int fera___init(struct descriptor_s *niddsc_ptr, InInitStruct *setup )
{
int status=1;
int num_dig;
int num_mem;
#define return_on_error(f,retstatus) if (!((status = f) & 1)) return retstatus;
last_nid = 0;
/* Reset the controler */
return_on_error(DevCamChk(CamPiow(setup->cntrl_name,0,9,0,16,0),&one,&one),status);
/* setup the digitizers */
if((num_dig = NElements(setup->head_nid + FERA_N_DIG_NAME))) {
int i;
for(i=0; i<num_dig; i++) {
char *name = ArrayRef(setup->head_nid + FERA_N_DIG_NAME, i);
if(name) {
static short ecl_cam = 0x2400;
return_on_error(DevCamChk(CamPiow(name,0,9,0,16,0),&one,&one),status); /* reset digitizer */
return_on_error(DevCamChk(CamPiow(name,0,16, &ecl_cam, 16, 0),&one,&one),status); /* goto ECL/CAM mode */
}
else {
status = FERA$_DIGNOTSTRARRAY;
break;
}
}
}
else
status = FERA$_NODIG;
/* setup the memory modules */
if (status&1) {
if ((num_mem = NElements(setup->head_nid + FERA_N_MEM_NAME)) ) {
int i;
for(i=0; i<num_mem; i++) {
char *name = ArrayRef(setup->head_nid + FERA_N_MEM_NAME, i);
if(name) {
return_on_error(DevCamChk(CamPiow(name, 1, 17, &one, 16, 0),&one,&one),status);
return_on_error(DevCamChk(CamPiow(name, 0, 24, 0, 16, 0),&one,&one),status);
return_on_error(DevCamChk(CamPiow(name, 0, 17, &zero, 16, 0),&one,&one),status);
return_on_error(DevCamChk(CamPiow(name, 1, 17, &three, 16, 0),&one,&one),status);
}
else {
status = FERA$_MEMNOTSTRARRAY;
break;
}
}
}
else
status = FERA$_NOMEM;
}
/* Reset the controler */
return_on_error(DevCamChk(CamPiow(setup->cntrl_name, 0, 9, 0, 16, 0),&one,&one),status);
return status;
}
int add_vector(GapIO *io, char *V, int level)
{
int err;
int vector;
int freerec;
GVectors v;
/* allocate vector name */
v.name = allocate(io,GT_Text);
err = TextWrite(io,v.name,V,strlen(V));
v.level = level;
/* find new vector number */
vector = ++io->db.Nvectors;
ArrayRef(io->vectors,vector-1);
/* add vector to list */
arr(GCardinal,io->vectors,vector-1) = freerec = allocate(io,GT_Vectors);
err = GT_Write(io,freerec,&v,sizeof(v),GT_Vectors);
/* write array */
err = ArrayDelay(io, io->db.vectors, io->db.Nvectors, io->vectors);
/* write db */
DBDelayWrite(io);
return vector;
}
int add_clone(GapIO *io, char *CN, char *CV)
{
int err;
int clone;
int freerec;
GClones c;
/* find vector record */
c.vector = find_vector(io,CV);
if (!c.vector) c.vector = add_vector(io,CV,1);
/* allocate clone name */
c.name = allocate(io,GT_Text);
err = TextWrite(io,c.name,CN,strlen(CN));
/* find new clone number */
clone = ++io->db.Nclones;
ArrayRef(io->clones,clone-1);
/* add clone to list */
arr(GCardinal,io->clones,clone-1) = freerec = allocate(io,GT_Clones);
err = GT_Write(io,freerec,&c,sizeof(c),GT_Clones);
/* write array */
err = ArrayDelay(io, io->db.clones, io->db.Nclones, io->clones);
/* write db */
DBDelayWrite(io);
return clone;
}
/*
* Adds a contig to a scaffold array.
* Gap size, type and evidence refer to the gap between this and the
* "previous" contig - ie the last in the scaffold. More complex
* scaffold manipulations will be handled elsewhere.
*
* Set these fields to 0 if you do not know them.
