darshan-convert.c 10.4 KB
Newer Older
1
/*
2 3 4
 * Copyright (C) 2015 University of Chicago.
 * See COPYRIGHT notice in top-level directory.
 *
5 6 7 8 9 10 11 12 13 14 15 16 17
 */

#include <stdio.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <unistd.h>
#include <string.h>
#include <zlib.h>
#include <time.h>
#include <stdlib.h>
#include <getopt.h>
#include <assert.h>
18
#include <errno.h>
19 20 21

#include "darshan-logutils.h"

22
extern uint32_t darshan_hashlittle(const void *key, size_t length, uint32_t initval);
23 24 25 26 27 28

int usage (char *exename)
{
    fprintf(stderr, "Usage: %s [options] <infile> <outfile>\n", exename);
    fprintf(stderr, "       Converts darshan log from infile to outfile.\n");
    fprintf(stderr, "       rewrites the log file into the newest format.\n");
29
    fprintf(stderr, "       --bzip2 Use bzip2 compression instead of zlib.\n");
30
    fprintf(stderr, "       --obfuscate Obfuscate items in the log.\n");
31 32
    fprintf(stderr, "       --key <key> Key to use when obfuscating.\n");
    fprintf(stderr, "       --annotate <string> Additional metadata to add.\n");
33
    fprintf(stderr, "       --file <hash> Limit output to specified (hashed) file only.\n");
34
    fprintf(stderr, "       --reset-md Reset old metadata during conversion.\n");
35 36 37 38

    exit(1);
}

39
void parse_args (int argc, char **argv, char **infile, char **outfile,
40 41
                 int *bzip2, int *obfuscate, int *reset_md, int *key,
                 char **annotate, uint64_t* hash)
42 43
{
    int index;
44 45
    int ret;

46 47
    static struct option long_opts[] =
    {
48
        {"bzip2", 0, NULL, 'b'},
49
        {"annotate", 1, NULL, 'a'},
50
        {"obfuscate", 0, NULL, 'o'},
51
        {"reset-md", 0, NULL, 'r'},
52
        {"key", 1, NULL, 'k'},
53
        {"file", 1, NULL, 'f'},
54 55
        {"help",  0, NULL, 0},
        { 0, 0, 0, 0 }
56 57
    };

58
    *bzip2 = 0;
59
    *obfuscate = 0;
60
    *reset_md = 0;
61
    *key = 0;
62 63
    *hash = 0;

64 65 66 67 68 69 70 71
    while(1)
    {
        int c = getopt_long(argc, argv, "", long_opts, &index);

        if (c == -1) break;

        switch(c)
        {
72 73 74
            case 'b':
                *bzip2 = 1;
                break;
75 76 77
            case 'a':
                *annotate = optarg;
                break;
78 79 80
            case 'o':
                *obfuscate = 1;
                break;
81 82 83
            case 'r':
                *reset_md = 1;
                break;
84 85 86
            case 'k':
                *key = atoi(optarg);
                break;
87 88 89 90 91
            case 'f':
                ret = sscanf(optarg, "%" PRIu64, hash);
                if(ret != 1)
                    usage(argv[0]);
                break;
92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112
            case 0:
            case '?':
            default:
                usage(argv[0]);
                break;
        }
    }

    if (optind + 2 == argc)
    {
        *infile = argv[optind];
        *outfile = argv[optind+1];
    }
    else
    {
        usage(argv[0]);
    }

    return;
}

113 114 115 116 117 118
static void reset_md_job(struct darshan_job *job)
{
    job->metadata[0] = '\0';
    return;
}

119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140
void obfuscate_job(int key, struct darshan_job *job)
{
    job->uid   = (int64_t) darshan_hashlittle(&job->uid, sizeof(job->uid), key);
    if (job->jobid != 0)
    {
        job->jobid = (int64_t) darshan_hashlittle(&job->jobid, sizeof(job->jobid), key);
    }

    return;
}

void obfuscate_exe(int key, char *exe)
{
    uint32_t hashed;

    hashed = darshan_hashlittle(exe, strlen(exe), key);
    memset(exe, 0, strlen(exe));
    sprintf(exe, "%u", hashed);

    return;
}

141
void obfuscate_filenames(int key, struct darshan_name_record_ref *name_hash)
142
{
143
    struct darshan_name_record_ref *ref, *tmp;
144
    uint32_t hashed;
145
    char tmp_string[128] = {0};
146
    darshan_record_id tmp_id;
147

