X-Git-Url: http://git.treefish.org/~alex/phys/latlib.git/blobdiff_plain/a55a5830c5c98e950d58593697dd4ee5e4384716..8d11db1cf57b103fcec701d71c9d620b1c381b73:/configcache.cpp?ds=inline

diff --git a/configcache.cpp b/configcache.cpp
index fcfdd8e..8687123 100644
--- a/configcache.cpp
+++ b/configcache.cpp
@@ -4,39 +4,74 @@
 #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_ostreambuf *out;
+};
 
 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;
 
+  dataReader = new datread(configMemSize, log);
+  
   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;
 
-  outBuffer = NULL;
-  inBuffer = NULL;
+  ioBuffers = new iobuffers;
+  ioBuffers->out = NULL;
 
   MODE = cachemode;
 
   refetchDataFiles = false;
+}
 
-  readHeaderData = NULL;
+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(const bool& shortid)
+string configcache::getFileId(int actnequi, const bool& superextended, const bool& shortid)
 {
   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;
+
+  if( superextended )
+    fileid << "_" << hash( paraString() );
+  else
+    fileid << paraString();
 
   return fileid.str();
 }
@@ -65,20 +100,27 @@ bool configcache::isValidInFile(const string& infile, infiledesc *filedesc)
   string truncIn, truncOut;
 
   filedesc->filename = infile;
-  filedesc->doVirtualEquilibration = false;
 
   if( infile.size() < 4 ) return false;
 
-  if( infile.substr(infile.size()-4) == ".dat" )
+  if( infile.substr(infile.size()-4) == ".dat" ) {
     filedesc->extended = false;
-  else if( infile.substr(infile.size()-4) == "edat" )
+    filedesc->superextended = false;
+  }
+  else if( infile.substr(infile.size()-4) == "edat" ) {
+    filedesc->extended = true;
+    filedesc->superextended = false;
+  }
+  else if( infile.substr(infile.size()-4) == "sdat" ) {
     filedesc->extended = true;
+    filedesc->superextended = true;
+  }
   else
     return false;
 
   inchar = new char [infile.size()+1];
   strcpy (inchar, infile.c_str());
-  
+
   inParts = strtok( inchar, "_" );
   for(int iPart=0; inParts!=NULL; iPart++)
     {
@@ -86,12 +128,10 @@ bool configcache::isValidInFile(const string& infile, infiledesc *filedesc)
 
       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: 
@@ -106,44 +146,20 @@ bool configcache::isValidInFile(const string& infile, infiledesc *filedesc)
 
   delete[] inchar;
 
-  if( truncIn.find( getFileId(true) + "_" ) == string::npos ) return false;
+  if( truncIn.find( getFileId(NEQUI, filedesc->superextended, true) + "_" ) == string::npos ) return false;
 
   return true;
 }
 
-bool configcache::readHeader()
-{
-  long unsigned int headersize;
-
-  if( readDataToMem((char*)&headersize, sizeof(long unsigned int)) == sizeof(long unsigned int) && inFile.is_open() )
-    {
-      if( readHeaderData != NULL ) free(readHeaderData);
-
-      readHeaderData = (char*) malloc(headersize);
-
-      if( readDataToMem(readHeaderData, headersize) == headersize && inFile.is_open() ) {
-	return true;
-      }
-      else {
-	if(out) *out->log << "CCACHE: Could not read header! Closing dat-file: " << openFileDesc.filename << endl << flush;
-	inFile.close();
-	return false;
-      }
-    }
-  else {
-    if(out) *out->log << "CCACHE: Could not read header size. Closing dat-file: " << openFileDesc.filename << endl << flush;
-    inFile.close();
-    return false;
-  }
-}
-
-void *configcache::getHeader() {
-  return readHeaderData;
+void * configcache::getHeader(const string& headerid) {
+  return dataReader->getHeader(headerid);
 }
 
