path: root/manual.txt

Lua CJSON Manual
================
Mark Pulford <mark@kyne.com.au>
v1.0.4, November 30 2011

Overview
--------

Lua CJSON provides fast JSON parsing and encoding support for Lua.

.Features
- More than 10x to 20x faster than the efficient pure Lua JSON modules.
- Full support for JSON with UTF-8, including decoding surrogate
  pairs.
- Optional run-time support for common exceptions to the JSON
  specification (NaN, Inf,..).

.Caveats
- UTF-16 and UTF-32 are not supported.
- Multi-threading within a single Lua state is not supported.
  However, this is not a recommended configuration under Lua.

Lua CJSON is covered by the MIT license. See the file +LICENSE+ for
details.

The latest version of this software is available from the 
http://www.kyne.com.au/~mark/software/lua-cjson.php[Lua CJSON website].

Feel free to email me if you have any patches, suggestions, or
comments.


Installation Methods
--------------------

Lua CJSON requires either http://www.lua.org[Lua] or
http://www.luajit.org[LuaJIT] to build.

There are 3 build methods available:

- Make: POSIX, OSX
- RPM: Various Linux distributions
- http://www.luarocks.org[LuaRocks]: POSIX, OSX, Windows


Make
~~~~

Review and update the included Makefile to suit your platform. Next,
build and install the module:

[source,sh]
-----------
make install
-----------

Or install manually:

[source,sh]
-----------
make
cp cjson.so $your_lua_module_directory
-----------


RPM
~~~

Linux distributions using RPM should be able to create a package via
the included RPM spec file. Install the +rpm-build+ package (or
similar) then:

[source,sh]
-----------
rpmbuild -tb lua-cjson-1.0.4.tar.gz
rpm -Uvh $newly_built_lua_cjson_rpm
-----------


LuaRocks
~~~~~~~~

http://luarocks.org[LuaRocks] can be used to install and manage Lua
modules on a wide range of platforms (including Windows).

Extract the Lua CJSON source package into a directory and run:

[source,sh]
-----------
cd lua-cjson-1.0.4
luarocks make
-----------

LuaRocks does not support platform specific configuration for Solaris.
On Solaris, you may need to manually uncomment +USE_INTERNAL_ISINF+ in
the rockspec before building this module.

See the http://luarocks.org/en/Documentation[LuaRocks documentation] for
further details.


Lua API
-------

Synopsis
~~~~~~~~

[source,lua]
------------
require "cjson"
-- Or:
local cjson = require "cjson"
 
-- Translate Lua value to/from JSON
text = cjson.encode(value)
value = cjson.decode(text)
 
-- Get and/or set Lua CJSON configuration
setting = cjson.refuse_invalid_numbers([setting])
depth = cjson.encode_max_depth([depth])
convert, ratio, safe = cjson.encode_sparse_array([convert[, ratio[, safe]]])
keep = cjson.encode_keep_buffer([keep])
------------


cjson.encode
~~~~~~~~~~~~

[source,lua]
------------
json_text = cjson.encode(value)
------------

+cjson.encode+ will serialise a Lua value into a string containing the
JSON representation.

+cjson.encode+ supports the following types:

- +boolean+
- +lightuserdata+ (NULL value only)
- +nil+
- +number+
- +string+
- +table+

The remaining Lua types cannot be serialised:

- +function+
- +lightuserdata+ (non-NULL values)
- +thread+
- +userdata+

Numbers are encoded using the standard Lua number format.

Lua CJSON will escape the following characters within each string:

- Control characters (ASCII 0 - 31)
- Double quote (ASCII 34)
- Forward slash (ASCII 47)
- Blackslash (ASCII 92)
- Delete (ASCII 127)

All other characters are passed transparently. Any UTF-8 error
checking must be done by the application.

Lua CJSON uses a heuristic to determine whether to encode a Lua table
as a JSON array or an object. A Lua table which only has positive
integers (>0) keys of type +number+ will be encoded as a JSON array.
All other tables will be encoded as a JSON object.

Missing entries from sparse Lua arrays are encoded as +null+. For
example:

.Lua input
[source,lua]
------------
{ [3] = "data" }
------------

.JSON output
[source,javascript]
-------------------
[null,null,"data"]
-------------------

[NOTE]
Standards compliant JSON must be encapsulated in either an
object (+{}+) or an array (+[]+). Hence a table must be passed to
+cjson.encode+ to generate standards compliant JSON output.

By default, the following will generate errors:

- Excessively <<sparse_arrays,sparse arrays>>
- More than 20 nested tables
- Invalid numbers (NaN, Infinity)

These defaults can be changed with:

- +cjson.encode_sparse_array+
- +cjson.encode_max_depth+
- +cjson.refuse_invalid_numbers+

.Example: encoding
[source,lua]
------------
data_obj = { true, { foo = "bar" } }
data_json = cjson.encode(data_obj)
------------


cjson.decode
~~~~~~~~~~~~

[source,lua]
------------
value = cjson.decode(json_text)
------------

+cjson.decode+ will deserialise any UTF-8 JSON string into a Lua value
or table. It may return any of the types that +cjson.encode+ supports.

UTF-16 and UTF-32 JSON strings are not supported.

+cjson.decode+ requires that any NULL (ASCII 0) and double quote
(ASCII 34) characters are escaped within strings. All escape codes
will be decoded and other characters will be passed transparently.
UTF-8 characters are not validated during decoding and should be
checked elsewhere if required.

JSON +null+ will be converted to a NULL +lightuserdata+ value. This
can be compared with +cjson.null+ for convenience.

