vault_t *ReadData(params_t *params)
{
vault_t *vault;
gk_csr_t *mat1, *mat2;
vault = (vault_t *)gk_malloc(sizeof(vault_t), "ReadData: vault");
/* read the document vectors */
printf("Reading documents...\n");
ASSERT(gk_fexists(params->infstem));
vault->mat = gk_csr_Read(params->infstem, GK_CSR_FMT_CSR, 1, 0);
vault->ndocs = vault->mat->nrows;
return vault;
}
params_t *parse_cmdline(int argc, char *argv[])
{
int i;
int c, option_index;
params_t *params;
params = (params_t *)gk_malloc(sizeof(params_t), "parse_cmdline: params");
/* initialize the params data structure */
params->niter = 100;
params->ppr = -1;
params->ntvs = -1;
params->eps = 1e-10;
params->lamda = 0.80;
params->infile = NULL;
params->outfile = NULL;
/* Parse the command line arguments */
while ((c = gk_getopt_long_only(argc, argv, "", long_options, &option_index)) != -1) {
switch (c) {
case CMD_NITER:
if (gk_optarg) params->niter = atoi(gk_optarg);
break;
case CMD_NTVS:
if (gk_optarg) params->ntvs = atoi(gk_optarg);
break;
case CMD_PPR:
if (gk_optarg) params->ppr = atoi(gk_optarg);
break;
case CMD_EPS:
if (gk_optarg) params->eps = atof(gk_optarg);
break;
case CMD_LAMDA:
if (gk_optarg) params->lamda = atof(gk_optarg);
break;
case CMD_HELP:
for (i=0; strlen(helpstr[i]) > 0; i++)
printf("%s\n", helpstr[i]);
exit(0);
break;
case '?':
default:
printf("Illegal command-line option(s)\nUse %s -help for a summary of the options.\n", argv[0]);
exit(0);
}
}
if (argc-gk_optind != 2) {
printf("Unrecognized parameters.");
for (i=0; strlen(shorthelpstr[i]) > 0; i++)
printf("%s\n", shorthelpstr[i]);
exit(0);
}
params->infile = gk_strdup(argv[gk_optind++]);
params->outfile = gk_strdup(argv[gk_optind++]);
if (!gk_fexists(params->infile))
errexit("input file %s does not exist.\n", params->infile);
if (params->ppr != -1 && params->ntvs != -1)
errexit("Only one of the -ppr and -ntvs options can be specified.\n");
return params;
}
gk_graph_t *gk_graph_Read(char *filename, int format, int isfewgts,
int isfvwgts, int isfvsizes)
{
ssize_t i, k, l;
size_t nfields, nvtxs, nedges, fmt, ncon, lnlen;
int32_t ival;
float fval;
int readsizes=0, readwgts=0, readvals=0, numbering=0;
char *line=NULL, *head, *tail, fmtstr[256];
FILE *fpin=NULL;
gk_graph_t *graph=NULL;
if (!gk_fexists(filename))
gk_errexit(SIGERR, "File %s does not exist!\n", filename);
if (format == GK_GRAPH_FMT_METIS) {
fpin = gk_fopen(filename, "r", "gk_graph_Read: fpin");
do {
if (gk_getline(&line, &lnlen, fpin) <= 0)
gk_errexit(SIGERR, "Premature end of input file: file:%s\n", filename);
} while (line[0] == '%');
fmt = ncon = 0;
nfields = sscanf(line, "%zu %zu %zu %zu", &nvtxs, &nedges, &fmt, &ncon);
if (nfields < 2)
gk_errexit(SIGERR, "Header line must contain at least 2 integers (#vtxs and #edges).\n");
nedges *= 2;
if (fmt > 111)
gk_errexit(SIGERR, "Cannot read this type of file format [fmt=%zu]!\n", fmt);
sprintf(fmtstr, "%03zu", fmt%1000);
readsizes = (fmtstr[0] == '1');
readwgts = (fmtstr[1] == '1');
readvals = (fmtstr[2] == '1');
numbering = 1;
ncon = (ncon == 0 ? 1 : ncon);
}
else {
gk_errexit(SIGERR, "Unrecognized format: %d\n", format);
}
graph = gk_graph_Create();
graph->nvtxs = nvtxs;
graph->xadj = gk_zmalloc(nvtxs+1, "gk_graph_Read: xadj");
graph->adjncy = gk_i32malloc(nedges, "gk_graph_Read: adjncy");
if (readvals) {
