NotionCommotion
-
Posts
2,446 -
Joined
-
Last visited
-
Days Won
10
Posts posted by NotionCommotion
-
-
I am working on a SQL backed document management API where often the documents will reside in one of several categories (i.e. subject, type, project which created it, etc).
I am looking into different ways to implement search functionality such as...
- Standard WHERE with LIKE and maybe regex.
- Full-Text queries (I happen to be using PostgreSQL)
- 3rd party Semantic search libraries such as https://github.com/neuml/txtai
- 3rd party machine learning libraries such as https://php-ml.readthedocs.io/en/latest/
- API calls to some 3rd party webservice.
- Other?
Is there any approach which is best for most applications? Or as I expect is the "best" approach based on the actual application requirements, and if so, can you please share your decision criteria?
-
Thanks, exactly what I was hoping for! Tried without the "from" but obviously didn't do what I wished for.
-
The following works, but seems like it has one more loop than it should. Better way to do so? I could get rid of the generator and merge to an array, but the generator seems like a better solution. Thanks
private function doSomething(File $file) { foreach($this->getTextFiles($file) as $textFile) { // ... } } private function getTextFiles(File $file):\Generator { foreach($file->getChildren() as $child) { if($child instanceof IsTextFile) { yield $child; } else { foreach($this->getTextFiles($child) as $grandChild) { yield $grandChild; } } } }
-
Thanks kicken, Yep, a stupid question and I wasn't thinking.
-
This is probably a stupid question.
I wish to get a collection of classes which exist in a given directory, and found this post which directed me to this class, but it is depreciated and appears that this class should now be used.
Great, I have Composer, and all I need to do is call Composer\Autoload\ClassMapGenerator::createMap($pathToDirectory)!
Or do I? The only files included in vendor/composer are below.
- autoload_classmap.php
- autoload_files.php
- autoload_namespaces.php
- autoload_psr4.php
- autoload_real.php
- autoload_static.php
- ClassLoader.php
- installed.json
- installed.php
- InstalledVersions.php
- LICENSE
- platform_check.php
I expect this is due to how I installed composer globally.
Should composer classes ever be used in an application other than for dependency management? Or should I just create my own class for this need?
-
43 minutes ago, kicken said:
If you need an actual file path, then tempnam() is the way to go. tmpfile is for if you need a temporary file resource, just a shortcut essentially for doing fopen('php://temp', 'w+');.
As an example, I have a html2pdf class that generates some temp files and calls an external program to generate the PDF. This is how that is coded (roughly).
public function generatePdf(){ $pdf = null; try { $source = tempnam(sys_get_temp_dir(), 'htmlpdf'); $destination = tempnam(sys_get_temp_dir(), 'htmlpdf'); file_put_contents($source, $this->html); $cmd = $this->createCommandLine($source, $destination); exec($cmd, $output, $ret); if ($ret !== 0){ throw new \RuntimeException('Failed to generate PDF with command [' . $cmd . '] Output: ' . implode(PHP_EOL, $output)); } $pdf = file_get_contents($destination); } finally { unlink($source); unlink($destination); } return $pdf; }
Generate the two temporary file locations, execute the command, then read the content out of the file. The temporary files are then cleaned up using a finally clause which will run whether the function when the function exits, whether that be by a successful return or by throwing an exception.
Thanks kicken, Was just trying to get out of the measly two unlink statements, and will stop trying.
-
1 hour ago, requinix said:
Why do it that way since all you're doing is reinventing tempnam?
Yes, but you missed the important part: "when there are no remaining references to the file handle". Which is exactly the case when you use the handle, call stream_get_meta_data on it, and then forget the handle existed.
I didn't think I was quite reinventing tempnam(). tempnam() doesn't delete the file when complete, correct? The example in the doc shows unlinking($tmpfname).
However, if calling stream_get_meta_data() deletes the file and I just create it back, and the PHP no longer thinks it is responsible to delete it, suppose I am reinventing.
-
I will be using exec() to execute an external program which requires a file output path and will create a file located at that path, and the file is only needed during the execution of the PHP script and should be deleted after the script ends.
Good or bad idea? If bad, why? Thanks
public function tmpFile():string { return stream_get_meta_data(tmpfile())['uri']; }
EDIT - Does "when the script ends" happen about the same time as the class's destructor?QuoteThe file is automatically removed when closed (for example, by calling fclose(), or when there are no remaining references to the file handle returned by tmpfile()), or when the script ends.
Caution
If the script terminates unexpectedly, the temporary file may not be deleted.
-
Thanks maxxd, I'll checkout the facade object/pattern, but likely won't use it for this use.
Thanks kicken, In hindsight, I agree chaining doesn't bring much value for this case. "If it did" (which it doesn't), guess I could just add another traditional method which would call the static method and directly return $this (but I won't).
-
Is it possible to return the instance of an object when a static method is called on it? For instance, (new SomeClass)->echo() returns $this and SomeClass::echo() return a newly created object? Thinking I could do so by making properties also static and then creating the new object and populating it (albeit it will be a clone and not the same object). Thanks
<?php class SomeClass { private $services = []; private $paramaters = []; public function Foo():self { echo(__METHOD__); return $this; } public static function echo(string $msg, array $data=[], bool $syslog = false):self|static { echo($msg.($data?(': data: '.json_encode($data)):'').PHP_EOL); return $syslog?self::syslog($msg, $data):$this->getSelfOrStatic(); } public static function syslog(string $msg, array $data=[], bool $echo = false):self|static { syslog(LOG_INFO, $msg.($data?(': data: '.json_encode($data)):'')); return $echo?self::echo($msg, $data):$this->getSelfOrStatic(); } private static function getSelfOrStatic():self|static { return isset($this)?$this:new static(); } }
-
Thanks maxxd, If I go through your configuration, would you be willing to answer some specific questions in order to confirm my understanding? If so, think it is best to fully understand the Dockerfile before moving to the docker-compose.yml file, or the other way around?
