e652a6ea8c
* Add elastic to our docker compose * add AND/OR/NOT search operations * add elastic and create article in elastic * change error code when elastic throws error * add search pages in elastic * add search by labels * Add elastic to GitHub Action * Update elastic version * Fix port for elastic * add url in search query * Set elastic features when running tests * add debug logs * Use localhost instead of service hostname * refresh elastic after create/update * update search labels query * add typescript support * search pages in elastic * fix search queries * use elastic for saving page * fix test failure * update getArticle api to use elastic * use generic get page function * add elastic migration python script * fix bulk helper param * save elastic page id in article_saving_request instead of postgres article_id * fix page archiving and deleting * add tests for deleteArticle * remove custom date type in elastic mappings which not exist in older version of elastic * fix timestamp format issue * add tests for save reading progress * add tests for save file * optimize search results * add alias to index * update migration script to receive env var as params * Add failing test to validate we don't decrease reading progress This test is failing with Elastic because we aren't fetching the reading progress from elastic here, and are fetching it from postgres. * Rename readingProgress to readingProgressPercent This is the name stored in elastic, so fixes issues pulling the value out. * Linting * Add failing test for creating highlights w/elastic This test fails because the highlight can't be looked up. Is there a different ID we should be passing in to query for highlights, or do we need to update the query to look for elastic_id? * add tests code coverage threshold * update nyc config * include more files in test coverage * change alias name * update updateContent to update pages in elastic * remove debug log * fix createhighlight test * search pages by alias in elastic * update set labels and delete labels in elastic * migration script enumeration * make BULK_SIZE an env var * fix pdf search indexing * debug github action exit issue * call pubsub when create/update/delete page in elastic * fix json parsing bug and reduce reading data from file * replace a depreciated pubsub api call * debug github action exit issue * debug github action exit issue * add handler to upload elastic page data to GCS * fix tests * Use http_auth instead of basic_auth * add index creation and existing postgres tables update in migration script * fix a typo to connect to elastic * rename readingProgress to readingProgressPercent * migrate elastic_page_id in highlights and article_saving_request tables * update migration script to include number of updated rows * update db migration query * read index mappings from file * fix upload pages to gcs * fix tests failure due to pageContext * fix upload file id not exist error * Handle savedAt & isArchived attributes w/out quering elastic * Fix prettier issues * fix content-type mismatching * revert pageId to linkId because frontend was not deployed yet * fix newsletters and attachment not saved in elastic * put linkId in article for setting labels * exclude orginalHtml in the result of searching to improve performace * exclude content in the result of searching to improve performace * remove score sorting * do not refresh immediately to reduce searching and indexing time * do not replace the backup data in gcs * fix no article id defined in articleSavingRequest * add logging of elastic api running time * reduce home feed pagination size to 15 * reduce home feed pagination size to 10 * stop revalidating first page * do not use a separate api to fetch reading progress * Remove unused comment * get reading progress if not exists * replace ngram tokenizer with standard tokenizer * fix tests * remove .env.local * add sort keyword in searching to sort by score Co-authored-by: Hongbo Wu <hongbo@omnivore.app>
Backend
This workspace is dedicated to API server that uses Apollo GraphQL server and Knex query builder to provide the app with data operations.
GraphQL schema
GraphQL schema is located in schema.ts. In order to use new types, queries or mutations you need to declare them there and then run yarn gql-typegen from the application root to create the necessary typings in order to write GQL queries from the app.
Apollo resolvers and data sources
We make use of Apollo resolvers and data sources. Resolvers typically contain business logic and handling of user domain errors, while data sources are ideally simple atomic database operations.
Interacting with the database
All operations on the database must be wrapped in Knex transaction on a resolver layer. This ensures data integrity and safety with no side effects on a failed operations.
Because we make use of Row Level Security in the database, - all operations typically begin with assuming the role for which policies exist via omnivore.set_claims database function.
Image Proxy (optional for local dev)
Backend API server returns article image links using image proxy
(/imageproxy). You will need to set the env with var IMAGE_PROXY_URL to point
to a running instance of image proxy along with env var IMAGE_PROXY_SECRET. The
same secret env var ought to be passed as config to the running image proxy
service. You can also use the docker-compose-dev.yml file to bring up just the
image proxy service alone (w/ env var for secret specified in the compose file)
by running: docker-compose -f docker-compose-dev.yml up -d imageproxy.
When running locally, use the .env.local file to set up the env variables in your environment.
Set up the database
Refer the using locally section from db README.
Copy .env.example file to .env file:
cp .env.example .env
Run the app
yarn dev