diff options
Diffstat (limited to 'data')
-rw-r--r-- | data/Makefile | 3 | ||||
-rw-r--r-- | data/make_paks.cc | 5 | ||||
-rw-r--r-- | data/make_paks2.cc | 121 |
3 files changed, 126 insertions, 3 deletions
diff --git a/data/Makefile b/data/Makefile index e56b229..24d85a3 100644 --- a/data/Makefile +++ b/data/Makefile @@ -1,6 +1,9 @@ make_paks: make_paks.cc g++ -std=c++11 make_paks.cc -I../msgpack-c/include/ ../msgpack-c/lib/libmsgpack.a -o make_paks +make_paks2: make_paks2.cc + g++ -std=c++11 make_paks2.cc -I../msgpack-c/include/ ../msgpack-c/lib/libmsgpack.a -o make_paks2 + clean: rm -f make_paks diff --git a/data/make_paks.cc b/data/make_paks.cc index 3477294..ca6c9b2 100644 --- a/data/make_paks.cc +++ b/data/make_paks.cc @@ -60,15 +60,14 @@ struct Hg { Vector weights; vector<Node> nodes; vector<Edge> edges; - vector<string> rules; - MSGPACK_DEFINE(weights, nodes, edges, rules); + MSGPACK_DEFINE(weights, nodes, edges); }; template<typename X> inline void serialize(jsoncpp::Stream<X>& stream, Hg& o) { - fields(o, stream, "weights", o.weights, "nodes", o.nodes, "edges", o.edges, "rules", o.rules); + fields(o, stream, "weights", o.weights, "nodes", o.nodes, "edges", o.edges); } template<typename X> inline void diff --git a/data/make_paks2.cc b/data/make_paks2.cc new file mode 100644 index 0000000..1b5895b --- /dev/null +++ b/data/make_paks2.cc @@ -0,0 +1,121 @@ +#include <iostream> +#include <fstream> +#include <string> +#include <msgpack.hpp> +#include <msgpack/fbuffer.h> +#include <msgpack/fbuffer.hpp> + + +/* + * https://github.com/ascheglov/json-cpp + * + */ +#include "../json-cpp.hpp" + +using namespace std; + + +struct Node { + int id; + string cat; + vector<int> span; + + MSGPACK_DEFINE(id, cat, span); +}; + +struct Vector { + double CountEF; + double EgivenFCoherent; + double Glue; + double IsSingletonF; + double IsSingletonFE; + double LanguageModel; + double LanguageModel_OOV; + double MaxLexFgivenE; + double MaxLexEgivenF; + double PassThrough; + double PassThrough_1; + double PassThrough_2; + double PassThrough_3; + double PassThrough_4; + double PassThrough_5; + double PassThrough_6; + double SampleCountF; + double WordPenalty; + + MSGPACK_DEFINE(CountEF, EgivenFCoherent, Glue, IsSingletonF, IsSingletonFE, LanguageModel, LanguageModel_OOV, MaxLexEgivenF, MaxLexFgivenE, PassThrough, PassThrough_1, PassThrough_2, PassThrough_3, PassThrough_4, PassThrough_5, PassThrough_6, SampleCountF, WordPenalty); +}; + +struct Edge { + int head; + string rule; + vector<int> tails; + Vector f; + double weight; + + MSGPACK_DEFINE(head, rule, tails, f, weight); +}; + +struct Hg { + Vector weights; + vector<Node> nodes; + vector<Edge> edges; + + MSGPACK_DEFINE(weights, nodes, edges); +}; + +template<typename X> inline void +serialize(jsoncpp::Stream<X>& stream, Hg& o) +{ + fields(o, stream, "weights", o.weights, "nodes", o.nodes, "edges", o.edges); +} + +template<typename X> inline void +serialize(jsoncpp::Stream<X>& stream, Edge& o) +{ + fields(o, stream, "head", o.head, "rule", o.rule, "tails", o.tails, "f", o.f, "weight", o.weight); +} + +template<typename X> inline void +serialize(jsoncpp::Stream<X>& stream, Vector& o) +{ + fields(o, stream, "EgivenFCoherent", o.EgivenFCoherent, "SampleCountF", o.SampleCountF, "CountEF", o.CountEF, "MaxLexFgivenE", o.MaxLexFgivenE, "MaxLexEgivenF", o.MaxLexEgivenF, "IsSingletonF", o.IsSingletonF, "IsSingletonFE", o.IsSingletonFE, "LanguageModel", o.LanguageModel, "LanguageModel_OOV", o.LanguageModel_OOV, "PassThrough", o.PassThrough, "PassThrough_1", o.PassThrough_1, "PassThrough_2", o.PassThrough_2, "PassThrough_3", o.PassThrough_3, "PassThrough_4", o.PassThrough_4, "PassThrough_5", o.PassThrough_5, "PassThrough_6", o.PassThrough_6, "WordPenalty", o.WordPenalty, "Glue", o.Glue); +} + +template<typename X> inline void +serialize(jsoncpp::Stream<X>& stream, Node& o) +{ + fields(o, stream, "id", o.id, "cat", o.cat, "span", o.span); +} + +int +main(int argc, char** argv) +{ + ifstream ifs(argv[1]); + string json_str((istreambuf_iterator<char>(ifs) ), + (istreambuf_iterator<char>())); + + Hg hg; + Vector w; + hg.weights = w; + vector<Node> nodes; + hg.nodes = nodes; + vector<Edge> edges; + hg.edges = edges; + jsoncpp::parse(hg, json_str); + + FILE* file = fopen(argv[2], "wb"); + msgpack::fbuffer fbuf(file); + msgpack::pack(fbuf, hg.nodes.size()); + msgpack::pack(fbuf, hg.edges.size()); + msgpack::pack(fbuf, hg.weights); + for (auto it = hg.nodes.begin(); it != hg.nodes.end(); it++) + msgpack::pack(fbuf, *it); + for (auto it = hg.edges.begin(); it != hg.edges.end(); it++) + msgpack::pack(fbuf, *it); + + fclose(file); + + return 0; +} + |