Coerces value v from a database type by calling a function returned by invoking amelinium.db/out-coercer multimethod on a qualified keyword table-column (or a qualified keyword made out of table and column). If there is no coercer attached for the keyword, returns unchanged v.

If a coercer can be obtained at compile-time, a coercion function-call form will be generated. If a coercer can be obtained at compile-time and the given value is statically convertable, value resulting from applying coercion function will be generated immediately.

Coerces value v to a database type by calling a function returned by invoking amelinium.db/in-coercer multimethod on a qualified keyword table-column (or a qualified keyword made out of table and column). If there is no coercer attached for the keyword, returns unchanged v.

If a coercer can be obtained at compile-time, a coercion function-call form will be generated. If a coercer can be obtained at compile-time and the given value is statically convertable, value resulting from applying coercion function will be generated immediately.

Coerces a sequence of values coll to database types by calling a function returned by invoking amelinium.db/in-coercer multimethod on a qualified keyword table-column (or a qualified keyword made out of table and column). If there is no coercer attached for the keyword, returns unchanged coll.

If both, a table and a column can be used to establish coercion function at compile-time, a mapping form will be generated which uses that function.

Magical macro which converts a sequence of values with optional table and column specifications to a database-suitable formats. Pre-processing of arguments is executed at compile-time, further processing is performed at run-time.

Any type of argument is accepted but literal vectors, including nested vectors, are control structures. Their first elements are table names (for the first one), column names (for nested vectors) or both (when expressed using fully-qualified keywords).

Example: (<<- 1 2 3 [:table-name [:column-name :val1 val2 (exp3) "val4"])

1, 2 and 3 are regular values, :table-name is literally expressed table name for coercer, :column-name is literally expressed column name for coercer, other expressions are values to be coerced. Table and column names can be dynamic, expressed with symbols or call forms.

All macro arguments are sequentially transformed with the following rules:

• If there is a literal vector at 1st level, its first element should be a table name. All elements of that vector will inherit that table name during conversion to a database-suitable format.

• If there is a literal vector nested within existing vector its first element will be considered a column name. All elements of that vector will inherit that column name during conversion to a database-suitable format.

• If the given literal vector contains a fully-qualified keyword at its first position then both table and column will be memorized to be applied during conversion of elements contained in that vector (table taken from a namespace and column from a name of the keyword).

• If there is a table name inherited but no column specified, a value is not converted but returned as is, with the exception: when the value is expressed as a literal, simple symbol then a column name will be derived from its name.

• A sub-vector may begin with an unspecified value nil. In such case it will group values to be converted but column name will not be set. The values will be left unconverted unless they are simple symbol forms; in such case column names will be derived from their names.

Values used to set column and table names at the beginnings of vectors can be expressed with any valid code. However, literal strings or literal keywords (or symbols in case of identifier-derived column names) will be pre-processed at compile time. So, if both column and table name are expressed that way, the conversion specifier will be generated ahead.

Moreover, if apart from the above, a value to be coerced is expressed as a literal number, string, keyword, boolean, or a nil, it will be converted at compile-time.

Conversion of repeating atomic expressions sharing the same table and column name identifiers (including symbolic identifiers) will be performed in ad-hoc created let block, and the results will be referenced using auto-generated symbols replacing the original expressions. Therefore, conversion for repeated symbols (and other atomic expressions) will be performed just once.

Examples:

(<<- [:users [:id id] [:email email]])

The above will convert values expressed by id and email symbol forms using a variant of coercion multimethod registered for :users/id and :users/email keywords, accordingly.

(<<- [:users/id id [:email e]])

The above will convert values of id and e symbol forms using a variant of coercion multimethod registered for :users/id and :users/email keywords, accordingly.

(<<- [:users [:id id] [:confirmations/email email [:expires expires]])

The above will convert values of id, email and expires symbol forms using a variant of coercion multimethod registered for :users/id, :confirmations/email, and :confirmations/expires keywords, accordingly. We can see that second vector changes the table name to confirmations in its scope so later expires derives it and is converted with :confirmations/expires specification.

