mouf / schema-analyzer
A package that offers utility tools to analyze database schemas (on top of Doctrine DBAL)
Installs: 191 999
Dependents: 2
Suggesters: 0
Security: 0
Stars: 6
Watchers: 14
Forks: 5
Open Issues: 0
Requires
- php: ^7.4 || ^8.0
- clue/graph: ~0.9.0
- doctrine/cache: ^1.4.1
- doctrine/dbal: ^3.0
- graphp/algorithms: ~0.8.0
Requires (Dev)
- php-coveralls/php-coveralls: ^2.7.0
- phpunit/phpunit: ^9.6.16
README
Schema analyzer for DBAL
This package offer utility functions to analyze database schemas. It is built on top of Doctrine DBAL.
In this package, you will find:
- Functions to automatically detect junction tables
- Functions to compute the shortest path between 2 tables based on the relationships stored in the schema.
Installation
You can install this package through Composer:
{ "require": { "mouf/schema-analyzer": "~1.0" } }
The packages adheres to the SemVer specification, and there will be full backward compatibility between minor versions.
Detecting junction tables
The starting point is always a DBAL Schema. Pass the schema manager to SchemaAnalyzer, and then, simply call the functions.
// $conn is the DBAL connection. $schemaAnalyzer = new SchemaAnalyzer($conn->getSchemaManager()); // Let's detect all junctions tables $tables = $schemaAnalyzer->detectJunctionTables(); // This will return an array of Doctrine\DBAL\Schema\Table objects
A junction table is a table:
- that has exactly 2 foreign keys
- that has only 2 columns (or 3 columns if the one of those is an autoincremented primary key).
There is an optional parameter you can use with detectJunctionTables
that will automatically ignore any junction
table that is referenced by a foreign key of another table.
// Get all junction tables except the ones that are references by a foreign key. $tables = $schemaAnalyzer->detectJunctionTables(true);
Detecting inheritance relationship between tables
About inheritance relationships
If a table "user" has a primary key that is also a foreign key pointing on table "contact", then table "user" is considered to be a child of table "contact". This is because you cannot create a row in "user" without having a row with the same ID in "contact".
Therefore, a "user" ID has to match a "contact", but a "contact" has not necessarily a "user" associated.
Detecting inheritance relationships
You can use SchemaAnalyzer
to detect parent / child relationships.
getParentRelationship
takes a table name in parameter and returns the DBALForeignKeyConstraint
representing the relationship between this table and its parent.getChildrenRelationships
takes a table name in parameter and returns an array of DBALForeignKeyConstraint
representing the relationship between this table and its children.
$parentKeyConstraint = $schemaAnalyzer->getParentRelationship("user"); /* @var $parentKeyConstraint ForeignKeyConstraint */ $parent = $parentKeyConstraint->getForeignTableName(); // This will return the "contact" table (as a string) $childrenKeyConstraints = $schemaAnalyzer->getChildrenRelationships("contact"); /* @var $childrenKeyConstraints ForeignKeyConstraint[] */ $children = array_map(function($item) { return $item->getLocalTableName(); }, $childrenKeyConstraints); // This will return an array of tables whose parent is contact: ["user"]
Computing the shortest path between 2 tables
Following foreign keys, the getShortestPath
function will try to find the shortest path between 2 tables.
It will return the list of foreign keys it used to link the 2 tables.
Internals:
- Each foreign key has a cost of 1
- Junction tables have a cost of 1.5, instead of 2 (one for each foreign key)
- Foreign keys representing an inheritance relationship (i.e. foreign keys binding the primary keys of 2 tables) have a cost of 0.1
// $conn is the DBAL connection. $schemaAnalyzer = new SchemaAnalyzer($conn->getSchemaManager()); // Let's detect the shortest path between 2 tables: $fks = $schemaAnalyzer->getShortestPath("users", "rights"); // This will return an array of Doctrine\DBAL\Schema\ForeignKeyConstraint objectsHeads up! The shortest path is based on the cost of the foreign keys. It is perfectly possible to have several shortest paths (if several paths have the same total cost). If there are several shortest paths, rather than choosing one path amongst the others, SchemaAnalyzer will throw a
ShortestPathAmbiguityException
. The exception message details all the possible shortest
paths.
Caching results
Analyzing the full data model and looking for shortest paths can take a long time. For anything that should run
in a production environment, it is recommended to cache the result. SchemaAnalyzer
can be passed a Doctrine cache,
along a cache prefix. The cache prefix is a string that will be used to prefix all cache keys. It is useful to
avoid cache collisions between several databases.
Usage:
// $conn is the DBAL connection. // Let's use the ApcCache (or any other Doctrine cache...) $cache = new ApcCache(); $schemaAnalyzer = new SchemaAnalyzer($conn->getSchemaManager(), $cache, "my_prefix");
Changing the cost of the foreign keys to alter the shortest path
If you are facing an ambiguity exception or if the shortest path simply does not suit you, you can alter the cost of the foreign keys.
$schemaAnalyzer->setForeignKeyCost($tableName, $columnName, $cost);
The $cost
can be any number. Remember that the default cost for a foreign key is 1.
SchemaAnalyzer comes with a set of default constants to help you work with costs:
SchemaAnalyzer::WEIGHT_IMPORTANT
(0.75) for foreign keys that should be followed in prioritySchemaAnalyzer::WEIGHT_IRRELEVANT
(2) for foreign keys that should be generally avoidedSchemaAnalyzer::WEIGHT_IGNORE
(Infinity) for foreign keys that should never be used as part of the shortest path
Another option is to add a cost modifier to a table. This will alter the cost of all foreign keys pointing to or originating from this table.
$schemaAnalyzer->setTableCostModifier($tableName, $cost);