-bool configcache::readConfig()
+void configcache::readConfig(bool *readnewconfig, int *nequileft, vector<unsigned long> *excludeFileHashes)
 {
-  if(DATADIR == "" || MODE == CACHE_MODE_DISABLED) return false;
+  *readnewconfig = false;
+
+  if( DATADIR == "" || !(MODE==CACHE_MODE_RO||MODE==CACHE_MODE_RW) ) return;
 
   if(refetchDataFiles){
     refetchDataFiles = false;
@@ -152,87 +168,138 @@ bool configcache::readConfig()
 
   while(true)
     {
-      if( (!inFile.is_open()) && inFiles.size() == 0 ) return false;
+      vector<infiledesc>::iterator inFileIt = getNextInfile(excludeFileHashes);
+      int iDidVirtualSkips;
+
+      if( (!dataReader->fisopen()) && inFileIt == inFiles.end() ) {
+	if (*readnewconfig)
+	  *nequileft = nequileft_internal;
+	return;
+      }
+
+      while( (!dataReader->fisopen()) && inFiles.size() > 0 ) {
+	string inFileParaString;
+
+	openFileDesc = *inFileIt;
+
+	if (openFileDesc.nequi < NEQUI)
+	  doVirtualEquilibration = true;
+	else
+	  doVirtualEquilibration = false;
 
-      while( (!inFile.is_open()) && inFiles.size() > 0 ) {
-	if(out) *out->log << "CCACHE: Opening dat-file: " << inFiles.back().filename << endl << flush;
+	firstUsedConfig = true;
 
-	openFileDesc = inFiles.back();
-	inFile.open( (DATADIR + "/" + inFiles.back().filename).c_str(), std::ios::binary );
-	inFiles.pop_back();
+	if(log) *log << "CCACHE: Opening dat-file: " << inFileIt->filename << endl << flush;
 
-	if( !inFile.is_open() ) continue;
+	dataReader->openFile( DATADIR + "/" + inFileIt->filename );
+		
+	if( openFileDesc.superextended ) {
+	  if( dataReader->getParaString() != paraString() ) {
+	    if(log) *log << "CCACHE: Parastring does not match. Closing dat-file..." << endl << flush;
+	    dataReader->closeFile();
+	  }
+	}
 
-	inBuffer = new boost::iostreams::filtering_istreambuf;
-	inBuffer->push( boost::iostreams::bzip2_decompressor() );
-	inBuffer->push(inFile);
+	inFiles.erase(inFileIt);
+	
+	if( !dataReader->fisopen() ) continue;
       }
 
-      if( inFile.is_open() ) 
+      if( dataReader->fisopen() ) 
 	{
-	  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( readFullBlock(tmpConfig, configSize) != 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( dataReader->readFullBlock(tmpConfig) < 0 )
 		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( readFullBlock(tmpConfig, configSize) == configSize && inFile.is_open() )
+	  if( dataReader->readFullBlock(tmpConfig) >= 0 )
 	    {
 	      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 {
-	    if(out) *out->log << "CCACHE: Could not read configuration. Closing dat-file: " << openFileDesc.filename << endl << flush;
-	    inFile.close();
-	  }
 	}
     }
 }
 
-void configcache::openOutFile()
-{
+void configcache::openOutFile(int actnequi)
+{ 
   time_t secstamp = time(NULL);
+  int iseq=0;
+  
+  while (true) {
+    outFileName.str("");
+    outFileName << DATADIR << "/" << secstamp << "." << iseq << "_" << getFileId(actnequi, true, false) << "_.sdat.tmp";
+
+    int tmpfd = open(outFileName.str().c_str(), O_CREAT | O_EXCL, 0644);
+
+    if ( tmpfd != -1 ) {
+      close(tmpfd);
+      break;
+    }
+    else if ( errno != EEXIST ) {
+      if(log) *log << "CCACHE: Could not create cachefile!" << endl << flush;
+      exit(1);
+    }
 
-  outFileName.str("");
-  outFileName << DATADIR << "/" << secstamp << "_" << getFileId() << "_.edat.tmp";    
+    iseq++;
+  }
+  
   outFile.open( outFileName.str().c_str(), std::ios::binary );
 
