xref: /petsc/src/sys/fileio/fretrieve.c (revision d6acfc2dd824932841bfdf20245c037310c77dc9)
1e5c89e4eSSatish Balay /*
2e5c89e4eSSatish Balay       Code for opening and closing files.
3e5c89e4eSSatish Balay */
4c6db04a5SJed Brown #include <petscsys.h>
5e5c89e4eSSatish Balay #if defined(PETSC_HAVE_PWD_H)
6e5c89e4eSSatish Balay   #include <pwd.h>
7e5c89e4eSSatish Balay #endif
8e5c89e4eSSatish Balay #include <ctype.h>
9e5c89e4eSSatish Balay #include <sys/stat.h>
10e5c89e4eSSatish Balay #if defined(PETSC_HAVE_UNISTD_H)
11e5c89e4eSSatish Balay   #include <unistd.h>
12e5c89e4eSSatish Balay #endif
13e5c89e4eSSatish Balay #if defined(PETSC_HAVE_SYS_UTSNAME_H)
14e5c89e4eSSatish Balay   #include <sys/utsname.h>
15e5c89e4eSSatish Balay #endif
16e5c89e4eSSatish Balay #include <fcntl.h>
17e5c89e4eSSatish Balay #include <time.h>
18e5c89e4eSSatish Balay #if defined(PETSC_HAVE_SYS_SYSTEMINFO_H)
19e5c89e4eSSatish Balay   #include <sys/systeminfo.h>
20e5c89e4eSSatish Balay #endif
215ea2b939SDuncan Campbell #include <petsc/private/petscimpl.h>
22e5c89e4eSSatish Balay 
23480cf27aSJed Brown /*
24480cf27aSJed Brown    Private routine to delete tmp/shared storage
25480cf27aSJed Brown 
2621532e8aSBarry Smith    This is called by MPI, not by users, when communicator attributes are deleted
27480cf27aSJed Brown 
28480cf27aSJed Brown */
29*d6acfc2dSPierre Jolivet static PetscMPIInt MPIAPI Petsc_DelTmpShared(MPI_Comm comm, PetscMPIInt keyval, void *count_val, void *extra_state)
30d71ae5a4SJacob Faibussowitsch {
31480cf27aSJed Brown   PetscFunctionBegin;
327c5b2466SBarry Smith   PetscCallReturnMPI(PetscInfo(NULL, "Deleting tmp/shared data in an MPI_Comm %ld\n", (long)comm));
337c5b2466SBarry Smith   PetscCallReturnMPI(PetscFree(count_val));
34480cf27aSJed Brown   PetscFunctionReturn(MPI_SUCCESS);
35480cf27aSJed Brown }
36e5c89e4eSSatish Balay 
3710450e9eSJacob Faibussowitsch // "Unknown section 'Environmental Variables'"
3810450e9eSJacob Faibussowitsch // PetscClangLinter pragma disable: -fdoc-section-header-unknown
39e5c89e4eSSatish Balay /*@C
4021532e8aSBarry Smith   PetscGetTmp - Gets the name of the "tmp" directory, often this is `/tmp`
41e5c89e4eSSatish Balay 
42d083f849SBarry Smith   Collective
43e5c89e4eSSatish Balay 
44e5c89e4eSSatish Balay   Input Parameters:
4521532e8aSBarry Smith + comm - MPI_Communicator that may share tmp
46e5c89e4eSSatish Balay - len  - length of string to hold name
47e5c89e4eSSatish Balay 
48f899ff85SJose E. Roman   Output Parameter:
49e5c89e4eSSatish Balay . dir - directory name
50e5c89e4eSSatish Balay 
51e5c89e4eSSatish Balay   Options Database Keys:
5221532e8aSBarry Smith + -shared_tmp     - indicates the directory is known to be shared among the MPI processes
5321532e8aSBarry Smith . -not_shared_tmp - indicates the directory is known to be not shared among the MPI processes
5421532e8aSBarry Smith - -tmp tmpdir     - name of the directory you wish to use as tmp
55e5c89e4eSSatish Balay 
56e5c89e4eSSatish Balay   Environmental Variables:
5721532e8aSBarry Smith + `PETSC_SHARED_TMP`     - indicates the directory is known to be shared among the MPI processes
5821532e8aSBarry Smith . `PETSC_NOT_SHARED_TMP` - indicates the directory is known to be not shared among the MPI processes
5921532e8aSBarry Smith - `PETSC_TMP`            - name of the directory you wish to use as tmp