*
* Returns 0 on success
* -1 on failure
*/
int scaffold_add(GapIO *io, tg_rec scaffold, tg_rec contig,
int gap_size, int gap_type, int evidence) {
scaffold_t *f;
contig_t *c;
scaffold_member_t *m;
int i;
/* Check if this contig is in a scaffold, if so remove now */
c = cache_search(io, GT_Contig, contig);
if (c->scaffold)
scaffold_remove(io, c->scaffold, contig);
if (!(f = cache_search(io, GT_Scaffold, scaffold)))
return -1;
/* Check if it already exists */
for (i = 0; i < ArrayMax(f->contig); i++) {
m = arrp(scaffold_member_t, f->contig, i);
if (m->rec == contig)
return 0;
}
/* Append */
f = cache_rw(io, f);
m = ArrayRef(f->contig, ArrayMax(f->contig)); // extend
m->rec = contig;
m->gap_size = ArrayMax(f->contig) > 1 ? gap_size : 0;
m->gap_type = gap_type;
m->evidence = evidence;
/* Update the contig record too */
c = cache_search(io, GT_Contig, contig);
c = cache_rw(io, c);
c->scaffold = scaffold;
#if 0
/* Add a scaffold link to the contig graph too */
if (ArrayMax(f->contig) >= 2) {
m = arrp(scaffold_member_t, f->contig, ArrayMax(f->contig)-2);
contig_link_t lnk;
lnk.rec1 = contig;
lnk.rec2 = m->rec;
/* Best guess */
lnk.pos1 = 0; lnk.end1 = 1;
lnk.pos2 = 0; lnk.end2 = 0;
lnk.orientation = 0;
lnk.size = 100;
lnk.type = CLINK_TYPE_SCAFFOLD;
lnk.score = 0;
contig_add_link(io, &lnk);
}
#endif
return 0;
}
/*
* Translates a Read structure and an Experiment file.
* The Read structure is left unchanged.
*
* Returns:
* A pointer to an allocated Exp_info structure upon success.
* NULL upon failure.
*/
Exp_info *read2exp(Read *read, char *EN) {
Exp_info *e;
char *t = trace_type_int2str(read->format), *p;
int l = strlen(EN)+1;
char *sq;
int i;
static char valid_bases[256];
static int valid_setup = 0;
if (!valid_setup) {
for (i = 0; i < 256; i++)
valid_bases[i] = '-';
/* IUBC codes */
for (sq = "acgturymkswbdhvnACGTURYMKSWBDHVN"; *sq; sq++)
valid_bases[(unsigned)*sq] = *sq;
valid_setup = 1;
}
if (NULL == (e = exp_create_info()))
return NULL;
/* Copy original exp file if present */
if (read->orig_trace && read->orig_trace_format == TT_EXP) {
int i, j, k;
Exp_info *re = (Exp_info *)read->orig_trace;
for (i = 0; i < MAXIMUM_EFLTS; i++) {
if (EFLT_SQ == i ||
EFLT_QL == i ||
EFLT_QR == i)
continue;
if (0 == (k = exp_Nentries(re, i)))
continue;
e->Nentries[i] = k;
ArrayRef(e->entries[i], e->Nentries[i]);
for (j = 0; j < k; j++) {
arr(char *, e->entries[i], j) =
strdup(arr(char *, re->entries[i], j));
}
}
/* Otherwise create our EN, ID, LN and LT lines */
} else {
GView g_new_view(GDB *gdb)
/*
* allocate a new view
*/
{
GView i;
i = gdb->free_view;
if (i==-1) {
(void)ArrayRef(gdb->view,gdb->Nview);
i = gdb->Nview++;
} else {
/* adjust freetree */
gdb->free_view = arr(View,gdb->view,i).next;
}
arr(View,gdb->view,i).next = -1;
arr(View,gdb->view,i).flags = G_VIEW_NEW;
arr(View,gdb->view,i).lcache.rec = -1;
return i;
}
GDB *g_open_database_(char *fns[], GCardinal Nfns, int read_only)
/*
* Open a database with a given file name
*/
{
GDB *gdb;
GCardinal i;
/* check arguments */
if (fns==NULL) {
(void) gerr_set(GERR_INVALID_ARGUMENTS);
return NULL;
}
if (NULL == (gdb = g_new_gdb()))
return NULL;
/*
* initialise data structs for clients
*/
gdb->Nclient = G_MAX_CLIENTS;
if ( (gdb->client = ArrayCreate(sizeof(Client),gdb->Nclient)) == NULL ) {
g_free_gdb(gdb);
(void)gerr_set(GERR_OUT_OF_MEMORY);
return NULL;
}
(void)ArrayRef(gdb->client,gdb->Nclient-1);
for (i=0;i<gdb->Nclient;i++) arr(Client,gdb->client,i).id = -1;
/*
* Open the file (now only 1 file allowed, we ignore the rest)
*/
gdb->gfile = g_open_file(fns[0], read_only);
if (NULL == gdb->gfile) {
g_free_gdb(gdb);
/* g_open_file sets gerrnum */
return NULL;
}
/*
* allocate views - assume using Gap4, which locks all records.
* FIXME: Changed to have max 1000 views at startup. It's reallocated
* as we go anyway.
