| Geant4 Cross Reference |
1 // Copyright (C) 2010, Guy Barrand. All rights reserved.
2 // See the file tools.license for terms.
3
4 #ifndef tools_rroot_file
5 #define tools_rroot_file
6
7 #include "ifile"
8
9 #include "directory"
10
11 #include "../platform"
12
13 #include "obj_list"
14 #include "info"
15 #include "streamer_fac"
16
17 #include <string>
18 #include <fcntl.h>
19 #include <errno.h>
20
21 #if defined(_MSC_VER) || defined(__MINGW32__)
22 #include <io.h>
23 #include <sys/stat.h>
24 #else
25 #include <unistd.h>
26 #endif
27
28 namespace tools {
29 namespace rroot {
30
31 class file : public virtual ifile {
32 file& get_me() {return *this;} //_MSC_VER : to avoid warning about the usage of "this" in the constructor.
33 static int not_open() {return -1;}
34 public:
35 static const std::string& s_class() {
36 static const std::string s_v("tools::rroot::file");
37 return s_v;
38 }
39 virtual const std::string& s_cls() const {return s_class();}
40 public: //ifile
41 virtual const std::string& path() const {return m_path;}
42
43 virtual bool verbose() const {return m_verbose;}
44 virtual std::ostream& out() const {return m_out;}
45
46 virtual bool byte_swap() const {return is_little_endian();}
47 virtual bool set_pos(seek a_offset = 0,from a_from = begin){
48 int whence = 0;
49 switch(a_from) {
50 case begin:
51 whence = SEEK_SET;
52 break;
53 case current:
54 whence = SEEK_CUR;
55 break;
56 case end:
57 whence = SEEK_END;
58 break;
59 }
60
61 #if defined(__linux__) && (__GLIBC__ == 2) && (__GLIBC_MINOR__ >= 2)
62 if (::lseek64(m_file, a_offset, whence) < 0) {
63 #elif defined(_MSC_VER) || defined(__MINGW32__)
64 if (::_lseeki64(m_file, a_offset, whence) < 0) {
65 #else
66 if (::lseek(m_file, a_offset, whence) < 0) {
67 #endif
68 m_out << "tools::rroot::file::set_pos :"
69 << " cannot set position " << a_offset
70 << " in file " << sout(m_path) << "."
71 << std::endl;
72 return false;
73 }
74 return true;
75 }
76 virtual bool read_buffer(char* a_buffer,uint32 a_length) {
77 // Read a buffer from the file.
78 // This is the basic low level read operation.
79 #ifdef _MSC_VER
80 typedef int ssize_t;
81 #endif
82 ssize_t siz;
83 while ((siz = ::read(m_file,a_buffer,a_length)) < 0 &&
84 error_number() == EINTR) reset_error_number();
85 if (siz < 0) {
86 m_out << "tools::rroot::file::read_buffer :"
87 << " error reading from file " << sout(m_path) << "."
88 << std::endl;
89 return false;
90 }
91 if (siz != ssize_t(a_length)) {
92 m_out << "tools::rroot::file::read_buffer :"
93 << " error reading all requested bytes from file "
94 << sout(m_path) << ", got " << long_out(siz)
95 << " of " << a_length
96 << std::endl;
97 return false;
98 }
99 m_bytes_read += siz;
100 return true;
101 }
102 virtual bool unziper(char a_key,decompress_func& a_func) const {
103 std::map<char,decompress_func>::const_iterator it = m_unzipers.find(a_key);
104 if(it==m_unzipers.end()) {
105 a_func = 0;
106 return false;
107 }
108 a_func = (*it).second;
109 return true;
110 }
111
112 virtual key& sinfos_key() {return m_streamer_infos_key;}
113
114 public:
115 file(std::ostream& a_out,const std::string& a_path,bool a_verbose = false)
116 :m_out(a_out)
117 ,m_path(a_path)
118 ,m_verbose(a_verbose)
119 ,m_file(not_open())
120 ,m_bytes_read(0)
121 ,m_root_directory(get_me())
122 ,m_streamer_infos_key(a_out)
123 ,m_streamer_fac(a_out)
124 ,m_streamer_infos(m_streamer_fac)
125 // begin of record :
126 ,m_version(0)
127 ,m_BEGIN(0)
128 ,m_END(0)
129 ,m_seek_free(0)
130 ,m_seek_info(0)
131 ,m_nbytes_free(0)
132 ,m_nbytes_info(0)
133 ,m_nbytes_name(0)
134 {
135 #ifdef TOOLS_MEM
136 mem::increment(s_class().c_str());
137 #endif
138
139 m_file = _open(a_path.c_str(),
140 #if defined(_MSC_VER) || defined(__MINGW32__)
141 O_RDONLY | O_BINARY,S_IREAD | S_IWRITE
142 #else
143 O_RDONLY,0644
144 #endif
145 );
146 if(m_file==not_open()) {
147 m_out << "tools::rroot::file::file :"
148 << " can't open " << sout(a_path) << "."
