Windows->Linux: finalizing reconciliation of Linux vs. Windows version of HTKMLFReader

DataReader/HTKMLFReader/latticearchive.h

@@ -934,16 +934,17 @@ public:
                     if (!isendworkaround.empty() && isendworkaround[k])       // secondary criterion to detect ends in broken lattices
                     {
                         k--;    // don't advance, since nothing to advance over
-                        goto skipscores;
                     }
-                    // this is a regular token: update it in-place
-                    auto & ai = uniquededgedatatokens[k];
-                    if (ai.unit >= idmap.size())
-                        throw std::runtime_error ("fread: broken-file heuristics failed");
-                    ai.updateunit (idmap);      // updates itself
-                    if (!ai.last)
-                        continue;
-                skipscores:
+                    else
+                    {
+                        // this is a regular token: update it in-place
+                        auto & ai = uniquededgedatatokens[k];
+                        if (ai.unit >= idmap.size())
+                            throw std::runtime_error ("fread: broken-file heuristics failed");
+                        ai.updateunit (idmap);      // updates itself
+                        if (!ai.last)
+                            continue;
+                    }
                     // if last then skip over the lm and ac scores
                     k += skipscoretokens;
                     uniquealignments++;

DataReader/HTKMLFReader_linux/HTKMLFReader.cpp

@@ -249,11 +249,11 @@ namespace Microsoft { namespace MSR { namespace CNTK {
             const std::string& randomizeString = readerConfig("randomize");
             if (randomizeString == "None")
             {
-                randomize = m_htkRandomizeDisable; // randomizeNone;
+                randomize = randomizeNone;
             }
             else if (randomizeString == "Auto")
             {
-                randomize = m_htkRandomizeAuto; // randomizeAuto
+                randomize = randomizeAuto;
             }
             else
             {
@@ -379,8 +379,12 @@ namespace Microsoft { namespace MSR { namespace CNTK {
         //std::vector<std::wstring> pagepath;
         foreach_index(i, mlfpathsmulti)
         {
+            const map<string,size_t>* wordmap = NULL;
+#ifdef WIN32
+            wordmap = unigram ? &unigramsymbols : (map<string,size_t>*) NULL;
+#endif
             msra::asr::htkmlfreader<msra::asr::htkmlfentry,msra::lattices::lattice::htkmlfwordsequence>  
-                labels(mlfpathsmulti[i], restrictmlftokeys, statelistpaths[i], /*unigram ? &unigramsymbols :*/(map<string,size_t>*)  NULL, (map<string,size_t>*) NULL, htktimetoframe);      // label MLF
+                labels(mlfpathsmulti[i], restrictmlftokeys, statelistpaths[i], wordmap, (map<string,size_t>*) NULL, htktimetoframe);      // label MLF
             // get the temp file name for the page file
             labelsmulti.push_back(labels);
         }

DataReader/HTKMLFReader_linux/fileutil.cpp

@@ -248,7 +248,6 @@ size_t filesize (FILE * f)
     {
     RuntimeError ("error determining file position: %s", strerror (errno));
     }
-    int rc = fseek (f, 0, SEEK_END);
     int rc = _fseeki64 (f, 0, SEEK_END);
     if (rc != 0)
     {

DataReader/HTKMLFReader_linux/htkfeatio.h

@@ -230,7 +230,7 @@ public:
     // We write to a tmp file first to ensure we don't leave broken files that would confuse make mode.
     template<class MATRIX> static void write (const wstring & path, const string & kindstr, unsigned int period, const MATRIX & feat)
     {
-        wstring tmppath = path + L""; // tmp path for make-mode compliant
+        wstring tmppath = path + L"$$"; // tmp path for make-mode compliant
         unlinkOrDie (path);             // delete if old file is already there
         // write it out
         size_t featdim = feat.rows();

DataReader/HTKMLFReader_linux/latticearchive.h

@@ -933,20 +933,20 @@ public:
                 const size_t skipscoretokens = info.hasacscores ? 2 : 1;
                 for (size_t k = skipscoretokens; k < uniquededgedatatokens.size(); k++)
                 {
-                    auto & ai = uniquededgedatatokens[k];
                     if (!isendworkaround.empty() && isendworkaround[k])       // secondary criterion to detect ends in broken lattices
                     {
                         k--;    // don't advance, since nothing to advance over
-                        goto skipscores;
                     }
-                    // this is a regular token: update it in-place
-                    
-                    if (ai.unit >= idmap.size())
-                        throw std::runtime_error ("fread: broken-file heuristics failed");
-                    ai.updateunit (idmap);      // updates itself
-                    if (!ai.last)
-                        continue;
-                skipscores:
+                    else
+                    {
+                        // this is a regular token: update it in-place
+                        auto & ai = uniquededgedatatokens[k];
+                        if (ai.unit >= idmap.size())
+                            throw std::runtime_error ("fread: broken-file heuristics failed");
+                        ai.updateunit (idmap);      // updates itself
+                        if (!ai.last)
+                            continue;
+                    }
                     // if last then skip over the lm and ac scores
                     k += skipscoretokens;
                     uniquealignments++;

DataReader/HTKMLFReader_linux/utterancesourcemulti.h

@@ -612,7 +612,8 @@ private:
             return sweep;
 
         currentsweep = sweep;
-        fprintf (stderr, "lazyrandomization: re-randomizing for sweep %zu in %s mode\n", currentsweep, framemode ? "frame" : "utterance");
+        if (verbosity>0)
+            fprintf (stderr, "lazyrandomization: re-randomizing for sweep %zu in %s mode\n", currentsweep, framemode ? "frame" : "utterance");
 
         const size_t sweepts = sweep * _totalframes;     // first global frame index for this sweep