By default, _invalid_ numbers (NaN, Infinity, Hexidecimal) will be
decoded. This default can be changed with
+cjson.refuse_invalid_numbers+.

.Example: decoding
[source,lua]
------------
data_json = '[ true, { "foo": "bar" } ]'
data_obj = cjson.decode(data_json)
------------


cjson.refuse_invalid_numbers
~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[source,lua]
------------
setting = cjson.refuse_invalid_numbers([setting])
-- "setting" must be on of:
--       false, "encode", "decode", "both", true
------------

Lua CJSON considers numbers which are outside the JSON specification to be
_invalid_:

- Infinity
- NaN
- Hexadecimal numbers

By default Lua CJSON will decode _invalid_ numbers, but will refuse to
encode them.

This setting can be configured separately for encoding and/or
decoding:

Enabled::
  an error will be generated if an _invalid_ number is found.
Disabled (encoding)::
  NaN and Infinity can be encoded.
Disabled (decoding)::
  All numbers supported by +strtod+(3) will be parsed.


[[sparse_arrays]]
cjson.encode_sparse_array
~~~~~~~~~~~~~~~~~~~~~~~~~

[source,lua]
------------
convert, ratio, safe = cjson.encode_sparse_array([convert[, ratio[, safe]]])
-- "convert" must be a boolean. Default: false.
-- "ratio" must be a positive integer (>0). Default: 2
-- "safe" must be a positive integer (>0). Default: 10
------------

A Lua array is sparse if it is missing a value for at least 1 index.
Lua CJSON encodes missing values as JSON +null+. Eg, Lua array:

.Lua input
[source,lua]  
------------
{ [3] = "sparse" }
------------

.JSON output
[source,javascript]
-------------------
[null,null,"sparse"]
-------------------

Lua CJSON detects excessively sparse arrays by comparing the number of
items in a Lua array with the maximum index. In particular:

-----
maximum index > safe AND maximum index > array_items * ratio
-----

By default, attempting to encode excessively sparse arrays will
generate an error.

If _convert_ is set to +true+, excessively sparse arrays will be
encoded as a JSON object:

.Lua input
[source,lua]
{ [1000] = "excessively sparse" }

.JSON output
[source,javascript]
{"1000":"excessively sparse"}

Setting +ratio+ to +0+ disables checking for excessively sparse arrays.


cjson.encode_max_depth
~~~~~~~~~~~~~~~~~~~~~~

[source,lua]
------------
depth = cjson.encode_max_depth([depth])
-- "depth" must be a positive integer (>0).
------------

By default, Lua CJSON will reject data structure with more than 20 nested
tables.

This check prevents a deeply nested or recursive data structure from
crashing the application.

.Example
[source,lua]
a = {}; b = { a }; a[1] = b


cjson.encode_number_precision
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[source,lua]
------------
precision = cjson.encode_number_precision([precision])
-- "precision" must be between 1 and 14 (inclusive)
------------

By default, Lua CJSON will output 14 significant digits when
converting a number to text.

Reducing number precision to _3_ can improve performance of number
heavy JSON conversions by up to 50%.


cjson.keep_encode_buffer
~~~~~~~~~~~~~~~~~~~~~~~~

[source,lua]
------------
keep = cjson.keep_encode_buffer([keep])
-- "keep" must be a boolean
------------

By default, Lua CJSON will reuse the JSON encoding buffer to improve
performance. The buffer will grow to the largest size required and is
not freed until the Lua CJSON module is garbage collected. Setting this
option to +false+ will cause the buffer to be freed after each call to
+cjson.encode+.


cjson.version
~~~~~~~~~~~~~

[source,lua]
------------
ver = cjson.version
-- variable containing the package version (1.0.4)
------------

FIXME


JSON and handling under Lua CJSON
---------------------------------

Nulls
~~~~~

Lua CJSON decodes JSON +null+ as a Lua lightuserdata NULL pointer.

As a convenience, +cjson.null+ is provided for comparison.


Table keys
~~~~~~~~~~

If all Lua table keys are type +number+ (not +string+), Lua CJSON will
encode the table as a JSON array. See
<<sparse_arrays,cjson.encode_sparse_array>> above for more details.

All other tables will be encoded as a JSON object.

JSON requires that object keys must be type string. +cjson.encode+
will convert numeric keys to a string, and other non-string types will
generate an error.

[CAUTION]
=========
Care must be taken when encoding/decoding objects which contain keys of type
+number+.
[source,lua]
------------
val = cjson.decode(cjson.encode{[1] = "1", a = "a"})
-- Returns: { ["1"] = "1", ...
-- NOT:     { [1] = "1", ...
------------
=========

Metamethods
~~~~~~~~~~~

Lua CJSON does not use metamethods when serialising tables.

- +rawget+ is used to iterate over Lua arrays.
- +next+ is used to iterate over Lua objects.


Functions, Userdata, Threads
~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Lua CJSON will generate an error if asked to serialise Lua functions,
userdata, lightuserdata or threads.


Locales
~~~~~~~

Lua CJSON uses +strtod+(3) and +snprintf+(3) to perform numeric conversion
as they are usually well supported, fast and bug free.

To ensure JSON encoding/decoding works correctly for locales using
comma decimal separators, Lua CJSON must be compiled with either
+USE_POSIX_USELOCALE+ or +USE_POSIX_SETLOCALE+. See the +Makefile+ or the
rockspec for details.


[sect1]
References
----------

- http://tools.ietf.org/html/rfc4627[RFC 4627]
- http://www.json.org/[JSON website]