ViewVC Help
View File | Revision Log | Show Annotations | Root Listing
root/cvsroot/UserCode/grimes/L1Menu/src/ReducedMenuSample.cpp
Revision: 1.9
Committed: Sat Jun 8 09:31:25 2013 UTC (11 years, 10 months ago) by grimes
Branch: MAIN
Changes since 1.8: +15 -1 lines
Log Message:
Changes to load very large files, and some fixes about histogram binning.

File Contents

# Content
1 #include "l1menu/ReducedMenuSample.h"
2
3 #include <vector>
4 #include <stdexcept>
5 #include <fcntl.h>
6 #include <algorithm>
7 #include <iostream>
8 #include "l1menu/IReducedEvent.h"
9 #include "l1menu/MenuSample.h"
10 #include "l1menu/TriggerMenu.h"
11 #include "l1menu/ITrigger.h"
12 #include "l1menu/tools.h"
13 #include "protobuf/l1menu.pb.h"
14 #include <google/protobuf/io/zero_copy_stream_impl.h>
15 #include <google/protobuf/io/gzip_stream.h>
16
17 namespace // unnamed namespace
18 {
19 class ReducedEventImplementation : public l1menu::IReducedEvent
20 {
21 public:
22 virtual ~ReducedEventImplementation() {}
23 virtual float parameterValue( size_t parameterNumber ) const { return pProtobufEvent->threshold(parameterNumber); }
24 virtual float weight() const { if( pProtobufEvent->has_weight() ) return pProtobufEvent->weight(); else return 1; }
25 void setWeight( float newWeight ) { pProtobufEvent->set_weight(newWeight); }
26 l1menuprotobuf::Event* pProtobufEvent;
27 };
28
29 /** @brief Sentry that closes a Unix file descriptor when it goes out of scope.
30 * @author Mark Grimes (mark.grimes@bristol.ac.uk)
31 * @date 07/Jun/2013
32 */
33 class UnixFileSentry
34 {
35 public:
36 UnixFileSentry( int fileDescriptor ) : fileDescriptor_(fileDescriptor) {}
37 ~UnixFileSentry() { close(fileDescriptor_); }
38 private:
39 int fileDescriptor_;
40 };
41 }
42
43 namespace l1menu
44 {
45 /** @brief Private members for the ReducedMenuSample class
46 *
47 * @author Mark Grimes (mark.grimes@bristol.ac.uk)
48 * @date 28/May/2013
49 */
50 class ReducedMenuSamplePrivateMembers
51 {
52 private:
53 l1menu::TriggerMenu mutableTriggerMenu_;
54 public:
55 ReducedMenuSamplePrivateMembers( const l1menu::TriggerMenu& newTriggerMenu );
56 ReducedMenuSamplePrivateMembers( const std::string& filename );
57 void copyMenuToProtobufSample();
58 ::ReducedEventImplementation event;
59 const l1menu::TriggerMenu& triggerMenu;
60 l1menuprotobuf::SampleHeader protobufSampleHeader;
61 // Protobuf doesn't implement move semantics so I'll use pointers
62 std::vector<std::unique_ptr<l1menuprotobuf::Run> > protobufRuns;
63 const static int EVENTS_PER_RUN;
64 const static char PROTOBUF_MESSAGE_DELIMETER;
65 const static std::string FILE_FORMAT_MAGIC_NUMBER;
66 };
67
68 const int ReducedMenuSamplePrivateMembers::EVENTS_PER_RUN=20000;
69 const char ReducedMenuSamplePrivateMembers::PROTOBUF_MESSAGE_DELIMETER='\n';
70 const std::string ReducedMenuSamplePrivateMembers::FILE_FORMAT_MAGIC_NUMBER="l1menuReducedMenuSample";
71 }
72
73 l1menu::ReducedMenuSamplePrivateMembers::ReducedMenuSamplePrivateMembers( const l1menu::TriggerMenu& newTriggerMenu )
74 : mutableTriggerMenu_( newTriggerMenu ), triggerMenu( mutableTriggerMenu_ )
75 {
76 GOOGLE_PROTOBUF_VERIFY_VERSION;
77
78 // I need to copy the details of the trigger menu into the protobuf storage.
79 // This means I'm holding a duplicate, but I need it to write the sample to a
80 // protobuf file, so I might as well do it now.