This is synonymous to:

(<<- [:users id] [:confirmations email expires])

As we can see, the id symbolic identifier is used to set the column name, same as email and expires symbols. The above code will be expanded to:

[(amelinium.db/<- :users/id id)
 (amelinium.db/<- :confirmations/email email)
 (amelinium.db/<- :confirmations/expires expires)]

And then to:

[(#<Fn@4d7e49d amelinium.model.user/id_to_db> id)
 (amelinium.db/coerce-in* :confirmations/email email)
 (#<Fn@3bf6fdc6 amelinium.model.confirmation/to_expiry> expires)]

We can see that coercers for id and expires symbols were resolved and function call forms were created at compile-time. That’s because :users/id and :confirmations/expires were recognized as existing dispatch values when calling in-coercer internally. A coercer for the email symbol (using :confirmations/email dispatch value) was not recognized at compile-time so the call to amelinium.db/coerce-in* was generated instead.

Let’s have a quick look at some real-world example:

(amelinium.db/<<-  [:confirmations id code token reason id-type expires id id])

And generated Clojure code (phase 1):

(let [DB__confirmations_id_id_54377141 (amelinium.db/<- :confirmations/id id)]
  [DB__confirmations_id_id_54377141
   (amelinium.db/<- :confirmations/code       code)
   (amelinium.db/<- :confirmations/token     token)
   (amelinium.db/<- :confirmations/reason   reason)
   (amelinium.db/<- :confirmations/id-type id-type)
   (amelinium.db/<- :confirmations/expires expires)
   DB__confirmations_id_id_54377141
   DB__confirmations_id_id_54377141])

And fully expanded:

(let* [DB__confirmations_id_id_54377141 (#<Fn@5a424a5a amelinium.identity/__GT_db> id)]
  [DB__confirmations_id_id_54377141
   (#<Fn@279d4dd9 io.randomseed.utils/safe_parse_long>       code)
   (#<Fn@7f1abd95 io.randomseed.utils/some_str>             token)
   (#<Fn@7f1abd95 io.randomseed.utils/some_str>            reason)
   (#<Fn@7f1abd95 io.randomseed.utils/some_str>           id-type)
   (#<Fn@4ac5d426 amelinium.model.confirmation/to_expiry> expires)
   DB__confirmations_id_id_54377141
   DB__confirmations_id_id_54377141])

A SQL query which uses the sequence of values presented above needs one of them (identified with the id) to be repeated. We can observe that the macro generated let binding for it to assign the result of calling amelinium.identity/->db on id to auto-generated symbol named DB__confirmations_id_id_54377141. This symbol is then re-used in output vector multiple times so the calculation is performed just once.

Also, other coercers were successfully resolved to function objects during macro expansion since we have static table and column specifiers given.

Rule of a thumb is: if you can express certain values or specifications with literal strings or keywords, it may speed things up.

Same as <<- but its last argument should be a sequence to be concatenated with the results without any pre-processing.

Calls function f, passing db as a first argument and a sequence of: - a query (being a result of transforming query with amelinium.db.sql/build-query-dynamic), - coerced parameters (being a result of transforming params with <<-).

Calls execute!, passing db as a first argument and a sequence of: - a query (being a result of transforming query with amelinium.db.sql/build-query-dynamic), - coerced parameters (being a result of transforming params with <<-).

Calls execute-one!, passing db as a first argument and a sequence of: - a query (being a result of transforming query with amelinium.db.sql/build-query-dynamic), - coerced parameters (being a result of transforming params with <<-).

Calls function f passing db as a first argument and a sequence of: - a query (being a result of transforming query with amelinium.db.sql/build-query), - coerced parameters (being a result of transforming params with <<-).

Simple wrapper around amelinium.db.sql/build-query-dynamic and <<- macros. First argument should be a query (possibly grouped with a vector, if multiple arguments need to be passed), all other arguments are passed to <<-.

Produces a sequence suitable to be used with execute-* family of functions (a parameterized query as its first element and coerced query parameters as other elements).

Intended to be used for dynamically generated database queries. Uses a bit slower but safer FIFO cache of default size (about 150k items).