60e5c89e4eSSatish Balay 
61e5c89e4eSSatish Balay   Level: developer
62e5c89e4eSSatish Balay 
63811af0c4SBarry Smith .seealso: `PetscSharedTmp()`, `PetscSharedWorkingDirectory()`, `PetscGetWorkingDirectory()`, `PetscGetHomeDirectory()`
64e5c89e4eSSatish Balay @*/
65d71ae5a4SJacob Faibussowitsch PetscErrorCode PetscGetTmp(MPI_Comm comm, char dir[], size_t len)
66d71ae5a4SJacob Faibussowitsch {
67ace3abfcSBarry Smith   PetscBool flg;
68e5c89e4eSSatish Balay 
69e5c89e4eSSatish Balay   PetscFunctionBegin;
709566063dSJacob Faibussowitsch   PetscCall(PetscOptionsGetenv(comm, "PETSC_TMP", dir, len, &flg));
7148a46eb9SPierre Jolivet   if (!flg) PetscCall(PetscStrncpy(dir, "/tmp", len));
723ba16761SJacob Faibussowitsch   PetscFunctionReturn(PETSC_SUCCESS);
73e5c89e4eSSatish Balay }
74e5c89e4eSSatish Balay 
7510450e9eSJacob Faibussowitsch // "Unknown section 'Environmental Variables'"
7610450e9eSJacob Faibussowitsch // PetscClangLinter pragma disable: -fdoc-section-header-unknown
77cc4c1da9SBarry Smith /*@
78e5c89e4eSSatish Balay   PetscSharedTmp - Determines if all processors in a communicator share a
7921532e8aSBarry Smith   tmp directory or have different ones.
80e5c89e4eSSatish Balay 
81d083f849SBarry Smith   Collective
82e5c89e4eSSatish Balay 
83811af0c4SBarry Smith   Input Parameter:
8421532e8aSBarry Smith . comm - MPI_Communicator that may share tmp
85e5c89e4eSSatish Balay 
86811af0c4SBarry Smith   Output Parameter:
87811af0c4SBarry Smith . shared - `PETSC_TRUE` or `PETSC_FALSE`
88e5c89e4eSSatish Balay 
89e5c89e4eSSatish Balay   Options Database Keys:
9021532e8aSBarry Smith + -shared_tmp     - indicates the directory is known to be shared among the MPI processes
9121532e8aSBarry Smith . -not_shared_tmp - indicates the directory is known to be not shared among the MPI processes
9221532e8aSBarry Smith - -tmp tmpdir     - name of the directory you wish to use as tmp
93e5c89e4eSSatish Balay 
94e5c89e4eSSatish Balay   Environmental Variables:
9521532e8aSBarry Smith + `PETSC_SHARED_TMP`     - indicates the directory is known to be shared among the MPI processes
9621532e8aSBarry Smith . `PETSC_NOT_SHARED_TMP` - indicates the directory is known to be not shared among the MPI processes
9721532e8aSBarry Smith - `PETSC_TMP`            - name of the directory you wish to use as tmp
98e5c89e4eSSatish Balay 
99e5c89e4eSSatish Balay   Level: developer
100e5c89e4eSSatish Balay 
101e5c89e4eSSatish Balay   Notes:
102e5c89e4eSSatish Balay   Stores the status as a MPI attribute so it does not have
103e5c89e4eSSatish Balay   to be redetermined each time.
104e5c89e4eSSatish Balay 
105e5c89e4eSSatish Balay   Assumes that all processors in a communicator either
10621532e8aSBarry Smith   1) have a common tmp or
10721532e8aSBarry Smith   2) each has a separate tmp
108e5c89e4eSSatish Balay   eventually we can write a fancier one that determines which processors
10921532e8aSBarry Smith   share a common tmp.
110e5c89e4eSSatish Balay 
111e5c89e4eSSatish Balay   This will be very slow on runs with a large number of processors since
112e5c89e4eSSatish Balay   it requires O(p*p) file opens.
113e5c89e4eSSatish Balay 
114bfbbc7b7SBarry Smith   If the environmental variable `PETSC_TMP` is set it will use this directory
11521532e8aSBarry Smith   as the "tmp" directory.