81 for( size_t triggerNumber=0; triggerNumber<triggerMenu.numberOfTriggers(); ++triggerNumber )
82 {
83 const l1menu::ITrigger& trigger=triggerMenu.getTrigger(triggerNumber);
84
85 l1menuprotobuf::Trigger* pProtobufTrigger=protobufSampleHeader.add_trigger();
86 pProtobufTrigger->set_name( trigger.name() );
87 pProtobufTrigger->set_version( trigger.version() );
88
89 // Record all of the parameters. It's not strictly necessary to record the values
90 // of the parameters that are recorded for each event, but I might as well so that
91 // the trigger menu is loaded exactly as it was saved.
92 const auto parameterNames=trigger.parameterNames();
93 for( const auto& parameterName : parameterNames )
94 {
95 l1menuprotobuf::Trigger_TriggerParameter* pProtobufParameter=pProtobufTrigger->add_parameter();
96 pProtobufParameter->set_name(parameterName);
97 pProtobufParameter->set_value( trigger.parameter(parameterName) );
98 }
99
100 // Make a note of the names of the parameters that are recorded for each event. For this
101 // I'm just recording the parameters that refer to the thresholds.
102 const auto thresholdNames=l1menu::getThresholdNames(trigger);
103 for( const auto& thresholdName : thresholdNames ) pProtobufTrigger->add_varying_parameter(thresholdName);
104
105 } // end of loop over triggers
106
107 // Always make sure there is at least one Run ready to be added to
108 std::unique_ptr<l1menuprotobuf::Run> pNewRun( new l1menuprotobuf::Run );
109 protobufRuns.push_back( std::move( pNewRun ) );
110
111 }
112
113 l1menu::ReducedMenuSamplePrivateMembers::ReducedMenuSamplePrivateMembers( const std::string& filename )
114 : triggerMenu(mutableTriggerMenu_)
115 {
116 GOOGLE_PROTOBUF_VERIFY_VERSION;
117
118 // Open the file with read ability
119 int fileDescriptor = open( filename.c_str(), O_RDONLY );
120 if( fileDescriptor==0 ) throw std::runtime_error( "ReducedMenuSample initialise from file - couldn't open file" );
121 ::UnixFileSentry fileSentry( fileDescriptor ); // Use this as an exception safe way of closing the input file
122 google::protobuf::io::FileInputStream fileInput( fileDescriptor );
123 google::protobuf::io::GzipInputStream gzipInput( &fileInput );
124 google::protobuf::io::CodedInputStream codedInput( &gzipInput );
125
126 // Disable warnings on this input stream (second parameter, -1). The
127 // first parameter is the default. I'll change this if necessary in
128 // the loop later.
129 size_t totalBytesLimit=67108864;
130 codedInput.SetTotalBytesLimit( totalBytesLimit, -1 );
131
132 // First read the magic number at the start of the file and make sure it
133 // matches what I expect. As a read buffer, I'll create a string the correct
134 // size (filled with an arbitrary character) and read straight into that.
135 std::string readMagicNumber;
136 if( !codedInput.ReadString( &readMagicNumber, FILE_FORMAT_MAGIC_NUMBER.size() ) ) throw std::runtime_error( "ReducedMenuSample initialise from file - error reading magic number" );
137 if( readMagicNumber!=FILE_FORMAT_MAGIC_NUMBER ) throw std::runtime_error( "ReducedMenuSample - tried to initialise with a file that is not the correct format" );
138
139 google::protobuf::uint32 fileformatVersion;
140 if( !codedInput.ReadVarint32( &fileformatVersion ) ) throw std::runtime_error( "ReducedMenuSample initialise from file - error reading file format version" );
141 // So far I only have (and ever expect to have) one version of the file
142 // format, imaginatively versioned "1". You never know though...
143 if( fileformatVersion>1 ) std::cerr << "Warning: Attempting to read a ReducedMenuSample with version " << fileformatVersion << " with code that only knows up to version 1." << std::endl;
144
145 google::protobuf::uint64 messageSize;
146
147 // Read the size of the header message
148 if( !codedInput.ReadVarint64( &messageSize ) ) throw std::runtime_error( "ReducedMenuSample initialise from file - error reading message size for header" );
149 google::protobuf::io::CodedInputStream::Limit readLimit=codedInput.PushLimit(messageSize);
150 if( !protobufSampleHeader.ParseFromCodedStream( &codedInput ) ) throw std::runtime_error( "ReducedMenuSample initialise from file - some unknown error while reading header" );
151 codedInput.PopLimit(readLimit);
152
153 // Keep looping until there is nothing more to be read from the file.