-
Super/sub type DB modelling provides a more structured approach for inheritance and centralizes the conditional statement logic and IMO provides polymorphism, however, be careful as if taken too far, will likely paint you into a corner that you didn't want to be in, so use sparingly when the benefits outweigh the risks.
-
I mostly figured out the SQL error. Appears this was being called by Dockerfile which added a table called user which I had been using, and I didn't realize it was called when the image was being created. Maybe got some technicalities wrong, but changing it got rid of the SQL errors.
# api/docker/php/docker-entrypoint.sh #!/bin/sh # bla bla bla... if [ "$( find ./migrations -iname '*.php' -print -quit )" ]; then php bin/console doctrine:migrations:migrate --no-interaction fi if [ "$APP_ENV" != 'prod' ]; then echo "Load fixtures" bin/console hautelook:fixtures:load --no-interaction fi fi fi exec docker-php-entrypoint "$@"
Still don't understand the composer part and the not found classes. For instance, I see stof/doctrine-extensions-bundle (v1.7.1) first being installed, but then I get Uncaught Error: Class "Stof\DoctrineExtensionsBundle\StofDoctrineExtensionsBundle" not found error. The only place I see that class being used is in bundles.php, and suppose I could comment it out like I did for libphonenumber\PhoneNumberUtil, but then suspect some other issue will pop up. Not sure where to go next on this... EDIT. Maybe I got lucky! Commented it out and at least PHP is not crashing. Still need to better understand how all this stuff works...
facdocs-consumer-1 exited with code 255 facdocs-php-1 | - Installing behat/transliterator (v1.5.0): Extracting archive facdocs-php-1 | - Installing brick/money (0.8.0): Extracting archive facdocs-php-1 | - Installing doctrine/doctrine-fixtures-bundle (3.4.2): Extracting archive facdocs-php-1 | - Installing fig/link-util (1.2.0): Extracting archive facdocs-php-1 | - Installing gesdinet/jwt-refresh-token-bundle (v1.1.1): Extracting archive facdocs-php-1 | - Installing giggsey/locale (2.3): Extracting archive facdocs-php-1 | - Installing laminas/laminas-code (4.8.0): Extracting archive facdocs-php-1 | - Installing symfony/translation (v6.1.11): Extracting archive facdocs-php-1 | - Installing nesbot/carbon (2.66.0): Extracting archive facdocs-php-1 | - Installing nilportugues/sql-query-formatter (v1.2.2): Extracting archive facdocs-php-1 | - Installing notion-commotion/attribute-validator (1.01): Extracting archive facdocs-php-1 | - Installing notion-commotion/attribute-validator-command (1.06): Extracting archive facdocs-php-1 | - Installing symfony/intl (v6.1.11): Extracting archive facdocs-php-1 | - Installing giggsey/libphonenumber-for-php (8.13.6): Extracting archive facdocs-php-1 | - Installing odolbeau/phone-number-bundle (v3.9.1): Extracting archive facdocs-php-1 | - Installing phpstan/phpdoc-parser (1.16.1): Extracting archive facdocs-php-1 | - Installing sebastian/version (3.0.2): Extracting archive facdocs-php-1 | - Installing sebastian/type (3.2.1): Extracting archive facdocs-php-1 | - Installing sebastian/resource-operations (3.0.3): Extracting archive facdocs-php-1 | - Installing sebastian/object-reflector (2.0.4): Extracting archive facdocs-php-1 | - Installing sebastian/object-enumerator (4.0.4): Extracting archive facdocs-php-1 | - Installing sebastian/global-state (5.0.5): Extracting archive facdocs-php-1 | - Installing sebastian/environment (5.1.5): Extracting archive facdocs-php-1 | - Installing sebastian/code-unit (1.0.8): Extracting archive facdocs-php-1 | - Installing sebastian/cli-parser (1.0.1): Extracting archive facdocs-php-1 | - Installing phpunit/php-timer (5.0.3): Extracting archive facdocs-php-1 | - Installing phpunit/php-text-template (2.0.4): Extracting archive facdocs-php-1 | - Installing phpunit/php-invoker (3.1.1): Extracting archive facdocs-php-1 | - Installing phpunit/php-file-iterator (3.0.6): Extracting archive facdocs-php-1 | - Installing theseer/tokenizer (1.2.1): Extracting archive facdocs-php-1 | - Installing sebastian/lines-of-code (1.0.3): Extracting archive facdocs-php-1 | - Installing sebastian/complexity (2.0.2): Extracting archive facdocs-php-1 | - Installing sebastian/code-unit-reverse-lookup (2.0.3): Extracting archive facdocs-php-1 | - Installing phpunit/php-code-coverage (9.2.24): Extracting archive facdocs-php-1 | - Installing phar-io/version (3.2.1): Extracting archive facdocs-php-1 | - Installing phar-io/manifest (2.0.3): Extracting archive facdocs-php-1 | - Installing phpunit/phpunit (9.6.3): Extracting archive facdocs-php-1 | - Installing phpstan/phpstan (1.10.2): Extracting archive facdocs-php-1 | - Installing rector/rector (0.15.18): Extracting archive facdocs-php-1 | - Installing gedmo/doctrine-extensions (v3.11.1): Extracting archive facdocs-php-1 | - Installing stof/doctrine-extensions-bundle (v1.7.1): Extracting archive facdocs-php-1 | - Installing symfony/polyfill-intl-idn (v1.27.0): Extracting archive facdocs-php-1 | - Installing symfony/mime (v6.1.11): Extracting archive facdocs-php-1 | - Installing egulias/email-validator (4.0.1): Extracting archive facdocs-php-1 | - Installing symfony/mailer (v6.1.11): Extracting archive facdocs-php-1 | - Installing symfony/polyfill-intl-icu (v1.27.0): Extracting archive facdocs-php-1 | - Installing friendsofphp/proxy-manager-lts (v1.0.14): Extracting archive facdocs-php-1 | - Installing symfony/proxy-manager-bridge (v6.1.11): Extracting archive facdocs-php-1 | - Installing symfony/polyfill-uuid (v1.27.0): Extracting archive facdocs-php-1 | - Installing symfony/uid (v6.1.11): Extracting archive facdocs-php-1 | - Installing symfony/templating (v6.1.11): Extracting archive facdocs-php-1 | - Installing symfony/form (v6.1.11): Extracting archive facdocs-php-1 | - Installing moneyphp/money (v3.3.3): Extracting archive facdocs-php-1 | - Installing tbbc/money-bundle (5.0.1): Extracting archive facdocs-php-1 | Package webmozart/path-util is abandoned, you should avoid using it. Use symfony/filesystem instead. facdocs-php-1 | Generating optimized autoload files facdocs-consumer-1 | 2023-02-24T18:05:33+00:00 [critical] Uncaught Error: Class "Stof\DoctrineExtensionsBundle\StofDoctrineExtensionsBundle" not found facdocs-consumer-1 | Symfony\Component\ErrorHandler\Error\ClassNotFoundError {#69 facdocs-consumer-1 | #message: """ facdocs-consumer-1 | Attempted to load class "StofDoctrineExtensionsBundle" from namespace "Stof\DoctrineExtensionsBundle".\n facdocs-consumer-1 | Did you forget a "use" statement for another namespace? facdocs-consumer-1 | """ facdocs-consumer-1 | #code: 0 facdocs-consumer-1 | #file: "./vendor/symfony/framework-bundle/Kernel/MicroKernelTrait.php" facdocs-consumer-1 | #line: 135 facdocs-consumer-1 | trace: { facdocs-consumer-1 | ./vendor/symfony/framework-bundle/Kernel/MicroKernelTrait.php:135 { …} facdocs-consumer-1 | ./vendor/symfony/http-kernel/Kernel.php:382 { …} facdocs-consumer-1 | ./vendor/symfony/http-kernel/Kernel.php:766 { …} facdocs-consumer-1 | ./vendor/symfony/http-kernel/Kernel.php:128 { …} facdocs-consumer-1 | ./vendor/symfony/framework-bundle/Console/Application.php:166 { …} facdocs-consumer-1 | ./vendor/symfony/framework-bundle/Console/Application.php:72 { …} facdocs-consumer-1 | ./vendor/symfony/console/Application.php:171 { …} facdocs-consumer-1 | ./vendor/symfony/runtime/Runner/Symfony/ConsoleApplicationRunner.php:54 { …} facdocs-consumer-1 | ./vendor/autoload_runtime.php:29 { …} facdocs-consumer-1 | ./bin/console:11 { facdocs-consumer-1 | › facdocs-consumer-1 | › require_once dirname(__DIR__).'/vendor/autoload_runtime.php'; facdocs-consumer-1 | › facdocs-consumer-1 | arguments: { facdocs-consumer-1 | "/srv/app/vendor/autoload_runtime.php" facdocs-consumer-1 | } facdocs-consumer-1 | } facdocs-consumer-1 | } facdocs-consumer-1 | } facdocs-consumer-1 exited with code 255
-
Thanks maxxd! Not quite production or even close, unfortunately... Current objective is to provide a workableness system to get frontend peps to better understand what they need to do on their end. I probably shouldn't be going down this dockers path as it seems like a blackhole to me, but the main reasons I am doing so is it allows me to better show the frontend people what they need to interface to by "automatically" creating a super-basic client app. I somewhat got things working, but then restarted the docker, and had SQL problems... I know it can't be magic, but not being able to identify what is initiating the script which is causing me grief makes me kind of hate dockers. I've made some progress and see that it is calling a PHP script to initiate things, but even after I comment out what I think are the culprits, still issues... Will try again tomorrow!
-
I ended up leaving the folder which are mounted to allow external editing alone and just updated the other files. At first, all seemed good, but then something went wrong.
I start the docker using docker compose up, and then I instantly start getting errors like the following. So, I expect that maybe autoload paths need to be regenerated or I have a issue with my composer.json file. Problem is I can update composer as the docker keeps on restarting, and am having a difficult time troubleshooting. I suspect that the docker-compose (and not the DockerFile, right?) file is initiating running some pre-check script and hoping I can temporarily bypass it, but it what initiates it escapes me. Any thoughts? Thank you
Cannot autowire service "App\Command\AddSystemUser": argument "$phoneNumberUtil" of method "App\Command\AbstractAddUser::__construct()" references class "libphonenumber\PhoneNumberUtil" but no such service exists.