149 << std::endl;
150 return;
151 }
152 initialize();
153 }
154 virtual ~file() {
155 close();
156 #ifdef TOOLS_MEM
157 mem::decrement(s_class().c_str());
158 #endif
159 }
160 protected:
161 file(const file& a_from)
162 :ifile(a_from)
163 ,m_out(a_from.m_out)
164 ,m_root_directory(get_me())
165 ,m_streamer_infos_key(a_from.m_out)
166 ,m_streamer_fac(a_from.m_out)
167 ,m_streamer_infos(m_streamer_fac)
168 {
169 #ifdef TOOLS_MEM
170 mem::increment(s_class().c_str());
171 #endif
172 }
173 file& operator=(const file&){return *this;}
174 public:
175 uint32 version() const {return m_version;}
176
177 bool is_open() const {
178 return (m_file==not_open()?false:true);
179 }
180
181 void close() {
182 if(m_file!=not_open()) ::close(m_file);
183 m_file = not_open();
184 m_root_directory.clear_keys();
185 }
186
187 directory& dir() {return m_root_directory;}
188 const directory& dir() const {return m_root_directory;}
189
190 bool add_unziper(char a_key,decompress_func a_func){
191 std::map<char,decompress_func>::const_iterator it = m_unzipers.find(a_key);
192 if(it!=m_unzipers.end()) {
193 //(*it).second = a_func; //override ?
194 return false;
195 } else {
196 m_unzipers[a_key] = a_func;
197 return true;
198 }
199 }
200
201 bool dump_streamer_infos() {
202 // read_streamer_infos_data() done here (and not in initialize) since it may need to have unziper declared.
203 if(m_streamer_infos.empty()) {if(!read_streamer_infos_data()) return false;}
204 tools_vforcit(iro*,m_streamer_infos,it) {
205 streamer_info* info = safe_cast<iro,streamer_info>(*(*it));
206 if(!info) return false;
207 info->out(m_out);
208 }
209 return true;
210 }
211 streamer_info* find_streamer_info(const std::string& a_class) {
212 // read_streamer_infos_data() done here (and not in initialize) since it may need to have unziper declared.
213 if(m_streamer_infos.empty()) {if(!read_streamer_infos_data()) return 0;}
214 tools_vforcit(iro*,m_streamer_infos,it) {
215 streamer_info* info = safe_cast<iro,streamer_info>(*(*it));
216 if(info) {
217 if(info->name()==a_class) return info;
218 }
219 }
220 return 0;
221 }
222
223 protected:
224 static int _open(const char* a_name,int a_flags,uint32 a_mode) {
225 #if defined(__linux__) && (__GLIBC__ == 2) && (__GLIBC_MINOR__ >= 2)
226 return ::open64(a_name,a_flags,a_mode);
227 #else
228 return ::open(a_name,a_flags,a_mode);
229 #endif
230 }
231 static std::string sout(const std::string& a_string) {return "\""+a_string+"\"";}
232 bool initialize() {
233 if(!read_header()) {
234 m_out << "tools::rroot::file::initialize :"
235 << " can't read header."
