]>
Commit | Line | Data |
---|---|---|
1 | // -*- mode: cpp; mode: fold -*- | |
2 | // Description /*{{{*/ | |
3 | // $Id: multicompress.cc,v 1.4 2003/02/10 07:34:41 doogie Exp $ | |
4 | /* ###################################################################### | |
5 | ||
6 | MultiCompressor | |
7 | ||
8 | This class is very complicated in order to optimize for the common | |
9 | case of its use, writing a large set of compressed files that are | |
10 | different from the old set. It spawns off compressors in parallel | |
11 | to maximize compression throughput and has a separate task managing | |
12 | the data going into the compressors. | |
13 | ||
14 | ##################################################################### */ | |
15 | /*}}}*/ | |
16 | // Include Files /*{{{*/ | |
17 | #include <config.h> | |
18 | ||
19 | #include <apt-pkg/fileutl.h> | |
20 | #include <apt-pkg/strutl.h> | |
21 | #include <apt-pkg/error.h> | |
22 | #include <apt-pkg/md5.h> | |
23 | #include <apt-pkg/aptconfiguration.h> | |
24 | #include <apt-pkg/hashsum_template.h> | |
25 | ||
26 | #include <ctype.h> | |
27 | #include <vector> | |
28 | #include <sys/types.h> | |
29 | #include <sys/stat.h> | |
30 | #include <sys/time.h> | |
31 | #include <unistd.h> | |
32 | ||
33 | #include "multicompress.h" | |
34 | #include <apti18n.h> | |
35 | /*}}}*/ | |
36 | ||
37 | using namespace std; | |
38 | ||
39 | ||
40 | // MultiCompress::MultiCompress - Constructor /*{{{*/ | |
41 | // --------------------------------------------------------------------- | |
42 | /* Setup the file outputs, compression modes and fork the writer child */ | |
43 | MultiCompress::MultiCompress(string const &Output,string const &Compress, | |
44 | mode_t const &Permissions,bool const &Write) : | |
45 | Permissions(Permissions) | |
46 | { | |
47 | Outputs = 0; | |
48 | Outputter = -1; | |
49 | UpdateMTime = 0; | |
50 | ||
51 | /* Parse the compression string, a space separated lists of compresison | |
52 | types */ | |
53 | string::const_iterator I = Compress.begin(); | |
54 | for (; I != Compress.end();) | |
55 | { | |
56 | for (; I != Compress.end() && isspace(*I); ++I); | |
57 | ||
58 | // Grab a word | |
59 | string::const_iterator Start = I; | |
60 | for (; I != Compress.end() && !isspace(*I); ++I); | |
61 | ||
62 | // Find the matching compressor | |
63 | std::vector<APT::Configuration::Compressor> Compressors = APT::Configuration::getCompressors(); | |
64 | std::vector<APT::Configuration::Compressor>::const_iterator Comp = Compressors.begin(); | |
65 | for (; Comp != Compressors.end(); ++Comp) | |
66 | if (stringcmp(Start,I,Comp->Name.c_str()) == 0) | |
67 | break; | |
68 | ||
69 | // Hmm.. unknown. | |
70 | if (Comp == Compressors.end()) | |
71 | { | |
72 | _error->Warning(_("Unknown compression algorithm '%s'"),string(Start,I).c_str()); | |
73 | continue; | |
74 | } | |
75 | ||
76 | // Create and link in a new output | |
77 | Files *NewOut = new Files; | |
78 | NewOut->Next = Outputs; | |
79 | Outputs = NewOut; | |
80 | NewOut->CompressProg = *Comp; | |
81 | NewOut->Output = Output+Comp->Extension; | |
82 | ||
83 | struct stat St; | |
84 | if (stat(NewOut->Output.c_str(),&St) == 0) | |
85 | NewOut->OldMTime = St.st_mtime; | |
86 | else | |
87 | NewOut->OldMTime = 0; | |
88 | } | |
89 | ||
