From 6da0534d0335e60e349582590e1aa60907c84171 Mon Sep 17 00:00:00 2001
From: Kenneth Heafield <github@kheafield.com>
Date: Tue, 8 May 2012 14:57:45 -0400
Subject: Output separate files for target hypergraphs.
---
decoder/decoder.cc | 4 ++--
decoder/hg_io.cc | 17 +++++++++++------
decoder/hg_io.h | 3 ++-
3 files changed, 15 insertions(+), 9 deletions(-)
(limited to 'decoder')
diff --git a/decoder/decoder.cc b/decoder/decoder.cc
index ec6f75f7..4ce2ba86 100644
--- a/decoder/decoder.cc
+++ b/decoder/decoder.cc
@@ -407,7 +407,7 @@ DecoderImpl::DecoderImpl(po::variables_map& conf, int argc, char** argv, istream
("show_partition,z", "Compute and show the partition (inside score)")
("show_conditional_prob", "Output the conditional log prob to STDOUT instead of a translation")
("show_cfg_search_space", "Show the search space as a CFG")
- ("show_target_graph", "Output the target hypergraph")
+ ("show_target_graph", po::value<string>(), "Directory to write the target hypergraphs to")
("coarse_to_fine_beam_prune", po::value<double>(), "Prune paths from coarse parse forest before fine parse, keeping paths within exp(alpha>=0)")
("ctf_beam_widen", po::value<double>()->default_value(2.0), "Expand coarse pass beam by this factor if no fine parse is found")
("ctf_num_widenings", po::value<int>()->default_value(2), "Widen coarse beam this many times before backing off to full parse")
@@ -816,7 +816,7 @@ bool DecoderImpl::Decode(const string& input, DecoderObserver* o) {
}
if (conf.count("show_target_graph"))
- HypergraphIO::WriteTarget(forest);
+ HypergraphIO::WriteTarget(conf["show_target_graph"].as<string>(), sent_id, forest);
for (int pass = 0; pass < rescoring_passes.size(); ++pass) {
const RescoringPass& rp = rescoring_passes[pass];
diff --git a/decoder/hg_io.cc b/decoder/hg_io.cc
index d416dbf6..734c2ce8 100644
--- a/decoder/hg_io.cc
+++ b/decoder/hg_io.cc
@@ -1,5 +1,6 @@
#include "hg_io.h"
+#include <fstream>
#include <sstream>
#include <iostream>
@@ -651,22 +652,26 @@ void HypergraphIO::WriteAsCFG(const Hypergraph& hg) {
* for each downward edge:
* RHS with [vertex_index] for NTs ||| scores
*/
-void HypergraphIO::WriteTarget(const Hypergraph& hg) {
- cout << hg.nodes_.size() << ' ' << hg.edges_.size() << '\n';
+void HypergraphIO::WriteTarget(const std::string &base, unsigned int id, const Hypergraph& hg) {
+ std::string name(base);
+ name += '/';
+ name += boost::lexical_cast<std::string>(id);
+ std::fstream out(name.c_str(), std::fstream::out);
+ out << hg.nodes_.size() << ' ' << hg.edges_.size() << '\n';
for (unsigned int i = 0; i < hg.nodes_.size(); ++i) {
const Hypergraph::EdgesVector &edges = hg.nodes_[i].in_edges_;
- cout << edges.size() << '\n';
+ out << edges.size() << '\n';
for (unsigned int j = 0; j < edges.size(); ++j) {
const Hypergraph::Edge &edge = hg.edges_[edges[j]];
const std::vector<WordID> &e = edge.rule_->e();
for (std::vector<WordID>::const_iterator word = e.begin(); word != e.end(); ++word) {
if (*word <= 0) {
- cout << '[' << edge.tail_nodes_[-*word] << "] ";
+ out << '[' << edge.tail_nodes_[-*word] << "] ";
} else {
- cout << TD::Convert(*word) << ' ';
+ out << TD::Convert(*word) << ' ';
}
}
- cout << "||| " << edge.rule_->scores_ << '\n';
+ out << "||| " << edge.rule_->scores_ << '\n';
}
}
}
diff --git a/decoder/hg_io.h b/decoder/hg_io.h
index 4e502a0c..58af8132 100644
--- a/decoder/hg_io.h
+++ b/decoder/hg_io.h
@@ -2,6 +2,7 @@
#define _HG_IO_H_
#include <iostream>
+#include <string>
#include "lattice.h"
class Hypergraph;
@@ -24,7 +25,7 @@ struct HypergraphIO {
static void WriteAsCFG(const Hypergraph& hg);
// Write only the target size information in bottom-up order.
- static void WriteTarget(const Hypergraph& hg);
+ static void WriteTarget(const std::string &base, unsigned int sent_id, const Hypergraph& hg);
// serialization utils
static void ReadFromPLF(const std::string& in, Hypergraph* out, int line = 0);
--
cgit v1.2.3