-  outBuffer = new boost::iostreams::filtering_ostreambuf;
-  outBuffer->push(boost::iostreams::bzip2_compressor());
-  outBuffer->push(outFile);
+  outFile << paraString() << endl;
 
-  headerWritten = false;
+  ioBuffers->out = new boost::iostreams::filtering_ostreambuf;
+  ioBuffers->out->push(boost::iostreams::bzip2_compressor());
+  ioBuffers->out->push(outFile);
 }
 
-void configcache::writeHeader(char *header, long unsigned int size) {
-  if( DATADIR == "" || MODE < 2 ) return;
+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;
 
   if(!outFile.is_open())
-    openOutFile();
+    openOutFile(actnequi);
 
-  boost::iostreams::write(*outBuffer, (char*)&size, sizeof(long unsigned int));
-  boost::iostreams::write(*outBuffer, header, size);
+  headeridhash = hash(headerid);
 
-  headerWritten = true;
+  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()
+void configcache::writeConfig(int actnequi)
 {
-  if ( DATADIR == "" || MODE < 2 ) return;
+  long unsigned int zeroheader=0;
+
+  if ( DATADIR == "" || !(MODE==CACHE_MODE_WO||MODE==CACHE_MODE_RW) ) return;
 
   if ( ! outFile.is_open() )
-    openOutFile();
+    openOutFile(actnequi);
   
-  if ( ! headerWritten ) {
-    long unsigned int zeroheader=0;
-    boost::iostreams::write(*outBuffer, (char*)&zeroheader, sizeof(long unsigned int));
-  }
+  boost::iostreams::write(*ioBuffers->out, (char*)&zeroheader, sizeof(long unsigned int));
 
-  boost::iostreams::write(*outBuffer, configMem, configSize);
-  headerWritten = false;
+  boost::iostreams::write(*ioBuffers->out, configMem, configSize);
 }
 
 void configcache::addPara(const string& parid, const double& val){
@@ -249,27 +316,27 @@ int configcache::getParIndex(const string& parid){
 
 void configcache::setPara(const string& parid, const double& value){
   Paras[getParIndex(parid)].val = value;
+
   finishOutFile();
-  if(inBuffer != NULL) { delete inBuffer; inBuffer=NULL; } 
-  inFile.close();
+  dataReader->closeFile();
   inFiles.clear();
 
   refetchDataFiles = true;
+  nequileft_internal = NEQUI;
 }
 
 configcache::~configcache()
 {
   finishOutFile();
-  delete inBuffer;
-  inBuffer = NULL;
+  delete dataReader;
 }
 
 void configcache::finishOutFile()
 {
-  if( outBuffer != NULL )
+  if( ioBuffers->out != NULL )
     {
-      delete outBuffer;
-      outBuffer = NULL;
+      delete ioBuffers->out;
+      ioBuffers->out = NULL;
     }
 
   if( outFile.is_open() )
@@ -279,36 +346,31 @@ void configcache::finishOutFile()
     }
 }
 
-int configcache::readFullBlock(char *tmpData, long unsigned int dataSize)
+unsigned long configcache::hash(const string& str)
 {
-  /* try to read header */
-  if ( openFileDesc.extended )
-    if ( ! readHeader() ) 
-      return -1;
+  unsigned long hash = 5381;
+
+  for(string::const_iterator it=str.begin();it!=str.end();it++) 
+    hash = ((hash << 5) + hash) + *it; /* hash * 33 + character */
 
-  /* read data */
-  return readDataToMem(tmpData, dataSize);
+  return hash;
 }
 
-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(out) *out->log << "CCACHE: Caught bzip2 exception with error code: " << error.error() << endl << flush;
-    inFile.close();
-  } 
-  catch (std::exception const& ex) {
-    if(out) *out->log << "CCACHE: Caught exception: " << ex.what() << endl << flush;
-    inFile.close();
-  }
-  catch( ... ) {
-    if(out) *out->log << "CCACHE: Caught unknown exception while reading." << endl << flush;
-    inFile.close();
-  }
+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;
+	}
 
-  return readturn;
+      if (excludethisfile)
+	continue;
+    }
+    return init;
+  }
+  return inFiles.end();
 }