PS. In addition to docker-compose.yml provided above, I also have docker-compose.override.yml.
version: "3.4" # Development environment override services: php: build: target: app_php_dev volumes: - ./api:/srv/app - ./api/docker/php/conf.d/app.dev.ini:/usr/local/etc/php/conf.d/app.dev.ini:ro # If you develop on Mac or Windows you can remove the vendor/ directory # from the bind-mount for better performance by enabling the next line: #- /srv/app/vendor environment: # See https://xdebug.org/docs/all_settings#mode XDEBUG_MODE: "${XDEBUG_MODE:-off}" extra_hosts: # Ensure that host.docker.internal is correctly defined on Linux - host.docker.internal:host-gateway consumer: build: target: app_php_dev volumes: - ./api:/srv/app - ./api/docker/php/conf.d/app.dev.ini:/usr/local/etc/php/conf.d/app.dev.ini:ro # If you develop on Mac or Windows you can remove the vendor/ directory # from the bind-mount for better performance by enabling the next line: #- /srv/app/vendor environment: # See https://xdebug.org/docs/all_settings#mode XDEBUG_MODE: "${XDEBUG_MODE:-off}" extra_hosts: # Ensure that host.docker.internal is correctly defined on Linux - host.docker.internal:host-gateway pwa: build: context: ./pwa target: dev volumes: - ./pwa:/srv/app environment: API_PLATFORM_CREATE_CLIENT_ENTRYPOINT: http://caddy API_PLATFORM_CREATE_CLIENT_OUTPUT: . caddy: volumes: - ./api/public:/srv/app/public:ro - ./api/docker/caddy/Caddyfile:/etc/caddy/Caddyfile:ro environment: MERCURE_EXTRA_DIRECTIVES: demo ###> doctrine/doctrine-bundle ### database: ports: - target: 5432 published: 5432 protocol: tcp ###< doctrine/doctrine-bundle ### ###> symfony/mercure-bundle ### ###< symfony/mercure-bundle ### -
8 hours ago, requinix said:
Whose classes? Are there maybe autoload paths that need to be regenerated?
Vendor classes. Maybe regenerated. How does one regenerate within a docker?
1 hour ago, maxxd said:What's your docker-compose file look like? We need to see that in order to say if it's an issue with the volume mounts.
docker-compose.yml
version: "3.4" services: php: build: context: ./api target: app_php depends_on: - database restart: unless-stopped volumes: - php_socket:/var/run/php healthcheck: interval: 10s timeout: 3s retries: 3 start_period: 30s environment: DATABASE_URL: postgresql://${POSTGRES_USER:-app}:${POSTGRES_PASSWORD:-!ChangeMe!}@database:5432/${POSTGRES_DB:-app}?serverVersion=${POSTGRES_VERSION:-14} TRUSTED_PROXIES: ${TRUSTED_PROXIES:-127.0.0.0/8,10.0.0.0/8,172.16.0.0/12,192.168.0.0/16} TRUSTED_HOSTS: ^${SERVER_NAME:-example\.com|localhost}|caddy$$ MERCURE_URL: ${CADDY_MERCURE_URL:-http://caddy/.well-known/mercure} MERCURE_PUBLIC_URL: https://${SERVER_NAME:-localhost}/.well-known/mercure MERCURE_JWT_SECRET: ${CADDY_MERCURE_JWT_SECRET:-!ChangeThisMercureHubJWTSecretKey!} consumer: build: context: ./api target: app_php entrypoint: docker-php-entrypoint command: bin/console messenger:consume depends_on: - database restart: unless-stopped healthcheck: test: ['CMD', 'ps', 'aux', '|', 'egrep', '"\d+:\d+ php bin/console messenger:consume"'] interval: 10s timeout: 3s retries: 3 start_period: 30s environment: DATABASE_URL: postgresql://${POSTGRES_USER:-app}:${POSTGRES_PASSWORD:-!ChangeMe!}@database:5432/${POSTGRES_DB:-app}?serverVersion=${POSTGRES_VERSION:-14} TRUSTED_PROXIES: ${TRUSTED_PROXIES:-127.0.0.0/8,10.0.0.0/8,172.16.0.0/12,192.168.0.0/16} TRUSTED_HOSTS: ^${SERVER_NAME:-example\.com|localhost}|caddy$$ MERCURE_URL: ${CADDY_MERCURE_URL:-http://caddy/.well-known/mercure} MERCURE_PUBLIC_URL: https://${SERVER_NAME:-localhost}/.well-known/mercure MERCURE_JWT_SECRET: ${CADDY_MERCURE_JWT_SECRET:-!ChangeThisMercureHubJWTSecretKey!} pwa: build: context: ./pwa target: prod environment: NEXT_PUBLIC_ENTRYPOINT: http://caddy caddy: build: context: api/ target: app_caddy depends_on: - php - pwa environment: PWA_UPSTREAM: pwa:3000 SERVER_NAME: ${SERVER_NAME:-localhost}, caddy:80 MERCURE_PUBLISHER_JWT_KEY: ${CADDY_MERCURE_JWT_SECRET:-!ChangeThisMercureHubJWTSecretKey!} MERCURE_SUBSCRIBER_JWT_KEY: ${CADDY_MERCURE_JWT_SECRET:-!ChangeThisMercureHubJWTSecretKey!