage

The command will produce output indicating any errors.

Navigating to http://localhost:6500/ should show an OPDS feed containing various books:

Troubleshooting

The ./bin/repair/where_are_my_books command can produce output that may indicate why books are not appearing in OPDS feeds. A working, correctly configured installation, at the time of writing, produces output such as this:

(circ) $ ./bin/repair/where_are_my_books
Checking library Hazelnut Peak
 Associated with collection Palace Bookshelf.
 Associated with 171 lanes.

0 feeds in cachedfeeds table, not counting grouped feeds.

Examining collection "Palace Bookshelf"
 7838 presentation-ready works.
 0 works not presentation-ready.
 7824 works in the search index, expected around 7838.

We can see from the above output that the vast majority of the books in the Open Bookshelf collection were indexed correctly.

Sitewide Settings

Some settings have been provided in the admin UI that configure or toggle various functions of the Circulation Manager. These can be found at /admin/web/config/SitewideSettings in the admin interface.

Push Notification Status

This setting is a toggle that may be used to turn on or off the ability for the the system to send the Loan and Hold reminders to the mobile applications.

Scheduled Jobs

All jobs are scheduled via cron, as specified in the docker/services/simplified_crontab file. This includes all the import and reaper jobs, as well as other necessary background tasks, such as maintaining the search index and feed caches.

Job Requirements

hold_notifications

Requires one of the Firebase Cloud Messaging credentials environment variables (described above) to be present and non-empty. In addition, the site-wide PUSH_NOTIFICATIONS_STATUS setting must be either unset or true.

loan_notifications

Requires one of the Firebase Cloud Messaging credentials environment variables (described above to be present and non-empty. In addition, the site-wide PUSH_NOTIFICATIONS_STATUS setting must be either unset or true.

Code Style

Code style on this project is linted using pre-commit. This python application is included in our pyproject.toml file, so if you have the applications requirements installed it should be available. pre-commit is run automatically on each push and PR by our CI System.

You can run it manually on all files with the command: pre-commit run --all-files.

You can also set it up, so that it runs automatically for you on each commit. Running the command pre-commit install will install the pre-commit script in your local repositories git hooks folder, so that pre-commit is run automatically on each commit.

Configuration

The pre-commit configuration file is named .pre-commit-config.yaml. This file configures the different lints that pre-commit runs.

Linters

Built in

Pre-commit ships with a number of lints out of the box, we are configured to use:

• trailing-whitespace - trims trailing whitespace.
• end-of-file-fixer - ensures that a file is either empty, or ends with one newline.
• check-yaml - checks yaml files for parseable syntax.
• check-json - checks json files for parseable syntax.
• check-ast - simply checks whether the files parse as valid python.
• check-shebang-scripts-are-executable - ensures that (non-binary) files with a shebang are executable.
• check-executables-have-shebangs - ensures that (non-binary) executables have a shebang.
• check-merge-conflict - checks for files that contain merge conflict strings.
• check-added-large-files - prevents giant files from being committed.
• mixed-line-ending - replaces or checks mixed line ending.

Black

We lint using the black code formatter, so that all of our code is formatted consistently.

isort

We lint to make sure our imports are sorted and correctly formatted using isort. Our isort configuration is stored in our tox.ini which isort automatically detects.

autoflake

We lint using autoflake to flag and remove any unused import statement. If an unused import is needed for some reason it can be ignored with a #noqa comment in the code.

Continuous Integration

This project runs all the unit tests through Github Actions for new pull requests and when merging into the default main branch. The relevant file can be found in .github/workflows/test-build.yml. When contributing updates or fixes, it's required for the test Github Action to pass for all Python 3 environments. Run the tox command locally before pushing changes to make sure you find any failing tests before committing them.

For each push to a branch, CI also creates a docker image for the code in the branch. These images can be used for testing the branch, or deploying hotfixes.

To install the tools used by CI run:

poetry install --only ci

Testing

The Github Actions CI service runs the unit tests against Python 3.10, and 3.11 automatically using tox.

Tox has an environment for each python version, the module being tested, and an optional -docker factor that will automatically use docker to deploy service containers used for the tests. You can select the environment you would like to test with the tox -e flag.

Factors

When running tox without an environment specified, it tests circulation and core using all supported Python versions with service dependencies running in docker containers.

Python version

Factor	Python Version
py310	Python 3.10
py311	Python 3.11

All of these environments are tested by default when running tox. To test one specific environment you can use the -e flag.

Test Python 3.8

tox -e py38

You need to have the Python versions you are testing against installed on your local system. tox searches the system for installed Python versions, but does not install new Python versions. If tox doesn't find the Python version its looking for it will give an InterpreterNotFound errror.

Pyenv is a useful tool to install multiple Python versions, if you need to install missing Python versions in your system for local testing.