*/
gdb->Nview = 1000 < gdb->gfile->header.num_records
? 1000 : gdb->gfile->header.num_records;
if ( (gdb->view = ArrayCreate(sizeof(View),gdb->Nview)) == NULL ) {
g_free_gdb(gdb);
(void)gerr_set(GERR_OUT_OF_MEMORY);
return NULL;
}
(void)ArrayRef(gdb->view,gdb->Nview-1);
/* initialise views */
for(i=0;i<gdb->Nview;i++) {
arr(View,gdb->view,i).next = (i-1);
arr(View,gdb->view,i).flags = G_VIEW_NEW;
}
gdb->free_view = gdb->Nview-1;
return gdb;
}
Exp_info *exp_read_staden_info(mFILE *fp, char *filename)
/*
* Read a staden file into an Exp_info data structure
*/
{
Exp_info *e;
char *seq;
char *fn;
/* find last / in filename */
for(fn=filename+strlen(filename)-1;fn>filename && *fn!='/';fn--);
if (*fn=='/') fn++;
if ( (e = exp_create_info()) != NULL ) {
char line[128];
/* an upper limit for the file size */
int max_seq_size = file_size(filename);
int left, right, len, i;
int lineno;
int formatIsBap;
int CS_from, CS_to;
int SR;
CS_from = CS_to = SR = 0;
/*ID*/
(void)ArrayRef(e->entries[EFLT_ID],e->Nentries[EFLT_ID]++);
exp_get_entry(e,EFLT_ID) = (char *)strdup(fn);
/*EN*/
(void)ArrayRef(e->entries[EFLT_EN],e->Nentries[EFLT_EN]++);
exp_get_entry(e,EFLT_EN) = (char *)strdup(fn);
/*CC*/
(void)ArrayRef(e->entries[EFLT_CC],e->Nentries[EFLT_CC]++);
exp_get_entry(e,EFLT_CC) = (char *)strdup("Created from a staden format sequence assembly file");
seq = (char *) xmalloc(max_seq_size+1);
if (!seq)
return NULL;
left = 0;
right = 0;
len = 0;
lineno = 0;
formatIsBap = 1;
while (mfgets(line,sizeof(line),fp)!=NULL) {
char *s;
lineno++;
if (lineno == 1) {
int pos, ret;
char *cp;
/*
* This maybe a fasta format file.
*/
if (line[0] == '>') {
if (cp = strchr(line, ' '))
*cp = 0;
if (cp = strchr(line, '\t'))
*cp = 0;
if (cp = strchr(line, '\n'))
*cp = 0;
exp_set_entry(e, EFLT_ID, strdup(line+1));
exp_set_entry(e, EFLT_EN, strdup(line+1));
continue;
}
/*
* This maybe a file created from 'output consensus',
* in which case it'll have the Staden format style of:
* " <-----T2.00047----->" header on the first line.
*
* We strip this off. Argh - why don't Suns have the
* ANSI function memmove() ?
*/
ret = sscanf(line, " <%*18s>%n", &pos);
if (ret && pos == 21) {
int i, e = sizeof(line)-21;
for (i=0; i < e; i++)
line[i] = line[i+21];
/* memmove((void *)line,
* (void *)(line + 21),
* sizeof(line)-21);
*/
}
}
if (line[0] == ';') {
/********************************************
* Title line parsing
*******************************************/
if (lineno==1 &&
!(line[1] == ';' || line[1] == '<' || line[1] == '>')
) {
/* format of line is:
* <ln> ::= ;<i6><i6><i6><c4><c+>
* <i6> ::= <s><s><s><s><s><d> | ... <d><d><d><d><d><d>
* <c4> ::= <C><C><C><s>
* <c+> ::= <a><c+> | <a>
* <a> is [a-zA-Z0-9.]
* <s> is ' '
* <d> is [0-9]
* <C> is [A-Z]
*/
int d;
if (sscanf(line,";%6d%6d%6d",&d,&d,&d)==3 &&
strlen(line)>23) {
/* trim off trailing white space */
trim_white_space(line+23);
/*LN*/
(void)ArrayRef(e->entries[EFLT_LN],e->Nentries[EFLT_LN]++);
exp_get_entry(e,EFLT_LN) = (char *)strdup(line+23); line[23]='\0';
/*LT*/
(void)ArrayRef(e->entries[EFLT_LT],e->Nentries[EFLT_LT]++);
/* trim off trace type white space */
trim_white_space(line+19);
exp_get_entry(e,EFLT_LT) = (char *)strdup(line+19);
}
} else if (formatIsBap) {
switch (line[1]) {
case '<':
for(s=&line[2];*s;s++) {
if(!isspace(*s) && isprint(*s))
seq[left++] = *s;
}
break;
case '>':
for(s=&line[2];*s;s++) {
if(!isspace(*s) && isprint(*s))
seq[max_seq_size-right++] = *s;
}
break;
case ';':
/*TG*/
#if 0
trim_white_space(line);
(void)ArrayRef(e->entries[EFLT_TG],e->Nentries[EFLT_TG]++);
/* convert format from Staden format to
* Experiment file format
*/
{
char *cp;
int pos, len;
char type[5];
cp = (char *)xmalloc(strlen(line)+20);
if (cp == NULL)
break;
sscanf(line, ";;%4s %6d %6d",
type, &pos, &len);
/*
* Need to add 'left' to each start position
* in tag. ASSUMPTION: 'left' has already been
* defined. Ie that the ;< lines are before
* any ;; lines.
