bugbug

[phys/latlib.git] / configcache.cpp

diff --git a/configcache.cpp b/configcache.cpp
index af0a206eb6c73ea126b392395ad1472f0b4d882a..02b6d94ec3521578d196f23f0b63cefb4c8fa874 100644 (file)
--- a/configcache.cpp
+++ b/configcache.cpp
@@ -5,15 +5,24 @@
 #include <time.h>
 #include <dirent.h>
 
 #include <time.h>
 #include <dirent.h>
 

+#define HEADER_READOK   0
+#define HEADER_READERR  1
+#define HEADER_READLAST 2
+

 configcache::configcache(const string& cacheid, const int& nequi, const int& nskip, const string& datadir, char **configmem, const int& configMemSize, const int& cachemode,
 configcache::configcache(const string& cacheid, const int& nequi, const int& nskip, const string& datadir, char **configmem, const int& configMemSize, const int& cachemode,

-                        writeout *out_a){
-  out = out_a;
+                        ostream *_log){
+  log = _log;

 
   NEQUI = nequi;
   NSKIP = nskip;
   DATADIR = datadir;
   CACHEID = cacheid;
 
 
   NEQUI = nequi;
   NSKIP = nskip;
   DATADIR = datadir;
   CACHEID = cacheid;
 

+  if ( cacheid.find("_") != -1 ) {
+    if(log) *log << "CCACHE: Invalid cacheid \"" << cacheid << "\" given. Cacheids must not contain underscores!" << endl << flush;
+    exit(1);
+  }
+

   configMem = (char*)malloc(configMemSize);
   tmpConfig = (char*)malloc(configMemSize);
 
   configMem = (char*)malloc(configMemSize);
   tmpConfig = (char*)malloc(configMemSize);
 


@@ -43,31 +52,40 @@ void configcache::fetchDataFiles()
 {
   struct dirent *de=NULL;
   DIR *d=NULL;
 {
   struct dirent *de=NULL;
   DIR *d=NULL;

+  static infiledesc filedesc;

   
   d=opendir(DATADIR.c_str());
   if(d != NULL){
     while(de = readdir(d)){
       string filename = de->d_name;
   
   d=opendir(DATADIR.c_str());
   if(d != NULL){
     while(de = readdir(d)){
       string filename = de->d_name;

-      if(isValidInFile(filename)) 
+      if(isValidInFile(filename, &filedesc)) 

        {
        {

-         inFiles.push_back(filename);
+         inFiles.push_back(filedesc);

        }
     }
   }
 }
 
        }
     }
   }
 }
 

-bool configcache::isValidInFile(const string& infile)
+bool configcache::isValidInFile(const string& infile, infiledesc *filedesc)

 {
   char *inchar, *inParts;
   string truncIn, truncOut;
 
 {
   char *inchar, *inParts;
   string truncIn, truncOut;
 

+  filedesc->filename = infile;
+  filedesc->doVirtualEquilibration = false;
+

   if( infile.size() < 4 ) return false;
 
   if( infile.size() < 4 ) return false;
 

-  if( infile.substr(infile.size()-3) != "dat" ) return false;
+  if( infile.substr(infile.size()-4) == ".dat" )
+    filedesc->extended = false;
+  else if( infile.substr(infile.size()-4) == "edat" )
+    filedesc->extended = true;
+  else
+    return false;

 
   inchar = new char [infile.size()+1];
   strcpy (inchar, infile.c_str());
 
   inchar = new char [infile.size()+1];
   strcpy (inchar, infile.c_str());

-  
+

   inParts = strtok( inchar, "_" );
   for(int iPart=0; inParts!=NULL; iPart++)
     {
   inParts = strtok( inchar, "_" );
   for(int iPart=0; inParts!=NULL; iPart++)
     {


