mcxIO* mcxIOnew
( const char* str
, const char* mode
)
{ if (!str || !mode)
{ mcxErr("mcxIOnew PBD", "void string or mode argument")
; return NULL
; }
return mcxIOrenew(NULL, str, mode)
; }
mcxstatus mcxIOnewName
( mcxIO* xf
, const char* newname
)
{ return mcxIOrenew(xf, newname, NULL) ? STATUS_OK : STATUS_FAIL
; }
int main
( int argc
, const char* argv[]
)
{ mcxIO* xfmx = mcxIOnew("-", "r"), *xfout = mcxIOnew("-", "w")
; mclx* mx = NULL
; mclv* mx_diag = NULL
; mcxstatus parseStatus = STATUS_OK
; mcxOption* opts, *opt
; dim N_edge = 0
; dim* offsets
; dim template_n_nodes = 0
; mcxbool plus = FALSE
; double e_min = 1.0
; double e_max = 0.0
; double skew = 0.0
; double radius = 0.0
; double n_sdev = 0.5
; double n_range = 2.0
; double g_radius = 0.0
; double g_mean = 0.0
; double g_sdev = 0.0
; double g_min = 1.0
; double g_max = 0.0
; mcxbool do_gaussian = FALSE
; dim i = 0
; dim N_remove = 0
; dim N_add = 0
; dim N_shuffle = 0
; unsigned long random_ignore = 0
; srandom(mcxSeed(2308947))
; mcxOptAnchorSortById(options, sizeof(options)/sizeof(mcxOptAnchor) -1)
; if
(!(opts = mcxOptParse(options, (char**) argv, argc, 1, 0, &parseStatus)))
exit(0)
; mcxLogLevel =
MCX_LOG_AGGR | MCX_LOG_MODULE | MCX_LOG_IO | MCX_LOG_GAUGE | MCX_LOG_WARN
; mclxIOsetQMode("MCLXIOVERBOSITY", MCL_APP_VB_YES)
; mclx_app_init(stderr)
; for (opt=opts;opt->anch;opt++)
{ mcxOptAnchor* anch = opt->anch
; switch(anch->id)
{ case MY_OPT_HELP
: case MY_OPT_APROPOS
: mcxOptApropos(stdout, me, syntax, 20, MCX_OPT_DISPLAY_SKIP, options)
; return 0
;
case MY_OPT_VERSION
: app_report_version(me)
; return 0
;
case MY_OPT_SKEW
: skew = atof(opt->val)
; break
;
case MY_OPT_GEN
: template_n_nodes = atoi(opt->val)
; break
;
case MY_OPT_IMX
: mcxIOrenew(xfmx, opt->val, NULL)
; break
;
case MY_OPT_PLUS
: case MY_OPT_WB
: plus = TRUE
; break
;
case MY_OPT_OUT
: mcxIOrenew(xfout, opt->val, NULL)
; break
;
case MY_OPT_E_MAX
: if (!strcmp(opt->val, "copy"))
e_max = -DBL_MAX
; else
e_max = atof(opt->val)
; break
;
case MY_OPT_E_MIN
: e_min = atof(opt->val)
; break
;
case MY_OPT_G_MIN
: g_min = atof(opt->val)
; break
;
case MY_OPT_G_MAX
: g_max = atof(opt->val)
; break
;
case MY_OPT_G_SDEV
: g_sdev = atof(opt->val)
; break
;
case MY_OPT_G_MEAN
: g_mean = atof(opt->val)
; do_gaussian = TRUE
; break
;
case MY_OPT_G_RADIUS
: g_radius = atof(opt->val)
; break
;
case MY_OPT_N_RANGE
: n_range = atof(opt->val)
; break
;
case MY_OPT_N_SDEV
: n_sdev = atof(opt->val)
; break
;
case MY_OPT_N_RADIUS
: radius = atof(opt->val)
; break