154 while( codedInput.ReadVarint64( &messageSize ) )
155 {
156 readLimit=codedInput.PushLimit(messageSize);
157
158 // Make sure the CodedInputStream doesn't refuse to read the message because it's
159 // read too much already. I'll also add an arbitrary 50 on to always make sure
160 // I can read the next messageSize if there is one.
161 if( gzipInput.ByteCount()+messageSize+50 > totalBytesLimit )
162 {
163 totalBytesLimit+=messageSize*5; // Might as well set it a little higher than necessary while I'm at it.
164 codedInput.SetTotalBytesLimit( totalBytesLimit, -1 );
165 }
166 std::unique_ptr<l1menuprotobuf::Run> pNewRun( new l1menuprotobuf::Run );
167 if( !pNewRun->ParseFromCodedStream( &codedInput ) ) throw std::runtime_error( "ReducedMenuSample initialise from file - some unknown error while reading run" );
168 protobufRuns.push_back( std::move( pNewRun ) );
169
170 codedInput.PopLimit(readLimit);
171 }
172
173
174 // Always make sure there is at least one Run ready to be added to. Later
175 // code assumes there is already a run there.
176 if( protobufRuns.empty() )
177 {
178 std::unique_ptr<l1menuprotobuf::Run> pNewRun( new l1menuprotobuf::Run );
179 protobufRuns.push_back( std::move( pNewRun ) );
180 }
181
182 // I have all of the information in the protobuf members, but I also need the trigger information
183 // in the form of l1menu::TriggerMenu. Copy out the required information.
184 for( int triggerNumber=0; triggerNumber<protobufSampleHeader.trigger_size(); ++triggerNumber )
185 {
186 const l1menuprotobuf::Trigger& inputTrigger=protobufSampleHeader.trigger(triggerNumber);
187
188 mutableTriggerMenu_.addTrigger( inputTrigger.name(), inputTrigger.version() );
189 // Get a reference to the trigger I just created
190 l1menu::ITrigger& trigger=mutableTriggerMenu_.getTrigger(triggerNumber);
191
192 // Run through all of the parameters and set them to what they were
193 // when the sample was made.
194 for( int parameterNumber=0; parameterNumber<inputTrigger.parameter_size(); ++parameterNumber )
195 {
196 const auto& inputParameter=inputTrigger.parameter(parameterNumber);
197 trigger.parameter(inputParameter.name())=inputParameter.value();
198 }
199
200 // I should probably check the threshold names exist. I'll do it another time.
201 }
202
203 }
204
205 l1menu::ReducedMenuSample::ReducedMenuSample( const l1menu::MenuSample& originalSample, const l1menu::TriggerMenu& triggerMenu )
206 : pImple_( new l1menu::ReducedMenuSamplePrivateMembers( triggerMenu ) )
207 {
208 addSample( originalSample );
209 }
210
211 l1menu::ReducedMenuSample::ReducedMenuSample( const l1menu::TriggerMenu& triggerMenu )
212 : pImple_( new l1menu::ReducedMenuSamplePrivateMembers( triggerMenu ) )
213 {
214 // No operation besides the initialiser list
215 }
216
217 l1menu::ReducedMenuSample::~ReducedMenuSample()
218 {
219 // No operation. Just need one defined otherwise the default one messes up
220 // the unique_ptr deletion because ReducedMenuSamplePrivateMembers isn't
221 // defined elsewhere.
222 }
223
224 void l1menu::ReducedMenuSample::addSample( const l1menu::MenuSample& originalSample )
225 {
226 l1menuprotobuf::Run* pCurrentRun=pImple_->protobufRuns.back().get();
227
228 for( size_t eventNumber=0; eventNumber<originalSample.numberOfEvents(); ++eventNumber )
229 {
230 // Split the events up into groups in arbitrary numbers. This is to get around
231 // a protobuf aversion to long messages.
232 if( pCurrentRun->event_size() >= pImple_->EVENTS_PER_RUN )
233 {
234 // Gone over the arbitrary limit, so create a new protobuf Run and start
235 // using that instead.