116e5c89e4eSSatish Balay 
117811af0c4SBarry Smith .seealso: `PetscGetTmp()`, `PetscSharedWorkingDirectory()`, `PetscGetWorkingDirectory()`, `PetscGetHomeDirectory()`
118e5c89e4eSSatish Balay @*/
119d71ae5a4SJacob Faibussowitsch PetscErrorCode PetscSharedTmp(MPI_Comm comm, PetscBool *shared)
120d71ae5a4SJacob Faibussowitsch {
121e5c89e4eSSatish Balay   PetscMPIInt size, rank, *tagvalp, sum, cnt, i;
122ace3abfcSBarry Smith   PetscBool   flg, iflg;
123e5c89e4eSSatish Balay   FILE       *fd;
124ed9cf6e9SBarry Smith   int         err;
125e5c89e4eSSatish Balay 
126e5c89e4eSSatish Balay   PetscFunctionBegin;
1279566063dSJacob Faibussowitsch   PetscCallMPI(MPI_Comm_size(comm, &size));
128e5c89e4eSSatish Balay   if (size == 1) {
129e5c89e4eSSatish Balay     *shared = PETSC_TRUE;
1303ba16761SJacob Faibussowitsch     PetscFunctionReturn(PETSC_SUCCESS);
131e5c89e4eSSatish Balay   }
132e5c89e4eSSatish Balay 
1339566063dSJacob Faibussowitsch   PetscCall(PetscOptionsGetenv(comm, "PETSC_SHARED_TMP", NULL, 0, &flg));
134e5c89e4eSSatish Balay   if (flg) {
135e5c89e4eSSatish Balay     *shared = PETSC_TRUE;
1363ba16761SJacob Faibussowitsch     PetscFunctionReturn(PETSC_SUCCESS);
137e5c89e4eSSatish Balay   }
138e5c89e4eSSatish Balay 
1399566063dSJacob Faibussowitsch   PetscCall(PetscOptionsGetenv(comm, "PETSC_NOT_SHARED_TMP", NULL, 0, &flg));
140e5c89e4eSSatish Balay   if (flg) {
141e5c89e4eSSatish Balay     *shared = PETSC_FALSE;
1423ba16761SJacob Faibussowitsch     PetscFunctionReturn(PETSC_SUCCESS);
143e5c89e4eSSatish Balay   }
144e5c89e4eSSatish Balay 
1455ea2b939SDuncan Campbell   if (Petsc_SharedTmp_keyval == MPI_KEYVAL_INVALID) PetscCallMPI(MPI_Comm_create_keyval(MPI_COMM_NULL_COPY_FN, Petsc_DelTmpShared, &Petsc_SharedTmp_keyval, NULL));
146e5c89e4eSSatish Balay 
1475ea2b939SDuncan Campbell   PetscCallMPI(MPI_Comm_get_attr(comm, Petsc_SharedTmp_keyval, (void **)&tagvalp, (int *)&iflg));
148e5c89e4eSSatish Balay   if (!iflg) {
149e5c89e4eSSatish Balay     char filename[PETSC_MAX_PATH_LEN], tmpname[PETSC_MAX_PATH_LEN];
150e5c89e4eSSatish Balay 
151e5c89e4eSSatish Balay     /* This communicator does not yet have a shared tmp attribute */
1529566063dSJacob Faibussowitsch     PetscCall(PetscMalloc1(1, &tagvalp));
1535ea2b939SDuncan Campbell     PetscCallMPI(MPI_Comm_set_attr(comm, Petsc_SharedTmp_keyval, tagvalp));
154e5c89e4eSSatish Balay 
1559566063dSJacob Faibussowitsch     PetscCall(PetscOptionsGetenv(comm, "PETSC_TMP", tmpname, 238, &iflg));
156e5c89e4eSSatish Balay     if (!iflg) {
157c6a7a370SJeremy L Thompson       PetscCall(PetscStrncpy(filename, "/tmp", sizeof(filename)));
158e5c89e4eSSatish Balay     } else {
159c6a7a370SJeremy L Thompson       PetscCall(PetscStrncpy(filename, tmpname, sizeof(filename)));
160e5c89e4eSSatish Balay     }
161e5c89e4eSSatish Balay 
162c6a7a370SJeremy L Thompson     PetscCall(PetscStrlcat(filename, "/petsctestshared", sizeof(filename)));
1639566063dSJacob Faibussowitsch     PetscCallMPI(MPI_Comm_rank(comm, &rank));
164e5c89e4eSSatish Balay 
165e5c89e4eSSatish Balay     /* each processor creates a /tmp file and all the later ones check */
166e5c89e4eSSatish Balay     /* this makes sure no subset of processors is shared */
167e5c89e4eSSatish Balay     *shared = PETSC_FALSE;
168e5c89e4eSSatish Balay     for (i = 0; i < size - 1; i++) {
169e5c89e4eSSatish Balay       if (rank == i) {
170e5c89e4eSSatish Balay         fd = fopen(filename, "w");
17128b400f6SJacob Faibussowitsch         PetscCheck(fd, PETSC_COMM_SELF, PETSC_ERR_FILE_OPEN, "Unable to open test file %s", filename);
172ed9cf6e9SBarry Smith         err = fclose(fd);
17328b400f6SJacob Faibussowitsch         PetscCheck(!err, PETSC_COMM_SELF, PETSC_ERR_SYS, "fclose() failed on file");
