Programming in Lua ¶

Programming YottaDB in Lua is provided by lua-yottadb, developed by Mitchell and Berwyn Hoyt, and sponsored by the University of Antwerp. We acknowledge their contribution and thank them for the value it adds to the YottaDB community.

lua-yottadb wraps the YottaDB Simple API to provide a Lua API.

Installation ¶

The YottaDB Lua API requires a minimum YottaDB release of r1.34 and is tested with a minimum Lua version of 5.1.

sudo apt install lua5.4
sudo apt install liblua5.4-dev
git clone https://github.com/anet-be/lua-yottadb.git
cd lua-yottadb
make
make test
sudo make install

Introduction by example ¶

Let's tinker with setting some database values in different ways:

 ydb = require 'yottadb'
 n = ydb.node('^hello')  -- create node object pointing to YottaDB global ^hello

 n:get()  -- get current value of the node in the database
 -- nil
 n:set('Hello World')
 n:get()
 -- Hello World

 -- Equivalent ways to create a new subnode object pointing to an added 'cowboy' subscript
 n2 = ydb.node('^hello')('cowboy')
 n2 = ydb.node('^hello', 'cowboy')
 n2 = n('cowboy')
 n2 = n['cowboy']
 n2 = n.cowboy

 n2:set('Howdy partner!')  -- set ^hello('cowboy') to 'Howdy partner!'
 n2, n2:get()
 -- ^hello("cowboy")  Howdy partner!

 n2.ranches:set(3)  -- create subnode object ydb.node('^hello', 'cowboy', 'ranches') and set to 3
 n2.ranches.__ = 3   -- same as :set() but 3x faster (ugly but direct access to value)

 n3 = n.chinese  -- add a second subscript to '^hello', creating a third object
 n3:set('你好世界!') -- value can be number or string, including UTF-8
 n3, n3:get()
 -- hello("chinese")  你好世界!

We can also use other methods of the node object like incr() name() has_value() has_key() lock_incr():

 n2.ranches:incr(2)  -- increment
 -- 5
 n2:name()
 -- cowboy
 n2:__name()  -- uglier but 15x faster access to node object methods
 -- cowboy

(Note: lua-yottadb is able to distinguish n:method(n) from subnode creation n.method. See details in the notes here.)

Now, let's try dump to see what we've got so far:

 n:dump()

The output will be:

 ^hello="Hello World"
 ^hello("chinese")="你好世界!"
 ^hello("cowboy")="Howdy partner!"
 ^hello("cowboy","ranches")="5"

We can delete a node – either its value or its entire tree:

 n:set(nil)  -- delete the value of '^hello', but not any of its child nodes
 n:get()
 -- nil
 n:dump()  -- the values of the child node are still in the database
 -- hello("chinese")="你好世界!!"
 -- hello("cowboy")="Howdy partner!"

 n:kill() -- delete both the value at the '^hello' node and all of its children
 n:dump()
 -- nil

Doing something useful

Let's use Lua to calculate the height of 3 oak trees, based on their shadow length and the angle of the sun. Method settree() is a handy way to enter literal data into the database from a Lua table constructor:

 trees = ydb.node('^oaks')  -- create node object pointing to YottaDB global ^oaks
 -- store initial data values into database subnodes ^oaks('1', 'shadow'), etc.
 trees:settree({{shadow=10, angle=30}, {shadow=13, angle=30}, {shadow=15, angle=45}})

 for oaktree, value, index in pairs(trees) do
      oaktree.height.__ = oaktree.shadow.__ * math.tan( math.rad(oaktree.angle.__) )
      print( string.format('Oak %s is %.1fm high', index, oaktree.height.__) )
   end
 -- Oak 1 is 5.8m high
 -- Oak 2 is 7.5m high
 -- Oak 3 is 15.0m high

You may also wish to look at node:gettree() which has multiple uses. On first appearances, it just loads a database tree into a Lua table (opposite of settree above), but it also allows you to iterate over a whole database tree and process each node through a filter function. For example, to use print as a filter function, do node:gettree(nil, print). Incidentally, lua-yottadb itself uses gettree, to implement node:dump().

Database transactions are also available

> Znode = ydb.node('^Ztest')
> transact = ydb.transaction(function(end_func)
  print("^Ztest starts as", Znode:get())
  Znode:set('value')
  end_func()
  end)

> transact(ydb.trollback)  -- perform a rollback after setting Znode
^Ztest starts as      nil
YDB Error: 2147483645: YDB_TP_ROLLBACK
> Znode.get()  -- see that the data didn't get set
nil

> tries = 2
> function trier()  tries=tries-1  if tries>0 then ydb.trestart() end  end
> transact(trier)  -- restart with initial dbase state and try again
^Ztest starts as      nil
^Ztest starts as      nil
> Znode:get()  -- check that the data got set after restart
value

> Znode:set(nil)
> transact(function() end)  -- end the transaction normally without restart
^Ztest starts as      nil
> Znode:get()  -- check that the data got set
value

Lua API ¶

Basic low level functions ¶

block_M_signals (bool)¶

Block or unblock YDB signals while M code is running. This function is designed to be passed as a callback to yottadb.init() as the signal_blocker parameter. The function accepts one boolean parameter: true for entering Lua and false for exiting Lua:

When true is passed to this function it blocks all the signals listed in BLOCKED_SIGNALS in callins.c using OS call sigprocmask(). SIGALRM is treated specially: instead of being blocked OS call sigaction() is used to set the SA_RESTART flag for SIGALRM. This makes the OS automatically restart IO calls that are interrupted by SIGALRM. The benefit of this over blocking is that the YottaDB SIGALRM handler does actually run, even during Lua code, allowing YottaDB to flush the database or IO as necessary (otherwise M code would need to run the M command VIEW "FLUSH" after calling Lua code).

