Using hashed parameter strings in cachefilenames.

[phys/latlib.git] / configcache.cpp

diff --git a/configcache.cpp b/configcache.cpp
index 96dc470ef8e247b740b2ab8fbcd8c8b670229121..bdc593e3c53131ff78e3a869707a7b3c8886c7e7 100644 (file)
--- a/configcache.cpp
+++ b/configcache.cpp
@@ -4,37 +4,71 @@
 #include <iostream>
 #include <time.h>
 #include <dirent.h>
 #include <iostream>
 #include <time.h>
 #include <dirent.h>

+#include <errno.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+
+#include <boost/iostreams/filtering_streambuf.hpp>
+#include <boost/iostreams/stream.hpp>
+#include <boost/iostreams/filter/bzip2.hpp>
+#include <boost/iostreams/device/array.hpp>
+#include <boost/iostreams/copy.hpp>
+
+#define HEADER_READOK   0
+#define HEADER_READERR  1
+#define HEADER_READLAST 2
+
+struct configcache::iobuffers
+{
+  boost::iostreams::filtering_istreambuf *in;
+  boost::iostreams::filtering_ostreambuf *out;
+};

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

-  outBuffer = NULL;
-  inBuffer = NULL;
+  ioBuffers = new iobuffers;
+  ioBuffers->in = NULL;
+  ioBuffers->out = NULL;

 
   MODE = cachemode;
 
   refetchDataFiles = false;
 }
 
 
   MODE = cachemode;
 
   refetchDataFiles = false;
 }
 

-string configcache::getFileId(const bool& shortid)
+string configcache::paraString() {
+  stringstream parastring;
+
+  for(int ipara=0; ipara<Paras.size(); ipara++)
+    parastring << "_" << Paras[ipara].id << Paras[ipara].val;
+  
+  return parastring.str();
+}
+
+string configcache::getFileId(int actnequi, const bool& shortid)

 {
   stringstream fileid;
 
 {
   stringstream fileid;
 

-  if(!shortid) fileid << CACHEID << "_" << NEQUI << "_" << NSKIP;
-  for(int ipara=0; ipara<Paras.size(); ipara++)
-    fileid << "_" << Paras[ipara].id << Paras[ipara].val;
+  if(!shortid) fileid << CACHEID << "_" << actnequi << "_" << NSKIP;
+
+  fileid << "_" << hash( paraString() );

 
   return fileid.str();
 }
 
   return fileid.str();
 }


@@ -63,15 +97,19 @@ bool configcache::isValidInFile(const string& infile, infiledesc *filedesc)
   string truncIn, truncOut;
 
   filedesc->filename = infile;
   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++)
     {


@@ -79,12 +117,10 @@ bool configcache::isValidInFile(const string& infile, infiledesc *filedesc)
 
       switch(iPart)
        {
 
       switch(iPart)
        {

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

          filedesc->nequi = atoi(inParts);
          break;
        case 3: 
          filedesc->nequi = atoi(inParts);
          break;
        case 3: 


@@ -99,14 +135,75 @@ bool configcache::isValidInFile(const string& infile, infiledesc *filedesc)
 
   delete[] inchar;
 
 
   delete[] inchar;
 

-  if( truncIn.find( getFileId(true) + "_" ) == string::npos ) return false;
+  if( truncIn.find( getFileId(NEQUI, true) + "_" ) == string::npos ) return false;

 
   return true;
 }
 
 
   return true;
 }
 

-bool configcache::readConfig()
+int configcache::readHeader()

 {
 {

-  if(DATADIR == "" || MODE == CACHE_MODE_DISABLED) return false;
+  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;
+}
+
+void configcache::readConfig(bool *readnewconfig, int *nequileft, vector<unsigned long> *excludeFileHashes)
+{
+  *readnewconfig = false;
+
+  if( DATADIR == "" || !(MODE==CACHE_MODE_RO||MODE==CACHE_MODE_RW) ) return;

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


@@ -115,64 +212,144 @@ bool configcache::readConfig()
 
   while(true)
     {
 
   while(true)
     {

-      if( (!inFile.is_open()) && inFiles.size() == 0 ) return false;
-
-      while( (!inFile.is_open()) && inFiles.size() > 0 )
-       {
-         if(out) *out->log << "CCACHE: Opening dat-file: " << inFiles.back().filename << endl << flush;
+      vector<infiledesc>::iterator inFileIt = getNextInfile(excludeFileHashes);
+      int iDidVirtualSkips;
+
+      if( (!inFile.is_open()) && inFileIt == inFiles.end() ) {
+       if (*readnewconfig)
+         *nequileft = nequileft_internal;
+       return;
+      }
+
+      while( (!inFile.is_open()) && inFiles.size() > 0 ) {
+       string inFileParaString;
+
+       openFileDesc = *inFileIt;
+
+       if (openFileDesc.nequi < NEQUI)
+         doVirtualEquilibration = true;
+       else
+         doVirtualEquilibration = false;
+
+       firstUsedConfig = true;
+
+       if(log) *log << "CCACHE: Opening dat-file: " << inFileIt->filename << endl << flush;
+       inFile.open( (DATADIR + "/" + inFileIt->filename).c_str(), std::ios::binary );
+       
+       getline( inFile, inFileParaString );
+       if( inFileParaString != paraString() ) {
+         if(log) *log << "CCACHE: Parastring does not match. Closing dat-file..." << endl << flush;
+         inFile.close();
+       }