@@ -75,9 +93,21 @@ bool configcache::isValidInFile(const string& infile)
 
       switch(iPart)
        {
 
       switch(iPart)
        {

-       case 1: if(inParts != CACHEID) return false; break;
-       case 2: if(atoi(inParts) != NEQUI) return false; break;
-       case 3: if(atoi(inParts) != NSKIP) return false; break;
+       case 1: if(inParts != CACHEID)
+           return false;
+         break;
+       case 2:
+         if (atoi(inParts) > NEQUI) 
+           return false; 
+         else if (atoi(inParts) < NEQUI)
+           filedesc->doVirtualEquilibration = true;
+         filedesc->nequi = atoi(inParts);
+         break;
+       case 3: 
+         if(atoi(inParts) != NSKIP) 
+           return false;
+         filedesc->nskip = atoi(inParts);
+         break;

        }
       inParts = strtok( NULL, "_");
     }
        }
       inParts = strtok( NULL, "_");
     }


@@ -90,9 +120,71 @@ bool configcache::isValidInFile(const string& infile)
   return true;
 }
 
   return true;
 }
 

-bool configcache::readConfig()
+int configcache::readHeader()
+{
+  long unsigned int headersize;
+  
+  if( readDataToMem((char *)&headersize, sizeof(long unsigned int)) == sizeof(long unsigned int) && inFile.is_open() ) {
+    if ( headersize == 0 )
+      return HEADER_READLAST;
+
+    pair<unsigned long, void *> newHeader;
+
+    if( readDataToMem((char *)&newHeader.first, sizeof(unsigned long)) == sizeof(unsigned long) && inFile.is_open() ) {
+      newHeader.second = malloc(headersize);
+
+      if( readDataToMem((char *)newHeader.second, headersize) == headersize && inFile.is_open() ) {
+       headerStore.push_back(newHeader);
+       return HEADER_READOK;
+      }
+      else {
+       if(log) *log << "CCACHE: Could not read heade-data! Closing dat-file: " << openFileDesc.filename << endl << flush;
+       inFile.close();
+       return HEADER_READERR;
+      }
+    }
+    else {
+      if(log) *log << "CCACHE: Could not read headerid-hash! Closing dat-file: " << openFileDesc.filename << endl << flush;
+      inFile.close();
+      return HEADER_READERR;
+    }
+  }
+  else {
+    if(log) *log << "CCACHE: Could not read header size. Closing dat-file: " << openFileDesc.filename << endl << flush;
+    inFile.close();
+    return HEADER_READERR;
+  }
+}
+
+bool configcache::readAllHeaders()
+{
+  int readHeaderStatus;
+
+  deleteHeaderStore();
+  
+  do {
+    readHeaderStatus = readHeader();
+  }
+  while ( readHeaderStatus == HEADER_READOK );
+
+  if ( readHeaderStatus == HEADER_READLAST ) return true;
+  else if ( readHeaderStatus == HEADER_READERR ) return false;
+}
+
+void * configcache::getHeader(const string& headerid) {
+  for (vector< pair<unsigned long, void *> >::iterator headerStoreIt = headerStore.begin(); headerStoreIt != headerStore.end(); ++headerStoreIt)
+    if ( headerStoreIt->first == hash(headerid) )
+      return headerStoreIt->second;
+  
+  return NULL;
+}
+
+/* returns number of equilibration-steps left */
+int configcache::readConfig(vector<unsigned long> *excludeFileHashes)

 {
 {

-  if(DATADIR == "" || MODE == CACHE_MODE_DISABLED) return false;
+  int nequileft = NEQUI;
+
+  if(DATADIR == "" || MODE == CACHE_MODE_DISABLED) return nequileft;

 
   if(refetchDataFiles){
     refetchDataFiles = false;
 
   if(refetchDataFiles){
     refetchDataFiles = false;


@@ -101,66 +193,92 @@ bool configcache::readConfig()
 
   while(true)
     {
 
   while(true)
     {

-      if( (!inFile.is_open()) && inFiles.size() == 0 ) return false;
+      vector<infiledesc>::iterator inFileIt = getNextInfile(excludeFileHashes);

 
 

-      while( (!inFile.is_open()) && inFiles.size() > 0 )
-       {
-         if(out) *out->log << "CCACHE: Opening dat-file: " << inFiles.back() << endl << flush;
+      if( (!inFile.is_open()) && inFileIt == inFiles.end() ) return nequileft;