When false is passed to this function it will reverse the effect, unblocking and clearing SA_RESTART for the same signals manipulated when true is passed.

Note: This function does take time, as OS calls are slow. Using it will increase the M calling overhead by a factor of 2-5 times the bare calling overhead (on the order of 1.4 microseconds on a typical 2020s x86_64 CPU; see make benchmarks)

bool: true to block; false to unblock YottaDB signals

Returns:: nothing

data (varname[, subsarray[, …]])¶

Return whether a node has a value or subtree.

varname: String of the database node (this can also be replaced by cachearray)
subsarray: (optional) Table of subscripts
...: (optional) List of subscripts to append after any elements in optional subsarray table

Returns:

0: (node has neither value nor subtree)
1: node has value, not subtree
10: node has no value, but does have a subtree
11: node has both value and subtree

Example:

-- include setup from example at yottadb.set()
ydb.data('^Population')
-- 10.0
ydb.data('^Population', {'USA'})
-- 11.0

delete_node (varname[, subsarray[, …]]) [deprecated v3.0]:¶

Deprecated and replaced by set(varname[, subsarray[, ...]], nil).

varname: String of the database node (this can also be replaced by cachearray)
subsarray: (optional) Table of subscripts
...: (optional) List of subscripts to append after any elements in optional subsarray table

Example:

ydb = require('yottadb')
ydb.set('^Population', {'Belgium'}, 1367000)
ydb.delete_node('^Population', {'Belgium'})  -- or ydb.set('^Population', {'Belgium'}, nil)
ydb.get('^Population', {'Belgium'})
-- nil

delete_tree (varname[, subsarray[, …]]) [deprecated v3.0]:¶

Deprecated and replaced by kill.

varname: String of the database node (this can also be replaced by cachearray)
subsarray: (optional) Table of subscripts
...: (optional) List of subscripts to append after any elements in optional subsarray table

get (varname[, subsarray[, …]])¶

Gets and returns the value of a database variable or node; or nil if the variable or node does not exist.

varname: String of the database node (this can also be replaced by cachearray)
subsarray: (optional) Table of subscripts
...: (optional) List of subscripts or table subscripts

Returns:

string value or nil

Example:

-- include setup from example at yottadb.set()
ydb.get('^Population')
-- nil
ydb.get('^Population', {'Belgium'})
-- 1367000
ydb.get('$zgbldir')
-- /home/ydbuser/.yottadb/r1.34_x86_64/g/yottadb.gld

get_error_code (message)¶

Get the YDB error code (if any) contained in the given error message.

message: String error message.

Returns:

the YDB error code (if any) for the given error message,
or nil if the message is not a YDB error.

Example:

ydb = require('yottadb')
ydb.get_error_code('YDB Error: -150374122: %YDB-E-ZGBLDIRACC, Cannot access global directory !AD!AD!AD.')
-- -150374122

grab (varname[, subsarray[, …[, timeout]]])¶

Alias for lock_incr to acquire or increment a lock named varname[(subsarray)]. Returns after timeout, if specified. Raises a yottadb.YDB_LOCK_TIMEOUT error if lock could not be acquired.

If timeout is not supplied or is nil, wait forever; timeout of zero means try only once. Caution: timeout is not optional if ... list of subscripts is provided, but it may be nil. Otherwise lock_incr cannot tell whether it is a subscript or a timeout.

varname: of database node (this can also be replaced by cachearray)
subsarray: (optional) Table of subscripts
...: (optional) List of subscripts or table subscripts
timeout: (optional) Number timeout in seconds to wait for the lock. Optional only if subscripts is a table.

Returns:: 0 (always)

incr (varname[, subsarray][, …], increment)¶

Increments the numeric value of a database variable or node. Raises an error on overflow.

Caution: increment is not optional if ... list of subscript is provided. Otherwise incr() cannot tell whether last parameter is a subscript or an increment.

varname: of database node (this can also be replaced by cachearray)
subsarray: (optional) Table of subscripts
...: (optional) List of subscripts or table subscripts
increment: Number or string amount to increment by (default=1)

Returns:

the new value

Example:

ydb = require('yottadb')
ydb.get('num')
-- 4
ydb.incr('num', 3)
-- 7
ydb.incr('num')
-- 8

init ([signal_blocker])¶

Initialize ydb and set blocking of M signals. If signal_blocker is specified, block M signals which could otherwise interrupt slow IO operations like reading from stdin or a pipe. Raise any errors. See also the notes on signals in the README.