236 << std::endl;
237 return false;
238 }
239 /*
240 fRootDirectory->setSeekDirectory(fBEGIN);
241 // Read Free segments structure if file is writable :
242 if (fWritable) {
243 if (fSeekFree > fBEGIN) {
244 if(!readFreeSegments()) {
245 m_out << "tools::rroot::file::initialize : Cannot read free segments."
246 << std::endl;
247 return false;
248 }
249 } else {
250 m_out << "tools::rroot::file::initialize : file \"" << fName
251 << "\" probably not closed, cannot read free segments" << std::endl;
252 }
253 }
254 */
255 // Read Directory info :
256 uint32 nbytes = m_nbytes_name + m_root_directory.record_size(m_version);
257 char* header = new char[nbytes];
258 char* buffer = header;
259 if(!set_pos(m_BEGIN)) {
260 m_out << "tools::rroot::file::initialize :"
261 << " can't set position."
262 << std::endl;
263 delete [] header;
264 return false;
265 }
266 if(!read_buffer(buffer,nbytes)) {
267 m_out << "tools::rroot::file::initialize :"
268 << " can't read buffer."
269 << std::endl;
270 delete [] header;
271 return false;
272 }
273 buffer = header+m_nbytes_name;
274 const char* eob = header+nbytes;
275 if(!m_root_directory.from_buffer(eob,buffer)) {
276 m_out << "tools::rroot::file::initialize :"
277 << " can't read buffer (2)."
278 << std::endl;
279 delete [] header;
280 return false;
281 }
282 uint32 nk = //size of Key
283 sizeof(int) + //Key::fNumberOfBytes
284 sizeof(short) + //Key::fVersion
285 2*sizeof(int) + //Key::fObjectSize, Date
286 2*sizeof(short) + //Key::fKeyLength,fCycle
287 2*sizeof(seek32); //Key::fSeekKey,fSeekParentDirectory
288 //WARNING : the upper is seek32 since at begin of file.
289 buffer = header+nk;
290 std::string cname;
291 rbuf rb(m_out,byte_swap(),eob,buffer);
292 // Should be "TFile".
293 if(!rb.read(cname)) {
294 m_out << "tools::rroot::file::initialize :"
295 << " can't read buffer (3)."
296 << std::endl;
297 delete [] header;
298 return false;
299 }
300 if(cname!="TFile") {
301 m_out << "tools::rroot::file::initialize : TFile expected." << std::endl;
302 delete [] header;
303 return false;
304 }
305 if(m_verbose) {
306 m_out << "tools::rroot::file::initialize :"
307 << " " << sout("TFile") << " found."
308 << std::endl;
309 }
310 if(!rb.read(cname)) {
311 m_out << "tools::rroot::file::initialize :"
312 << " can't read buffer (4)."
313 << std::endl;
314 delete [] header;
315 return false;
316 }
317 if(m_verbose) {
318 m_out << "tools::rroot::file::initialize :"
319 << " found file name " << sout(cname)
320 << std::endl;
321 }
322 if(!rb.read(m_title)) {
323 m_out << "tools::rroot::file::initialize :"
324 << " can't read buffer (5)."
325 << std::endl;
326 delete [] header;
327 return false;
328 }
329 delete [] header;
330 if(m_verbose) {
331 m_out << "tools::rroot::file::initialize :"
332 << " found title " << sout(m_title)
333 << std::endl;
334 }
335 uint32 dirNbytesName = m_root_directory.nbytes_name();
336 if (dirNbytesName < 10 || dirNbytesName > 1000) {
337 m_out << "tools::rroot::file::initialize :"
338 << " can't read directory info."
339 << std::endl;
340 return false;
341 }
342 // Read keys of the top directory :
343 if(m_root_directory.seek_keys() > m_BEGIN) {
344 uint32 n;
345 if(!m_root_directory.read_keys(n)) {
346 m_out << "tools::rroot::file::initialize :"
347 << " can't read keys."