 
 

-         inFile.open( (DATADIR + "/" + inFiles.back()).c_str(), std::ios::binary );
-         inFiles.pop_back();
+      while( (!inFile.is_open()) && inFiles.size() > 0 ) {
+       openFileDesc = *inFileIt;

 
 

-         if( !inFile.is_open() ) continue;
+       if(log) *log << "CCACHE: Opening dat-file: " << inFileIt->filename << endl << flush;
+       inFile.open( (DATADIR + "/" + inFileIt->filename).c_str(), std::ios::binary );
+       
+       inFiles.erase(inFileIt);
+       
+       if( !inFile.is_open() ) continue;

 
 

-         std::streampos fsize = inFile.tellg();
-          inFile.seekg( 0, std::ios::end );
-          fsize = inFile.tellg() - fsize;
-         if( fsize == 0 ) {
-           *out->log << "CCACHE: dat-file has zero length! Skipping." << endl << flush;
-           inFile.close();
-           continue;
-         }
-         inFile.seekg(0);
-
-         inBuffer = new boost::iostreams::filtering_istreambuf;
-         inBuffer->push( boost::iostreams::bzip2_decompressor() );
-         inBuffer->push(inFile);
-       }
+       inBuffer = new boost::iostreams::filtering_istreambuf;
+       inBuffer->push( boost::iostreams::bzip2_decompressor() );
+       inBuffer->push(inFile);
+      }

 
       if( inFile.is_open() ) 
        {
 
       if( inFile.is_open() ) 
        {

-         int readturn = -1;
-         try { readturn = boost::iostreams::read(*inBuffer, tmpConfig, configSize); }
-         catch (std::exception const& ex) {
-           if(out) *out->log << "CCACHE: Caught exception: " << ex.what() << endl << flush;
-           inFile.close();
+         if (openFileDesc.doVirtualEquilibration) {
+           if(log) *log << "CCACHE: Trying virtual equilibration." << endl << flush;
+           openFileDesc.doVirtualEquilibration = false;
+           for (int iskip=0; iskip < (NEQUI-openFileDesc.nequi)/openFileDesc.nskip; iskip++) {
+             if( readFullBlock(tmpConfig, configSize) != configSize || ! inFile.is_open() )
+               break;
+             else if ( (NEQUI-openFileDesc.nequi) - (iskip+1)*openFileDesc.nskip < nequileft ) {
+               memcpy(configMem, tmpConfig, configSize);
+               nequileft = (NEQUI-openFileDesc.nequi) - (iskip+1)*openFileDesc.nskip;
+             }
+           }

          }
          }

-         
-         if( readturn == configSize && inFile.is_open() )
+
+         if( readFullBlock(tmpConfig, configSize) == configSize && inFile.is_open() )

            {
              memcpy(configMem, tmpConfig, configSize);
            {
              memcpy(configMem, tmpConfig, configSize);

-             return true;
+             return -1;

            }
            }

-         else inFile.close(); 
+         else {
+           if(log) *log << "CCACHE: Could not read configuration. Closing dat-file: " << openFileDesc.filename << endl << flush;
+           inFile.close();
+         }

        }
     }
 }
 
        }
     }
 }
 

-void configcache::writeConfig()
+void configcache::openOutFile()

 {
 {

+  time_t secstamp = time(NULL);
+
+  outFileName.str("");
+  outFileName << DATADIR << "/" << secstamp << "_" << getFileId() << "_.edat.tmp";
+
+  outFile.open( outFileName.str().c_str(), std::ios::binary );
+
+  outBuffer = new boost::iostreams::filtering_ostreambuf;
+  outBuffer->push(boost::iostreams::bzip2_compressor());
+  outBuffer->push(outFile);
+}
+
+void configcache::writeHeader(const string& headerid, const char *header, long unsigned int size) {
+  unsigned long headeridhash;
+

   if( DATADIR == "" || MODE < 2 ) return;
 
   if( DATADIR == "" || MODE < 2 ) return;
 