Note: any calls to the YDB API also initialize YDB; any subsequent call here will set signal_blocker but not re-init YDB.

signal_blocker: (optional) Specifies a Lua callback CFunction (e.g. yottadb.block_M_signals()) which will be called with its one parameter set to false on entry to M, and with true on exit from M, so as to unblock YDB signals while M is in use. Setting signal_blocker to nil switches off signal blocking.

Note: Changing this to support a generic Lua function as callback would be possible but slow, as it would require fetching the function pointer from a C closure, and using lua_call().

Returns:: nothing

kill (varname[, subsarray[, …]])¶

Deletes a database variable tree (node and subnodes) or a node subtree.

varname: String of the database node (this can also be replaced by cachearray)
subsarray: (optional) Table of subscripts
...: (optional) List of subscripts to append after any elements in optional subsarray table

Example:

-- include setup from example at yottadb.set()
ydb.get('^Population', {'USA'})
-- 325737000
ydb.get('^Population', {'USA', '17900802'})
-- 3929326
ydb.get('^Population', {'USA', '18000804'})
-- 5308483
ydb.kill('^Population', {'USA'})
ydb.data('^Population', {'USA'})
-- 0.0

lock ([nodes[, timeout]])¶

Releases all locks held and attempts to acquire all requested locks. Returns after timeout, if specified. If timeout is not supplied or is nil, wait forever; timeout of zero means try only once. Raises an error yottadb.YDB_LOCK_TIMEOUT if a lock could not be acquired.

nodes: (optional) Table array of {varname[, subs]} elements or node objects that specify the lock names to lock.
timeout: (optional) Number timeout in seconds to wait for the lock.

Returns:: 0 (always)

lock_decr (varname[, subsarray[, …]])¶

Decrements a lock of the same name as varname[(subsarray)], releasing it if zero. Releasing a lock cannot create an error unless the varname/subsarray names are invalid.

varname: String of the database node (this can also be replaced by cachearray)
subsarray: (optional) Table of subscripts
...: (optional) List of subscripts to append after any elements in optional subsarray table

Returns:: 0 (always)

lock_incr (varname[, subsarray[, …[, timeout]]])¶

Attempts to acquire or increment a lock named varname[(subsarray)]. Returns after timeout, if specified. Raises a yottadb.YDB_LOCK_TIMEOUT error if lock could not be acquired.

If timeout is not supplied or is nil, wait forever; timeout of zero means try only once. Caution: timeout is not optional if ... list of subscripts is provided, but it may be nil. Otherwise lock_incr cannot tell whether it is a subscript or a timeout.

varname: of database node (this can also be replaced by cachearray)
subsarray: (optional) Table of subscripts
...: (optional) List of subscripts or table subscripts
timeout: (optional) Number timeout in seconds to wait for the lock. Optional only if subscripts is a table.

Returns:: 0 (always)

node_next (varname[, subsarray[, …]])¶

Returns the full subscript list of the next node after a database variable or node. A next node chain started from varname will eventually reach all nodes under that varname in order.

Note: node:gettree() or node:subscripts() may be a better way to iterate a node tree

varname: String of the database node (this can also be replaced by cachearray)
subsarray: (optional) Table of subscripts
...: (optional) List of subscripts to append after any elements in optional subsarray table

Returns:

list of subscripts for the node, or nil if there isn't a next node

Examples:

-- include setup from example at yottadb.set()
print(table.concat(ydb.node_next('^Population'), ', '))
-- Belgium
print(table.concat(ydb.node_next('^Population', {'Belgium'}), ', '))
-- Thailand
print(table.concat(ydb.node_next('^Population', {'Thailand'}), ', '))
-- USA
print(table.concat(ydb.node_next('^Population', {'USA'}), ', '))
-- USA, 17900802
print(table.concat(ydb.node_next('^Population', {'USA', '17900802'}), ', '))
-- USA, 18000804

-- Note: The format used above to print the next node will give an error if there is no next node, i.e., the value returned is nil.
-- This case will have to be handled gracefully. The following code snippet is one way to handle nil as the return value:

local ydb = require('yottadb')
next = ydb.node_next('^Population', {'USA', '18000804'})
if next ~= nil then
  print(table.concat(next, ', '))
else
  print(next)
end

node_previous (varname[, subsarray[, …]])¶

Returns the full subscript list of the previous node after a database variable or node. A previous node chain started from varname will eventually reach all nodes under that varname in reverse order.