90 | if (Write == false) | |
91 | return; | |
92 | ||
93 | /* Open all the temp files now so we can report any errors. File is | |
94 | made unreable to prevent people from touching it during creating. */ | |
95 | for (Files *I = Outputs; I != 0; I = I->Next) | |
96 | I->TmpFile.Open(I->Output + ".new", FileFd::WriteOnly | FileFd::Create | FileFd::Empty, FileFd::Extension, 0600); | |
97 | if (_error->PendingError() == true) | |
98 | return; | |
99 | ||
100 | if (Outputs == 0) | |
101 | { | |
102 | _error->Error(_("Compressed output %s needs a compression set"),Output.c_str()); | |
103 | return; | |
104 | } | |
105 | ||
106 | Start(); | |
107 | } | |
108 | /*}}}*/ | |
109 | // MultiCompress::~MultiCompress - Destructor /*{{{*/ | |
110 | // --------------------------------------------------------------------- | |
111 | /* Just erase the file linked list. */ | |
112 | MultiCompress::~MultiCompress() | |
113 | { | |
114 | Die(); | |
115 | ||
116 | for (; Outputs != 0;) | |
117 | { | |
118 | Files *Tmp = Outputs->Next; | |
119 | delete Outputs; | |
120 | Outputs = Tmp; | |
121 | } | |
122 | } | |
123 | /*}}}*/ | |
124 | // MultiCompress::GetStat - Get stat information for compressed files /*{{{*/ | |
125 | // --------------------------------------------------------------------- | |
126 | /* This checks each compressed file to make sure it exists and returns | |
127 | stat information for a random file from the collection. False means | |
128 | one or more of the files is missing. */ | |
129 | bool MultiCompress::GetStat(string const &Output,string const &Compress,struct stat &St) | |
130 | { | |
131 | /* Parse the compression string, a space separated lists of compresison | |
132 | types */ | |
133 | string::const_iterator I = Compress.begin(); | |
134 | bool DidStat = false; | |
135 | for (; I != Compress.end();) | |
136 | { | |
137 | for (; I != Compress.end() && isspace(*I); ++I); | |
138 | ||
139 | // Grab a word | |
140 | string::const_iterator Start = I; | |
141 | for (; I != Compress.end() && !isspace(*I); ++I); | |
142 | ||
143 | // Find the matching compressor | |
144 | std::vector<APT::Configuration::Compressor> Compressors = APT::Configuration::getCompressors(); | |
145 | std::vector<APT::Configuration::Compressor>::const_iterator Comp = Compressors.begin(); | |
146 | for (; Comp != Compressors.end(); ++Comp) | |
147 | if (stringcmp(Start,I,Comp->Name.c_str()) == 0) | |
148 | break; | |
149 | ||
150 | // Hmm.. unknown. | |
151 | if (Comp == Compressors.end()) | |
152 | continue; | |
153 | ||
154 | string Name = Output+Comp->Extension; | |
155 | if (stat(Name.c_str(),&St) != 0) | |
156 | return false; | |
157 | DidStat = true; | |
158 | } | |
159 | return DidStat; | |
160 | } | |
161 | /*}}}*/ | |
162 | // MultiCompress::Start - Start up the writer child /*{{{*/ | |
163 | // --------------------------------------------------------------------- | |
164 | /* Fork a child and setup the communication pipe. */ | |
165 | bool MultiCompress::Start() | |
166 | { | |
167 | // Create a data pipe | |
168 | int Pipe[2] = {-1,-1}; | |
169 | if (pipe(Pipe) != 0) | |
170 | return _error->Errno("pipe",_("Failed to create IPC pipe to subprocess")); | |
171 | for (int I = 0; I != 2; I++) | |
172 | SetCloseExec(Pipe[I],true); | |
173 | ||