Example:

(<q ["select %(id) from %[u] where %(id) = ? AND %(t) = ?"
     {:id :users/id, :u :users/id, :t :users/account-type}]
    [:users [:id "42"] [:account-type :user]])

The above will return:

("select `id` from `users` where `id` = ? AND `account_type` = ?" 42 "user")

Note that "42" and :user are values which are to be coerced (first with a coercer registered for :users/id and second with a coercer registered for :users/account-type). After the coercion resulting values (42 and "user") are placed in a sequence to become query parameters.

Calls execute!, passing db as a first argument and a sequence of: - a query (being a result of transforming query with amelinium.db.sql/build-query), - coerced parameters (being a result of transforming params with <<-).

Calls execute-one!, passing db as a first argument and a sequence of: - a query (being a result of transforming query with amelinium.db.sql/build-query), - coerced parameters (being a result of transforming params with <<-).

Simple wrapper around amelinium.db.sql/build-query and <<- macros. First argument should be a query (possibly grouped with a vector, if multiple arguments need to be passed), all other arguments are passed to <<-.

Produces a sequence suitable to be used with execute-* family of functions (a parameterized query as its first element and coerced query parameters as other elements).

Example:

(<q ["select %(id) from %[u] where %(id) = ? AND %(t) = ?"
     {:id :users/id, :u :users/id, :t :users/account-type}]
    [:users [:id "42" :account-type :user]])

The above will return:

("select `id` from `users` where `id` = ? AND `account_type` = ?" 42 "user")

Note that "42" and :user are values which are to be coerced (first with a coercer registered for :users/id and second with a coercer registered for :users/account-type). After the coercion resulting values (42 and "user") are placed in a sequence to become query parameters.

Returns a bindable, auto-generated symbol for the table/column (from qs, which should be a QSlot record) and a value v. The unique identifier (obtained using gen-qs-keyword) must exists in bindings map. Otherwise, nil is returned.

Returns true if a bindable, auto-generated symbol for the table/column (from qs, which should be a QSlot record) and a value v exist in bindings map. Otherwise it returns false.

Creates a cache object of the given TTL and/or queue size. Optionally it can get an initial map of entries. Returns cache object encapsulated in an atom.

Removes entry or entries from the cache. Returns the updated cache from the atom.

Looks for the entry of the given ID in a cache which should be a cache object encapsulated in an atom. For multiple IDs, calls cache-lookup-coll. If the entry was not found, returns :io.randomseed.utils.db/not-found.

Looks for a collection of entries identified by the given ID in a cache which should be a cache object encapsulated in an atom. Returns a map with identifiers as keys and values for all found entries. Entries which are missing in the cache are grouped under the :io.randomseed.utils.db/not-found key as a list.

Prepares a cache object of the given TTL and/or queue size. Optionally it can get an initial map of entries. Returns a cache object.

Deletes the cached result of calling the given setting deleter. Purges cache entry after operation succeeded.

Gets the cached result of calling the given setting getter. Updates cache when necessary.

Sets the cached result of calling the given setting setter. Purges cache entry after operation succeeded.

Calls .close on obj if it implements java.io.Closeable interface. Otherwise uses reflection to check if there is unary .close method, and if it is found, calls it passing obj.

Closes the database connection. Calls configured finalized (:finalizer key) before.

Closes database connection used for migration. Expects config to be an argument-less function returning database configuration.

Closes connection pool ds.

Coerces the given value v to a database type by calling a function returned by invoking amelinium.db/in-coercer multimethod on a qualified keyword table-column (or a qualified keyword made out of table and column). If there is no coercer attached for the keyword, tries with a keyword created using the column-kw function. If there is no coercer, returns unchanged v.

Will immediately return the given value v if the coercer exists but it is set to false.

Same as coerce-in but table-column must be a lisp-cased keyword (already a result of colspec-kw or column-kw).

Coerces the given value v from a database type by calling a function returned by invoking amelinium.db/out-coercer multimethod on a qualified keyword table-column (or a qualified keyword made out of table and column). If there is no coercer attached for the keyword, tries with a keyword created using the column-kw function (to use column alone). If there is no coercer, returns unchanged v.

Will immediately return the given value v if the coercer exists but it is set to false.

Same as coerce-out but table-column must be a lisp-cased keyword (already a result of colspec-kw or column-kw).