148
    HASH_ITER(hlink, name_hash, ref, tmp)
149
    {
150 151 152
        tmp_id = ref->name_record->id;
        hashed = darshan_hashlittle(ref->name_record->name,
            strlen(ref->name_record->name), key);
153
        sprintf(tmp_string, "%u", hashed);
154 155 156 157 158 159
        free(ref->name_record);
        ref->name_record = malloc(sizeof(struct darshan_name_record) +
            strlen(tmp_string));
        assert(ref->name_record);
        ref->name_record->id = tmp_id;
        strcpy(ref->name_record->name, tmp_string);
160
    }
161 162 163 164

    return;
}

165 166 167 168 169
void add_annotation (char *annotation,
                     struct darshan_job *job)
{
    char *token;
    char *save;
170
    int len;
171
    
172 173 174 175 176 177 178 179 180 181 182
    /* check for newline in existing metadata, insert if needed */
    len = strlen(job->metadata);
    if(len > 0 && len < sizeof(job->metadata))
    {
        if(job->metadata[len-1] != '\n')
        {
            job->metadata[len] = '\n';
            job->metadata[len+1] = '\0';
        }
    }

183 184
    /* determine remaining space in metadata string */
    int remaining = sizeof(job->metadata) - strlen(job->metadata);
185 186 187 188 189

    for(token=strtok_r(annotation, "\t", &save);
        token != NULL;
        token=strtok_r(NULL, "\t", &save))
    {
190 191 192 193 194 195 196 197 198 199
        if ((strlen(token)+1) < remaining)
        {
            strcat(job->metadata, token);
            strcat(job->metadata, "\n");
            remaining -= (strlen(token)+1);
        }
        else
        {
            fprintf(stderr,
                    "not enough space left in metadata for: current=%s token=%s (remain=%d:need=%d)\n",
200 201 202 203 204 205 206
                    job->metadata, token, remaining-1, (int)strlen(token)+1);
        }
    }

    return;
}

207 208
static void remove_hash_recs(struct darshan_name_record_ref **name_hash,
    darshan_record_id hash)
209
{
210
    struct darshan_name_record_ref *ref, *tmp;
211

212
    HASH_ITER(hlink, *name_hash, ref, tmp)
213
    {
214
        if(ref->name_record->id != hash)
215
        {
216 217
            HASH_DELETE(hlink, *name_hash, ref);
            free(ref->name_record);
218
            free(ref);
219
        }
220 221 222 223 224
    }

    return;
}

225 226 227 228 229 230
int main(int argc, char **argv)
{
    int ret;
    char *infile_name;
    char *outfile_name;
    struct darshan_job job;
231
    char tmp_string[4096] = {0};
232 233 234 235
    darshan_fd infile;
    darshan_fd outfile;
    int i;
    int mount_count;
236
    struct darshan_mnt_info *mnt_data_array;
237 238
    struct darshan_name_record_ref *name_hash = NULL;
    struct darshan_name_record_ref *ref, *tmp;
239
    char mod_buf[DEF_MOD_BUF_SIZE];
240
    enum darshan_comp_type comp_type;
241 242 243
    int bzip2;
    int obfuscate;
    int key;
244
    char *annotation = NULL;
245 246
    darshan_record_id hash;
    int reset_md;
247

248 249
    parse_args(argc, argv, &infile_name, &outfile_name, &bzip2, &obfuscate,
               &reset_md, &key, &annotation, &hash);
250

251
    infile = darshan_log_open(infile_name);
252 253 254
    if(!infile)
        return(-1);
 
255
    comp_type = bzip2 ? DARSHAN_BZIP2_COMP : DARSHAN_ZLIB_COMP;
256
    outfile = darshan_log_create(outfile_name, comp_type, infile->partial_flag);
257 258
    if(!outfile)
    {
259 260 261 262
        darshan_log_close(infile);
        return(-1);
    }

263
    /* read job info */
264
    ret = darshan_log_get_job(infile, &job);
265 266 267
    if(ret < 0)
    {
        darshan_log_close(infile);
268
        darshan_log_close(outfile);
269
        unlink(outfile_name);
270 271 272
        return(-1);
    }

273
    if (reset_md) reset_md_job(&job);
274
    if (obfuscate) obfuscate_job(key, &job);
275
    if (annotation) add_annotation(annotation, &job);
276

