iskra / jsonx Goto Github PK
View Code? Open in Web Editor NEWJSONX is an Erlang library for efficient decode and encode JSON, written in C.
License: Other
JSONX is an Erlang library for efficient decode and encode JSON, written in C.
License: Other
It seems that there is a bug which exhibits itself under heavy load. In my case this means parsing and encoding huge JSON data ~50mb in parallel by ~10 erlang processes. Oh yes that was a Solaris OS, it might be the reason of a bug.
Anyway it might be interesting for you to see the following (from core dump):
# mdb core
Loading modules: [ libc.so.1 libuutil.so.1 ld.so.1 ]
> ::stack
libc_psr.so.1`memcpy+0x58(f977be50, 1e46132e, f977bd78, 4000013, 224000, 44dcc)
jsonx.so`match_string+0x1c(f977be50, 1e46132e, f977bd78, f977ba54, 1e461344, 2)
jsonx.so`match_pair+0x60(f977be50, 1e461342, f977bd78, f977bd94, 640bfffb, 2c)
jsonx.so`match_proplist+0x158(f977be50, 1e460de9, f977bd78, f977bb5c, 1e460fb4, 1e460fb4)
jsonx.so`match_tuple+0xd0(f977be50, 1e460fb2, f977bd78, 0, f977bd78, 62080028)
jsonx.so`match_term+0x154(f977be50, 1e460fb2, f977bd78, f977bd94, 640bfd79, 2c)
jsonx.so`match_list+0x1ac(f977be50, 1a800019, f977bd78, 0, e3279240, 419bc)
jsonx.so`match_term+0xec(f977be50, 1a800019, f977bd78, 7cafc, 0, f977bd78)
jsonx.so`encode_nif+0x174(f977be50, 1, fce86180, 3e7, f80e3b98, 27a84388)
process_main+0x6c4c(4, 4, fa216410, 8, fe435788, 1)
sched_thread_func+0x248(fca7f400, f9d00080, f9d00080, 0, f9d00880, 3)
thr_wrapper+0x70(ffbff128, f977c000, 0, 0, 1a0040, 0)
libc.so.1`_lwp_start(0, 0, 0, 0, 0, 0)
Note: this very long issue contains three different examples, that's why it's so long :)
I'm working with Neo4J REST API. The JSON they return for some of the objects may contain optional fields.
Erlang R16B02
For example, this is JSON for a node which is not in an index:
{
"extensions" : {
},
"paged_traverse" : "http://localhost:7474/db/data/node/376/paged/traverse/{returnType}{?pageSize,leaseTime}",
"outgoing_relationships" : "http://localhost:7474/db/data/node/376/relationships/out",
"traverse" : "http://localhost:7474/db/data/node/376/traverse/{returnType}",
"all_typed_relationships" : "http://localhost:7474/db/data/node/376/relationships/all/{-list|&|types}",
"property" : "http://localhost:7474/db/data/node/376/properties/{key}",
"all_relationships" : "http://localhost:7474/db/data/node/376/relationships/all",
"self" : "http://localhost:7474/db/data/node/376",
"outgoing_typed_relationships" : "http://localhost:7474/db/data/node/376/relationships/out/{-list|&|types}",
"properties" : "http://localhost:7474/db/data/node/376/properties",
"incoming_relationships" : "http://localhost:7474/db/data/node/376/relationships/in",
"incoming_typed_relationships" : "http://localhost:7474/db/data/node/376/relationships/in/{-list|&|types}",
"create_relationship" : "http://localhost:7474/db/data/node/376/relationships",
"data" : {
}
}
And this is JSON for a node that has been added to an index:
{
"extensions" : {
},
"paged_traverse" : "http://localhost:7474/db/data/node/8/paged/traverse/{returnType}{?pageSize,leaseTime}",
"outgoing_relationships" : "http://localhost:7474/db/data/node/8/relationships/out",
"traverse" : "http://localhost:7474/db/data/node/8/traverse/{returnType}",
"all_typed_relationships" : "http://localhost:7474/db/data/node/8/relationships/all/{-list|&|types}",
"property" : "http://localhost:7474/db/data/node/8/properties/{key}",
"all_relationships" : "http://localhost:7474/db/data/node/8/relationships/all",
"self" : "http://localhost:7474/db/data/node/8",
"outgoing_typed_relationships" : "http://localhost:7474/db/data/node/8/relationships/out/{-list|&|types}",
"properties" : "http://localhost:7474/db/data/node/8/properties",
"incoming_relationships" : "http://localhost:7474/db/data/node/8/relationships/in",
"incoming_typed_relationships" : "http://localhost:7474/db/data/node/8/relationships/in/{-list|&|types}",
"create_relationship" : "http://localhost:7474/db/data/node/8/relationships",
"data" : {
},
"indexed" : "http://localhost:7474/db/data/index/node/favorites/some-key/some%20value/8"
}
The only difference is the "indexed" field.
Given this record:
-record(neo4j_node, { extensions
, paged_traverse
, outgoing_relationships
, traverse
, all_typed_relationships
, property
, all_relationships
, self
, outgoing_typed_relationships
, properties
, incoming_relationships
, incoming_typed_relationships
, create_relationship
, data
, labels
}
).
and
> Decoder = jsonx:decoder( [{neo4j_node, record_info(fields, neo4j_node)}]
, [{format, proplist}]).
