From e834fd628f61aca04f98691eb56e9808c33c6787 Mon Sep 17 00:00:00 2001 From: Patrick Simianer
Date: Thu, 12 Jun 2014 14:39:37 +0200 Subject: cleanup --- util/cdec2json.py | 82 +++++++++++++++++++++++++++++++++++++++++++++++++ util/cdec_hg_to_json.py | 82 ------------------------------------------------- 2 files changed, 82 insertions(+), 82 deletions(-) create mode 100755 util/cdec2json.py delete mode 100755 util/cdec_hg_to_json.py (limited to 'util') diff --git a/util/cdec2json.py b/util/cdec2json.py new file mode 100755 index 0000000..76e2cd4 --- /dev/null +++ b/util/cdec2json.py @@ -0,0 +1,82 @@ +#!/usr/bin/env python2 + +import cdec +import sys, argparse + + +def hg2json(hg, weights): + """ + output a JSON representation of a cdec hypegraph + """ + res = '' + res += "{\n" + res += '"weights":{'+"\n" + a = [] + for i in weights: + if i[1] != 0: + a.append( '"%s":%s'%(i[0], i[1]) ) + res += ", ".join(a)+"\n" + res += "},\n" + res += '"nodes":'+"\n" + res += "[\n" + a = [] + a.append( '{ "id":-1, "cat":"root", "span":[-1,-1] }' ) + for i in hg.nodes: + a.append('{ "id":%d, "cat":"%s", "span":[%d,%d] }'%(i.id, i.cat, i.span[0], i.span[1])) + res += ",\n".join(a)+"\n" + res += "],\n" + res += '"edges":'+"\n" + res += "[\n" + a = [] + for i in hg.edges: + s = "{" + s += '"head":%d'%(i.head_node.id) + s += ', "rule":"%s"'%(i.trule) + # f + xs = ' "f":{' + b = [] + for j in i.feature_values: + b.append( '"%s":%s'%(j[0], j[1]) ) + xs += ", ".join(b) + xs += "}," + # tails + if len(list(i.tail_nodes)) > 0: + s += ', "tails":[ %s ],'%(",".join([str(n.id) for n in i.tail_nodes])) + else: + s += ', "tails":[ -1 ],' + s += xs + s += ' "weight":%s }'%(i.prob) + a.append(s) + res += ",\n".join(a)+"\n" + res += "]\n" + res += "}\n" + return res + +def main(): + parser = argparse.ArgumentParser(description='get a proper json representation of cdec hypergraphs') + parser.add_argument('-c', '--config', required=True, help='decoder configuration') + parser.add_argument('-w', '--weights', required=True, help='feature weights') + args = parser.parse_args() + with open(args.config) as config: + config = config.read() + decoder = cdec.Decoder(config) + decoder.read_weights(args.weights) + ins = sys.stdin.readline().strip() + hg = decoder.translate(ins) + + sys.stderr.write( "input:\n '%s'\n"%(ins) ) + sys.stderr.write( "viterbi translation:\n '%s'\n"%(hg.viterbi()) ) + num_nodes = 0 + for i in hg.nodes: num_nodes+=1 + sys.stderr.write( "# nodes = %s\n"%(num_nodes) ) + num_edges = 0 + for i in hg.edges: num_edges+=1 + sys.stderr.write( "# edges = %s\n"%(num_edges) ) + sys.stderr.write( "viterbi score = %s\n"%(round(hg.viterbi_features().dot(decoder.weights), 2)) ) + + print hg2json(hg, decoder.weights) + + +if __name__=="__main__": + main() + diff --git a/util/cdec_hg_to_json.py b/util/cdec_hg_to_json.py deleted file mode 100755 index 2fcc409..0000000 --- a/util/cdec_hg_to_json.py +++ /dev/null @@ -1,82 +0,0 @@ -#!/usr/bin/env python2 - -import cdec -import sys, argparse - -def hg2json(hg, weights): - """ - output a JSON representation of a cdec hypegraph - (see http://aclweb.org/aclwiki/index.php?title=Hypergraph_Format ) - """ - res = '' - res += "{\n" - res += '"weights":{'+"\n" - a = [] - for i in weights: - if i[1] != 0: - a.append( '"%s":%s'%(i[0], i[1]) ) - res += ", ".join(a)+"\n" - res += "},\n" - res += '"nodes":'+"\n" - res += "[\n" - a = [] - a.append( '{ "id":-1, "cat":"root", "span":[-1,-1] }' ) - for i in hg.nodes: - a.append('{ "id":%d, "cat":"%s", "span":[%d,%d] }'%(i.id, i.cat, i.span[0], i.span[1])) - res += ",\n".join(a)+"\n" - res += "],\n" - res += '"edges":'+"\n" - res += "[\n" - a = [] - for i in hg.edges: - s = "{" - s += '"head":%d'%(i.head_node.id) - s += ', "rule":"%s"'%(i.trule) - # f - xs = ' "f":{' - b = [] - for j in i.feature_values: - b.append( '"%s":%s'%(j[0], j[1]) ) - xs += ", ".join(b) - xs += "}," - # tails - if len(list(i.tail_nodes)) > 0: - s += ', "tails":[ %s ],'%(",".join([str(n.id) for n in i.tail_nodes])) - else: - s += ', "tails":[ -1 ],' - s += xs - s += ' "weight":%s }'%(i.prob) - a.append(s) - res += ",\n".join(a)+"\n" - res += "]\n" - res += "}\n" - return res - -def main(): - parser = argparse.ArgumentParser(description='get a proper json representation of cdec hypergraphs') - parser.add_argument('-c', '--config', required=True, help='decoder configuration') - parser.add_argument('-w', '--weights', required=True, help='feature weights') - args = parser.parse_args() - with open(args.config) as config: - config = config.read() - decoder = cdec.Decoder(config) - decoder.read_weights(args.weights) - ins = sys.stdin.readline().strip() - hg = decoder.translate(ins) - - sys.stderr.write( "input:\n '%s'\n"%(ins) ) - sys.stderr.write( "viterbi translation:\n '%s'\n"%(hg.viterbi()) ) - num_nodes = 0 - for i in hg.nodes: num_nodes+=1 - sys.stderr.write( "# nodes = %s\n"%(num_nodes) ) - num_edges = 0 - for i in hg.edges: num_edges+=1 - sys.stderr.write( "# edges = %s\n"%(num_edges) ) - sys.stderr.write( "viterbi score = %s\n"%(round(hg.viterbi_features().dot(decoder.weights), 2)) ) - - print hg2json(hg, decoder.weights) - - -if __name__=="__main__": - main() - -- cgit v1.2.3