174e5c89e4eSSatish Balay       }
1759566063dSJacob Faibussowitsch       PetscCallMPI(MPI_Barrier(comm));
176e5c89e4eSSatish Balay       if (rank >= i) {
177e5c89e4eSSatish Balay         fd = fopen(filename, "r");
178a297a907SKarl Rupp         if (fd) cnt = 1;
179a297a907SKarl Rupp         else cnt = 0;
180e5c89e4eSSatish Balay         if (fd) {
181ed9cf6e9SBarry Smith           err = fclose(fd);
18228b400f6SJacob Faibussowitsch           PetscCheck(!err, PETSC_COMM_SELF, PETSC_ERR_SYS, "fclose() failed on file");
183e5c89e4eSSatish Balay         }
184a297a907SKarl Rupp       } else cnt = 0;
185a297a907SKarl Rupp 
186462c564dSBarry Smith       PetscCallMPI(MPIU_Allreduce(&cnt, &sum, 1, MPI_INT, MPI_SUM, comm));
187a297a907SKarl Rupp       if (rank == i) unlink(filename);
188e5c89e4eSSatish Balay 
189e5c89e4eSSatish Balay       if (sum == size) {
190e5c89e4eSSatish Balay         *shared = PETSC_TRUE;
191e5c89e4eSSatish Balay         break;
19208401ef6SPierre Jolivet       } else PetscCheck(sum == 1, PETSC_COMM_SELF, PETSC_ERR_SUP_SYS, "Subset of processes share /tmp ");
193e5c89e4eSSatish Balay     }
194e5c89e4eSSatish Balay     *tagvalp = (int)*shared;
19557508eceSPierre Jolivet     PetscCall(PetscInfo(NULL, "processors %s %s\n", *shared ? "share" : "do NOT share", iflg ? tmpname : "/tmp"));
196a297a907SKarl Rupp   } else *shared = (PetscBool)*tagvalp;
1973ba16761SJacob Faibussowitsch   PetscFunctionReturn(PETSC_SUCCESS);
198e5c89e4eSSatish Balay }
199e5c89e4eSSatish Balay 
20010450e9eSJacob Faibussowitsch // "Unknown section 'Environmental Variables'"
20110450e9eSJacob Faibussowitsch // PetscClangLinter pragma disable: -fdoc-section-header-unknown
202cc4c1da9SBarry Smith /*@
203f1a722f8SMatthew G. Knepley   PetscSharedWorkingDirectory - Determines if all processors in a communicator share a working directory or have different ones.
204e5c89e4eSSatish Balay 
205d083f849SBarry Smith   Collective
206e5c89e4eSSatish Balay 
2076b867d5aSJose E. Roman   Input Parameter:
208e5c89e4eSSatish Balay . comm - MPI_Communicator that may share working directory
209e5c89e4eSSatish Balay 
2106b867d5aSJose E. Roman   Output Parameter:
211811af0c4SBarry Smith . shared - `PETSC_TRUE` or `PETSC_FALSE`
212e5c89e4eSSatish Balay 
213e5c89e4eSSatish Balay   Options Database Keys:
21421532e8aSBarry Smith + -shared_working_directory     - indicates the directory is known to be shared among the MPI processes
21521532e8aSBarry Smith - -not_shared_working_directory - indicates the directory is known to be not shared among the MPI processes
216e5c89e4eSSatish Balay 
217e5c89e4eSSatish Balay   Environmental Variables:
21821532e8aSBarry Smith + `PETSC_SHARED_WORKING_DIRECTORY`     - indicates the directory is known to be shared among the MPI processes
21921532e8aSBarry Smith - `PETSC_NOT_SHARED_WORKING_DIRECTORY` - indicates the directory is known to be not shared among the MPI processes
220e5c89e4eSSatish Balay 
221e5c89e4eSSatish Balay   Level: developer
222e5c89e4eSSatish Balay 
223e5c89e4eSSatish Balay   Notes:
224f1a722f8SMatthew G. Knepley   Stores the status as a MPI attribute so it does not have to be redetermined each time.
225e5c89e4eSSatish Balay 
226e5c89e4eSSatish Balay   Assumes that all processors in a communicator either
227bfbbc7b7SBarry Smith .vb
228bfbbc7b7SBarry Smith    1) have a common working directory or
229bfbbc7b7SBarry Smith    2) each has a separate working directory
230bfbbc7b7SBarry Smith .ve
231f1a722f8SMatthew G. Knepley   eventually we can write a fancier one that determines which processors share a common working directory.
232e5c89e4eSSatish Balay 
233f1a722f8SMatthew G. Knepley   This will be very slow on runs with a large number of processors since it requires O(p*p) file opens.