Note: node:gettree() or node:subscripts() may be a better way to iterate a node tree

varname: String of the database node (this can also be replaced by cachearray)
subsarray: (optional) Table of subscripts
...: (optional) List of subscripts to append after any elements in optional subsarray table

Returns:

list of subscripts for the node, or nil if there isn't a previous node

Examples:

-- include setup from example at yottadb.set()
print(table.concat(ydb.node_previous('^Population', {'USA', '18000804'}), ', '))
-- USA, 17900802
print(table.concat(ydb.node_previous('^Population', {'USA', '17900802'}), ', '))
-- USA
print(table.concat(ydb.node_previous('^Population', {'USA'}), ', '))
-- Thailand
print(table.concat(ydb.node_previous('^Population', {'Thailand'}), ', '))
-- Belgium

-- Note: See the note on handling nil return values in node_next() which applies to node_previous() as well.

release (varname[, subsarray[, …]])¶

Alias for lock_decr to decrement a lock of the same name as varname[(subsarray)], releasing it if zero. Releasing a lock cannot create an error unless the varname/subsarray names are invalid.

varname: String of the database node (this can also be replaced by cachearray)
subsarray: (optional) Table of subscripts
...: (optional) List of subscripts to append after any elements in optional subsarray table

Returns:: 0 (always)

set (varname[, subsarray][, …], value)¶

Sets the value of a database variable or node.

varname: of database node (this can also be replaced by cachearray)
subsarray: (optional) Table of subscripts
...: (optional) List of subscripts or table subscripts
value: The value to assign to the node. If this is a number, it is converted to a string. If it is nil, the node's value, if any, is deleted.

Returns:

value

Example:

ydb = require('yottadb')
ydb.set('^Population', {'Belgium'}, 1367000)
ydb.set('^Population', {'Thailand'}, 8414000)
ydb.set('^Population', {'USA'}, 325737000)
ydb.set('^Population', {'USA', '17900802'}, 3929326)
ydb.set('^Population', {'USA', '18000804'}, 5308483)

str2zwr (s)¶

Returns the zwrite-formatted version of the given string.

s: String to format.

Returns:

formatted string

Example:

ydb=require('yottadb')
str='The quick brown dog\b\b\bfox jumps over the lazy fox\b\b\bdog.'
print(str)
-- The quick brown fox jumps over the lazy dog.
ydb.str2zwr(str)
-- "The quick brown dog"_$C(8,8,8)_"fox jumps over the lazy fox"_$C(8,8,8)_"dog."

subscript_next (varname[, subsarray[, …]])¶

Returns the next subscript for a database variable or node; or nil if there isn't one.

varname: of database node (this can also be replaced by cachearray)
subsarray: (optional) Table of subscripts
...: (optional) List of subscripts or table subscripts

Returns:

string subscript name, or nil if there are no more subscripts

Example:

-- include setup from example at yottadb.set()
ydb.subscript_next('^Population', {''})
-- Belgium
ydb.subscript_next('^Population', {'Belgium'})
-- Thailand
ydb.subscript_next('^Population', {'Thailand'})
-- USA

subscript_previous (varname[, subsarray[, …]])¶

Returns the previous subscript for a database variable or node; or nil if there isn't one.

varname: of database node (this can also be replaced by cachearray)
subsarray: (optional) Table of subscripts
...: (optional) List of subscripts or table subscripts

Returns:

string subscript name, or nil if there are no previous subscripts

Example:

-- include setup from example at yottadb.set()
ydb.subscript_previous('^Population', {'USA', ''})
-- 18000804
ydb.subscript_previous('^Population', {'USA', '18000804'})
-- 17900802
ydb.subscript_previous('^Population', {'USA', '17900802'})
-- nil
ydb.subscript_previous('^Population', {'USA'})
-- Thailand

subscripts (varname[, subsarray[, …[, reverse]]])¶

Returns an iterator for iterating over database sibling subscripts starting from the node referenced by varname and subarray.

Note: this starts from the given location and gives the next sibling subscript in the M collation sequence. It operates differently than node:subscipts() which yields all subscripts that are children of the given node, and which you may consider to be preferable.

varname: of database node (this can also be replaced by cachearray)
subsarray: (optional) Table of subscripts
...: (optional) List of subscripts or table subscripts
reverse: (optional) Flag that indicates whether to iterate backwards. Not optional when '…' is provided

Returns:: iterator

ydb_eintr_handler ()¶

Lua function to call ydb_eintr_handler(). Code intended to handle EINTR errors, instead of blocking signals, should call ydb_eintr_handler() when it gets an EINTR return code, before re-issuing the interrupted system call.

Returns:: YDB_OK on success, and greater than zero on error (with message in ZSTATUS)

zwr2str (s)¶

Returns the string described by the given zwrite-formatted string.

s: String in zwrite format.

Returns:

string

Example:

ydb=require('yottadb')
str1='The quick brown dog\b\b\bfox jumps over the lazy fox\b\b\bdog.'
zwr_str=ydb.str2zwr(str1)
print(zwr_str)
-- "The quick brown dog"_$C(8,8,8)_"fox jumps over the lazy fox"_$C(8,8,8)_"dog."
str2=ydb.zwr2str(zwr_str)
print(str2)
-- The quick brown fox jumps over the lazy dog.
str1==str2
-- true

Transactions ¶

tp ([id][, varnames], f[, …])¶

Initiates a transaction (low level function). Restarts are subject to $ZMAXTPTIME after which they cause error %YDB-E-TPTIMEOUT

id: (optional) optional string transaction id. For special ids BA or BATCH, see Transaction Processing.
varnames: (optional) optional table of local M variable names to restore on transaction restart (or {'*'} for all locals) Restoration applies to rollback.
f: Function to call. The transaction's affected globals are:

Committed if the function returns nothing or yottadb.YDB_OK.