} MERCURE_EXTRA_DIRECTIVES: demo restart: unless-stopped volumes: - php_socket:/var/run/php - caddy_data:/data - caddy_config:/config ports: # HTTP - target: 80 published: ${HTTP_PORT:-80} protocol: tcp # HTTPS - target: 443 published: ${HTTPS_PORT:-443} protocol: tcp # HTTP/3 - target: 443 published: ${HTTP3_PORT:-443} protocol: udp ###> doctrine/doctrine-bundle ### database: image: postgres:${POSTGRES_VERSION:-14}-alpine environment: - POSTGRES_DB=${POSTGRES_DB:-app} # You should definitely change the password in production - POSTGRES_PASSWORD=${POSTGRES_PASSWORD:-!ChangeMe!} - POSTGRES_USER=${POSTGRES_USER:-app} volumes: - db_data:/var/lib/postgresql/data # you may use a bind-mounted host directory instead, so that it is harder to accidentally remove the volume and lose all your data! # - ./api/docker/db/data:/var/lib/postgresql/data ###< doctrine/doctrine-bundle ### # Mercure is installed as a Caddy module, prevent the Flex recipe from installing another service ###> symfony/mercure-bundle ### ###< symfony/mercure-bundle ### volumes: php_socket: caddy_data: caddy_config: ###> doctrine/doctrine-bundle ### db_data: ###< doctrine/doctrine-bundle ### ###> symfony/mercure-bundle ### ###< symfony/mercure-bundle ###
api/Dockerfile#syntax=docker/dockerfile:1.4 # Adapted from https://github.com/dunglas/symfony-docker # Prod image FROM php:8.1-fpm-alpine AS app_php ENV APP_ENV=prod WORKDIR /srv/app # persistent / runtime deps RUN apk add --no-cache \ acl \ fcgi \ file \ gettext \ git \ gnu-libiconv \ ; # install gnu-libiconv and set LD_PRELOAD env to make iconv work fully on Alpine image. # see https://github.com/docker-library/php/issues/240#issuecomment-763112749 ENV LD_PRELOAD /usr/lib/preloadable_libiconv.so RUN set -eux; \ apk update; \ apk upgrade -U -a --purge; \ apk add --no-cache --virtual .build-deps \ $PHPIZE_DEPS \ icu-data-full \ icu-dev \ libzip-dev \ zlib-dev \ ; \ \ docker-php-ext-configure zip; \ docker-php-ext-install -j$(nproc) \ intl \ zip \ ; \ pecl install \ apcu \ ; \ pecl clear-cache; \ docker-php-ext-enable \ apcu \ opcache \ ; \ \ runDeps="$( \ scanelf --needed --nobanner --format '%n#p' --recursive /usr/local/lib/php/extensions \ | tr ',' '\n' \ | sort -u \ | awk 'system("[ -e /usr/local/lib/" $1 " ]") == 0 { next } { print "so:" $1 }' \ )"; \ apk add --no-cache --virtual .app-phpexts-rundeps $runDeps; \ \ apk del .build-deps ###> recipes ### ###> doctrine/doctrine-bundle ### RUN apk add --no-cache --virtual .pgsql-deps postgresql-dev; \ docker-php-ext-install -j$(nproc) pdo_pgsql; \ apk add --no-cache --virtual .pgsql-rundeps so:libpq.so.5; \ apk del .pgsql-deps ###< doctrine/doctrine-bundle ### ###< recipes ### RUN mv "$PHP_INI_DIR/php.ini-production" "$PHP_INI_DIR/php.ini" COPY --link docker/php/conf.d/app.ini $PHP_INI_DIR/conf.d/ COPY --link docker/php/conf.d/app.prod.ini $PHP_INI_DIR/conf.d/ COPY --link docker/php/php-fpm.d/zz-docker.conf /usr/local/etc/php-fpm.d/zz-docker.conf RUN mkdir -p /var/run/php COPY --link docker/php/docker-healthcheck.sh /usr/local/bin/docker-healthcheck RUN chmod +x /usr/local/bin/docker-healthcheck HEALTHCHECK --interval=10s --timeout=3s --retries=3 CMD ["docker-healthcheck"] COPY --link docker/php/docker-entrypoint.sh /usr/local/bin/docker-entrypoint RUN chmod +x /usr/local/bin/docker-entrypoint ENTRYPOINT ["docker-entrypoint"] CMD ["php-fpm"] # https://getcomposer.org/doc/03-cli.md#composer-allow-superuser ENV COMPOSER_ALLOW_SUPERUSER=1 ENV PATH="${PATH}:/root/.composer/vendor/bin" COPY --from=composer:2 --link /usr/bin/composer /usr/bin/composer # prevent the reinstallation of vendors at every changes in the source code COPY composer.* symfony.* ./ RUN set -eux; \ composer install --prefer-dist --no-dev --no-autoloader --no-scripts --no-progress; \ composer clear-cache # copy sources COPY --link . . RUN rm -Rf docker/ RUN set -eux; \ mkdir -p var/cache var/log; \ composer dump-autoload --classmap-authoritative --no-dev; \ composer dump-env prod; \ composer run-script --no-dev post-install-cmd; \ chmod +x bin/console; sync VOLUME /srv/app/config/jwt/ # Dev image FROM app_php AS app_php_dev ENV APP_ENV=dev XDEBUG_MODE=off VOLUME /srv/app/var/ RUN rm $PHP_INI_DIR/conf.d/app.prod.ini; \ mv "$PHP_INI_DIR/php.ini" "$PHP_INI_DIR/php.ini-production"; \ mv "$PHP_INI_DIR/php.ini-development" "$PHP_INI_DIR/php.ini" COPY --link docker/php/conf.d/app.dev.ini $PHP_INI_DIR/conf.d/ RUN set -eux; \ apk add --no-cache --update --virtual .build-deps linux-headers $PHPIZE_DEPS; \ pecl install xdebug; \ docker-php-ext-enable xdebug; \ apk del .build-deps RUN rm -f .env.local.php # Build Caddy with the Mercure and Vulcain modules FROM caddy:2-builder-alpine AS app_caddy_builder RUN xcaddy build \ --with github.com/dunglas/mercure \ --with github.com/dunglas/mercure/caddy \ --with github.com/dunglas/vulcain \ --with github.com/dunglas/vulcain/caddy # Caddy image FROM caddy:2-alpine AS app_caddy WORKDIR /srv/app COPY --from=app_caddy_builder --link /usr/bin/caddy /usr/bin/caddy COPY --from=app_php --link /srv/app/public public/ COPY --link docker/caddy/Caddyfile /etc/caddy/Caddyfile