234811af0c4SBarry Smith 
235811af0c4SBarry Smith .seealso: `PetscGetTmp()`, `PetscSharedTmp()`, `PetscGetWorkingDirectory()`, `PetscGetHomeDirectory()`
236e5c89e4eSSatish Balay @*/
237d71ae5a4SJacob Faibussowitsch PetscErrorCode PetscSharedWorkingDirectory(MPI_Comm comm, PetscBool *shared)
238d71ae5a4SJacob Faibussowitsch {
239e5c89e4eSSatish Balay   PetscMPIInt size, rank, *tagvalp, sum, cnt, i;
240ace3abfcSBarry Smith   PetscBool   flg, iflg;
241e5c89e4eSSatish Balay   FILE       *fd;
242ed9cf6e9SBarry Smith   int         err;
243e5c89e4eSSatish Balay 
244e5c89e4eSSatish Balay   PetscFunctionBegin;
2459566063dSJacob Faibussowitsch   PetscCallMPI(MPI_Comm_size(comm, &size));
246e5c89e4eSSatish Balay   if (size == 1) {
247e5c89e4eSSatish Balay     *shared = PETSC_TRUE;
2483ba16761SJacob Faibussowitsch     PetscFunctionReturn(PETSC_SUCCESS);
249e5c89e4eSSatish Balay   }
250e5c89e4eSSatish Balay 
2519566063dSJacob Faibussowitsch   PetscCall(PetscOptionsGetenv(comm, "PETSC_SHARED_WORKING_DIRECTORY", NULL, 0, &flg));
252e5c89e4eSSatish Balay   if (flg) {
253e5c89e4eSSatish Balay     *shared = PETSC_TRUE;
2543ba16761SJacob Faibussowitsch     PetscFunctionReturn(PETSC_SUCCESS);
255e5c89e4eSSatish Balay   }
256e5c89e4eSSatish Balay 
2579566063dSJacob Faibussowitsch   PetscCall(PetscOptionsGetenv(comm, "PETSC_NOT_SHARED_WORKING_DIRECTORY", NULL, 0, &flg));
258e5c89e4eSSatish Balay   if (flg) {
259e5c89e4eSSatish Balay     *shared = PETSC_FALSE;
2603ba16761SJacob Faibussowitsch     PetscFunctionReturn(PETSC_SUCCESS);
261e5c89e4eSSatish Balay   }
262e5c89e4eSSatish Balay 
2635ea2b939SDuncan Campbell   if (Petsc_SharedWD_keyval == MPI_KEYVAL_INVALID) PetscCallMPI(MPI_Comm_create_keyval(MPI_COMM_NULL_COPY_FN, Petsc_DelTmpShared, &Petsc_SharedWD_keyval, NULL));
264e5c89e4eSSatish Balay 
2655ea2b939SDuncan Campbell   PetscCallMPI(MPI_Comm_get_attr(comm, Petsc_SharedWD_keyval, (void **)&tagvalp, (int *)&iflg));
266e5c89e4eSSatish Balay   if (!iflg) {
267e5c89e4eSSatish Balay     char filename[PETSC_MAX_PATH_LEN];
268e5c89e4eSSatish Balay 
269e5c89e4eSSatish Balay     /* This communicator does not yet have a shared  attribute */
2709566063dSJacob Faibussowitsch     PetscCall(PetscMalloc1(1, &tagvalp));
2715ea2b939SDuncan Campbell     PetscCallMPI(MPI_Comm_set_attr(comm, Petsc_SharedWD_keyval, tagvalp));
272e5c89e4eSSatish Balay 
273bf31d2d3SBarry Smith     PetscCall(PetscGetWorkingDirectory(filename, sizeof(filename) - 16));
274c6a7a370SJeremy L Thompson     PetscCall(PetscStrlcat(filename, "/petsctestshared", sizeof(filename)));
2759566063dSJacob Faibussowitsch     PetscCallMPI(MPI_Comm_rank(comm, &rank));
276e5c89e4eSSatish Balay 
277e5c89e4eSSatish Balay     /* each processor creates a  file and all the later ones check */
278e5c89e4eSSatish Balay     /* this makes sure no subset of processors is shared */
279e5c89e4eSSatish Balay     *shared = PETSC_FALSE;
280e5c89e4eSSatish Balay     for (i = 0; i < size - 1; i++) {