 
 

-         openFileDesc = inFiles.back();
-         inFile.open( (DATADIR + "/" + inFiles.back().filename).c_str(), std::ios::binary );
-         inFiles.pop_back();
+       inFiles.erase(inFileIt);
+       
+       if( !inFile.is_open() ) continue;

 
 

-         if( !inFile.is_open() ) continue;
-
-         inBuffer = new boost::iostreams::filtering_istreambuf;
-         inBuffer->push( boost::iostreams::bzip2_decompressor() );
-         inBuffer->push(inFile);
-       }
+       ioBuffers->in = new boost::iostreams::filtering_istreambuf;
+       ioBuffers->in->push( boost::iostreams::bzip2_decompressor() );
+       ioBuffers->in->push(inFile);
+      }

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

-         if (openFileDesc.doVirtualEquilibration) {
-           if(out) *out->log << "CCACHE: Trying virtual equilibration." << endl << flush;
-           openFileDesc.doVirtualEquilibration = false;
-           for (int iskip=0; iskip < (NEQUI-openFileDesc.nequi)/openFileDesc.nskip; iskip++) {
-             if( readConfigToMem(tmpConfig) != configSize || ! inFile.is_open() )
+         if (doVirtualEquilibration) {
+           if(log) *log << "CCACHE: Trying virtual equilibration." << endl << flush;
+           doVirtualEquilibration = false;
+           for (iDidVirtualSkips=0; iDidVirtualSkips < (NEQUI-openFileDesc.nequi)/openFileDesc.nskip; iDidVirtualSkips++) {
+             if( readFullBlock(tmpConfig, configSize) != configSize || ! inFile.is_open() )

                break;
                break;

+             else if ( (NEQUI-openFileDesc.nequi) - (iDidVirtualSkips+1)*openFileDesc.nskip < nequileft_internal ) {
+               memcpy(configMem, tmpConfig, configSize);
+               nequileft_internal = NEQUI - openFileDesc.nequi - (iDidVirtualSkips+1)*openFileDesc.nskip;
+               *readnewconfig = true;
+               firstUsedConfig = false;
+             }

            }
          }
 
            }
          }
 

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

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

-             return true;
+             *readnewconfig = true;
+             if (firstUsedConfig) {
+               firstUsedConfig = false;
+               if (openFileDesc.nequi < NEQUI)
+                 nequileft_internal = NEQUI - openFileDesc.nequi - iDidVirtualSkips*openFileDesc.nskip;
+               else
+                 nequileft_internal = NEQUI - openFileDesc.nequi;
+             }
+             nequileft_internal -= openFileDesc.nskip;
+             *nequileft = nequileft_internal;
+             return;

            }
          else {
            }
          else {

-           if(out) *out->log << "CCACHE: Closing dat-file: " << openFileDesc.filename << endl << flush;
+           if(log) *log << "CCACHE: Could not read configuration. Closing dat-file: " << openFileDesc.filename << endl << flush;

            inFile.close();
          }
        }
     }
 }
 
            inFile.close();
          }
        }
     }
 }
 

-void configcache::writeConfig()
-{
-  if( DATADIR == "" || MODE < 2 ) return;
+void configcache::openOutFile(int actnequi)
+{ 
+  time_t secstamp = time(NULL);
+  int iseq=0;
+  
+  while (true) {
+    outFileName.str("");
+    outFileName << DATADIR << "/" << secstamp << "." << iseq << "_" << getFileId(actnequi) << "_.edat.tmp";

 
 

-  if(!outFile.is_open()){
-    time_t secstamp = time(NULL);
+    int tmpfd = open(outFileName.str().c_str(), O_CREAT | O_EXCL, 0644);

 
 

-    outFileName.str("");
-    outFileName << DATADIR << "/" << secstamp << "_" << getFileId() << "_.dat.tmp";    
-    outFile.open( outFileName.str().c_str(), std::ios::binary );
+    if ( tmpfd != -1 ) {
+      close(tmpfd);
+      break;
+    }
+    else if ( errno != EEXIST ) {
+      if(log) *log << "CCACHE: Could not create cachefile!" << endl << flush;
+      exit(1);
+    }

 
 

-    outBuffer = new boost::iostreams::filtering_ostreambuf;
-    outBuffer->push(boost::iostreams::bzip2_compressor());
-    outBuffer->push(outFile);
+    iseq++;