348 << std::endl;
349 return false;
350 }
351 } else {
352 m_out << "tools::rroot::file::initialize :"
353 << " file " << sout(m_path)
354 << " probably not closed."
355 << std::endl;
356 return false;
357 }
358
359 // Create StreamerInfo index
360 if(m_seek_info > m_BEGIN) {
361 if(!read_streamer_infos_key()) {
362 m_out << "tools::rroot::file::initialize :"
363 << " read_streamer_infos_key() failed."
364 << std::endl;
365 return false;
366 }
367 } else {
368 m_out << "tools::rroot::file::initialize :"
369 << " file " << sout(m_path)
370 << " probably not closed."
371 << std::endl;
372 return false;
373 }
374
375 return true;
376 }
377 bool read_header() {
378 static const uint32 kBegin = 64;
379 char header[kBegin];
380 if(!set_pos()) return false;
381 if(!read_buffer(header,kBegin)) return false;
382 // make sure this is a root file
383 if(::strncmp(header, "root", 4)) {
384 m_out << "tools::rroot::file::read_header :"
385 << " " << sout(m_path) << " not a file at the CERN-ROOT format."
386 << std::endl;
387 return false;
388 }
389 if(m_verbose) {
390 m_out << "tools::rroot::file::read_header :"
391 << " file signature is " << sout("root")
392 << std::endl;
393 }
394 char* buffer = header + 4; // skip the "root" file identifier
395 const char* eob = header + kBegin;
396 rbuf rb(m_out,byte_swap(),eob,buffer);
397 {int v;
398 if(!rb.read(v)) return false;
399 m_version = v;}
400 {seek32 i;
401 if(!rb.read(i)) return false;
402 m_BEGIN = i;}
403 if(m_version>1000000) {
404 if(!rb.read(m_END)) return false;
405 if(!rb.read(m_seek_free)) return false;
406 } else {
407 {seek32 i;
408 if(!rb.read(i)) return false;
409 m_END = i;}
410 {seek32 i;
411 if(!rb.read(i)) return false;
412 m_seek_free = i;}
413 }
414 if(m_verbose) {
415 m_out << "tools::rroot::file::read_header :"
416 << " begin " << m_BEGIN
417 << " end " << m_END
418 << std::endl;
419 }
420 {int v;
421 if(!rb.read(v)) return false;
422 m_nbytes_free = v;}
423 int nfree = 0;
424 if(!rb.read(nfree)) return false;
425 {int v;
426 if(!rb.read(v)) return false;
427 m_nbytes_name = v;}
428 //m_out << "debug : 1002 " << m_nbytes_name << std::endl;
429 {char fUnits;
430 if(!rb.read(fUnits)) return false;}
431 {int fCompress;
432 if(!rb.read(fCompress)) return false;}
433 if(m_version>1000000) {
434 if(!rb.read(m_seek_info)) return false;
435 } else {
436 {seek32 i;
437 if(!rb.read(i)) return false;
438 m_seek_info = i;}
439 }
440 if(!rb.read(m_nbytes_info)) return false;
441 //m_out << "debug : seek_info " << m_seek_info << " nbytes_info " << m_nbytes_info << std::endl;
442 return true;
443 }
444
445 bool read_streamer_infos_key() {
446 // Read the list of StreamerInfo from this file
447 // The key with name holding the list of TStreamerInfo objects is read.
448 // The corresponding TClass objects are updated.