Restarted if the function returns yottadb.YDB_TP_RESTART (f will be called again).

Not committed if the function returns yottadb.YDB_TP_ROLLBACK or errors out.

...: (optional) arguments to pass to f

Examples:

local ydb = require('yottadb')

function transfer_to_savings(t)
   local ok, e = pcall(ydb.incr, '^checking', -t)
   if (ydb.get_error_code(e) == ydb.YDB_TP_RESTART) then
      return ydb.YDB_TP_RESTART
   end
   if (not ok or tonumber(e)<0) then
      return ydb.YDB_TP_ROLLBACK
   end
   local ok, e = pcall(ydb.incr, '^savings', t)
   if (ydb.get_error_code(e) == ydb.YDB_TP_RESTART) then
      return ydb.YDB_TP_RESTART
   end
   if (not ok) then
      return ydb.YDB_TP_ROLLBACK
   end
   return ydb.YDB_OK
end

ydb.set('^checking', 200)
ydb.set('^savings', 85000)

print("Amount currently in checking account: $" .. ydb.get('^checking'))
print("Amount currently in savings account: $" .. ydb.get('^savings'))

print("Transferring $10 from checking to savings")
local ok, e = pcall(ydb.tp, '', {'*'}, transfer_to_savings, 10)
if (not e) then
   print("Transfer successful")
elseif (ydb.get_error_code(e) == ydb.YDB_TP_ROLLBACK) then
   print("Transfer not possible. Insufficient funds")
end

print("Amount in checking account: $" .. ydb.get('^checking'))
print("Amount in savings account: $" .. ydb.get('^savings'))

print("Transferring $1000 from checking to savings")
local ok, e = pcall(ydb.tp, '', {'*'}, transfer_to_savings, 1000)
if (not e) then
   print("Transfer successful")
elseif (ydb.get_error_code(e) == ydb.YDB_TP_ROLLBACK) then
   print("Transfer not possible. Insufficient funds")
end

print("Amount in checking account: $" .. ydb.get('^checking'))
print("Amount in savings account: $" .. ydb.get('^savings'))

Output:
  Amount currently in checking account: $200
  Amount currently in savings account: $85000
  Transferring $10 from checking to savings
  Transfer successful
  Amount in checking account: $190
  Amount in savings account: $85010
  Transferring $1000 from checking to savings
  Transfer not possible. Insufficient funds
  Amount in checking account: $190
  Amount in savings account: $85010

transaction ([id][, varnames], f)¶

Returns a high-level transaction-safe version of the given function. It will be called within a YottaDB transaction and the database globals restored on error or yottadb.trollback()

id: (optional) optional string transaction id. For special ids BA or BATCH, see Transaction Processing.
varnames: (optional) optional table of local M variable names to restore on transaction trestart() (or {'*'} for all locals). Restoration applies to rollback.
f: Function to call. The transaction's affected globals are:

Committed if the function returns nothing or yottadb.YDB_OK.

Restarted if the function returns yottadb.YDB_TP_RESTART (f will be called again). Restarts are subject to $ZMAXTPTIME after which they cause error %YDB-E-TPTIMEOUT

Not committed if the function returns yottadb.YDB_TP_ROLLBACK or errors out.

Returns:

transaction-safe function.

Example:

Znode = ydb.node('^Ztest')
transact = ydb.transaction(function(end_func)
  print("^Ztest starts as", Znode:get())
  Znode:set('value')
  end_func()
  end)

transact(ydb.trollback)  -- perform a rollback after setting Znode
-- ^Ztest starts as       nil
-- YDB Error: 2147483645: YDB_TP_ROLLBACK
-- stack traceback:
--   [C]: in function '_yottadb.tp' ...
Znode:get()  -- see that the data didn't get set
-- nil

tries = 2
function trier()  tries=tries-1  if tries>0 then ydb.trestart() end  end
transact(trier)  -- restart with initial dbase state and try again
-- ^Ztest starts as       nil
-- ^Ztest starts as       nil
Znode:get()  -- check that the data got set after restart
-- value

Znode:set(nil)
transact(function() end)  -- end the transaction normally without restart
-- ^Ztest starts as       nil
Znode:get()  -- check that the data got set
-- value

trestart ()¶

Make the currently running transaction function restart immediately.

trollback ()¶

Make the currently running transaction function rollback immediately with a YDB_TP_ROLLBACK error.

High level functions ¶

dump (node[, …[, maxlines=30]])¶

Dump the specified node tree.

node: Either a node object with ... subscripts or glvn varname with ... subsarray
...: (optional) Either a table or a list of subscripts to add to node
maxlines: (default: 30) Maximum number of lines to output before stopping dump

Returns:

dump as a string

Examples:

ydb.dump(node, [...[, maxlines]])

ydb.dump('^MYVAR', 'people')

require (Mprototypes, debug)¶

Import M routines as Lua functions specified in a ydb 'call-in' file.

See example call-in file arithmetic.ci and matching M file arithmetic.m.