;
case MY_OPT_SHUFFLE
: N_shuffle = atoi(opt->val)
; break
;
case MY_OPT_ADD
: N_add = atoi(opt->val)
; break
;
case MY_OPT_REMOVE
: N_remove = atoi(opt->val)
; break
; }
}
/* hitting y% in vi tells me the size of this block */
{ if (template_n_nodes)
mx = mclxAllocZero
( mclvCanonical(NULL, template_n_nodes, 1.0)
, mclvCanonical(NULL, template_n_nodes, 1.0)
)
; else
mx = mclxReadx
( xfmx
, EXIT_ON_FAIL
, MCLX_REQUIRE_GRAPH
)
; mx_diag = mclxDiagValues(mx, MCL_VECTOR_COMPLETE)
; if (N_shuffle)
mclxAdjustLoops(mx, mclxLoopCBremove, NULL)
; else
mclxSelectUpper(mx)
/* ^ apparently we always work on single arc representation (docme andsoon) */
; offsets = mcxAlloc(sizeof offsets[0] * N_COLS(mx), EXIT_ON_FAIL)
; N_edge = 0
; for (i=0;i<N_COLS(mx);i++)
{ offsets[i] = N_edge
; N_edge += mx->cols[i].n_ivps
; }
if (N_edge < N_remove)
{ mcxErr
( me
, "removal count %ld exceeds edge count %ld"
, (long) N_remove
, (long) N_edge
)
; N_remove = N_edge
; }
random_ignore = RAND_MAX - (N_edge ? RAND_MAX % N_edge : 0)
; if (RAND_MAX / 2 < N_edge)
mcxDie(1, me, "graph too large!")
; if (N_shuffle)
{ do_the_shuffle(mx, N_shuffle, offsets, N_edge, random_ignore)
; mx_readd_diagonal(mx, mx_diag)
; mclxWrite(mx, xfout, MCLXIO_VALUE_GETENV, RETURN_ON_FAIL)
; exit(0)
; }
; if (N_remove)
{ dim n_remove = do_remove(mx, N_remove, offsets, N_edge, random_ignore)
/* Need to recompute N_edge and random_ignore.
* NOTE we work with *upper* matrix; this counts graph edges.
*/
; N_edge = mclxNrofEntries(mx) - n_remove
; random_ignore = RAND_MAX - (RAND_MAX % N_COLS(mx))
; }
if (g_mean)
{ if (!g_radius)
{ if (g_sdev)
g_radius = 2 * g_sdev
; mcxWarn(me, "set radius to %.5f\n", g_radius)
; }
}
; if (N_add)
N_edge += do_add
( mx
, N_add , N_edge
, do_gaussian ? &g_mean : NULL, g_radius , g_sdev , g_min , g_max
, skew
, e_min , e_max
)
; if (radius)
{ for (i=0;i<N_COLS(mx);i++)
{ mclp* ivp = mx->cols[i].ivps, *ivpmax = ivp + mx->cols[i].n_ivps
;if(DEBUG)fprintf(stderr, "here %d\n", (int) i)
; while (ivp < ivpmax)
{ double val = ivp->val
; double r = mcxNormalCut(n_range * n_sdev, n_sdev)
; double newval = val + radius * (r / (n_range * n_sdev))
; if (e_min < e_max && newval >= e_min && newval <= e_max)
; ivp->val = newval
; ivp++
; }
}
}
mclxUnary(mx, fltxCopy, NULL) /* remove zeroes */
; mclxAddTranspose(mx, 0.0)
; mx_readd_diagonal(mx, mx_diag)
; if (plus)
mclxbWrite(mx, xfout, RETURN_ON_FAIL)
; else
mclxWrite(mx, xfout, MCLXIO_VALUE_GETENV, RETURN_ON_FAIL)
; }
return 0
; }
int main
( int argc
, const char* argv[]