-
I have the following PHP project, and I DO care about keeping the git history:
michael@ubuntu-hp-laptop-original:/var/www/facdocs/api$ ls -la total 1076 drwxr-xr-x 18 michael michael 4096 Feb 11 07:52 . drwxr-xr-x 37 michael michael 4096 Feb 22 18:42 .. drwxrwxr-x 2 michael michael 4096 Sep 16 05:34 bin -rw-rw-r-- 1 michael michael 3814 Feb 15 05:38 composer.json -rw-rw-r-- 1 michael michael 477849 Feb 15 05:38 composer.lock -rw-rw-r-- 1 michael michael 470716 Sep 16 05:25 composer.lock2.tmp drwxr-xr-x 5 michael michael 4096 Feb 22 09:48 config lrwxrwxrwx 1 michael michael 7 Oct 20 04:53 core -> ../core -rw-rw-r-- 1 michael michael 2026 Feb 22 09:40 .env -rw-rw-r-- 1 michael michael 320 Nov 22 06:23 .env.test drwxrwxr-x 2 michael michael 4096 Dec 11 10:29 fixtures drwxrwxr-x 8 michael michael 4096 Feb 22 16:17 .git -rw-rw-r-- 1 michael michael 704 Dec 11 10:29 .gitignore drwxr-xr-x 2 michael michael 4096 Oct 30 15:23 index drwxrwxr-x 2 michael michael 4096 Feb 3 07:01 migrations lrwxrwxrwx 1 michael michael 8 Aug 6 2022 notes -> ../notes -rw-rw-r-- 1 michael michael 197 Sep 16 05:33 .php-cs-fixer.dist.php -rw-rw-r-- 1 michael michael 30546 Feb 3 16:59 .phpunit.result.cache -rw-rw-r-- 1 michael michael 1431 Dec 13 04:14 phpunit.xml.dist drwxrwxr-x 4 michael michael 4096 Feb 22 09:29 public -rw-rw-r-- 1 michael michael 2689 Jul 17 2022 rector.php drwxr-xr-x 3 michael michael 4096 Jul 12 2022 resources drwxr-xr-x 2 michael michael 4096 Oct 10 05:33 source-data drwxrwxr-x 30 michael michael 4096 Nov 9 07:53 src -rw-rw-r-- 1 michael michael 10924 Dec 11 10:29 symfony.lock drwxrwxr-x 2 michael michael 4096 Jun 18 2022 templates drwxrwxr-x 5 michael michael 4096 Feb 15 05:35 testing drwxrwxr-x 2 michael michael 4096 Feb 2 05:50 tests -rw-r--r-- 1 michael michael 1415 Jan 21 11:00 TODO.md drwxrwxr-x 2 michael michael 4096 Sep 16 05:33 translations drwxrwxrwx 6 facdocs michael 4096 Jun 19 2022 var drwxrwxr-x 49 michael michael 4096 Feb 15 05:38 vendor michael@ubuntu-hp-laptop-original:/var/www/facdocs/api$
I have another project which uses dockers and PHP as well as other stuff, and I don't care about keeping the git history:michael@apiplatform-HP-EliteBook-830-G6:~/demo$ ls -la total 64 drwxrwxr-x 7 michael michael 4096 Feb 17 08:35 . drwxr-x--- 26 michael michael 4096 Feb 22 19:08 .. drwxrwxr-x 13 michael michael 4096 Feb 17 09:48 api -rw-rw-r-- 1 michael michael 1859 Feb 17 08:35 docker-compose.override.yml -rw-rw-r-- 1 michael michael 396 Feb 17 08:35 docker-compose.prod.yml -rw-rw-r-- 1 michael michael 3796 Feb 17 08:35 docker-compose.yml -rw-rw-r-- 1 michael michael 1203 Feb 17 08:35 .editorconfig drwxrwxr-x 8 michael michael 4096 Feb 22 19:08 .git -rw-rw-r-- 1 michael michael 319 Feb 17 08:35 .gitattributes drwxrwxr-x 3 michael michael 4096 Feb 17 08:35 .github -rw-rw-r-- 1 michael michael 72 Feb 17 08:35 .gitignore drwxrwxr-x 3 michael michael 4096 Feb 17 08:35 helm -rw-rw-r-- 1 michael michael 1066 Feb 17 08:35 LICENSE drwxrwxr-x 12 michael michael 4096 Feb 22 19:06 pwa -rwxrwxr-x 1 michael michael 2739 Feb 17 08:35 README.md -rwxrwxr-x 1 michael michael 433 Feb 17 08:35 update-deps.sh michael@apiplatform-HP-EliteBook-830-G6:~/demo$ cd api michael@apiplatform-HP-EliteBook-830-G6:~/demo/api$ ls -la total 520 drwxrwxr-x 13 michael michael 4096 Feb 17 09:48 . drwxrwxr-x 7 michael michael 4096 Feb 17 08:35 .. drwxrwxr-x 2 michael michael 4096 Feb 17 08:35 bin -rw-rw-r-- 1 michael michael 3330 Feb 22 19:06 composer.json -rw-rw-r-- 1 michael michael 419938 Feb 22 19:05 composer.lock drwxrwxr-x 5 michael michael 4096 Feb 17 08:35 config drwxrwxr-x 4 michael michael 4096 Feb 17 08:35 docker -rw-rw-r-- 1 michael michael 3977 Feb 17 08:35 Dockerfile -rw-rw-r-- 1 michael michael 390 Feb 17 08:35 .dockerignore -rw-rw-r-- 1 michael michael 2811 Feb 17 09:48 .env -rw-rw-r-- 1 michael michael 215 Feb 17 09:48 .env.test drwxrwxr-x 2 michael michael 4096 Feb 17 