Mprototypes: A list of lines in the format of ydb 'call-in' files required by ydb_ci(). If the string contains : it is considered to be the call-in specification itself; otherwise it is treated as the filename of a call-in file to be opened and read.
debug: When neither false nor nil, tell Lua not to delete the temporary preprocessed call-in table file it created. The name of this file will be stored in the __ci_filename field of the returned table.

Returns:

A table of functions analogous to a Lua module. Each function in the table will call an M routine specified in Mprototypes.

Example:

$ export ydb_routines=examples   # put arithmetic.m into ydb path
$ lua -lyottadb
arithmetic = yottadb.require('examples/arithmetic.ci')
arithmetic.add_verbose("Sum is:", 2, 3)
-- Sum is: 5
-- Sum is: 5
arithmetic.sub(5,7)
-- -2

Node class operations ¶

inherit (node_func[, metatable])¶

Create a new node class that inherits from the yottadb.node superclass. The returned function will generate nodes of a new class (a new metatable). You may add or override class methods by adding them to new_metatable either before or after calling inherit. Any class method the user does not define in new_metatable will default to the class method of the superclass. If desired, the new class may be further subclassed, also using inherit. Refer to the example below, and further explanation in the README on class inheritance.

node_func: yottadb.node or a descendent that inherits from it using inherit
metatable: (optional) to use for the new node class. Defaults to {}.

Returns:

function that generates node of the new type
subclass metatable (modified)
superclass metatable used by the original node_func

Example:

-- To average all data stored in a particular database node:
averaged_node, class, superclass = ydb.inherit(ydb.node)
function class:set(value)  sum=sum+value  writes=writes+1  return superclass.set(self, value)  end

-- Now use the new class
sum, writes = 0, 0
shoesize = averaged_node('shoesize')
shoesize:set(5)
shoesize:set(10)
shoesize.__ = 15  -- overriding set() also changes the behaviour of: .__ =
print('Average', sum/writes)
-- Average 10.0

isnode (object)¶

Tests whether object is a node object or inherits from a node object.

object: to test

Returns:: boolean true or false

node (varname[, subsarray][, …], node)¶

Creates an object that represents a YottaDB node. This node has all of the class methods defined below. Calling the returned node with one or more string parameters returns a new node further subscripted by those strings. Calling this on an existing node yottadb.node(node) creates an (immutable) copy of node.

Notes:

Several standard Lua operators work on nodes. These are: + - = pairs() tostring()

Although the syntax node:method() is pretty, be aware that it is slow. If you need speed, prefix the node method with two underscores, node:__method(), which is equivalent, but 15x faster. The former is slow because in Lua, node:method(...) is syntactic sugar which expands to node.method(node, ...), causing lua-yottadb to create an intermediate node object node.method. It is only when this new object gets called with (node, ...), and the first parameter is of type node, that the __call metamethod detects it was supposed to be a method access and invokes node.__method(), discarding the intermediate subnode object it created.

Because the __ prefix accesses method names (as above), it cannot access database subnode names starting with __. Instead, use mynode('__nodename') to access a database node named __nodename.

This __ prefix handling also means that object method names that start with two underscores, like __tostring, are only accessible with an additional __ prefix; for example, node:____tostring().

varname: String variable name.
subsarray: (optional) table of subscripts
...: (optional) list of subscripts to append after any elements in optional subsarray table
node: |key: is an existing node or key to copy into a new object (you can turn a key type into a node type this way)

Returns:

node object with metatable yottadb.node

Example:

yottadb.node('varname'[, {subsarray}][, ...])
yottadb.node(node|key[, {}][, ...])
yottadb.node('varname')('sub1', 'sub2')
yottadb.node('varname', 'sub1', 'sub2')
yottadb.node('varname', {'sub1', 'sub2'})
yottadb.node('varname').sub1.sub2
yottadb.node('varname')['sub1']['sub2']

Node class ¶

node:__ipairs ()¶

Not implemented: use pairs(node) or node:__pairs() instead. See alternative usage below. This is not implemented because Lua >=5.3 implements ipairs via __index(). This would mean that __index() would have to treat integer subscript lookup specially, so:

Although node['abc'] => produces a new node so that node.abc.def.ghi works.

node[1] => would have to produce value node(1).__ so ipairs() works.

Since ipairs() will be little used anyway, the consequent inconsistency discourages implementation.

Alternatives using pairs() are as follows:

Examples:

for k,v in pairs(node) do   if not tonumber(k) break end   <do_your_stuff with k,v>   end
 -- this works since M sorts numbers first by default. The order may be changed by specifying a non-default collation on the database

for i=1,1/0 do   v=node[i].__  if not v break then   <do_your_stuff with k,v>   end
 -- alternative that ensures integer keys

node:__pairs ([reverse])¶

Implement pairs() by iterating over the children of a given node. At each child, yielding the triplet: subnode, subnode value (or nil), and subscript. You can use either pairs(node) or node:pairs(). If you need to iterate in reverse (or in Lua 5.1), use node:pairs(reverse) instead of pairs(node).