236 std::unique_ptr<l1menuprotobuf::Run> pNewRun( new l1menuprotobuf::Run );
237 pImple_->protobufRuns.push_back( std::move( pNewRun ) );
238 pCurrentRun=pImple_->protobufRuns.back().get();
239 }
240
241 const l1menu::IEvent& event=originalSample.getEvent( eventNumber );
242 l1menuprotobuf::Event* pProtobufEvent=pCurrentRun->add_event();
243
244 // Loop over all of the triggers
245 for( size_t triggerNumber=0; triggerNumber<pImple_->triggerMenu.numberOfTriggers(); ++triggerNumber )
246 {
247 std::unique_ptr<l1menu::ITrigger> pTrigger=pImple_->triggerMenu.getTriggerCopy(triggerNumber);
248 std::vector<std::string> thresholdNames=getThresholdNames(*pTrigger);
249
250 try
251 {
252 setTriggerThresholdsAsTightAsPossible( event, *pTrigger, 0.001 );
253 // Set all of the parameters to match the thresholds in the trigger
254 for( const auto& thresholdName : thresholdNames )
255 {
256 pProtobufEvent->add_threshold( pTrigger->parameter(thresholdName) );
257 }
258 }
259 catch( std::exception& error )
260 {
261 // setTriggerThresholdsAsTightAsPossible() couldn't find thresholds so record
262 // -1 for everything.
263 // Range based for loop gives me a warning because I don't use the thresholdName.
264 for( size_t index=0; index<thresholdNames.size(); ++index ) pProtobufEvent->add_threshold(-1);
265 } // end of try block that sets the trigger thresholds
266
267 } // end of loop over triggers
268
269 } // end of loop over events
270 }
271
272 l1menu::ReducedMenuSample::ReducedMenuSample( const std::string& filename )
273 : pImple_( new l1menu::ReducedMenuSamplePrivateMembers( filename ) )
274 {
275 // No operation except the initialiser list
276 }
277
278 void l1menu::ReducedMenuSample::saveToFile( const std::string& filename ) const
279 {
280 // Open the file. Parameters are filename, write ability and create, rw-r--r-- permissions.
281 int fileDescriptor = open( filename.c_str(), O_WRONLY | O_CREAT, 0644 );
282 if( fileDescriptor==0 ) throw std::runtime_error( "ReducedMenuSample save to file - couldn't open file" );
283 ::UnixFileSentry fileSentry( fileDescriptor ); // Use this as an exception safe way of closing the output file
284
285 // Setup the protobuf file handlers
286 google::protobuf::io::FileOutputStream fileOutput( fileDescriptor );
287 google::protobuf::io::GzipOutputStream gzipOutput( &fileOutput );
288 google::protobuf::io::CodedOutputStream codedOutput( &gzipOutput );
289
290 // Write a magic number at the start of all files
291 codedOutput.WriteString( pImple_->FILE_FORMAT_MAGIC_NUMBER );
292 // Write an integer that specifies what version of the file format I'm using. I
293 // have no intention of changing the format but I might as well keep the option
294 // open.
295 codedOutput.WriteVarint32( 1 );
296
297 // Write the size of the header message into the file...
298 codedOutput.WriteVarint64( pImple_->protobufSampleHeader.ByteSize() );
299 // ...and then write the header
300 pImple_->protobufSampleHeader.SerializeToCodedStream( &codedOutput );
301
302 // Now go through each of the runs and do the same for those
303 for( const auto& pRun : pImple_->protobufRuns )
304 {
305 codedOutput.WriteVarint64( pRun->ByteSize() );
306 pRun->SerializeToCodedStream( &codedOutput );
307 }
308
309 }
310
311 size_t l1menu::ReducedMenuSample::numberOfEvents() const
312 {
313 size_t numberOfEvents=0;
314 for( const auto& pRun : pImple_->protobufRuns ) numberOfEvents+=pRun->event_size();
315 return numberOfEvents;
316 }
317
318 const l1menu::IReducedEvent& l1menu::ReducedMenuSample::getEvent( size_t eventNumber ) const
319 {
320 for( const auto& pRun : pImple_->protobufRuns )
321 {
322 if( eventNumber<static_cast<size_t>(pRun->event_size()) )
323 {
324 pImple_->event.pProtobufEvent=pRun->mutable_event(eventNumber);
325 return pImple_->event;
326 }
327 // Event must be in a later run, so reduce the number by how many events
328 // were in this run and look again.
329 eventNumber-=pRun->event_size();
330 }
331
332 // Should always find the event before getting to this point, so throw an
333 // exception if this happens.