   }
   }

+  
+  outFile.open( outFileName.str().c_str(), std::ios::binary );
+
+  outFile << paraString() << endl;
+
+  ioBuffers->out = new boost::iostreams::filtering_ostreambuf;
+  ioBuffers->out->push(boost::iostreams::bzip2_compressor());
+  ioBuffers->out->push(outFile);
+}
+
+void configcache::writeHeader(const string& headerid, const char *header, long unsigned int size, int actnequi) {
+  unsigned long headeridhash;
+
+  if( DATADIR == "" || !(MODE==CACHE_MODE_WO||MODE==CACHE_MODE_RW) ) return;

 
 

-  boost::iostreams::write(*outBuffer, configMem, configSize);
+  if(!outFile.is_open())
+    openOutFile(actnequi);
+
+  headeridhash = hash(headerid);
+
+  boost::iostreams::write(*ioBuffers->out, (char*)&size, sizeof(long unsigned int));
+  boost::iostreams::write(*ioBuffers->out, (char*)&headeridhash, sizeof(unsigned long));
+  boost::iostreams::write(*ioBuffers->out, header, size);
+}
+
+void configcache::writeConfig(int actnequi)
+{
+  long unsigned int zeroheader=0;
+
+  if ( DATADIR == "" || !(MODE==CACHE_MODE_WO||MODE==CACHE_MODE_RW) ) return;
+
+  if ( ! outFile.is_open() )
+    openOutFile(actnequi);
+  
+  boost::iostreams::write(*ioBuffers->out, (char*)&zeroheader, sizeof(long unsigned int));
+
+  boost::iostreams::write(*ioBuffers->out, configMem, configSize);

 }
 
 void configcache::addPara(const string& parid, const double& val){
 }
 
 void configcache::addPara(const string& parid, const double& val){


@@ -189,27 +366,29 @@ int configcache::getParIndex(const string& parid){
 
 void configcache::setPara(const string& parid, const double& value){
   Paras[getParIndex(parid)].val = value;
 
 void configcache::setPara(const string& parid, const double& value){
   Paras[getParIndex(parid)].val = value;

+

   finishOutFile();
   finishOutFile();

-  if(inBuffer != NULL) { delete inBuffer; inBuffer=NULL; } 
+  if(ioBuffers->in != NULL) { delete ioBuffers->in; ioBuffers->in=NULL; } 

   inFile.close();
   inFiles.clear();
 
   refetchDataFiles = true;
   inFile.close();
   inFiles.clear();
 
   refetchDataFiles = true;

+  nequileft_internal = NEQUI;

 }
 
 configcache::~configcache()
 {
   finishOutFile();
 }
 
 configcache::~configcache()
 {
   finishOutFile();

-  delete inBuffer;
-  inBuffer = NULL;
+  delete ioBuffers->in;
+  ioBuffers->in = NULL;

 }
 
 void configcache::finishOutFile()
 {
 }
 
 void configcache::finishOutFile()
 {

-  if( outBuffer != NULL )
+  if( ioBuffers->out != NULL )

     {
     {

-      delete outBuffer;
-      outBuffer = NULL;
+      delete ioBuffers->out;
+      ioBuffers->out = NULL;

     }
 
   if( outFile.is_open() )
     }
 
   if( outFile.is_open() )


@@ -219,23 +398,73 @@ void configcache::finishOutFile()
     }
 }
 
     }
 }
 

-int configcache::readConfigToMem(char *tmpConfig)
+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;
 
 {
   int readturn = -1;
 

-  try { readturn = boost::iostreams::read(*inBuffer, tmpConfig, configSize); }
+  if ( dataSize == 0 ) return 0;
+
+  try { readturn = boost::iostreams::read(*ioBuffers->in, tmpData, dataSize); }

   catch(boost::iostreams::bzip2_error& error) { 
   catch(boost::iostreams::bzip2_error& error) { 

-    if(out) *out->log << "CCACHE: Caught bzip2 exception with error code: " << error.error() << endl << flush;
+    if(log) *log << "CCACHE: Caught bzip2 exception with error code: " << error.error() << endl << flush;

     inFile.close();
   } 
   catch (std::exception const& ex) {
     inFile.close();
   } 
   catch (std::exception const& ex) {

-    if(out) *out->log << "CCACHE: Caught exception: " << ex.what() << endl << flush;
+    if(log) *log << "CCACHE: Caught exception: " << ex.what() << endl << flush;

     inFile.close();
   }
   catch( ... ) {
     inFile.close();
   }
   catch( ... ) {

-    if(out) *out->log << "CCACHE: Caught unknown exception while reading." << endl << flush;
+    if(log) *log << "CCACHE: Caught unknown exception while reading." << endl << flush;

     inFile.close();
   }
 
   return readturn;
 }
     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();
+}