281e5c89e4eSSatish Balay       if (rank == i) {
282e5c89e4eSSatish Balay         fd = fopen(filename, "w");
28328b400f6SJacob Faibussowitsch         PetscCheck(fd, PETSC_COMM_SELF, PETSC_ERR_FILE_OPEN, "Unable to open test file %s", filename);
284ed9cf6e9SBarry Smith         err = fclose(fd);
28528b400f6SJacob Faibussowitsch         PetscCheck(!err, PETSC_COMM_SELF, PETSC_ERR_SYS, "fclose() failed on file");
286e5c89e4eSSatish Balay       }
2879566063dSJacob Faibussowitsch       PetscCallMPI(MPI_Barrier(comm));
288e5c89e4eSSatish Balay       if (rank >= i) {
289e5c89e4eSSatish Balay         fd = fopen(filename, "r");
290a297a907SKarl Rupp         if (fd) cnt = 1;
291a297a907SKarl Rupp         else cnt = 0;
292e5c89e4eSSatish Balay         if (fd) {
293ed9cf6e9SBarry Smith           err = fclose(fd);
29428b400f6SJacob Faibussowitsch           PetscCheck(!err, PETSC_COMM_SELF, PETSC_ERR_SYS, "fclose() failed on file");
295e5c89e4eSSatish Balay         }
296a297a907SKarl Rupp       } else cnt = 0;
297a297a907SKarl Rupp 
298462c564dSBarry Smith       PetscCallMPI(MPIU_Allreduce(&cnt, &sum, 1, MPI_INT, MPI_SUM, comm));
299a297a907SKarl Rupp       if (rank == i) unlink(filename);
300e5c89e4eSSatish Balay 
301e5c89e4eSSatish Balay       if (sum == size) {
302e5c89e4eSSatish Balay         *shared = PETSC_TRUE;
303e5c89e4eSSatish Balay         break;
30408401ef6SPierre Jolivet       } else PetscCheck(sum == 1, PETSC_COMM_SELF, PETSC_ERR_SUP_SYS, "Subset of processes share working directory");
305e5c89e4eSSatish Balay     }
306e5c89e4eSSatish Balay     *tagvalp = (int)*shared;
307a297a907SKarl Rupp   } else *shared = (PetscBool)*tagvalp;
3089566063dSJacob Faibussowitsch   PetscCall(PetscInfo(NULL, "processors %s working directory\n", (*shared) ? "shared" : "do NOT share"));
3093ba16761SJacob Faibussowitsch   PetscFunctionReturn(PETSC_SUCCESS);
310e5c89e4eSSatish Balay }
311e5c89e4eSSatish Balay 
312e5c89e4eSSatish Balay /*@C
313bfbbc7b7SBarry Smith   PetscFileRetrieve - Obtains a file from a URL or a compressed file
314e5c89e4eSSatish Balay   and copies into local disk space as uncompressed.
315e5c89e4eSSatish Balay 
316d083f849SBarry Smith   Collective
317e5c89e4eSSatish Balay 
318d8d19677SJose E. Roman   Input Parameters:
3190c4f890aSBarry Smith + comm - processors accessing the file
3200c4f890aSBarry Smith . url  - name of file, including entire URL (with or without .gz)
32121532e8aSBarry Smith - llen - length of `localname`
322e5c89e4eSSatish Balay 
323d8d19677SJose E. Roman   Output Parameters:
32408fb59bfSBarry Smith + localname - name of local copy of file - valid on only process zero
32508fb59bfSBarry Smith - found     - if found or retrieved the file - valid on all processes
326e5c89e4eSSatish Balay 
32721532e8aSBarry Smith   Level: developer
3282fe279fdSBarry Smith 
329811af0c4SBarry Smith   Note:
33021532e8aSBarry Smith   if the file already exists locally this function just returns without downloading it.