Decoding the first JSON will yield a correct record:
#neo4j_node{
extensions = [],
paged_traverse =
<<"http://localhost:7474/db/data/node/189/paged/traverse/{returnType}{?pageSize,leaseTime}">>,
outgoing_relationships =
<<"http://localhost:7474/db/data/node/189/relationships/out">>,
traverse =
<<"http://localhost:7474/db/data/node/189/traverse/{returnType}">>,
all_typed_relationships =
<<"http://localhost:7474/db/data/node/189/relationships/all/{-list|&|types}">>,
property =
<<"http://localhost:7474/db/data/node/189/properties/{key}">>,
all_relationships =
<<"http://localhost:7474/db/data/node/189/relationships/all">>,
self = <<"http://localhost:7474/db/data/node/189">>,
outgoing_typed_relationships =
<<"http://localhost:7474/db/data/node/189/relationships/out/{-list|&|types}">>,
properties =
<<"http://localhost:7474/db/data/node/189/properties">>,
incoming_relationships =
<<"http://localhost:7474/db/data/node/189/relationships/in">>,
incoming_typed_relationships =
<<"http://localhost:7474/db/data/node/189/relationships/in/{-list|&|types}">>,
create_relationship =
<<"http://localhost:7474/db/data/node/189/relationships">>,
data = [{<<"born">>,1940},{<<"name">>,<<"Al Pacino">>}],
labels =
<<"http://localhost:7474/db/data/node/189/labels">>}
however, it will fail on the second JSON:
{error,invalid_json,1}
Even though I expected it to return a proplist (as it does for some other structures).
If I add "indexed" to record definition, decoder will fail with {error,invalid_json,1}
for the first JSON and not for the second.
The service root may or may not contain a field called "reference_node":
{
"extensions" : {
},
"node" : "http://localhost:7474/db/data/node",
"reference_node" : "http://localhost:7474/db/data/node/371",
"node_index" : "http://localhost:7474/db/data/index/node",
"relationship_index" : "http://localhost:7474/db/data/index/relationship",
"extensions_info" : "http://localhost:7474/db/data/ext",
"relationship_types" : "http://localhost:7474/db/data/relationship/types",
"batch" : "http://localhost:7474/db/data/batch",
"cypher" : "http://localhost:7474/db/data/cypher",
"neo4j_version" : "1.9.5"
}
Given this record definition:
-record(neo4j_root, { extensions
, node
, reference_node
, node_index
, relationship_index
, extensions_info
, relationship_types
, batch
, cypher
, transaction
, neo4j_version
}
).
and
Decoder = jsonx:decoder( [{neo4j_root, record_info(fields, neo4j_root)}]
, [{format, proplist}]).
this will fail with {error,invalid_json,1}
if reference_node
isn't in the JSON.
For some other JSON this works as expected. JSON for indices can be either
{
"template" : "http://localhost:7474/db/data/index/node/favorites/{key}/{value}"
}
{
"template" : "http://localhost:7474/db/data/index/node/fulltext/{key}/{value}",
"type" : "fulltext",
"provider" : "lucene"
}
Given
-record(neo4j_index, { template
, type
, provider
}
).
...
Decoder = jsonx:decoder( [{neo4j_index, record_info(fields, neo4j_index)}]
, [{format, proplist}]).
the first JSON will be decoded as
[{<<"template">>, <<"http://localhost:7474/db/data/index/node/fulltext/{key}/{value}">>}]
and the second one as
#neo4j_index{template = <<"http://localhost:7474/db/data/index/node/aa/{key}/{value}">>,
type = <<"exact">>,provider = <<"lucene">>}
That is, as expected.
I have no idea why this happens :)
This error does not occur every time, but it occurs.
Don't know how to reproduce it.
=ERROR REPORT==== 15-Apr-2013::06:31:34 ===
The on_load function for module jsonx returned {error,
{load,
"Library load-call unsuccessful."}}
-module(json_t).
-record(json_t, {
name :: binary(),
age = 0 :: pos_integer(),
spouse :: #json_t{}
}).
-record(json_t2, {
name :: binary(),
age = 0 :: pos_integer(),
spouse :: #json_t{}
}).
encoder1() ->
jsonx:encoder([{json_t, record_info(fields, json_t)},
{json_t2, record_info(fields, json_t2)} ],
[{ignore, [undefined]}]).
a() ->
R = #json_t2{ name = <<"John">>, age = 32, spouse = [#json_t{}] },
Encoder = encoder1(),
Res = Encoder(R),
io:format("~p ~n",[Res]).
RESULT:
<<"{"name": "John","age": 32,"spouse": [{,"age": 0}]}">>
If first field in record == undefined.
If one tries to encode large proplist, it is difficult to find property with error because jsonx always points to first tuple:
3> jsonx:encode([{<<"b">>, false},{<<"a">>, 999999999999999999999999999999999999999999999999999}]).
{no_match,{<<"b">>,false}}
Pointing to tuple with error makes error finding easier.
Returns {error, invalid_string, _} when trying to parse string with \ud83d
Just found jsonx can't work with really big numbers:
jsonx:encode([{<<"a">>, 999999999999999999999999999999999999999999999999999999}]).
{no_match,{<<"a">>,
999999999999999999999999999999999999999999999999999999}}
Is there any ways to encode big numbers to valid json?
When I use rebar3 to compile jsonx,got error like this:
=ERROR REPORT==== 19-Apr-2017::11:36:37 ===
Error in process <0.207.0> with exit value: {{badmatch,{error,{load_failed,"Failed to load NIF library: '/home/dasudian/Proj/swdportal/deps/jsonx/_build/test/lib/jsonx/priv/jsonx.so: cannot open shared object file: No such file or directory'"}}},[{jsonx...
Everything is ok when I compile with rebar...
Is there any advice?Thanks
There are a few issues when compiling on MSVC compiler. Specifically:
#ifdef _MSC_VER
#define inline __inline
#endif
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.