Coerces a sequence of values coll to database types by calling a function returned by invoking amelinium.db/in-coercer multimethod on a qualified keyword table-column (or a qualified keyword made out of table and column). If there is no coercer attached for the keyword, tries with a keyword created using the column-kw function (to use column alone). If there is no coercer, returns unchanged coll.

Same as coerce-seq-in but table-column must be a lisp-cased keyword (already a result of colspec-kw or column-kw).

Coerces a sequence of values coll from database types by calling a function returned by invoking amelinium.db/out-coercer multimethod on a qualified keyword table-column (or a qualified keyword made out of table and column). If there is no coercer attached for the keyword, tries with a keyword created using the column-kw function (to use column alone). If there is no coercer, returns unchanged coll.

Same as coerce-seq-out but table-column must be a lisp-cased keyword (already a result of colspec-kw or column-kw).

Returns true if a value of the given argument is an instance of DBConfig record type.

Obtains the database (data source) key name from the given configuration data structure by using known patterns.

Obtains the database (data source) name from the given configuration data structure by using known patterns.

Finds a database name in the given map m or by taking it from v if it is a string or ident.

Finds a database identifier in the given map m or by taking it from v if it is a string or ident.

Logs database migration event described by database identifier db-k-name, data source ds, operation (:up or :down) and migration identifier id.

Defines input and output coercions for a database table table. The specs should be an argument list consisting of triples in a form of column-name, input-coercer, output-coercer.

For each definition 4 multimethod implementations will be emitted, identified by a keyword having a namespace the same as the given table, and a name the same as currently processed column name, both in 2 variants: one for snake, and one for lisp case. Two multimethod definitions will be created for amelinium.db/in-coercer and two for amelinium.db/out-coercer.

If the given coercer is a literal nil or false value, it will be marked as undefined using false value. It is advised to use that approach instead of assigning identity function to express that coercion is not needed since it can cause compile-time calculations to short-circuit instead of generating fallback code.

Example:

(defcoercions :users :some-identifier str keyword)

The above will expand the following code:

(defmethod amelinium.db/in-coercer  :users/some-identifier [_] str)
(defmethod amelinium.db/in-coercer  :users/some_identifier [_] str)
(defmethod amelinium.db/out-coercer :users/some-identifier [_] keyword)
(defmethod amelinium.db/out-coercer :users/some_identifier [_] keyword)

This will allow specialized database coercion functions to transformed values which are exchanged with a database.

Optionally coercions can be defined without a table name. In such case the table should be set to either nil, false or :amelinium.db/any.

Gets the data source from the DBConfig record. If the given argument is not an instance of DBConfig, it simply returns it.

Given a table name, a vector of column names, and a vector of row values (each row is a vector of its values), return a vector of the full INSERT SQL string and its parameters. Applies any :table-fn / :column-fn supplied in the options. If :suffix is provided in opts, that string is appended to the INSERT IGNORE ... statement. The IGNORE part can be replaced by supplying :alt-clause option key.

Given a table name and a hash map of column names and their values, return a vector of the full INSERT OR IGNORE SQL string and its parameters. Applies any :table-fn / :column-fn supplied in the options. If :suffix is provided in opts, that string is appended to the INSERT ... statement. If :alt-clause is provided in opts, it will replace the default IGNORE string.

Given a table name and a hash map of column names and their values, return a vector of the full REPLACE SQL string and its parameters. Applies any :table-fn / :column-fn supplied in the options. If :suffix is provided in opts, that string is appended to the INSERT ... statement.

Given a table name, a vector of column names, and a vector of row values (each row is a vector of its values), return a vector of the full REPLACE SQL string and its parameters. Applies any :table-fn / :column-fn supplied in the options. If :suffix is provided in opts, that string is appended to the REPLACE ... statement.

Generates result set builder on a basis of the given builder rs-builder. Uses amelinium.db/column-by-index-fn to coerce the results.

Generates result set builder on a basis of the given builder rs-builder. Uses amelinium.db/delayed-column-by-index-fn to coerce the results.

Generates unique but deterministic symbolic name for t (presumably table name), c (column name) and v (value, being an identifier). Returns a keyword named like DB__[t]_[c]_[v]_[nnnnnnn] where [t], [c] and [v] are string representations of the given argument values, and [nnnnnnn] is a numeric representation of combined hash of all values given as arguments.

