Files
artcraft/codegen_database.sh
T

145 lines
5.4 KiB
Bash
Executable File

#!/bin/bash
# This is complicated since we're now in a more complex Rust workspace with
# multiple projects and targets.
# NB(2022-01-31): For some reason, 'storyteller-web', despite importing the obs queries
# via the `database_queries` shared library, started to exclude these queries. It was
# working, but mysteriously stopped. I'm using this more complicated setup with jq-based
# json merge as a stop gap.
# NB: This format can be used without changing directory:
# SQLX_OFFLINE=true cargo sqlx prepare -- --bin storyteller-web --manifest-path crates/service/storyteller_web/Cargo.toml
set -euxo pipefail
# NB: Database used by desktop app for AiChatBot sidecar.
# Make sure it has been migrated (use aichatbot-sidecar app directory as root).
SQLITE_DATABASE_PATH=./runtime_data/database.db
# NB: OpenSSL is broken on Ubuntu 22.04 (this is documented elsewhere)
# See https://askubuntu.com/a/1403961
# For some reason, the choice of "rustls" or the vendored openssl is not persistent between calls.
#cargo install sqlx-cli --features openssl-vendored,mysql
build_shared_mysql_database_library() {
# NB: For now, this is our monolithic DB that serves all of our microservices (gross)
# It's a single package so that queries can be shared (again, gross)
pushd crates/lib/database_queries
SQLX_OFFLINE=true cargo sqlx prepare --merged
popd
}
build_shared_sqlite_database_library() {
# NB(1): These are *SQLite* queries for the desktop AiChatBot app.
# NB(2): Make sure to run migrations first.
# NB(3): SqLite has trouble with relative paths, so we copy the file to /
cp $SQLITE_DATABASE_PATH /tmp
pushd crates/lib/sqlite_queries
DATABASE_URL=sqlite:///tmp/database.db cargo sqlx prepare --merged
popd
}
build_storyteller_web_app() {
# NB: This imports the inference/upload job queries
# It should also import the shared database lib queries, but something(???) broke.
pushd crates/service/web/storyteller_web
SQLX_OFFLINE=true cargo sqlx prepare --merged -- --bin storyteller-web # NB: Because multiple binary targets
popd
}
build_download_job() {
# NB: This imports the inference/upload job queries
# It should also import the shared database lib queries, but something(???) broke.
pushd crates/service/job/download_job
SQLX_OFFLINE=true cargo sqlx prepare --merged -- --bin download-job
popd
}
build_tts_download_job() {
# NB: This imports the inference/upload job queries
# It should also import the shared database lib queries, but something(???) broke.
pushd crates/service/job/tts_download_job
SQLX_OFFLINE=true cargo sqlx prepare --merged -- --bin tts-download-job
popd
}
build_inference_job() {
# NB: This imports the inference/upload job queries
# It should also import the shared database lib queries, but something(???) broke.
pushd crates/service/job/inference_job
SQLX_OFFLINE=true cargo sqlx prepare --merged -- --bin inference-job
popd
}
build_tts_inference_job() {
# NB: This imports the inference/upload job queries
# It should also import the shared database lib queries, but something(???) broke.
pushd crates/service/job/tts_inference_job
SQLX_OFFLINE=true cargo sqlx prepare --merged -- --bin tts-inference-job
popd
}
build_w2l_download_job() {
# NB: This imports the inference/upload job queries
# It should also import the shared database lib queries, but something(???) broke.
pushd crates/service/job/w2l_download_job
SQLX_OFFLINE=true cargo sqlx prepare --merged -- --bin w2l-download-job
popd
}
build_w2l_inference_job() {
# NB: This imports the inference/upload job queries
# It should also import the shared database lib queries, but something(???) broke.
pushd crates/service/job/w2l_inference_job
SQLX_OFFLINE=true cargo sqlx prepare --merged -- --bin w2l-inference-job
popd
}
combine_sqlx_queries() {
#jq: error: Could not open file crates/programs/aichatbot-sidecar/sqlx-data.json: No such file or directory
#jq: error: Could not open file crates/service/web/storyteller_web/sqlx-data.json: No such file or directory
# Merge multiple JSON files into a single dictionary.
# https://stackoverflow.com/a/24904276
jq -s '.[0] * .[1]' \
crates/lib/database_queries/sqlx-data.json \
crates/lib/sqlite_queries/sqlx-data.json \
crates/service/web/storyteller_web/sqlx-data.json \
crates/service/job/download_job/sqlx-data.json \
crates/service/job/tts_download_job/sqlx-data.json \
crates/service/job/inference_job/sqlx-data.json \
crates/service/job/tts_inference_job/sqlx-data.json \
crates/service/job/w2l_download_job/sqlx-data.json \
crates/service/job/w2l_inference_job/sqlx-data.json \
> sqlx-data.json
}
cleanup_temp_files() {
rm crates/lib/database_queries/sqlx-data.json \
crates/lib/sqlite_queries/sqlx-data.json \
crates/service/web/storyteller_web/sqlx-data.json \
crates/service/job/download_job/sqlx-data.json \
crates/service/job/tts_download_job/sqlx-data.json \
crates/service/job/inference_job/sqlx-data.json \
crates/service/job/tts_inference_job/sqlx-data.json \
crates/service/job/w2l_download_job/sqlx-data.json \
crates/service/job/w2l_inference_job/sqlx-data.json
}
build_shared_mysql_database_library
build_shared_sqlite_database_library
build_storyteller_web_app
build_download_job
build_tts_download_job
build_inference_job
build_tts_inference_job
build_w2l_download_job
build_w2l_inference_job
combine_sqlx_queries
cleanup_temp_files