08:35 fixtures -rw-rw-r-- 1 michael michael 517 Feb 17 09:04 .gitignore drwxrwxr-x 2 michael michael 4096 Feb 17 08:35 migrations -rw-rw-r-- 1 michael michael 197 Feb 17 08:35 .php-cs-fixer.dist.php -rw-rw-r-- 1 michael michael 1367 Feb 17 09:48 phpunit.xml.dist -rw-rw-r-- 1 michael michael 444 Feb 17 08:35 psalm.xml.dist drwxrwxr-x 3 michael michael 4096 Feb 17 08:35 public drwxrwxr-x 10 michael michael 4096 Feb 17 09:04 src -rw-rw-r-- 1 michael michael 16462 Feb 17 09:48 symfony.lock drwxrwxr-x 2 michael michael 4096 Feb 17 08:35 templates drwxrwxr-x 5 michael michael 4096 Feb 17 09:48 tests drwxr-xr-x 2 root root 4096 Feb 17 08:35 var drwxr-xr-x 42 root root 4096 Feb 22 19:05 vendor michael@apiplatform-HP-EliteBook-830-G6:~/demo/api$
I wish to replace the content in the api folder of the second project with the content in the first one, and either:- Keep the second project's git history and make the api folder a git submodule.
- Get rid of the second project's git history mov api/.git to . so I all the second project's files appear as new in the git history and I have only one git repo.
I thought this would be more straight forward, but repeatedly get PHP composer errors about classes not being found. I think it has to do with how some of the folders are mounted, but not positive. Note that I didn't just replace the content in the api folder, but actually replaced the folder, and maybe that is breaking things? And along those lines, maybe not replace app/var and app/vendor either?
Thank much
-
20 hours ago, kicken said:
To me, it sounds like that means your data provider will be fully iterated before the unit test is run, which would match the behavior you seem to be having.
It's my understanding of unit tests that you generally should avoid having some state / external dependency. You should instead create mock objects as necessary to return the data you want.
Similarly, if one executes the same provider multiple times, they will receive the same results.
I have also read so and recognize that external dependencies add additional dimensions, however, see some benefits of doing so and haven't gotten my head around mock objects.
15 hours ago, requinix said:Yielding the same object multiple times in a generator creates the sorts of problems...
I recommend cloning the $obj each time - it's a test so who cares about performance and memory usage?
I see your point about issues with yielding the same object, and was thinking of cloning as you suggest.
Assuming all agree that I should not attempt to update the database to a given state within the provider but right before actually performing each test.
-
I wish to test an API whether users with various permissions can perform actions on resources with various access policies.
I first envisioned using a provider such as the following, however, quickly found it had at least one major flaw and I am certain more.
/** * @dataProvider userResourceProvider */ public function testUserResource(User $user, object $resource, int $expectedHttpStatusCode):void { $client = $this->createClient($user); $client->makeSomeHttpRequest($resource); $this->assertStatusCode($expectedHttpStatusCode); } public function userResourceProvider(): \Generator { $user = new User(); $this->updateUserToInitialState($user); $this->updateDatabase($user); foreach($this->resourceClasses as $resourceClass) { $resource = new $resourceClass(); $this->updateResourceToInitialState($resource); $this->updateDatabase($resource); foreach($this->getResourcePolicy() as $policy) { $resource->setPolicy($policy); $this->updateDatabase($resource); foreach($this->getUserPermission() as $permission) { $user->setPermission($permission); $this->updateDatabase($user); yield [$user, $resource, $this->getExpectedHttpStatusCode($user, $resource)]; } } } }
Instead of passing the $user and the $resource in their current states to the test method, it will their last state after the second and third foreach loop which I confirmed using the following.public function someProvider(): \Generator { $obj = new SomeObject(); foreach([4,3,6,1] as $i) { $obj->setIndex($i); yield [$i, $obj]; } }
Not only will the state of the objects passed to the test be incorrect, I am almost sure the state of the database storing those objects be incorrect as (I think) all database updates be perform before executing the test methods.