Module

Factor	Module
core	core tests
api	api tests

Docker

If you install tox-docker tox will take care of setting up all the service containers necessary to run the unit tests and pass the correct environment variables to configure the tests to use these services. Using tox-docker is not required, but it is the recommended way to run the tests locally, since it runs the tests in the same way they are run on the Github Actions CI server. tox-docker is automatically included when installing the ci dependency group.

The docker functionality is included in a docker factor that can be added to the environment. To run an environment with a particular factor you add it to the end of the environment.

Test with Python 3.8 using docker containers for the services.

tox -e "py38-{api,core}-docker"

Local services

If you already have elastic search or postgres running locally, you can run them instead by setting the following environment variables:

• SIMPLIFIED_TEST_DATABASE
• PALACE_TEST_SEARCH_URL

Make sure the ports and usernames are updated to reflect the local configuration.

# Set environment variables
export SIMPLIFIED_TEST_DATABASE="postgresql://simplified_test:test@localhost:9005/simplified_circulation_test"
export SIMPLIFIED_TEST_OPENSEARCH="http://localhost:9200"

# Run tox
tox -e "py38-{api,core}"

Override pytest arguments

If you wish to pass additional arguments to pytest you can do so through tox. Every argument passed after a -- to the tox command line will the passed to pytest, overriding the default.

Only run the test_google_analytics_provider tests with Python 3.8 using docker.

tox -e "py38-api-docker" -- tests/api/test_google_analytics_provider.py

Coverage Reports

Code coverage is automatically tracked with pytest-cov when tests are run. When the tests are run with github actions, the coverage report is automatically uploaded to codecov and the results are added to the relevant pull request.

When running locally, the results from each individual run can be collected and combined into an HTML report using the report tox environment. This can be run on its own after running the tests, or as part of the tox environment selection.

# Run core and api tests under Python 3.8, using docker
# containers for dependencies, and generate code coverage report
tox -e "py38-{core,api}-docker,report"

Usage with Docker

Check out the Docker README in the /docker directory for in-depth information on optionally running and developing the Circulation Manager locally with Docker, or for deploying the Circulation Manager with Docker.

Performance Profiling

There are three different profilers included to help measure the performance of the application. They can each be enabled by setting environment variables while starting the application.

AWS XRay

Environment Variables

• PALACE_XRAY: Set to enable X-Ray tracing on the application.
• PALACE_XRAY_NAME: The name of the service shown in x-ray for these traces.
• PALACE_XRAY_ANNOTATE_: Any environment variable starting with this prefix will be added to to the trace as an annotation.
  • For example setting PALACE_XRAY_ANNOTATE_KEY=value will set the annotation key=value on all xray traces sent from the application.
• PALACE_XRAY_INCLUDE_BARCODE: If this environment variable is set to true then the tracing code will try to include the patrons barcode in the user parameter of the trace, if a barcode is available.

Additional environment variables are provided by the X-Ray Python SDK.

cProfile

This profiler uses the werkzeug ProfilerMiddleware to profile the code. This uses the cProfile module under the hood to do the profiling.

Environment Variables

• PALACE_CPROFILE: Profiling will the enabled if this variable is set. The saved profile data will be available at path specified in the environment variable.
• The profile data will have the extension .prof.
• The data can be accessed using the pstats.Stats class.
• Example code to print details of the gathered statistics:

  import os
  from pathlib import Path
  from pstats import SortKey, Stats
  
  path = Path(os.environ.get("PALACE_CPROFILE"))
  for file in path.glob("*.prof"):
      stats = Stats(str(file))
      stats.sort_stats(SortKey.CUMULATIVE, SortKey.CALLS)
      stats.print_stats()

PyInstrument

This profiler uses PyInstrument to profile the code.

Profiling tests suite

PyInstrument can also be used to profile the test suite. This can be useful to identify slow tests, or to identify performance regressions.

To profile the core test suite, run the following command:

pyinstrument -m pytest --no-cov tests/core/

To profile the API test suite, run the following command:

pyinstrument -m pytest --no-cov tests/api/

Environment Variables

• PALACE_PYINSTRUMENT: Profiling will the enabled if this variable is set. The saved profile data will be available at path specified in the environment variable.
  • The profile data will have the extension .pyisession.
  • The data can be accessed with the pyinstrument.session.Session class.
  • Example code to print details of the gathered statistics:

    import os
    from pathlib import Path
    
    from pyinstrument.renderers import HTMLRenderer
    from pyinstrument.session import Session
    
    path = Path(os.environ.get("PALACE_PYINSTRUMENT"))
    for file in path.glob("*.pyisession"):
        session = Session.load(file)
        renderer = HTMLRenderer()
        renderer.open_in_browser(session)

Other Environment Variables

• SIMPLIFIED_SIRSI_DYNIX_APP_ID: The Application ID for the SirsiDynix Authentication API (optional)