277
    ret = darshan_log_put_job(outfile, &job);
278 279
    if (ret < 0)
    {
280
        darshan_log_close(infile);
281 282 283 284
        darshan_log_close(outfile);
        return(-1);
    }

285
    ret = darshan_log_get_exe(infile, tmp_string);
286 287 288
    if(ret < 0)
    {
        darshan_log_close(infile);
289
        darshan_log_close(outfile);
290
        unlink(outfile_name);
291 292
        return(-1);
    }
293

294 295
    if (obfuscate) obfuscate_exe(key, tmp_string);

296
    ret = darshan_log_put_exe(outfile, tmp_string);
297 298
    if(ret < 0)
    {
299
        darshan_log_close(infile);
300 301 302
        darshan_log_close(outfile);
        return(-1);
    }
303

304
    ret = darshan_log_get_mounts(infile, &mnt_data_array, &mount_count);
305 306 307
    if(ret < 0)
    {
        darshan_log_close(infile);
308
        darshan_log_close(outfile);
309
        unlink(outfile_name);
310 311 312
        return(-1);
    }

313
    ret = darshan_log_put_mounts(outfile, mnt_data_array, mount_count);
314 315
    if(ret < 0)
    {
316
        darshan_log_close(infile);
317 318 319 320
        darshan_log_close(outfile);
        return(-1);
    }

321
    ret = darshan_log_get_namehash(infile, &name_hash);
322 323
    if(ret < 0)
    {
324 325
        darshan_log_close(infile);
        darshan_log_close(outfile);
326
        unlink(outfile_name);
327
        return(-1);
328
    }
329

330 331 332
    /* NOTE: obfuscating filepaths breaks the ability to map files
     * to the corresponding FS & mount info maintained by darshan
     */
333 334
    if(obfuscate) obfuscate_filenames(key, name_hash);
    if(hash) remove_hash_recs(&name_hash, hash);
335

336
    ret = darshan_log_put_namehash(outfile, name_hash);
337
    if(ret < 0)
338
    {
339 340 341
        darshan_log_close(infile);
        darshan_log_close(outfile);
        return(-1);
342 343
    }

344
    /* loop over each module and convert it's data to the new format */
345
    for(i=0; i<DARSHAN_MAX_MODS; i++)
346
    {
347
        struct darshan_base_record *base_rec;
348

349
        /* check each module for any data */
350
        if(infile->mod_map[i].len == 0)
351
            continue;
352
        else if(!mod_logutils[i])
353 354
        {
            fprintf(stderr, "Warning: no log utility handlers defined "
355
                "for module %s, SKIPPING.\n", darshan_module_names[i]);
356 357 358
            continue;
        }

359
        /* we have module data to convert */
360
        memset(mod_buf, 0, DEF_MOD_BUF_SIZE);
361

362
        ret = mod_logutils[i]->log_get_record(infile, mod_buf);
363
        if(ret != 1)
364
        {
365
            fprintf(stderr, "Error: failed to parse the first %s module record.\n",
366 367 368
                darshan_module_names[i]);
            darshan_log_close(infile);
            darshan_log_close(outfile);
369
            unlink(outfile_name);
370 371
            return(-1);
        }
372

373 374 375
        /* loop over each of the module's records and convert */
        do
        {
376 377 378
            base_rec = (struct darshan_base_record *)mod_buf;

            if(!hash || hash == base_rec->id)
379
            {
380
                ret = mod_logutils[i]->log_put_record(outfile, mod_buf, infile->mod_ver[i]);
381 382 383 384 385 386 387 388 389
                if(ret < 0)
                {
                    darshan_log_close(infile);
                    darshan_log_close(outfile);
                    return(-1);
                }

                memset(mod_buf, 0, DEF_MOD_BUF_SIZE);
            }
390
        } while((ret = mod_logutils[i]->log_get_record(infile, mod_buf)) == 1);
391 392 393 394 395
    }

    darshan_log_close(infile);
    darshan_log_close(outfile);

396
    if(mount_count > 0)
397
        free(mnt_data_array);
398

399
    HASH_ITER(hlink, name_hash, ref, tmp)
400
    {
401 402
        HASH_DELETE(hlink, name_hash, ref);
        free(ref->name_record);
403 404 405
        free(ref);
    }

406
    return(ret);
407 408
}

409 410 411 412 413 414 415 416
/*
 * Local variables:
 *  c-indent-level: 4
 *  c-basic-offset: 4
 * End:
 *
 * vim: ts=8 sts=4 sw=4 expandtab
 */