Returns a (possibly cached) sequence of maps requested using db-getter for the given IDs. A database connection and a table name can be passed to be used with the standard get-id getter function instead. When multiple IDs are given, calls get-cached-coll.

Returns a (possibly cached) sequence of maps requested using db-getter for the given IDs. When called with a table name it will be passed as a second argument to the given getter function. When no getter is given the standard one (get-ids) is used. Use make-getter-coll to create getter (with or without predefined table name).

Uses get-cached-coll to retrieve a map with keys being IDs and values being requested properties. If there is no data for the given ID, corresponding entry is not added to the resulting map. If the property does not exist, nil is added.

Same as get-cached but retrieves a single property from the result by using the get function. When multiple IDs are given it calls get-cached-coll-prop to handle it.

Same as get-cached-prop but when there is no entry for the given ID, it returns the given default value.

Returns true if getting from a database failed in post-processing phase (e.g. de-serialization) and the data were broken.

Gets properties of the given ID from a database table. For multiple IDs, calls get-ids.

Gets a map of ID-to-properties from a database for the given IDs and a table. Assumes each result will be related to a single, unique ID.

Tries to obtain a database coercion function by calling in-coercer multimethod for column and table specified with table and column, or by a single table-column. Transforms arguments to a lisp-cased keyword. If there is no coercer for table and column, tries to getting one using the column alone.

Returns coercer when it is found. Returns false when a coercer is found but explicitly set to undefined. Returns nil when there is no coercer.

Same as get-in-coercer but trusts that table-column is a fully-qualified keyword (already a result of colspec-kw).

Tries to obtain a database coercion function by calling out-coercer multimethod for column and table specified with table and column, or by a single table-column. Transforms arguments to a lisp-cased keyword. If there is no coercer for table and column, falls back to getting one using the column alone.

Returns coercer when it is found. Returns false when a coercer is found but explicitly set to undefined. Returns nil when there is no coercer.

Same as get-out-coercer but trusts that table-column is a lisp-cased keyword (already a result of colspec-kw).

Converts the given ID retrieved from a database to a value suitable to be used in Clojure programs. If v is a number or a keyword, it is returned as is. Otherwise it is converted to a keyword.

Converts the given ID to a value suitable to be stored in a database. If v is a number, it is passed as is. Otherwise it is converted to a string.

Returns a coercer suitable for transforming the given argument v to a database-suitable value, assuming table and column specified by the given qualified keyword table-column.

Initializes single cache by parsing TTL and queue size.

Initializes in-memory caches by resetting atoms associated with cache parameters.

Initializes database configuration config for the configuration key k.

Initializes migrators given in a config sequence. Calls each function found or init-mig if it’s not a function but migration configuration map.

Syntactic sugar over execute! to make inserting columns/rows easier. Same as insert-or! but supports multiple rows to be inserted at once.

Syntactic sugar over execute-one! to make inserting hash maps easier. Given a connectable object, a table name, and a data hash map, inserts the data as a single row in the database and attempts to return a map of generated keys. By default it uses INSERT OR IGNORE but the IGNORE can be changed to anything by supplying :alt-clause option in opts map.

Syntactic sugar over execute-one! to make inserting hash maps easier. Given a connectable object, a table name, and a data hash map, inserts the data as a single row in the database and attempts to return a map of generated keys.

Syntactic sugar over execute! to make inserting columns/rows easier. Same as insert-multi! but supports :alt-clause option key.

Syntactic sugar over execute-one! to make inserting hash maps easier. Given a connectable object, a table name, and a data hash map, inserts the data as a single row in the database and attempts to return a map of generated keys.

Syntactic sugar over execute! to make inserting columns/rows easier. Same as insert-multi! but supports :alt-clause option key.

Invalidates cache associated with memoized function f. If key-params are given it should be a cache key sequence or other structure (usually function arguments). If only function is given, it clears the whole cache.

Invalidates cache associated with memoized function f. If key-params are given it should be a cache key sequence or other structure (usually function arguments), internally transformed to vector. If only function is given, it clears the whole cache. Supports functions memoized with memoize+.