*/
pos += left;
values2tag(cp, type, pos, pos + len - 1,
2, &line[20]);
exp_get_entry(e,EFLT_TG) = cp;
}
if (strncmp(line+2,"IGNC",4)==0 ||
strncmp(line+2,"CVEC",4)==0) {
CS_from = atoi(&line[2+4+1]);
} else if (strncmp(line+2,"IGNS",4)== 0 ||
strncmp(line+2,"SVEC",4)== 0) {
SR = 1;
}
#endif
break;
default:
break;
}
}
} else {
/********************************************
* The actual sequence bit
*******************************************/
formatIsBap = 0; /* turn off title line parsing stuff */
for (s=line;*s;s++) {
if(!isspace(*s) && isprint(*s))
seq[left+len++] = *s;
}
}
}
/*
* The right cutoff has been stashed into the end of the array
* Move to correct place
*/
for(i=(max_seq_size-(left+len))/2;i>=0;i--) {
char temp;
/* swap */
temp = seq[left+len+i];
seq[left+len+i] = seq[max_seq_size-i];
seq[max_seq_size-i] = temp;
}
/* null terminate */
seq[left+len+right] = '\0';
/*SQ*/
(void)ArrayRef(e->entries[EFLT_SQ],e->Nentries[EFLT_SQ]++);
exp_get_entry(e,EFLT_SQ) = seq;
/*SL*/
sprintf(line,"%d",left);
(void)ArrayRef(e->entries[EFLT_SL],e->Nentries[EFLT_SL]++);
exp_get_entry(e,EFLT_SL) = (char *)strdup(line);
/*SR*/
if (SR) {
sprintf(line,"%d",left+len+1);
(void)ArrayRef(e->entries[EFLT_SR],e->Nentries[EFLT_SR]++);
exp_get_entry(e,EFLT_SR) = (char *)strdup(line);
}
/*CS*/
if (CS_from) {
if (CS_from == 1) {
CS_to = left;
} else {
CS_from = left + len + 1;
CS_to = left + len + right;
}
sprintf(line,"%d..%d",CS_from,CS_to);
(void)ArrayRef(e->entries[EFLT_CS],e->Nentries[EFLT_CS]++);
exp_get_entry(e,EFLT_CS) = (char *)strdup(line);
}
/*QR*/
if (!SR && !CS_from) {
sprintf(line,"%d",left+len+1);
(void)ArrayRef(e->entries[EFLT_QR],e->Nentries[EFLT_QR]++);
exp_get_entry(e,EFLT_QR) = (char *)strdup(line);
}
#if 0
/*TG*/
{
int i;
/* need to add LEFT to each start position in tag */
for(i=0;i<e->Nentries[EFLT_TG];i++) {
sprintf(line,"%4.4s %6d%s",
arr(char *,e->entries[EFLT_TG],i),
atoi(arr(char *,e->entries[EFLT_TG],i)+5)+left,
arr(char *,e->entries[EFLT_TG],i)+11);
xfree(arr(char *,e->entries[EFLT_TG],i));
arr(char *,e->entries[EFLT_TG],i) = (char *)strdup(line);
}
}
#endif
}
/*
* A recursive break contig function.
* bin_num The current bin being moved or split.
* pos The contig break point.
* offset The absolute positional offset of this bin in original contig
* pleft The parent bin/contig record num in the left new contig
* pright The parent bin/contig record num in the right new contig
* child_no 0 or 1 - whether this bin is the left/right child of its parent
*/
static int break_contig_recurse(GapIO *io, HacheTable *h,
contig_t *cl, contig_t *cr,
tg_rec bin_num, int pos, int offset,
int level, tg_rec pleft, tg_rec pright,
int child_no, int complement) {
int i, j, f_a, f_b;
tg_rec rbin;
bin_index_t *bin = get_bin(io, bin_num), *bin_dup ;
//int bin_min, bin_max;
int nseqs;
tg_rec opright; /* old pright, needed if we revert back */
cache_incr(io, bin);
if (bin->flags & BIN_COMPLEMENTED) {
complement ^= 1;
}
if (complement) {
f_a = -1;
f_b = offset + bin->size-1;
} else {
f_a = +1;
f_b = offset;
}
printf("%*sBreak offset %d pos %d => test bin %"PRIrec": %d..%d\n",
level*4, "",
offset, pos, bin->rec,
NMIN(bin->start_used, bin->end_used),
NMAX(bin->start_used, bin->end_used));
bin = cache_rw(io, bin);
nseqs = bin->nseqs;
bin->nseqs = 0;
/* Invalidate any cached data */
bin_invalidate_track(io, bin, TRACK_ALL);
if (bin->flags & BIN_CONS_VALID) {
bin->flags |= BIN_BIN_UPDATED;
bin->flags &= ~BIN_CONS_VALID;
}
//bin_min = bin->rng ? NMIN(bin->start_used, bin->end_used) : offset;
//bin_max = bin->rng ? NMAX(bin->start_used, bin->end_used) : offset;
/*
* Add to right parent if this bin is to the right of pos,
* or if the used portion is to the right and we have no left child.