449 if(m_seek_info<=0) return false;
450 if(m_seek_info>=m_END) return false;
451 if(!set_pos(m_seek_info)) return false;
452 char* buffer = new char[m_nbytes_info+1];
453 if(!read_buffer(buffer,m_nbytes_info)) {delete [] buffer;return false;}
454 char* buf = buffer;
455 if(!m_streamer_infos_key.from_buffer(byte_swap(),buffer+m_nbytes_info,buf,m_verbose)) {
456 delete [] buffer;
457 return false;
458 }
459 delete [] buffer;
460 return true;
461 }
462
463 bool read_streamer_infos_data() {
464 key& k = m_streamer_infos_key;
465 if(k.object_class()!="TList") {
466 m_out << "tools::rroot::file::read_streamer_infos_data : key not a TList." << std::endl;
467 return false;
468 }
469 unsigned int sz;
470 char* buf = k.get_object_buffer(*this,sz); //we don't have ownership of buf.
471 if(!buf) {
472 m_out << "tools::rroot::file::read_streamer_infos :"
473 << " can't get data buffer of " << k.object_name() << "."
474 << std::endl;
475 return false;
476 }
477 buffer b(m_out,byte_swap(),sz,buf,k.key_length(),false);
478 return m_streamer_infos.stream(b);
479 }
480
481 #if defined(__sun) && !defined(__linux__) && (__SUNPRO_CC > 0x420)
482 int error_number() {return ::errno;}
483 void reset_error_number() {::errno = 0;}
484 #else
485 int error_number() {return errno;}
486 void reset_error_number() {errno = 0;}
487 #endif
488
489 protected:
490 std::ostream& m_out;
491 std::string m_path;
492 bool m_verbose;
493 int m_file;
494 uint64 m_bytes_read; //Number of bytes read from this file
495 directory m_root_directory;
496 key m_streamer_infos_key;
497 streamer_fac m_streamer_fac;
498 obj_list m_streamer_infos;
499 std::map<char,decompress_func> m_unzipers;
500 std::string m_title;
501 // begin of record :
502 uint32 m_version; //File format version
503 seek m_BEGIN; //First used byte in file
504 seek m_END; //Last used byte in file
505 seek m_seek_free; //Location on disk of free segments structure
506 seek m_seek_info; //Location on disk of StreamerInfo record
507 uint32 m_nbytes_free; //Number of bytes for free segments structure
508 uint32 m_nbytes_info; //Number of bytes for StreamerInfo record
509 //int nfree
510 uint32 m_nbytes_name; //Number of bytes in TNamed at creation time
511 };
512
513
514 }}
515
516 #endif
517
518 //doc
519 //
520 // A ROOT file is a suite of consecutive data records with the following
521 // format (see also the TKey class);
522 // TKey ---------------------
523 // byte 1->4 Nbytes = Length of compressed object (in bytes)
524 // 5->6 Version = TKey version identifier
525 // 7->10 ObjLen = Length of uncompressed object
526 // 11->14 Datime = Date and time when object was written to file
527 // 15->16 KeyLen = Length of the key structure (in bytes)
528 // 17->18 Cycle = Cycle of key
529 // 19->22 SeekKey = Pointer to record itself (consistency check)
530 // 23->26 SeekPdir = Pointer to directory header
531 // 27->27 lname = Number of bytes in the class name
532 // 28->.. ClassName = Object Class Name
533 // ..->.. lname = Number of bytes in the object name
534 // ..->.. Name = lName bytes with the name of the object
535 // ..->.. lTitle = Number of bytes in the object title
536 // ..->.. Title = Title of the object
537 // -----> DATA = Data bytes associated to the object
538 //
539 // The first data record starts at byte fBEGIN (currently set to kBegin)
540 // Bytes 1->kBegin contain the file description:
541 // byte 1->4 "root" = Root file identifier
542 // 5->8 fVersion = File format version
543 // 9->12 fBEGIN = Pointer to first data record
544 // 13->16 fEND = Pointer to first free word at the EOF
545 // 17->20 fSeekFree = Pointer to FREE data record
546 // 21->24 fNbytesFree = Number of bytes in FREE data record
547 // 25->28 nfree = Number of free data records
548 // 29->32 fNbytesName = Number of bytes in TNamed at creation time
549 // 33->33 fUnits = Number of bytes for file pointers
550 // 34->37 fCompress = Zip compression level
551 //