From 02184018ff661833dd7b10b463e5cb1ac2b6e98c Mon Sep 17 00:00:00 2001 From: Chris Dyer Date: Fri, 13 Jun 2014 00:23:19 -0400 Subject: tree fragment accepts StringPiece instead of string --- decoder/tree_fragment.cc | 10 +++++----- decoder/tree_fragment.h | 6 ++++-- 2 files changed, 9 insertions(+), 7 deletions(-) diff --git a/decoder/tree_fragment.cc b/decoder/tree_fragment.cc index 696c8601..aad0b2c4 100644 --- a/decoder/tree_fragment.cc +++ b/decoder/tree_fragment.cc @@ -8,7 +8,7 @@ using namespace std; namespace cdec { -TreeFragment::TreeFragment(const string& tree, bool allow_frontier_sites) { +TreeFragment::TreeFragment(const StringPiece& tree, bool allow_frontier_sites) { int bal = 0; const unsigned len = tree.size(); unsigned cur = 0; @@ -50,7 +50,7 @@ void TreeFragment::DebugRec(unsigned cur, ostream* out) const { // cp is the character index in the tree // np keeps track of the nodes (nonterminals) that have been built // symp keeps track of the terminal symbols that have been built -void TreeFragment::ParseRec(const string& tree, bool afs, unsigned cp, unsigned symp, unsigned np, unsigned* pcp, unsigned* psymp, unsigned* pnp) { +void TreeFragment::ParseRec(const StringPiece& tree, bool afs, unsigned cp, unsigned symp, unsigned np, unsigned* pcp, unsigned* psymp, unsigned* pnp) { if (tree[cp] != '(') { cerr << "Expected ( at " << cp << endl; abort(); @@ -79,12 +79,12 @@ void TreeFragment::ParseRec(const string& tree, bool afs, unsigned cp, unsigned // TODO: add a terminal symbol to the current edge const bool is_terminal = tree[t_start] != '[' || (t_end - t_start < 3 || tree[t_end - 1] != ']'); if (is_terminal) { - const unsigned term = TD::Convert(tree.substr(t_start, t_end - t_start)); + const unsigned term = TD::Convert(tree.substr(t_start, t_end - t_start).as_string()); rhs.push_back(term); // cerr << "T='" << TD::Convert(term) << "'\n"; ++terminals; } else { // frontier site (NT but no recursion) - const unsigned nt = TD::Convert(tree.substr(t_start + 1, t_end - t_start - 2)) | FRONTIER_BIT; + const unsigned nt = TD::Convert(tree.substr(t_start + 1, t_end - t_start - 2).as_string()) | FRONTIER_BIT; rhs.push_back(nt); ++frontier_sites; // cerr << "FRONT-NT=[" << TD::Convert(nt & ALL_MASK) << "]\n"; @@ -97,7 +97,7 @@ void TreeFragment::ParseRec(const string& tree, bool afs, unsigned cp, unsigned } // continuent has completed, cp is at ), build node const unsigned j = symp; // span from (i,j) // add an internal non-terminal symbol - const unsigned nt = TD::Convert(tree.substr(nt_start, nt_end - nt_start)) | RHS_BIT; + const unsigned nt = TD::Convert(tree.substr(nt_start, nt_end - nt_start).as_string()) | RHS_BIT; nodes[np] = TreeFragmentProduction(nt, rhs); //cerr << np << " production(" << i << "," << j << ")= " << TD::Convert(nt & ALL_MASK) << " -->"; //for (auto& x : rhs) { diff --git a/decoder/tree_fragment.h b/decoder/tree_fragment.h index 79722b5a..8bb7251a 100644 --- a/decoder/tree_fragment.h +++ b/decoder/tree_fragment.h @@ -8,6 +8,8 @@ #include #include +#include "string_piece.hh" + namespace cdec { class BreadthFirstIterator; @@ -52,7 +54,7 @@ class TreeFragment { public: TreeFragment() : frontier_sites(), terminals() {} // (S (NP a (X b) c d) (VP (V foo) (NP (NN bar)))) - explicit TreeFragment(const std::string& tree, bool allow_frontier_sites = false); + explicit TreeFragment(const StringPiece& tree, bool allow_frontier_sites = false); void DebugRec(unsigned cur, std::ostream* out) const; typedef DepthFirstIterator iterator; typedef ptrdiff_t difference_type; @@ -73,7 +75,7 @@ class TreeFragment { // cp is the character index in the tree // np keeps track of the nodes (nonterminals) that have been built // symp keeps track of the terminal symbols that have been built - void ParseRec(const std::string& tree, bool afs, unsigned cp, unsigned symp, unsigned np, unsigned* pcp, unsigned* psymp, unsigned* pnp); + void ParseRec(const StringPiece& tree, bool afs, unsigned cp, unsigned symp, unsigned np, unsigned* pcp, unsigned* psymp, unsigned* pnp); public: unsigned root; unsigned char frontier_sites; -- cgit v1.2.3