*
* FIXME: Not a valid assumption!
* The used portion of a bin is not a placeholder for the used portion
* of all the the children beneath it. Therefore if the used portion of
* this bin is > pos (and we have no left child) it still doesn't mean
* that the absolute positions of the used portion of the right child
* won't be < pos.
*/
if (offset >= pos /*|| (bin_min >= pos && !bin->child[0])*/) {
printf("%*sADD_TO_RIGHT pl=%"PRIrec" pr=%"PRIrec"\n",
level*4, "", pleft, pright);
if (0 != break_contig_move_bin(io, bin,
cl, pleft, cr, pright,
child_no))
return -1;
bin_incr_nseq(io, bin, nseqs);
cache_decr(io, bin);
return 0;
}
/*
* Add to left parent if this bin is entirely to the left of pos,
* or if the used portion is to the left and we have no right child.
*/
if (offset + bin->size < pos /*|| (bin_max < pos && !bin->child[1])*/) {
printf("%*sADD_TO_LEFT\n", level*4, "");
//if (0 != break_contig_move_bin(io, bin, cr, pright, cl, pleft, child_no))
//return -1;
bin_incr_nseq(io, bin, nseqs);
cache_decr(io, bin);
return 0;
}
/*
* Nominally the bin overlaps both left and right and so needs duplicating.
* There are cases though at the roots of our trees where duplicating is
* unnecessary as it leads to empty bins at the root. In this case
* we skip creating a duplicate for the right, or alternatively steal
* the left root bin and use that instead.
*
* Similarly the range_t array will either be left where it is, moved to
* the right contig, or split in half (creating a new one for the right).
*
* FIXED: always need this. Eg:
*
* |-------------empty--------------|
* |----------------|---------------|
* |--------|-------|--------|------|
* ^
* |
* break here
*
* In this case we need to duplicate the parent as it overlaps the left
* bin, which may (or may not) have data that needs to end up in the right
* hand contig. Just duplicate for now and free later on if needed.
*/
if (1 /* always! */ || pright != cr->rec ||
(bin->rng && NMAX(bin->start_used, bin->end_used) >= pos)) {
//printf("NMAX=%d >= %d\n", NMAX(bin->start_used, bin->end_used), pos);
rbin = 0;
/* Possibly steal left contig's bin */
if (pleft == cl->rec && NMIN(bin->start_used, bin->end_used) >= pos) {
#if 0
/* Currently this doesn't always work */
if (bin->child[1]) {
bin_index_t *ch = get_bin(io, bin->child[1]);
if (NMIN(ch->pos, ch->pos + ch->size-1) >= pos) {
rbin = cl->bin;
cl->bin = bin->child[0];
}
}
#else
pleft = bin->rec;
#endif
} else {
pleft = bin->rec;
}
/* Create new bin, or use root of contig if it's unused so far */
if (!rbin && pright == cr->rec) {
rbin = cr->bin;
}
/* Otherwise we genuingly need a duplicate */
if (!rbin)
rbin = bin_new(io, 0, 0, 0, GT_Bin);
/* Initialise with duplicate values from left bin */
bin_dup = get_bin(io, rbin);
bin_dup = cache_rw(io, bin_dup);
bin_dup->size = bin->size;
bin_dup->pos = bin->pos;
bin_dup->parent = pright;
bin_dup->parent_type = (pright == cr->rec ? GT_Contig : GT_Bin);
bin_dup->flags = bin->flags | BIN_BIN_UPDATED;
bin_dup->start_used = bin->start_used;
bin_dup->end_used = bin->end_used;
/*
* Shift bin to offset if it's the contig root.
* It'll be shifted back by the correct amount later.