Generates invalidation function for the given memoized function f. The invalidation function takes arguments in the same way as memoized function. If they match a value in the associated cache, the entry is evicted. Supports functions memoized with memoize and memoize+.

Like next.jdbc/get-by-id but supports lazy maps.

Creates a database getter suitable for use with get-cached-coll- family of functions. The returned function should accept an argument containing multiple identifiers.

Creates a setting deleter on a basis of the given table and entity column.

Returns a function which gets a setting for the given entity and de-serializes it to a Clojure data structure. Table name and entity column name must be quoted if needed before passing to this function.

Returns a function which stores one or more settings for a given entity in a database. Max. object size is 32 KB. Table name and entity column name must be quoted if needed before passing to this function.

Manual cache updater for functions memoized with clojure.core.memoize. Sets a key k to a value v in a map being a cached result of prior calling memoized function f. Will not associate any value if the caching key does not exist. The key should be passed as a vector in key.

Creates memoized version of a database accessing or other function. With only 1 argument defaults to a FIFO cache with length of 256 and TTL cache with expiration of 150 seconds. When 2 arguments are given it only creates FIFO cache of the given length, without TTL. When queue-size is nil or <= 0, the FIFO cache will not be created. When ttl is nil or <= 0, the TTL cache will not be created.

Wrapper around clojure.core.memoize/fifo which uses fast, map-based memoization for cache size up to level1-size size, and then switches to slower memoization based on FIFO cache of level2-size size. The first level of cache is never purged automatically; once an element is there, it stays.

If only one size is given (size), the level-1 cache size will be set to 70% of it, and level-2 cache to 30% of it.

If no size is given, the level-1 cache will have a size of 89000 elements and level-2 cache will have a size of 39000 elements; both having a size of 128000 elements in total.

Be aware that caching key in level-1 cache is a vector of all arguments unless the argument count is greater than 8. In such case all extra arguments (expressed with native Clojure structure used for variadic function arguments) are grouped as 9th element of this vector.

To control caches associated with original function, use a metadata of the returned function object.

Creates a memoized functions with predefined TTL and queue size taken from config. If the function is not given it will try to dereference symbol present in the config under the :memoizer key. Uses io.randomseed.utils.db/memoize to initialize caches.

Migrates all databases (or a database specified by a migrator function passed as an argument) up to the latest migration. Optional map of options can be passed which will be merged with each migration options.

Calls migrator-obj function without any arguments.

Returns distinct identifiers of all migration databases found in config sequence of functions by calling each function and extracting a value under :dbkey key of a returned map.

Gets a current value of ragtime-repl/migration-indexes.

Takes a migrators vector (migrators-vec) or uses a default migrators vector (from a global variable amelinium.db/migrators) and calls all of them (without passing any arguments) gathering returned results in a vector.

Returns true when the given value equals to :io.randomseed.utils.db/not-found.

Returns a coercer suitable for transforming the given argument v read from a database, assuming table and column specified by the given qualified keyword table-column.

Returns connection pool (HikariDataSource) object obtained from db-props.

Prepares database configuration.

Removes cache structures.

Syntactic sugar over execute-one! to make inserting hash maps easier. Given a connectable object, a table name, and a data hash map, inserts the data as a single row in the database and attempts to return a map of generated keys. By default it uses REPLACE.

Syntactic sugar over execute! to make inserting columns/rows easier. Same as replace! but supports multiple rows to be inserted at once.

Resumes the database connection.

Resumes connection pool ds.

Rolls back all databases or a database specified by a migrator function passed as an argument. Optional map of options can be passed which will be merged with each migration options. If a value is passed instead of a map or a function it will be used as an additional argument meaning a number of migrations or a migration ID.

Coerces a sequence of values coll from database types by calling a function returned by invoking amelinium.db/out-coercer multimethod on a qualified keyword table-column (or a qualified keyword made out of table and column). If there is no coercer attached for the keyword, returns unchanged coll.

If both, a table and a column can be used to establish coercion function at compile-time, a mapping form will be generated which uses that function.

Suspends the database connection.

Suspends connection pool ds.

Tries to create a database described by config map if it does not exist yet.