)
{ mcxIO* xf_tab = NULL
; mcxIO* xf_tabr = NULL
; mcxIO* xf_tabc = NULL
; mcxIO* xf_restrict_tab = NULL
; mcxIO* xf_restrict_tabr = NULL
; mcxIO* xf_restrict_tabc = NULL
; mcxIO* xf_mx = mcxIOnew("-", "r")
; mcxIO* xfout = NULL
; const char* fndump = "-"
; mclTab* tabr = NULL
; mclTab* tabc = NULL
; mclTab* restrict_tabr = NULL
; mclTab* restrict_tabc = NULL
; mcxbool transpose = FALSE
; mcxbool lazy_tab = FALSE
; mcxbool write_tabc = FALSE
; mcxbool write_tabr = FALSE
; mcxbool cat = FALSE
; mcxbool tree = FALSE
; mcxbool skel = FALSE
; mcxbool newick = FALSE
; mcxbits newick_bits = 0
; mcxbits cat_bits = 0
; dim catmax = 1
; dim n_max = 0
; dim table_nlines = 0
; dim table_nfields = 0
; int split_idx = 1
; int split_inc = 1
; const char* split_stem = NULL
; const char* sort_mode = NULL
; mcxTing* line = mcxTingEmpty(NULL, 10)
; mcxbits modes = MCLX_DUMP_VALUES
; mcxbits mode_dump = MCLX_DUMP_PAIRS
; mcxbits mode_part = 0
; mcxbits mode_loop = MCLX_DUMP_LOOP_ASIS
; mcxbits mode_matrix = 0
; int digits = MCLXIO_VALUE_GETENV
; mcxOption* opts, *opt
; mcxstatus parseStatus = STATUS_OK
; mcxLogLevel =
MCX_LOG_AGGR | MCX_LOG_MODULE | MCX_LOG_IO | MCX_LOG_GAUGE | MCX_LOG_WARN
; mclxIOsetQMode("MCLXIOVERBOSITY", MCL_APP_VB_YES)
; mclx_app_init(stderr)
; mcxOptAnchorSortById(options, sizeof(options)/sizeof(mcxOptAnchor) -1)
; opts = mcxOptParse(options, (char**) argv, argc, 1, 0, &parseStatus)
; if (!opts)
exit(0)
; for (opt=opts;opt->anch;opt++)
{ mcxOptAnchor* anch = opt->anch
; switch(anch->id)
{ case MY_OPT_HELP
: case MY_OPT_APROPOS
: mcxOptApropos(stdout, me, syntax, 0, 0, options)
; return 0
;
case MY_OPT_VERSION
: app_report_version(me)
; return 0
;
case MY_OPT_TAB
: xf_tab = mcxIOnew(opt->val, "r")
; break
;
case MY_OPT_TABC
: xf_tabc = mcxIOnew(opt->val, "r")
; break
;
case MY_OPT_TABR
: xf_tabr = mcxIOnew(opt->val, "r")
; break
;
case MY_OPT_OUTPUT
: fndump = opt->val
; break
;
case MY_OPT_SEP_LEAD
: sep_lead_g = opt->val
; break
;
case MY_OPT_SEP_FIELD
: sep_row_g = opt->val
; break
;
case MY_OPT_SEP_CAT
: sep_cat_g = opt->val
; break
;
case MY_OPT_SEP_VAL
: sep_val_g = opt->val
; break
;
case MY_OPT_PREFIXC
: prefixc_g = opt->val
; break
;
case MY_OPT_RESTRICT_TAB
: xf_restrict_tab = mcxIOnew(opt->val, "r")
; break
;
case MY_OPT_RESTRICT_TABC
: xf_restrict_tabc = mcxIOnew(opt->val, "r")
; break
;
case MY_OPT_RESTRICT_TABR
: xf_restrict_tabr = mcxIOnew(opt->val, "r")
; break
;
case MY_OPT_LAZY_TAB
: lazy_tab = TRUE
; break
;
case MY_OPT_NO_VALUES