*/
if (pright == cr->rec) {
printf("moving root bin to offset=%d comp=%d\n", offset, complement);
bin_dup->pos = offset;
}
printf("%*sCreated dup for right, rec %"PRIrec"\n",
level*4,"", bin_dup->rec);
break_contig_move_bin(io, bin_dup, cl, 0, cr, pright, child_no);
opright = pright;
pright = bin_dup->rec;
} else {
bin_dup = NULL;
pleft = bin->rec;
}
if (!bin->rng) {
/* Empty bin */
printf("%*sEMPTY range\n", level*4, "");
bin->start_used = bin->end_used = 0;
bin->flags |= BIN_BIN_UPDATED;
if (bin_dup) {
bin_dup->start_used = bin_dup->end_used = 0;
bin_dup->flags |= BIN_BIN_UPDATED;
}
} else if (NMIN(bin->start_used, bin->end_used) >= pos) {
/* Move range to right contig */
printf("%*sDUP %"PRIrec", MOVE Array to right\n",
level*4, "", bin_dup->rec);
bin_dup->rng = bin->rng;
bin_dup->rng_rec = bin->rng_rec;
bin_dup->rng_free = bin->rng_free;
if (bin_dup->rng_rec)
bin_dup->flags |= BIN_RANGE_UPDATED;
if (bin->rec != bin_dup->rec) {
bin->rng = NULL;
bin->rng_rec = 0;
bin->rng_free = -1;
bin->flags |= BIN_BIN_UPDATED;
}
bin->start_used = bin->end_used = 0;
break_contig_reparent_seqs(io, bin_dup);
if (bin_dup->rng) {
int n = ArrayMax(bin_dup->rng);
for (i = j = 0; i < n; i++) {
range_t *r = arrp(range_t, bin_dup->rng, i), *r2;
if (r->flags & GRANGE_FLAG_UNUSED)
continue;
if ((r->flags & GRANGE_FLAG_ISMASK) != GRANGE_FLAG_ISANNO) {
HacheData hd; hd.i = 1;
HacheTableAdd(h, (char *)&r->rec, sizeof(r->rec), hd,NULL);
j++;
}
}
bin_incr_nseq(io, bin_dup, j);
}
} else if (NMAX(bin->start_used, bin->end_used) < pos) {
/* Range array already in left contig, so do nothing */
printf("%*sMOVE Array to left\n", level*4, "");
if (bin_dup)
bin_dup->start_used = bin_dup->end_used = 0;
if (bin->rng) {
int n = ArrayMax(bin->rng);
for (i = j = 0; i < n; i++) {
range_t *r = arrp(range_t, bin->rng, i);
if (r->flags & GRANGE_FLAG_UNUSED)
continue;
if ((r->flags & GRANGE_FLAG_ISMASK) != GRANGE_FLAG_ISANNO) {
HacheData hd; hd.i = 0;
HacheTableAdd(h, (char *)&r->rec, sizeof(r->rec), hd,NULL);
j++;
}
}
bin_incr_nseq(io, bin, j);
}
} else {
/* Range array covers pos, so split in two */
int n, nl = 0, nr = 0;
int lmin = bin->size, lmax = 0, rmin = bin->size, rmax = 0;
printf("%*sDUP %"PRIrec", SPLIT array\n", level*4, "", bin_dup->rec);
bin->flags |= BIN_RANGE_UPDATED;
bin_dup->flags |= BIN_RANGE_UPDATED;
bin_dup->rng = ArrayCreate(sizeof(range_t), 0);
bin_dup->rng_free = -1;
/* Pass 1 - hash sequences */
n = ArrayMax(bin->rng);
for (i = 0; i < n; i++) {
range_t *r = arrp(range_t, bin->rng, i);
int cstart; /* clipped sequence positions */
seq_t *s;
if (r->flags & GRANGE_FLAG_UNUSED)
continue;
if ((r->flags & GRANGE_FLAG_ISMASK) == GRANGE_FLAG_ISANNO)
continue;
s = (seq_t *)cache_search(io, GT_Seq, r->rec);
if ((s->len < 0) ^ complement) {
cstart = NMAX(r->start, r->end) - (s->right-1);
} else {
cstart = NMIN(r->start, r->end) + s->left-1;
}
if (cstart >= pos) {
HacheData hd; hd.i = 1;
HacheTableAdd(h, (char *)&r->rec, sizeof(r->rec), hd, NULL);
} else {
HacheData hd; hd.i = 0;
HacheTableAdd(h, (char *)&r->rec, sizeof(r->rec), hd, NULL);
}
}
/* Pass 2 - do the moving of anno/seqs */
n = ArrayMax(bin->rng);
for (i = j = 0; i < n; i++) {
range_t *r = arrp(range_t, bin->rng, i), *r2;
int cstart; /* clipped sequence positions */
if (r->flags & GRANGE_FLAG_UNUSED)
continue;
if ((r->flags & GRANGE_FLAG_ISMASK) == GRANGE_FLAG_ISANNO) {
cstart = NMAX(r->start, r->end);
} else {
seq_t *s = (seq_t *)cache_search(io, GT_Seq, r->rec);
if ((s->len < 0) ^ complement) {
cstart = NMAX(r->start, r->end) - (s->right-1);
} else {
cstart = NMIN(r->start, r->end) + s->left-1;
}
}
if (cstart >= pos &&
((r->flags & GRANGE_FLAG_ISMASK) == GRANGE_FLAG_ISANNO)) {
anno_ele_t *a = (anno_ele_t *)cache_search(io,
GT_AnnoEle,
r->rec);
/* If it's an annotation on a sequence < pos then we
* still don't move.