-  if(!outFile.is_open()){
-    time_t secstamp = time(NULL);
+  if(!outFile.is_open())
+    openOutFile();

 
 

-    outFileName.str("");
-    outFileName << DATADIR << "/" << secstamp << "_" << getFileId() << "_.dat.tmp";    
-    outFile.open( outFileName.str().c_str(), std::ios::binary );
+  headeridhash = hash(headerid);

 
 

-    outBuffer = new boost::iostreams::filtering_ostreambuf;
-    outBuffer->push(boost::iostreams::bzip2_compressor());
-    outBuffer->push(outFile);
-  }
+  boost::iostreams::write(*outBuffer, (char*)&size, sizeof(long unsigned int));
+  boost::iostreams::write(*outBuffer, (char*)&headeridhash, sizeof(unsigned long));
+  boost::iostreams::write(*outBuffer, header, size);
+}
+
+void configcache::writeConfig()
+{
+  long unsigned int zeroheader=0;
+
+  if ( DATADIR == "" || MODE < 2 ) return;
+
+  if ( ! outFile.is_open() )
+    openOutFile();
+  
+  boost::iostreams::write(*outBuffer, (char*)&zeroheader, sizeof(long unsigned int));

 
   boost::iostreams::write(*outBuffer, configMem, configSize);
 }
 
   boost::iostreams::write(*outBuffer, configMem, configSize);
 }


@@ -208,3 +326,74 @@ void configcache::finishOutFile()
       rename( outFileName.str().c_str(), outFileName.str().substr(0, outFileName.str().size()-4).c_str() );
     }
 }
       rename( outFileName.str().c_str(), outFileName.str().substr(0, outFileName.str().size()-4).c_str() );
     }
 }

+
+int configcache::readFullBlock(char *tmpData, long unsigned int dataSize)
+{
+  /* try to read header */
+  if ( openFileDesc.extended )
+    if ( ! readAllHeaders() ) 
+      return -1;
+
+  /* read data */
+  return readDataToMem(tmpData, dataSize);
+}
+
+int configcache::readDataToMem(char *tmpData, long unsigned int dataSize)
+{
+  int readturn = -1;
+
+  if ( dataSize == 0 ) return 0;
+
+  try { readturn = boost::iostreams::read(*inBuffer, tmpData, dataSize); }
+  catch(boost::iostreams::bzip2_error& error) { 
+    if(log) *log << "CCACHE: Caught bzip2 exception with error code: " << error.error() << endl << flush;
+    inFile.close();
+  } 
+  catch (std::exception const& ex) {
+    if(log) *log << "CCACHE: Caught exception: " << ex.what() << endl << flush;
+    inFile.close();
+  }
+  catch( ... ) {
+    if(log) *log << "CCACHE: Caught unknown exception while reading." << endl << flush;
+    inFile.close();
+  }
+
+  return readturn;
+}
+
+unsigned long configcache::hash(const string& str)
+{
+  unsigned long hash = 5381;
+
+  for(string::const_iterator it=str.begin();it!=str.end();it++) 
+    hash = ((hash << 5) + hash) + *it; /* hash * 33 + character */
+
+  return hash;
+}
+
+void configcache::deleteHeaderStore()
+{
+  while ( headerStore.size() > 0 ) {
+    free(headerStore.back().second);
+    headerStore.pop_back();
+  }
+}
+
+vector<infiledesc>::iterator configcache::getNextInfile(vector<unsigned long> *excludeFileHashes) {
+  for (vector<infiledesc>::iterator init = inFiles.begin(); init != inFiles.end(); ++init) {
+    if (excludeFileHashes != NULL) {
+      bool excludethisfile = false;
+
+      for (vector<unsigned long>::iterator exit = excludeFileHashes->begin(); exit != excludeFileHashes->end(); ++exit)
+       if ( *exit == hash(init->filename) ) {
+         excludethisfile = true;
+         break;
+       }
+
+      if (excludethisfile)
+       continue;
+    }
+    return init;
+  }
+  return inFiles.end();
+}