174 | // The child.. | |
175 | Outputter = fork(); | |
176 | if (Outputter == 0) | |
177 | { | |
178 | close(Pipe[1]); | |
179 | Child(Pipe[0]); | |
180 | if (_error->PendingError() == true) | |
181 | { | |
182 | _error->DumpErrors(); | |
183 | _exit(100); | |
184 | } | |
185 | _exit(0); | |
186 | }; | |
187 | ||
188 | close(Pipe[0]); | |
189 | if (Input.OpenDescriptor(Pipe[1], FileFd::WriteOnly, true) == false) | |
190 | return false; | |
191 | ||
192 | if (Outputter == -1) | |
193 | return _error->Errno("fork",_("Failed to fork")); | |
194 | return true; | |
195 | } | |
196 | /*}}}*/ | |
197 | // MultiCompress::Die - Clean up the writer /*{{{*/ | |
198 | // --------------------------------------------------------------------- | |
199 | /* */ | |
200 | bool MultiCompress::Die() | |
201 | { | |
202 | if (Input.IsOpen() == false) | |
203 | return true; | |
204 | ||
205 | Input.Close(); | |
206 | bool Res = ExecWait(Outputter,_("Compress child"),false); | |
207 | Outputter = -1; | |
208 | return Res; | |
209 | } | |
210 | /*}}}*/ | |
211 | // MultiCompress::Finalize - Finish up writing /*{{{*/ | |
212 | // --------------------------------------------------------------------- | |
213 | /* This is only necessary for statistics reporting. */ | |
214 | bool MultiCompress::Finalize(unsigned long long &OutSize) | |
215 | { | |
216 | OutSize = 0; | |
217 | if (Input.IsOpen() == false || Die() == false) | |
218 | return false; | |
219 | ||
220 | time_t Now; | |
221 | time(&Now); | |
222 | ||
223 | // Check the mtimes to see if the files were replaced. | |
224 | bool Changed = false; | |
225 | for (Files *I = Outputs; I != 0; I = I->Next) | |
226 | { | |
227 | struct stat St; | |
228 | if (stat(I->Output.c_str(),&St) != 0) | |
229 | return _error->Error(_("Internal error, failed to create %s"), | |
230 | I->Output.c_str()); | |
231 | ||
232 | if (I->OldMTime != St.st_mtime) | |
233 | Changed = true; | |
234 | else | |
235 | { | |
236 | // Update the mtime if necessary | |
237 | if (UpdateMTime > 0 && | |
238 | (Now - St.st_mtime > (signed)UpdateMTime || St.st_mtime > Now)) | |
239 | { | |
240 | utimes(I->Output.c_str(), NULL); | |
241 | Changed = true; | |
242 | } | |
243 | } | |
244 | ||
245 | // Force the file permissions | |
246 | if (St.st_mode != Permissions) | |
247 | chmod(I->Output.c_str(),Permissions); | |
248 | ||
249 | OutSize += St.st_size; | |
250 | } | |
251 | ||
252 | if (Changed == false) | |
253 | OutSize = 0; | |
254 | ||
255 | return true; | |
256 | } | |
257 | /*}}}*/ | |
258 | // MultiCompress::OpenOld - Open an old file /*{{{*/ | |
259 | // --------------------------------------------------------------------- | |
260 | /* This opens one of the original output files, possibly decompressing it. */ | |
261 | bool MultiCompress::OpenOld(FileFd &Fd) | |
262 | { | |
263 | Files *Best = Outputs; | |
264 | for (Files *I = Outputs; I != 0; I = I->Next) | |
265 | if (Best->CompressProg.Cost > I->CompressProg.Cost) | |
266 | Best = I; | |
267 | ||
268 | // Open the file | |
269 | return Fd.Open(Best->Output, FileFd::ReadOnly, FileFd::Extension); | |
270 | } | |
271 | /*}}}*/ | |
272 | // MultiCompress::Child - The writer child /*{{{*/ | |
273 | // --------------------------------------------------------------------- | |