Caution: for the sake of speed, the iterator supplies a mutable node. This means it can re-use the same node for each iteration by changing its last subscript, making it faster. But if your loop needs to retain a reference to the node after loop iteration, it should create an immutable copy of that node using ydb.node(node). Mutability can be tested for using node:ismutable()

Notes:

pairs() order is guaranteed to equal the M collation sequence order (even though pairs() order is not normally guaranteed for Lua tables). This means that pairs() is a reasonable substitute for ipairs which is not implemented.

This is very slightly slower than node:subscripts() which only iterates subscript names without fetching the node value.

reverse: (optional) Boolean flag iterates in reverse if true

Returns:

3 values: subnode_object, subnode_value_or_nil, subscript

Example:

for subnode,value[,subscript] in pairs(node) do  subnode:incr(value)  end
 -- to double the values of all subnodes of node

node:__repr ()¶

Return raw representation of node's unique memory address.

Returns:: string in hexadecimal format, starting with 0x.

node:delete_tree () [deprecated v3.0]:¶

Deprecated and replaced by kill.

node:dump ([maxlines=30])¶

Dump the specified node tree.

maxlines: (default: 30) Maximum number of lines to output before stopping dump

Returns:: dump as a string

node:get ([default])¶

Get node's value. Equivalent to node.__, but 2.5x slower. If node is subclassed, then node.__ invokes the subclass's node:__get() if it exists.

default: (optional) specify the value to return if the node has no data; if not supplied, nil is the default

Returns:: value of the node

node:gettree ([maxdepth[, filter[, _value[, _depth]]]])¶

Fetch database node and subtree and return a Lua table of it.

Notes:

special field name __ in the returned table indicates the value of the node itself.

Lua tables do not preserve the order YDB subtrees.

maxdepth: (optional) Subscript depth to fetch. A value of nil fetches subscripts of arbitrary depth, i.e. all levels in the tree. A value of 1 fetches the first layer of subscript values only.
filter: (optional) Either nil or a function matching the prototype function(node, node_top_subscript_name, value, recurse, depth)

If filter is nil, all values are fetched unfiltered.

If filter is a function it is invoked on every subscript to allow it to cast/alter every value and recurse flag; note that at node root (depth=0), subscript passed to filter is the empty string "".

Filter may optionally return two items: value and recurse, which must either be the input parameters value and recurse or may be altered:

If filter returns value then gettree() will store it in the table for that database subscript/value; or store nothing if value=nil.

If filter returns recurse=false, it will prevent recursion deeper into that particular subscript. If it returns nil, it will use the original value of recurse.

_value: (optional) For internal use only (to avoid duplicate value fetches, for speed).
_depth: (optional) For internal use only (to record depth of recursion) and must start unspecified (nil).

Returns:

Lua table containing data

Example:

n = ydb.node('^oaks')
n:settree({__='treedata', {shadow=10,angle=30}, {shadow=13,angle=30}})
n:gettree(nil, print)
-- ^oaks          treedata        true    0
-- ^oaks(1)       1       nil     true    1
-- ^oaks(1,"angle")       angle   30      false   2
-- ^oaks(1,"shadow")      shadow  10      false   2
-- ^oaks(2)       2       nil     true    1
-- ^oaks(2,"angle")       angle   30      false   2
-- ^oaks(2,"shadow")      shadow  13      false   2

-- now fetch the tree into a Lua table
tbl = n:gettree()

node:grab ([timeout])¶

Alias for node:lock_incr to acquire or increment a lock matching this node. Returns after timeout, if specified. If timeout is not supplied or is nil, wait forever; timeout of zero means try only once.

timeout: (optional) Number timeout in seconds to wait for the lock.

node:incr ([increment=1])¶

Increment node's value.

increment: (default: 1) Amount to increment by (negative to decrement)

Returns:: the new value

node:kill ()¶

Delete database tree (node and subnodes) pointed to by node object.

node:lock ([timeout])¶

Releases all locks held and attempts to acquire a lock matching only this node. Returns after timeout, if specified. If timeout is not supplied or is nil, wait forever; timeout of zero means try only once.

timeout: (optional) Number timeout in seconds to wait for the lock.

node:lock_decr ()¶

Decrements a lock matching this node, releasing it if zero.

node:lock_incr ([timeout])¶

Attempts to acquire or increment a lock matching this node. Returns after timeout, if specified. If timeout is not supplied or is nil, wait forever; timeout of zero means try only once.

timeout: (optional) Number timeout in seconds to wait for the lock.

node:release ()¶

Alias for node:lock_decr to decrement a lock matching this node, releasing it if zero.

node:set (value)¶

Set node's value. Equivalent to node.__ = x, but 4x slower. If node is subclassed, then node.__ = x invokes the subclass's node:__set(x) if it exists.

value: New value or nil to delete node

Returns:: value

node:settree (tbl[, filter[, _seen]])¶

Populate database from a table. In its simplest form:

n = ydb.node('var')
n:settree({__='berwyn', weight=78, ['!@#$']='junk', appearance={__='handsome', eyes='blue', hair='blond'}, age=ydb.DELETE})

tbl: The table to store into the database:

Special field name tbl.__ sets the value of the node itself, as opposed to a subnode.