*
* FIXME: we have no guarantee that the sequence being
* annotated is in the same bin as this annotation, as
* they may be different sizes and end up in different
* bins. (Should we enforce anno always in same bin as seq?
* If so, consensus annos fit anywhere?)
*/
if (a->obj_type == GT_Seq) {
HacheItem *hi = HacheTableSearch(h,
(char *)&r->pair_rec,
sizeof(r->pair_rec));
if (hi) {
if (hi->data.i == 0)
cstart = pos-1;
} else {
puts("FIXME: annotation for seq in unknown place - "
"work out correct location and move if needed.");
}
}
}
if (cstart >= pos) {
r2 = (range_t *)ArrayRef(bin_dup->rng, ArrayMax(bin_dup->rng));
*r2 = *r;
if (rmin > r->start) rmin = r->start;
if (rmin > r->end) rmin = r->end;
if (rmax < r->start) rmax = r->start;
if (rmax < r->end) rmax = r->end;
if ((r->flags & GRANGE_FLAG_ISMASK) == GRANGE_FLAG_ISSEQ)
nr++;
} else {
if (lmin > r->start) lmin = r->start;
if (lmin > r->end) lmin = r->end;
if (lmax < r->start) lmax = r->start;
if (lmax < r->end) lmax = r->end;
if (j != i) {
r2 = arrp(range_t, bin->rng, j);
*r2 = *r;
}
j++;
if ((r->flags & GRANGE_FLAG_ISMASK) == GRANGE_FLAG_ISSEQ)
nl++;
}
}
bin_incr_nseq(io, bin, nl);
bin_incr_nseq(io, bin_dup, nr);
ArrayMax(bin->rng) = j;
#if 0
/*
* Right now this causes problems, but I'm not sure why. Try again
* after we've fixed the bin->nseqs issues and other deallocation
* woes.
*/
if (ArrayMax(bin_dup->rng) == 0 && bin_dup->parent_type == GT_Bin) {
/* We didn't need it afterall! Odd. */
bin_index_t *pb;
printf("Purging bin %d that we didn't need afterall\n",
bin_dup->rec);
cache_rec_deallocate(io, GT_Bin, bin_dup->rec);
pb = cache_search(io, GT_Bin, bin_dup->parent);
if (pb->child[0] == bin_dup->rec)
pb->child[0] = 0;
if (pb->child[1] == bin_dup->rec)
pb->child[1] = 0;
bin_dup = NULL;
pright = opright;
}
#endif
if (bin_dup)
break_contig_reparent_seqs(io, bin_dup);
if (lmin < lmax) {
bin->start_used = lmin;
bin->end_used = lmax;
} else {
/* No data left in bin */
bin->start_used = 0;
bin->end_used = 0;
}
printf("%*sLeft=>%d..%d right=>%d..%d\n", level*4, "",
lmin, lmax, rmin, rmax);
if (bin_dup) {
if (rmin < rmax) {
bin_dup->start_used = rmin;
bin_dup->end_used = rmax;
} else {
/* No data moved in bin */
bin_dup->start_used = 0;
bin_dup->end_used = 0;
}
}
}
/* Recurse */
for (i = 0; i < 2; i++) {
bin_index_t *ch;
if (!bin->child[i])
continue;
ch = get_bin(io, bin->child[i]);
if (0 != break_contig_recurse(io, h, cl, cr, bin->child[i], pos,
NMIN(ch->pos, ch->pos + ch->size-1),
level+1, pleft, pright,
i, complement))
return -1;
}
cache_decr(io, bin);
// if (bin_dup)
// cache_decr(io, bin_dup);
return 0;
}
/*
* Dumps a G database in an ASCII readable form. Modelled around the start
* of the g_open_file() routine.