334 throw std::runtime_error( "ReducedMenuSample::getEvent(eventNumber) was asked for an invalid eventNumber" );
335 }
336
337 const l1menu::TriggerMenu& l1menu::ReducedMenuSample::getTriggerMenu() const
338 {
339 return pImple_->triggerMenu;
340 }
341
342 bool l1menu::ReducedMenuSample::containsTrigger( const l1menu::ITrigger& trigger, bool allowOlderVersion ) const
343 {
344 // Loop over all of the triggers in the menu, and see if there is one
345 // where the name and version match.
346 for( size_t triggerNumber=0; triggerNumber<pImple_->triggerMenu.numberOfTriggers(); ++triggerNumber )
347 {
348 const l1menu::ITrigger& triggerInMenu=pImple_->triggerMenu.getTrigger(triggerNumber);
349 if( triggerInMenu.name()!=trigger.name() ) continue;
350 if( allowOlderVersion )
351 {
352 if( triggerInMenu.version()>trigger.version() ) continue;
353 }
354 else
355 {
356 if( triggerInMenu.version()!=trigger.version() ) continue;
357 }
358
359 // If control got this far then there is a trigger with the required name
360 // and sufficient version. I now need to check all of the non threshold parameters
361 // to make sure they match, i.e. make sure the ReducedSample was made with the same
362 // eta cuts or whatever.
363 // I don't care if the thresholds don't match because that's what's stored in the
364 // ReducedMenuSample.
365 std::vector<std::string> parameterNames=getNonThresholdParameterNames( trigger );
366 bool allParametersMatch=true;
367 for( const auto& parameterName : parameterNames )
368 {
369 if( trigger.parameter(parameterName)!=triggerInMenu.parameter(parameterName) ) allParametersMatch=false;
370 }
371
372 if( allParametersMatch ) return true;
373 } // end of loop over triggers
374
375 // If control got this far then no trigger was found that matched
376 return false;
377 }
378
379 const std::map<std::string,size_t> l1menu::ReducedMenuSample::getTriggerParameterIdentifiers( const l1menu::ITrigger& trigger, bool allowOlderVersion ) const
380 {
381 std::map<std::string,size_t> returnValue;
382
383 // Need to find out how many parameters there are for each event. Basically the sum
384 // of the number of thresholds for all triggers.
385 size_t parameterNumber=0;
386 bool triggerWasFound=true;
387 for( size_t triggerNumber=0; triggerNumber<pImple_->triggerMenu.numberOfTriggers(); ++triggerNumber )
388 {
389 const l1menu::ITrigger& triggerInMenu=pImple_->triggerMenu.getTrigger(triggerNumber);
390
391 triggerWasFound=true; // Set to true, then back to false if any of the tests fail
392 // See if this trigger in the menu is the same as the one passed as a parameter
393 if( triggerInMenu.name()!=trigger.name() ) triggerWasFound=false;
394 if( allowOlderVersion )
395 {
396 if( triggerInMenu.version()>trigger.version() ) triggerWasFound=false;
397 }
398 else
399 {
400 if( triggerInMenu.version()!=trigger.version() ) triggerWasFound=false;
401 }
402
403 // If control got this far then there is a trigger with the required name
404 // and sufficient version. I now need to check all of the non threshold parameters
405 // to make sure they match, i.e. make sure the ReducedSample was made with the same
406 // eta cuts or whatever.
407 // I don't care if the thresholds don't match because that's what's stored in the
408 // ReducedMenuSample.
409 if( triggerWasFound ) // Trigger can still fail, but no point doing this check if it already has
410 {
411 std::vector<std::string> parameterNames=getNonThresholdParameterNames( trigger );
412 for( const auto& parameterName : parameterNames )
413 {
414 if( trigger.parameter(parameterName)!=triggerInMenu.parameter(parameterName) ) triggerWasFound=false;
415 }
416 }
417
418 std::vector<std::string> thresholdNames=l1menu::getThresholdNames(triggerInMenu);
419 if( triggerWasFound )
420 {
421 for( const auto& thresholdName : thresholdNames )
422 {
423 returnValue[thresholdName]=parameterNumber;
424 ++parameterNumber;
425 }
426 break;
427 }
428 else parameterNumber+=thresholdNames.size();
429 }
430
431 // There could conceivably be a trigger that was found but has no thresholds
432 // (I guess - it would be a pretty pointless trigger though). To indicate the
433 // difference between that and a trigger that wasn't found I'll respectively
434 // return the empty vector or throw an exception.
435 if( !triggerWasFound ) throw std::runtime_error( "l1menu::ReducedMenuSample::getTriggerParameterIdentifiers() called for a trigger that was not used to create the sample" );
436
437 return returnValue;
438 }