: BIT_OFF(modes, MCLX_DUMP_VALUES)
; break
;
case MY_OPT_DUMP_RLINES
: mode_dump = MCLX_DUMP_LINES
; BIT_ON(modes, MCLX_DUMP_NOLEAD)
; break
;
case MY_OPT_DUMP_VLINES
: mode_dump = MCLX_DUMP_LINES
; BIT_ON(modes, MCLX_DUMP_LEAD_VALUE)
; break
;
case MY_OPT_DUMP_LINES
: mode_dump = MCLX_DUMP_LINES
; break
;
case MY_OPT_OMIT_EMPTY
: BIT_ON(modes, MCLX_DUMP_OMIT_EMPTY)
; break
;
case MY_OPT_SORT
: sort_mode = opt->val
; break
;
case MY_OPT_NO_LOOPS
: mode_loop = MCLX_DUMP_LOOP_NONE
; break
;
case MY_OPT_CAT_LIMIT
: n_max = atoi(opt->val)
; break
;
case MY_OPT_SPLIT_STEM
: split_stem = opt->val
; sep_cat_g = NULL
; break
;
case MY_OPT_FORCE_LOOPS
: mode_loop = MCLX_DUMP_LOOP_FORCE
; break
;
case MY_OPT_SKEL
: skel = TRUE
; break
;
case MY_OPT_WRITE_TABC
: write_tabc = TRUE
; break
;
case MY_OPT_DIGITS
: digits = strtol(opt->val, NULL, 10)
; break
;
case MY_OPT_WRITE_TABR
: write_tabr = TRUE
; break
;
case MY_OPT_DUMP_RDOM
: transpose = TRUE
; skel = TRUE
; mode_dump = MCLX_DUMP_LINES
; break
;
case MY_OPT_DUMP_CDOM
: skel = TRUE
; mode_dump = MCLX_DUMP_LINES
; break
;
case MY_OPT_IMX
: mcxIOnewName(xf_mx, opt->val)
; break
;
case MY_OPT_ICL
: mcxIOnewName(xf_mx, opt->val)
; mode_dump = MCLX_DUMP_LINES
; BIT_ON(modes, MCLX_DUMP_NOLEAD)
; BIT_OFF(modes, MCLX_DUMP_VALUES)
; break
;
case MY_OPT_TREECAT
: mcxIOnewName(xf_mx, opt->val)
; tree = TRUE
; cat_bits |= MCLX_PRODUCE_DOMSTACK
; break
;
case MY_OPT_CAT
: mcxIOnewName(xf_mx, opt->val)
; cat = TRUE
; break
;
case MY_OPT_DUMP_MATRIX
: mode_matrix |= MCLX_DUMP_MATRIX
; break
;
case MY_OPT_TRANSPOSE
: transpose = TRUE
; break
;
case MY_OPT_DUMP_UPPER
: mode_part = MCLX_DUMP_PART_UPPER
; break
;
case MY_OPT_DUMP_UPPERI
: mode_part = MCLX_DUMP_PART_UPPERI
; break
;
case MY_OPT_DUMP_LOWER
: mode_part = MCLX_DUMP_PART_LOWER
; break
;
case MY_OPT_DUMP_LOWERI
: mode_part = MCLX_DUMP_PART_LOWERI
; break
;
case MY_OPT_DUMP_NOLEAD
: BIT_ON(modes, MCLX_DUMP_NOLEAD)
; break
;
case MY_OPT_NEWICK_MODE
: if (strchr(opt->val, 'N'))
newick_bits |= (MCLX_NEWICK_NONL | MCLX_NEWICK_NOINDENT)
; if (strchr(opt->val, 'I'))
newick_bits |= MCLX_NEWICK_NOINDENT
; if (strchr(opt->val, 'B'))
newick_bits |= MCLX_NEWICK_NONUM
; if (strchr(opt->val, 'S'))
newick_bits |= MCLX_NEWICK_NOPTHS
; newick = TRUE
; break
;
case MY_OPT_DUMP_NEWICK
: newick = TRUE
; break
;
case MY_OPT_DUMP_TABLE
: mode_dump = MCLX_DUMP_TABLE
; break
;
case MY_OPT_TABLE_NFIELDS
: table_nfields = atoi(opt->val)
; break
;
case MY_OPT_TABLE_NLINES
: table_nlines = atoi(opt->val)
; break
;
case MY_OPT_DUMP_PAIRS