Set any table value to yottadb.DELETE to have settree() delete the value of the associated database node. You cannot delete the whole subtree.

filter: (optional) Function of the form function(node, key, value) or nil

If filter is nil, all values are set unfiltered.

If filter is a function(node, key, value) it is invoked on every node to allow it to cast/alter every key name and value.

Filter must return the same or altered: key, value.

Type errors can be handled (or ignored) using this function, too.

If filter returns yottadb.DELETE as value, the key is deleted.

If filter returns nil as key or value, settree() will simply not update the current database value.

_seen: (optional) For internal use only (to prevent accidental duplicate sets: bad because order setting is not guaranteed).

Examples:

n = ydb.node('^oaks')
n:settree({__='treedata', {shadow=10,angle=30}, {shadow=13,angle=30}})
n:dump()

-- outputs:
^oaks="treedata"
^oaks("1","angle")="30"
^oaks("1","shadow")="10"
^oaks("2","angle")="30"
^oaks("2","shadow")="13"

node:subscripts ([reverse])¶

Return iterator over the child subscript names of a node (in M terms, collate from "" to ""). Unlike yottadb.subscripts(), node:subscripts() returns all child subscripts, not subsequent sibling subscripts in the same level.

Very slightly faster than node:__pairs() because it iterates subscript names without fetching the node value.

Note that subscripts() order is guaranteed to equal the M collation sequence.

reverse: (optional) set to true to iterate in reverse order

Returns:

iterator over child subscript names of a node, which returns a sequence of subscript name strings

Example:

ydb = require 'yottadb'
node = ydb.node('^myvar', 'subs1')
for subscript in node:subscripts() do  print subscript  end

Node class properties ¶

node:data ()¶

Fetch the 'data' bitfield of the node that describes whether the node has a data value or subtrees.

Returns:

yottadb.YDB_DATA_UNDEF (no value or subtree) or

yottadb.YDB_DATA_VALUE_NODESC (value, no subtree) or

yottadb.YDB_DATA_NOVALUE_DESC (no value, subtree) or

yottadb.YDB_DATA_VALUE_DESC (value and subtree)

node:depth ()¶

Fetch the depth of the node: how many subscripts it has.

node:has_tree ()¶

Return true if the node has a tree; otherwise false.

node:has_value ()¶

Return true if the node has a value; otherwise false.

node:ismutable ()¶

Return true if the node is mutable; otherwise false.

node:name ()¶

Fetch the name of the node: the rightmost subscript.

node:subsarray ()¶

Return node's subsarray of subscript strings as a table.

node:varname ()¶

Fetch the varname of the node: the leftmost subscript.

Key class ¶

_property_ [deprecated v1.0]:¶

Properties of key object that are accessed with a dot. These properties, listed below, are unlike object methods, which are accessed with a colon. This kind of property access is for backward compatibility.

For example, access data property with: key.data

name: equivalent to node:name()
data: equivalent to node:data()
has_value: equivalent to node:has_value()
has_tree: equivalent to node:has_tree()
value: equivalent to node.__
__varname: database variable name string – for compatibility with a previous version
__subsarray: table array of database subscript name strings – for compatibility with a previous version

key (varname[, subsarray]) [deprecated v1.0]:¶

Creates an object that represents a YDB node; deprecated after v0.1.

key() is a subclass of node() designed to implement deprecated property names for backward compatibility, as follows:

name (this node's subscript or variable name)

value (this node's value in the YottaDB database)

data (see data())

has_value (whether or not this node has a value)

has_tree (whether or not this node has a subtree)

__varname database variable name string – for compatibility with a previous version

__subsarray table array of database subscript name strings – for compatibility with a previous version and deprecated definitions of key:subscript(), key:subscript_next(), key:subscript_previous()

varname: String variable name.
subsarray: (optional) list of subscripts or table subscripts

Returns:: key object of the specified node with metatable yottadb._key

key:delete_node () [deprecated v1.0]:¶

Deprecated way to delete database node value pointed to by node object. Prefer node:set(nil)

key:subscript_next ([reset[, reverse]]) [deprecated v1.0]:¶

Deprecated way to get next sibling subscript.

Note: this starts from the given location and gives the next sibling subscript in the M collation sequence. It operates differently than node:subscripts() which yields all subscripts that are children of the given node. Deprecated because:

It keeps dangerous state in the object, causing bugs when stale references attempt to access defunct state.

It is more Lua-esque to iterate all subscripts in the node (think table) using pairs().

If sibling access becomes a common use-case, it should be reimplemented as an iterator.

reset: (optional) If true, resets to the original subscript before any calls to subscript_next()
reverse: (optional) If true then get previous instead of next

key:subscript_previous ([reset]) [deprecated v1.0]:¶

Deprecated way to get previous sibling subscript. See notes for subscript_previous()

reset: (optional) If true, resets to the original subscript before any calls to subscript_next() or subscript_previous()

key:subscripts ([reverse]) [deprecated v1.0]:¶

Deprecated way to get same-level subscripts from this node onward. Deprecated because:

pairs() is more Lua-esque.

It was non-intuitive that key:subscripts() iterates only subsequent subscripts, not all child subscripts.

reverse: (optional) When set to true, iterates in reverse