From 851e389dffdd6996ea32d70defb8906de80b9edc Mon Sep 17 00:00:00 2001
From: Chris Dyer <redpony@gmail.com>
Date: Mon, 14 Dec 2009 20:35:11 -0500
Subject: few small fixes of alignment tools, add new orthographic similarity
 feature for word aligner, final naming of directories, libraries in cdec

---
 decoder/stringlib.cc | 97 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 97 insertions(+)
 create mode 100644 decoder/stringlib.cc

(limited to 'decoder/stringlib.cc')

diff --git a/decoder/stringlib.cc b/decoder/stringlib.cc
new file mode 100644
index 00000000..3ed74bef
--- /dev/null
+++ b/decoder/stringlib.cc
@@ -0,0 +1,97 @@
+#include "stringlib.h"
+
+#include <cstdlib>
+#include <cassert>
+#include <iostream>
+#include <map>
+
+#include "lattice.h"
+
+using namespace std;
+
+void ParseTranslatorInput(const string& line, string* input, string* ref) {
+  size_t hint = 0;
+  if (line.find("{\"rules\":") == 0) {
+    hint = line.find("}}");
+    if (hint == string::npos) {
+      cerr << "Syntax error: " << line << endl;
+      abort();
+    }
+    hint += 2;
+  }
+  size_t pos = line.find("|||", hint);
+  if (pos == string::npos) { *input = line; return; }
+  ref->clear();
+  *input = line.substr(0, pos - 1);
+  string rline = line.substr(pos + 4);
+  if (rline.size() > 0) {
+    assert(ref);
+    *ref = rline;
+  }
+}
+
+void ParseTranslatorInputLattice(const string& line, string* input, Lattice* ref) {
+  string sref;
+  ParseTranslatorInput(line, input, &sref);
+  if (sref.size() > 0) {
+    assert(ref);
+    LatticeTools::ConvertTextOrPLF(sref, ref);
+  }
+}
+
+void ProcessAndStripSGML(string* pline, map<string, string>* out) {
+  map<string, string>& meta = *out;
+  string& line = *pline;
+  string lline = LowercaseString(line);
+  if (lline.find("<seg")!=0) return;
+  size_t close = lline.find(">");
+  if (close == string::npos) return; // error
+  size_t end = lline.find("</seg>");
+  string seg = Trim(lline.substr(4, close-4));
+  string text = line.substr(close+1, end - close - 1);
+  for (size_t i = 1; i < seg.size(); i++) {
+    if (seg[i] == '=' && seg[i-1] == ' ') {
+      string less = seg.substr(0, i-1) + seg.substr(i);
+      seg = less; i = 0; continue;
+    }
+    if (seg[i] == '=' && seg[i+1] == ' ') {
+      string less = seg.substr(0, i+1);
+      if (i+2 < seg.size()) less += seg.substr(i+2);
+      seg = less; i = 0; continue;
+    }
+  }
+  line = Trim(text);
+  if (seg == "") return;
+  for (size_t i = 1; i < seg.size(); i++) {
+    if (seg[i] == '=') {
+      string label = seg.substr(0, i);
+      string val = seg.substr(i+1);
+      if (val[0] == '"') {
+        val = val.substr(1);
+        size_t close = val.find('"');
+        if (close == string::npos) {
+          cerr << "SGML parse error: missing \"\n";
+          seg = "";
+          i = 0;
+        } else {
+          seg = val.substr(close+1);
+          val = val.substr(0, close);
+          i = 0;
+        }
+      } else {
+        size_t close = val.find(' ');
+        if (close == string::npos) {
+          seg = "";
+          i = 0;
+        } else {
+          seg = val.substr(close+1);
+          val = val.substr(0, close);
+        }
+      }
+      label = Trim(label);
+      seg = Trim(seg);
+      meta[label] = val;
+    }
+  }
+}
+
-- 
cgit v1.2.3