: mode_dump = MCLX_DUMP_PAIRS
; break
; }
}
; if (skel)
cat_bits |= MCLX_READ_SKELETON
; modes |= mode_loop | mode_dump | mode_part | mode_matrix
; xfout = mcxIOnew(fndump, "w")
; mcxIOopen(xfout, EXIT_ON_FAIL)
; mcxIOopen(xf_mx, EXIT_ON_FAIL)
; if (cat || tree)
catmax = n_max ? n_max : 0
; if ((write_tabc || write_tabr) && !xf_tab)
mcxDie(1, me, "need a single tab file (-tab option) with --write-tabc or --write-tabr")
; if (xf_tab && mcxIOopen(xf_tab, RETURN_ON_FAIL))
mcxDie(1, me, "no tab")
; else
{ if (xf_tabr && mcxIOopen(xf_tabr, RETURN_ON_FAIL))
mcxDie(1, me, "no tabr")
; if (xf_tabc && mcxIOopen(xf_tabc, RETURN_ON_FAIL))
mcxDie(1, me, "no tabc")
; }
{ if (xf_restrict_tab && mcxIOopen(xf_restrict_tab, RETURN_ON_FAIL))
mcxDie(1, me, "no restriction tab")
; else
{ if (xf_restrict_tabr && mcxIOopen(xf_restrict_tabr, RETURN_ON_FAIL))
mcxDie(1, me, "no restriction tabr")
; if (xf_restrict_tabc && mcxIOopen(xf_restrict_tabc, RETURN_ON_FAIL))
mcxDie(1, me, "no restriction tabc")
; }
/* fixme: below is pretty boilerplate, happens in other places as well */
if (xf_restrict_tab)
{ if (!(restrict_tabr = mclTabRead (xf_restrict_tab, NULL, RETURN_ON_FAIL)))
mcxDie(1, me, "error reading restriction tab")
; restrict_tabc = restrict_tabr
; mcxIOclose(xf_restrict_tab)
; }
else
{ if (xf_restrict_tabr)
{ if (!(restrict_tabr = mclTabRead(xf_restrict_tabr, NULL, RETURN_ON_FAIL)))
mcxDie(1, me, "error reading restriction tabr")
; mcxIOclose(xf_restrict_tabr)
; }
if (xf_restrict_tabc)
{ if (!(restrict_tabc = mclTabRead(xf_restrict_tabc, NULL, RETURN_ON_FAIL)))
mcxDie(1, me, "error reading restriction tabc")
; mcxIOclose(xf_restrict_tabc)
; }
}
}
/* fixme: restructure code to include bit below */
if (write_tabc || write_tabr)
{ mclv* dom_cols = mclvInit(NULL)
; mclv* dom_rows = mclvInit(NULL)
; mclv* dom = write_tabc ? dom_cols : dom_rows
; if (!(tabc = mclTabRead(xf_tab, NULL, RETURN_ON_FAIL)))
mcxDie(1, me, "error reading tab file")
; if (mclxReadDomains(xf_mx, dom_cols, dom_rows))
mcxDie(1, me, "error reading matrix file")
; mcxIOclose(xf_mx)
/* fixme check status */
; mclTabWrite(tabc, xfout, dom, RETURN_ON_FAIL)
; mcxIOclose(xfout)
; return 0
; }
if (newick)
{ mcxTing* thetree
; mclxCat cat
; if (xf_tab && !(tabr = mclTabRead(xf_tab, NULL, RETURN_ON_FAIL)))
mcxDie(1, me, "error reading tab file")
; mclxCatInit(&cat)
; if
( mclxCatRead
( xf_mx
, &cat
, 0
, NULL
, tabr ? tabr->domain : NULL
, MCLX_CATREAD_CLUSTERTREE | MCLX_ENSURE_ROOT
)
)
mcxDie(1, me, "failure reading file")