if (isfewgts)
graph->fadjwgt = gk_fmalloc(nedges, "gk_graph_Read: fadjwgt");
else
graph->iadjwgt = gk_i32malloc(nedges, "gk_graph_Read: iadjwgt");
}
if (readsizes) {
if (isfvsizes)
graph->fvsizes = gk_fmalloc(nvtxs, "gk_graph_Read: fvsizes");
else
graph->ivsizes = gk_i32malloc(nvtxs, "gk_graph_Read: ivsizes");
}
if (readwgts) {
if (isfvwgts)
graph->fvwgts = gk_fmalloc(nvtxs*ncon, "gk_graph_Read: fvwgts");
else
graph->ivwgts = gk_i32malloc(nvtxs*ncon, "gk_graph_Read: ivwgts");
}
/*----------------------------------------------------------------------
* Read the sparse graph file
*---------------------------------------------------------------------*/
numbering = (numbering ? - 1 : 0);
for (graph->xadj[0]=0, k=0, i=0; i<nvtxs; i++) {
do {
if (gk_getline(&line, &lnlen, fpin) == -1)
gk_errexit(SIGERR, "Pregraphure end of input file: file while reading row %d\n", i);
} while (line[0] == '%');
head = line;
tail = NULL;
/* Read vertex sizes */
if (readsizes) {
if (isfvsizes) {
#ifdef __MSC__
graph->fvsizes[i] = (float)strtod(head, &tail);
#else
graph->fvsizes[i] = strtof(head, &tail);
#endif
if (tail == head)
gk_errexit(SIGERR, "The line for vertex %zd does not have size information\n", i+1);
if (graph->fvsizes[i] < 0)
gk_errexit(SIGERR, "The size for vertex %zd must be >= 0\n", i+1);
}
else {
graph->ivsizes[i] = strtol(head, &tail, 0);
if (tail == head)
gk_errexit(SIGERR, "The line for vertex %zd does not have size information\n", i+1);
if (graph->ivsizes[i] < 0)
gk_errexit(SIGERR, "The size for vertex %zd must be >= 0\n", i+1);
}
head = tail;
}
/* Read vertex weights */
if (readwgts) {
for (l=0; l<ncon; l++) {
if (isfvwgts) {
#ifdef __MSC__
graph->fvwgts[i*ncon+l] = (float)strtod(head, &tail);
#else
graph->fvwgts[i*ncon+l] = strtof(head, &tail);
#endif
if (tail == head)
gk_errexit(SIGERR, "The line for vertex %zd does not have enough weights "
"for the %d constraints.\n", i+1, ncon);
if (graph->fvwgts[i*ncon+l] < 0)
gk_errexit(SIGERR, "The weight vertex %zd and constraint %zd must be >= 0\n", i+1, l);
}
else {
graph->ivwgts[i*ncon+l] = strtol(head, &tail, 0);
if (tail == head)
gk_errexit(SIGERR, "The line for vertex %zd does not have enough weights "
"for the %d constraints.\n", i+1, ncon);
if (graph->ivwgts[i*ncon+l] < 0)
gk_errexit(SIGERR, "The weight vertex %zd and constraint %zd must be >= 0\n", i+1, l);
}
head = tail;
}
}
/* Read the rest of the row */
while (1) {
ival = (int)strtol(head, &tail, 0);
if (tail == head)
break;
head = tail;
if ((graph->adjncy[k] = ival + numbering) < 0)
gk_errexit(SIGERR, "Error: Invalid column number %d at row %zd.\n", ival, i);
if (readvals) {
if (isfewgts) {
#ifdef __MSC__
fval = (float)strtod(head, &tail);
#else
fval = strtof(head, &tail);
#endif
if (tail == head)
gk_errexit(SIGERR, "Value could not be found for edge! Vertex:%zd, NNZ:%zd\n", i, k);
graph->fadjwgt[k] = fval;
}
else {
ival = strtol(head, &tail, 0);
if (tail == head)
gk_errexit(SIGERR, "Value could not be found for edge! Vertex:%zd, NNZ:%zd\n", i, k);
graph->iadjwgt[k] = ival;
}
head = tail;
}
k++;
}
graph->xadj[i+1] = k;
}
if (k != nedges)
gk_errexit(SIGERR, "gk_graph_Read: Something wrong with the number of edges in "
"the input file. nedges=%zd, Actualnedges=%zd.\n", nedges, k);
gk_fclose(fpin);
gk_free((void **)&line, LTERM);
return graph;
}