331e5c89e4eSSatish Balay 
33210450e9eSJacob Faibussowitsch .seealso: `PetscDLLibraryRetrieve()`
333e5c89e4eSSatish Balay @*/
334d71ae5a4SJacob Faibussowitsch PetscErrorCode PetscFileRetrieve(MPI_Comm comm, const char url[], char localname[], size_t llen, PetscBool *found)
335d71ae5a4SJacob Faibussowitsch {
336bbcf679cSJacob Faibussowitsch   char        buffer[PETSC_MAX_PATH_LEN], *par = NULL, *tlocalname = NULL, name[PETSC_MAX_PATH_LEN];
337e5c89e4eSSatish Balay   FILE       *fp;
338e5c89e4eSSatish Balay   PetscMPIInt rank;
339e5c89e4eSSatish Balay   size_t      len = 0;
34008fb59bfSBarry Smith   PetscBool   flg1, flg2, flg3, flg4, download, compressed = PETSC_FALSE;
341e5c89e4eSSatish Balay 
342e5c89e4eSSatish Balay   PetscFunctionBegin;
3439566063dSJacob Faibussowitsch   PetscCallMPI(MPI_Comm_rank(comm, &rank));
344dd400576SPatrick Sanan   if (rank == 0) {
345e5c89e4eSSatish Balay     *found = PETSC_FALSE;
346e5c89e4eSSatish Balay 
3479566063dSJacob Faibussowitsch     PetscCall(PetscStrstr(url, ".gz", &par));
34808fb59bfSBarry Smith     if (par) {
3499566063dSJacob Faibussowitsch       PetscCall(PetscStrlen(par, &len));
35008fb59bfSBarry Smith       if (len == 3) compressed = PETSC_TRUE;
35108fb59bfSBarry Smith     }
352e5c89e4eSSatish Balay 
3539566063dSJacob Faibussowitsch     PetscCall(PetscStrncmp(url, "ftp://", 6, &flg1));
3549566063dSJacob Faibussowitsch     PetscCall(PetscStrncmp(url, "http://", 7, &flg2));
3559566063dSJacob Faibussowitsch     PetscCall(PetscStrncmp(url, "file://", 7, &flg3));
3569566063dSJacob Faibussowitsch     PetscCall(PetscStrncmp(url, "https://", 8, &flg4));
357a4772d12SBarry Smith     download = (PetscBool)(flg1 || flg2 || flg3 || flg4);
35808fb59bfSBarry Smith 
35908fb59bfSBarry Smith     if (!download && !compressed) {
3609566063dSJacob Faibussowitsch       PetscCall(PetscStrncpy(localname, url, llen));
3619566063dSJacob Faibussowitsch       PetscCall(PetscTestFile(url, 'r', found));
362487e5849SBarry Smith       if (*found) {
3639566063dSJacob Faibussowitsch         PetscCall(PetscInfo(NULL, "Found file %s\n", url));
364487e5849SBarry Smith       } else {
3659566063dSJacob Faibussowitsch         PetscCall(PetscInfo(NULL, "Did not find file %s\n", url));
366487e5849SBarry Smith       }
36708fb59bfSBarry Smith       goto done;
368734f99bcSBarry Smith     }
369734f99bcSBarry Smith 
37005698389SBarry Smith     /* look for uncompressed file in requested directory */
37105698389SBarry Smith     if (compressed) {
3729566063dSJacob Faibussowitsch       PetscCall(PetscStrncpy(localname, url, llen));
3739566063dSJacob Faibussowitsch       PetscCall(PetscStrstr(localname, ".gz", &par));
37405698389SBarry Smith       *par = 0; /* remove .gz extension */
3759566063dSJacob Faibussowitsch       PetscCall(PetscTestFile(localname, 'r', found));
37605698389SBarry Smith       if (*found) goto done;
37705698389SBarry Smith     }
37805698389SBarry Smith 
37905698389SBarry Smith     /* look for file in current directory */
3809566063dSJacob Faibussowitsch     PetscCall(PetscStrrchr(url, '/', &tlocalname));
3819566063dSJacob Faibussowitsch     PetscCall(PetscStrncpy(localname, tlocalname, llen));
38208fb59bfSBarry Smith     if (compressed) {
3839566063dSJacob Faibussowitsch       PetscCall(PetscStrstr(localname, ".gz", &par));
38408fb59bfSBarry Smith       *par = 0; /* remove .gz extension */
38508fb59bfSBarry Smith     }
3869566063dSJacob Faibussowitsch     PetscCall(PetscTestFile(localname, 'r', found));
38708fb59bfSBarry Smith     if (*found) goto done;
388e5c89e4eSSatish Balay 
38908fb59bfSBarry Smith     if (download) {
39008fb59bfSBarry Smith       /* local file is not already here so use curl to get it */
3919566063dSJacob Faibussowitsch       PetscCall(PetscStrncpy(localname, tlocalname, llen));
392c6a7a370SJeremy L Thompson       PetscCall(PetscStrncpy(buffer, "curl --fail --silent --show-error ", sizeof(buffer)));
393c6a7a370SJeremy L Thompson       PetscCall(PetscStrlcat(buffer, url, sizeof(buffer)));
394c6a7a370SJeremy L Thompson       PetscCall(PetscStrlcat(buffer, " > ", sizeof(buffer)));