; thetree = mclxCatNewick(&cat, tabr, newick_bits)
; fwrite(thetree->str, 1, thetree->len, xfout->fp)
; fputc('\n', xfout->fp)
; mcxIOclose(xfout)
; return 0
; }
while (1)
{ mclxIOdumper dumper
; mclxCat cat
; dim i
; if (xf_tab && !lazy_tab)
cat_bits |= MCLX_REQUIRE_GRAPH
; mclxCatInit(&cat)
; if (mclxCatRead(xf_mx, &cat, catmax, NULL, NULL, cat_bits))
break
; for (i=0;i<cat.n_level;i++)
{ mclx* mx = cat.level[i].mx
; if (restrict_tabr || restrict_tabc)
{ mclx* sub
; sub
= mclxSub
( mx
, restrict_tabc
? restrict_tabc->domain
: mx->dom_cols
, restrict_tabr
? restrict_tabr->domain
: mx->dom_rows
)
; mx = sub
; }
/* noteme fixme dangersign mx now may violate some 'cat' invariant */
if (sort_mode)
{ if (!strcmp(sort_mode, "size-ascending"))
mclxColumnsRealign(mx, mclvSizeCmp)
; else if (!strcmp(sort_mode, "size-descending"))
mclxColumnsRealign(mx, mclvSizeRevCmp)
; else
mcxErr(me, "unknown sort mode <%s>", sort_mode)
; if (catmax != 1)
mcxErr(me, "-sort option and cat mode may fail or corrupt")
; }
if (xf_tab && !tabr)
{ if (!( tabr = mclTabRead
(xf_tab, lazy_tab ? NULL : mx->dom_rows, RETURN_ON_FAIL)
) )
mcxDie(1, me, "consider using --lazy-tab option")
; tabc = tabr
; mcxIOclose(xf_tab)
; }
else
{ if (!tabr && xf_tabr)
{ if (!(tabr = mclTabRead
(xf_tabr, lazy_tab ? NULL : mx->dom_rows, RETURN_ON_FAIL)
) )
mcxDie(1, me, "consider using --lazy-tab option")
; mcxIOclose(xf_tabr)
; }
if (!tabc && xf_tabc)
{ if (!( tabc = mclTabRead
(xf_tabc, lazy_tab ? NULL : mx->dom_cols, RETURN_ON_FAIL)
) )
mcxDie(1, me, "consider using --lazy-tab option")
; mcxIOclose(xf_tabc)
; }
}
; if (transpose)
{ mclx* tp = mclxTranspose(mx)
; mclxFree(&mx)
; mx = tp
; if (tabc || tabr)
{ mclTab* tabt = tabc
; tabc = tabr
; tabr = tabt
; }
}
if (mode_dump == MCLX_DUMP_TABLE)
BIT_ON(modes, MCLX_DUMP_TABLE_HEADER)
; mclxIOdumpSet(&dumper, modes, sep_lead_g, sep_row_g, sep_val_g)
; dumper.table_nlines = table_nlines
; dumper.table_nfields = table_nfields
; dumper.prefixc = prefixc_g
; if (split_stem)
{ mcxTing* ting = mcxTingPrint(NULL, "%s.%03d", split_stem, split_idx)
; mcxIOclose(xfout)
; mcxIOrenew(xfout, ting->str, "w")
; split_idx += split_inc
; }
if
( mclxIOdump
( mx
, xfout
, &dumper
, tabc
, tabr
, digits
, RETURN_ON_FAIL
) )
mcxDie(1, me, "something suboptimal")
; mclxFree(&mx)
; if (sep_cat_g && i+1 < cat.n_level)
fprintf(xfout->fp, "%s\n", sep_cat_g)
; }
break
; }
mcxIOfree(&xf_mx)
; mcxIOfree(&xfout)
; mcxIOfree(&xf_tab)
; mcxIOfree(&xf_tabr)
; mcxIOfree(&xf_tabc)
; mcxTingFree(&line)
; return 0
; }