274 | /* The child process forks a bunch of compression children and takes | |
275 | input on FD and passes it to all the compressor child. On the way it | |
276 | computes the MD5 of the raw data. After this the raw data in the | |
277 | original files is compared to see if this data is new. If the data | |
278 | is new then the temp files are renamed, otherwise they are erased. */ | |
279 | bool MultiCompress::Child(int const &FD) | |
280 | { | |
281 | /* Okay, now we just feed data from FD to all the other FDs. Also | |
282 | stash a hash of the data to use later. */ | |
283 | SetNonBlock(FD,false); | |
284 | unsigned char Buffer[32*1024]; | |
285 | unsigned long long FileSize = 0; | |
286 | MD5Summation MD5; | |
287 | while (1) | |
288 | { | |
289 | WaitFd(FD,false); | |
290 | int Res = read(FD,Buffer,sizeof(Buffer)); | |
291 | if (Res == 0) | |
292 | break; | |
293 | if (Res < 0) | |
294 | continue; | |
295 | ||
296 | MD5.Add(Buffer,Res); | |
297 | FileSize += Res; | |
298 | for (Files *I = Outputs; I != 0; I = I->Next) | |
299 | { | |
300 | if (I->TmpFile.Write(Buffer, Res) == false) | |
301 | { | |
302 | _error->Errno("write",_("IO to subprocess/file failed")); | |
303 | break; | |
304 | } | |
305 | } | |
306 | } | |
307 | ||
308 | if (_error->PendingError() == true) | |
309 | return false; | |
310 | ||
311 | /* Now we have to copy the files over, or erase them if they | |
312 | have not changed. First find the cheapest decompressor */ | |
313 | bool Missing = false; | |
314 | for (Files *I = Outputs; I != 0; I = I->Next) | |
315 | { | |
316 | if (I->OldMTime == 0) | |
317 | { | |
318 | Missing = true; | |
319 | break; | |
320 | } | |
321 | } | |
322 | ||
323 | // Check the MD5 of the lowest cost entity. | |
324 | while (Missing == false) | |
325 | { | |
326 | FileFd CompFd; | |
327 | if (OpenOld(CompFd) == false) | |
328 | { | |
329 | _error->Discard(); | |
330 | break; | |
331 | } | |
332 | ||
333 | // Compute the hash | |
334 | MD5Summation OldMD5; | |
335 | unsigned long long NewFileSize = 0; | |
336 | while (1) | |
337 | { | |
338 | unsigned long long Res = 0; | |
339 | if (CompFd.Read(Buffer,sizeof(Buffer), &Res) == false) | |
340 | return _error->Errno("read",_("Failed to read while computing MD5")); | |
341 | if (Res == 0) | |
342 | break; | |
343 | NewFileSize += Res; | |
344 | OldMD5.Add(Buffer,Res); | |
345 | } | |
346 | CompFd.Close(); | |
347 | ||
348 | // Check the hash | |
349 | if (OldMD5.Result() == MD5.Result() && | |
350 | FileSize == NewFileSize) | |
351 | { | |
352 | for (Files *I = Outputs; I != 0; I = I->Next) | |
353 | { | |
354 | I->TmpFile.Close(); | |
355 | RemoveFile("MultiCompress::Child", I->TmpFile.Name()); | |
356 | } | |
357 | return !_error->PendingError(); | |
358 | } | |
359 | break; | |
360 | } | |
361 | ||
362 | // Finalize | |
363 | for (Files *I = Outputs; I != 0; I = I->Next) | |
364 | { | |
365 | // Set the correct file modes | |
366 | chmod(I->TmpFile.Name().c_str(),Permissions); | |
367 | ||
368 | if (rename(I->TmpFile.Name().c_str(),I->Output.c_str()) != 0) | |
369 | _error->Errno("rename",_("Failed to rename %s to %s"), | |
370 | I->TmpFile.Name().c_str(),I->Output.c_str()); | |
371 | I->TmpFile.Close(); | |
372 | } | |
373 | ||
374 | return !_error->PendingError(); | |
375 | } | |
376 | /*}}}*/ | |
377 |