Questions:
- Could/should any portions of modifying the state of the objects and the state of the database be updated in the provider (for instance, updateUserToInitialState and updateResourceToInitialState), or should all be repeated each time within the test method?
- What are the implications of database transactions when performing tests?
- Any other common gochas when performing tests on objects of different state?
-
Actually, I think there is an Option 4, and while it goes down the inheritance path, it doesn't bring all the baggage. Instead of extending some big complicated class, only extend small simple classes which are less likely to require future changes. I also recognize that this solution does as you recommend and uses an intermediary table. Still isn't an alternative to inheritance, but at least makes inheritance less painful.
I went back and forth whether Project or ProjectPermissionPolicy should hold the other's PK as neither should exist on its own without the presence of the other. I was going to come up with some awesome abstract Person/Heart analogies, but fortunately decided that when the first rule of thumb doesn't apply, the next rule of thumb should be that the parent table is the one which most logically would be delete.
Project - Id (PK) - Data Asset - Id (PK) - Data AbstractPermissionPolicy - Id (PK) - discriminator - Data ProjectPermissionPolicy extends AbstractPermissionPolicy - Id (PK, FK to AbstractPermissionPolicy.id) - resourceId (FK to project.id, delete cascade) AssetPermissionPolicy extends AbstractPermissionPolicy - Id (PK, FK to AbstractPermissionPolicy.id) - resourceId (FK to asset.id, delete cascade) Member - PermissionPolicyId (PK, FK to AbstractPermissionPolicy.Id) - UserId (PK, FK to User.Id) - MemberPermissionData User - Id (PK) - GeneralUserPermissionData - OtherData
-
On 11/6/2022 at 7:31 PM, maxxd said:
As to the primary key/foreign key part - just for me - unless I know it's going to be a one-to-one relationship forever I'll assume it's going to turn into a many-to-many at some point. It's probably just my innate desire/tendency to over-engineer so, so many things, but I have gotten to a point where it's no harder to reason about than a simple one-to-one link. It's just me and YMMV, but I personally have never found it harder to code on a one-to-one with an intermediary table than it is to refactor a one-to-one relationship into any other relationship type.
Thanks Maxxd, I didn't think of it until you brought it up, but vaguely recall reading some blog awhile ago where they promoted always using an intermediary table. Don't know if I agree with "always", but I do think it helps for some situations. Will give it some thought and definitely consider. Thanks!
-
15 hours ago, maxxd said:
Why do you need a PermissionPolicy to be referenced by only 1 project or asset? Seems to me that permissions would span projects and assets at least, unless there's a very specific set of business logic that dictates this. I have to admit I can't imagine what that would be, but I'm not a business person, so...
Projects and assets act as "document container" and the PermissionPolicy which can be referenced by only 1 project or asset applies to the documents, and right or wrong it isn't desirable to have changes to one container affect another.
16 hours ago, maxxd said:Is this a concern? If there's a possibility of needing a "many-to-one" relationship, IMO (obviously) there's no need to think about a one-to-one and it's probably better to consider the possibility of a many-to-many relationship just in terms of future-proofing. It doesn't add that much cognitive load to have a linking table even if it may not be strictly necessary at the time, but it gives teams the possibility of expanding scope without too much ... well, cussing to be honest.
No concerns about any added work if I ever need to change, but some concerns that it is going against everything I've ever read regarding proper database design. I did a quick google search to make sure I was not just imaging it, and all feverishly state that the FK of the parent table must be stored in child table, and that the parent table is the one which can exists on its own without the presence of the child table. But what do they know, so I pulled out my Simple SQL book to see what Rudy says, and he says that the foreign key goes in the table on the many side of the relationship, and for one-to-ones, it should be the same if it did have a many side (potentially, I misinterpreted the part about one-and-ones, and even if I didn't, he was not as adamant on that part) .
Forgetting about my specific use case and a different way of asking this question. Have you ever had the need to move the foreign key from the child table to the parent table? Did you have any issues?
-
Thanks Maxxd,
Glad you didn't totally bash the first schema as it is currently what I am going with! It is just different than anything I've done in the past and I think (thought?) different than what others do, and usually that means I am doing it wrong.
Appreciate your response!
-
Pretty please?
Current search functionality strategies
in PHP Coding Help
Posted
Definitely, metadata will be searched and "maybe" content will be searched for some document types, but not sure.
Per your link, "Solr is the popular, blazing-fast, open source enterprise search platform built on Apache Lucene", and it appears that Apache Lucene is "just?" a Java library used for the full text search of documents. If only full text searching, think it is necessary or just use PostgreSQL's? Also, not sure yet whether my referenced textai is much more, and really haven't spent much time learning about full text searching.
Trying go understand how the app database and searching using a dedicated server work together. I't my understanding that I send stuff I wish to later search for to Solr/Lucerne/TextAi which in turn indexes it so it later be searched for. Seems like the same metadata would then be both in my SQL DB as well as the search engine DB which seems wierd.