Add copy_strings feature

Some users of Jiffy have experienced issues when decoding large JSON
documents. Normally Jiffy expects smallish documents and returns any
strings as sub-binaries. When dealing with large documents these
sub-binary references can keep a large amount of RAM around unless the
user goes through and applies `binary:copy/1` on every string returned
from Jiffy. This however causes a large amount of CPU usage to do
something that Jiffy could do as it builds the JSON structure.

The `copy_strings` decoder option does exactly this. Instead of
returning sub-binaries Jiffy now copies every string into a newly
allocated binary. Users report that this fixes the memory issues while
also not negatively affecting performance significantly.

README.md

@@ -52,6 +52,12 @@ The options for decode are:
   will ensure that the parsed object only contains a single entry
   containing the last value seen. This mirrors the parsing beahvior
   of virtually every other JSON parser.
+* `copy_strings` - Normaly when strings are decoded they are created
+  as sub-binaries of the input data. With some workloads this can lead
+  to an undeseriable bloating of memory when a few small strings in JSON
+  keep a reference to the full JSON document alive. Setting this option
+  will instead allocate new binaries for each string to avoid keeping
+  the original JSON document around after garbage collection.
 * `{bytes_per_red, N}` where N >= 0 - This controls the number of
   bytes that Jiffy will process as an equivalent to a reduction. Each
   20 reductions we consume 1% of our allocated time slice for the current

c_src/decoder.c

@@ -54,6 +54,7 @@ typedef struct {
     int             return_maps;
     int             return_trailer;
     int             dedupe_keys;
+    int             copy_strings;
     ERL_NIF_TERM    null_term;
 
     char*           p;
@@ -84,6 +85,7 @@ dec_new(ErlNifEnv* env)
     d->return_maps = 0;
     d->return_trailer = 0;
     d->dedupe_keys = 0;
+    d->copy_strings = 0;
     d->null_term = d->atoms->atom_null;
 
     d->p = NULL;
@@ -291,9 +293,14 @@ dec_string(Decoder* d, ERL_NIF_TERM* value)
     return 0;
 
 parse:
-    if(!has_escape) {
+    if(!has_escape && !d->copy_strings) {
         *value = enif_make_sub_binary(d->env, d->arg, st, (d->i - st - 1));
         return 1;
+    } else if(!has_escape) {
+        ulen = d->i - 1 - st;
+        chrbuf = (char*) enif_make_new_binary(d->env, ulen, value),
+        memcpy(chrbuf, &(d->p[st]), ulen);
+        return 1;
     }
 
     hi = 0;
@@ -684,6 +691,8 @@ decode_init(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
             d->return_trailer = 1;
         } else if(enif_compare(val, d->atoms->atom_dedupe_keys) == 0) {
             d->dedupe_keys = 1;
+        } else if(enif_compare(val, d->atoms->atom_copy_strings) == 0) {
+            d->copy_strings = 1;
         } else if(enif_compare(val, d->atoms->atom_use_nil) == 0) {
             d->null_term = d->atoms->atom_nil;
         } else if(get_null_term(env, val, &(d->null_term))) {

c_src/jiffy.c

@@ -33,6 +33,7 @@ load(ErlNifEnv* env, void** priv, ERL_NIF_TERM info)
     st->atom_null_term = make_atom(env, "null_term");
     st->atom_escape_forward_slashes = make_atom(env, "escape_forward_slashes");
     st->atom_dedupe_keys = make_atom(env, "dedupe_keys");
+    st->atom_copy_strings = make_atom(env, "copy_strings");
 
     // Markers used in encoding
     st->ref_object = make_atom(env, "$object_ref$");

c_src/jiffy.h

@@ -36,6 +36,7 @@ typedef struct {
     ERL_NIF_TERM    atom_null_term;
     ERL_NIF_TERM    atom_escape_forward_slashes;
     ERL_NIF_TERM    atom_dedupe_keys;
+    ERL_NIF_TERM    atom_copy_strings;
 
     ERL_NIF_TERM    ref_object;
     ERL_NIF_TERM    ref_array;

test/jiffy_17_copy_strings_tests.erl

@@ -0,0 +1,35 @@
+% This file is part of Jiffy released under the MIT license.
+% See the LICENSE file for more information.
+
+-module(jiffy_17_copy_strings_tests).
+
+-include_lib("eunit/include/eunit.hrl").
+
+
+check_binaries({Props}) when is_list(Props) ->
+    lists:all(fun({Key, Value}) ->
+        check_binaries(Key) andalso check_binaries(Value)
+    end, Props);
+check_binaries(Values) when is_list(Values) ->
+    lists:all(fun(Value) ->
+        check_binaries(Value)
+    end, Values);
+check_binaries(Bin) when is_binary(Bin) ->
+    io:format("~s :: ~p ~p", [Bin, byte_size(Bin), binary:referenced_byte_size(Bin)]),
+    byte_size(Bin) == binary:referenced_byte_size(Bin);
+check_binaries(Bin) ->
+    true.
+
+
+copy_strings_test_() ->
+    Opts = [copy_strings],
+    Cases = [
+        <<"\"foo\"">>,
+        <<"[\"bar\"]">>,
+        <<"{\"foo\":\"bar\"}">>,
+        <<"{\"foo\":[\"bar\"]}">>
+    ],
+    {"Test copy_strings", lists:map(fun(Json) ->
+        EJson = jiffy:decode(Json, Opts),
+        ?_assert(check_binaries(EJson))
+    end, Cases)}.