395c6a7a370SJeremy L Thompson       PetscCall(PetscStrlcat(buffer, localname, sizeof(buffer)));
396e5c89e4eSSatish Balay #if defined(PETSC_HAVE_POPEN)
3979566063dSJacob Faibussowitsch       PetscCall(PetscPOpen(PETSC_COMM_SELF, NULL, buffer, "r", &fp));
3989566063dSJacob Faibussowitsch       PetscCall(PetscPClose(PETSC_COMM_SELF, fp));
399e5c89e4eSSatish Balay #else
400e32f2f54SBarry Smith       SETERRQ(PETSC_COMM_SELF, PETSC_ERR_SUP_SYS, "Cannot run external programs on this machine");
401e5c89e4eSSatish Balay #endif
4029566063dSJacob Faibussowitsch       PetscCall(PetscTestFile(localname, 'r', found));
4036e3a5469SBarry Smith       if (*found) {
4046e3a5469SBarry Smith         FILE *fd;
4056e3a5469SBarry Smith         char  buf[1024], *str, *substring;
4066e3a5469SBarry Smith 
4076e3a5469SBarry Smith         /* check if the file didn't exist so it downloaded an HTML message instead */
4086e3a5469SBarry Smith         fd = fopen(localname, "r");
40928b400f6SJacob Faibussowitsch         PetscCheck(fd, PETSC_COMM_SELF, PETSC_ERR_PLIB, "PetscTestFile() indicates %s exists but fopen() cannot open it", localname);
4106e3a5469SBarry Smith         str = fgets(buf, sizeof(buf) - 1, fd);
4116e3a5469SBarry Smith         while (str) {
4129566063dSJacob Faibussowitsch           PetscCall(PetscStrstr(buf, "<!DOCTYPE html>", &substring));
41328b400f6SJacob Faibussowitsch           PetscCheck(!substring, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Unable to download %s it does not appear to exist at this URL, dummy HTML file was downloaded", url);
4149566063dSJacob Faibussowitsch           PetscCall(PetscStrstr(buf, "Not Found", &substring));
41528b400f6SJacob Faibussowitsch           PetscCheck(!substring, PETSC_COMM_SELF, PETSC_ERR_PLIB, "Unable to download %s it does not appear to exist at this URL, dummy HTML file was downloaded", url);
4166e3a5469SBarry Smith           str = fgets(buf, sizeof(buf) - 1, fd);
4176e3a5469SBarry Smith         }
4186e3a5469SBarry Smith         fclose(fd);
4196e3a5469SBarry Smith       }
42008fb59bfSBarry Smith     } else if (compressed) {
4219566063dSJacob Faibussowitsch       PetscCall(PetscTestFile(url, 'r', found));
42208fb59bfSBarry Smith       if (!*found) goto done;
4239566063dSJacob Faibussowitsch       PetscCall(PetscStrncpy(localname, url, llen));
42408fb59bfSBarry Smith     }
42508fb59bfSBarry Smith     if (compressed) {
4269566063dSJacob Faibussowitsch       PetscCall(PetscStrrchr(localname, '/', &tlocalname));
4279566063dSJacob Faibussowitsch       PetscCall(PetscStrncpy(name, tlocalname, PETSC_MAX_PATH_LEN));
4289566063dSJacob Faibussowitsch       PetscCall(PetscStrstr(name, ".gz", &par));
42908fb59bfSBarry Smith       *par = 0; /* remove .gz extension */
43008fb59bfSBarry Smith       /* uncompress file */
431c6a7a370SJeremy L Thompson       PetscCall(PetscStrncpy(buffer, "gzip -c -d ", sizeof(buffer)));
432c6a7a370SJeremy L Thompson       PetscCall(PetscStrlcat(buffer, localname, sizeof(buffer)));
433c6a7a370SJeremy L Thompson       PetscCall(PetscStrlcat(buffer, " > ", sizeof(buffer)));
434c6a7a370SJeremy L Thompson       PetscCall(PetscStrlcat(buffer, name, sizeof(buffer)));
43508fb59bfSBarry Smith #if defined(PETSC_HAVE_POPEN)
4369566063dSJacob Faibussowitsch       PetscCall(PetscPOpen(PETSC_COMM_SELF, NULL, buffer, "r", &fp));
4379566063dSJacob Faibussowitsch       PetscCall(PetscPClose(PETSC_COMM_SELF, fp));
43808fb59bfSBarry Smith #else
43908fb59bfSBarry Smith       SETERRQ(PETSC_COMM_SELF, PETSC_ERR_SUP_SYS, "Cannot run external programs on this machine");
44008fb59bfSBarry Smith #endif
4419566063dSJacob Faibussowitsch       PetscCall(PetscStrncpy(localname, name, llen));
4429566063dSJacob Faibussowitsch       PetscCall(PetscTestFile(localname, 'r', found));
443e5c89e4eSSatish Balay     }
444e5c89e4eSSatish Balay   }
445955d42a0SBarry Smith done:
4469566063dSJacob Faibussowitsch   PetscCallMPI(MPI_Bcast(found, 1, MPIU_BOOL, 0, comm));
4476497c311SBarry Smith   PetscCallMPI(MPI_Bcast(localname, (PetscMPIInt)llen, MPI_CHAR, 0, comm));
4483ba16761SJacob Faibussowitsch   PetscFunctionReturn(PETSC_SUCCESS);
449e5c89e4eSSatish Balay }
450