*/
void g_dump_file(char *fn) {
GFile *gfile = NULL;
char fnaux[1024];
AuxIndex aux_ind;
int i;
#define ABORT(E)\
{\
g_free_gfile(gfile); \
gfile = NULL; \
(void)gerr_set(E); \
perror("ABORT"); \
return; \
}
/* check file name isn't too long */
if (strlen(fn) + strlen(G_AUX_SUFFIX) >= sizeof(fnaux))
ABORT(GERR_NAME_TOO_LONG);
strcpy(fnaux, fn);
strcat(fnaux, G_AUX_SUFFIX);
/* allocate new data structure - GFile */
gfile = g_new_gfile();
if (gfile == NULL)
ABORT(GERR_OUT_OF_MEMORY);
/* set file name */
if ((gfile->fname = (char *)xmalloc(strlen(fn)+1)) != NULL)
strcpy(gfile->fname, fn);
/* open file and its aux */
/* LOW LEVEL IO HERE */
if ((gfile->fd = open(fn, O_RDONLY)) == -1)
ABORT(GERR_OPENING_FILE);
/* LOW LEVEL IO HERE */
if ((gfile->fdaux = open(fnaux, O_RDONLY)) == -1)
ABORT(GERR_OPENING_FILE);
/* LOW LEVEL IO HERE */
if (-1 == lseek(gfile->fdaux, 0, 0))
ABORT(GERR_SEEK_ERROR);
if (g_read_aux_header(gfile->fdaux, &gfile->header))
ABORT(GERR_READ_ERROR);
printf("** \n");
printf("** Opening file %s\n",fn);
printf("** file_size = %"PRIGImage"\n", gfile->header.file_size);
printf("** block_size = %"PRIGCardinal"\n", gfile->header.block_size);
printf("** num_records = %"PRIGCardinal"\n", gfile->header.num_records);
printf("** max_records = %"PRIGCardinal"\n", gfile->header.max_records);
printf("** last_time = %"PRIGCardinal"\n", gfile->header.last_time);
printf("** flags = %"PRIGHFlags"\n", gfile->header.flags);
printf("** \n");
/* allocate index */
gfile->Nidx = gfile->header.num_records;
if ((gfile->idx = ArrayCreate(sizeof(Index), gfile->Nidx)) == NULL )
ABORT(GERR_OUT_OF_MEMORY);
(void) ArrayRef(gfile->idx, gfile->Nidx-1);
for(i = 0; i < gfile->Nidx; i++)
arr(Index, gfile->idx, i).flags = G_INDEX_NEW;
/* read aux index and initialise */
/* LOW LEVEL IO HERE */
if (-1 == lseek(gfile->fdaux, sizeof(AuxHeader), 0))
ABORT(GERR_SEEK_ERROR);
/* force Array.max field to be updated */
(void)ArrayRef(gfile->idx, gfile->header.num_records-1);
printf("global_time %08x\n", gfile->header.last_time);
for (i = 0; i < gfile->header.num_records; i++) {
char buf[MAX_BUF];
int toggle, len, len_r;
/* Load index for this record */
if (g_read_aux_index(gfile->fdaux, &aux_ind))
ABORT(GERR_READ_ERROR);
/* Compute toggle */
toggle = g_toggle_state(gfile->header.last_time, &aux_ind);
/* LOW LEVEL IO HERE */
if (-1 == lseek(gfile->fd, aux_ind.image[toggle], 0))
ABORT(GERR_SEEK_ERROR);
len = MIN(aux_ind.used[toggle], MAX_BUF);
/* LOW LEVEL IO HERE */
if (-1 == (len_r = read(gfile->fd, buf, len)))
ABORT(GERR_READ_ERROR);
if (len_r != len) {
fprintf(stderr, "WARNING: Read too short. Requested %d, got %d\n",
len, len_r);
}
printf("record %05d pos %020"PRIGImage" len %08d : %08x",
i, aux_ind.image[toggle], aux_ind.used[toggle],
(((((buf[0] << 8) + buf[1]) << 8) + buf[2]) << 8) + buf[3]);
if (len > 4)
printf(" %c%c%c%c%c%c%c%c\n",
isprint(buf[4])?buf[4]:'.',
isprint(buf[5])?buf[5]:'.',
isprint(buf[6])?buf[6]:'.',
isprint(buf[7])?buf[7]:'.',
isprint(buf[8])?buf[8]:'.',
isprint(buf[9])?buf[9]:'.',
isprint(buf[10])?buf[10]:'.',
isprint(buf[11])?buf[11]:'.');
else
putchar('\n');
}
#undef ABORT
g_free_gfile(gfile);
return;
}
int fera___store(struct descriptor_s *niddsc_ptr, InStoreStruct *setup)
{
int mem_status = 1;
int put_status = 1;
int status;
int num_dig;
int num_mem;
int i;
int num_chan;
int total_data;
int *num_pts;
unsigned short *buffer;
int ind;
#define return_on_error(f,retstatus) if (!((status = f) & 1)) return retstatus;
last_nid = 0;
num_dig = NElements(setup->head_nid + FERA_N_DIG_NAME);
if (num_dig == 0) return FERA$_NODIG;
num_chan = num_dig*16;
total_data = 0;
num_mem = NElements(setup->head_nid+FERA_N_MEM_NAME);
if (num_mem == 0) return FERA$_NOMEM;
num_pts = malloc(sizeof(int)*num_mem);
for(i=0; i<num_mem; i++) {
char *name = ArrayRef(setup->head_nid+FERA_N_MEM_NAME, i);
num_pts[i] = 0;
if (name) {
return_on_error(DevCamChk(CamPiow(name, 1, 17, &one, 16, 0),&one,&one),status);
return_on_error(DevCamChk(CamPiow(name, 0, 2, &num_pts[i], 16, 0),&one,&one),status);
if (num_pts[i] >= 16*1024) {
num_pts[i] = 16*1024;
if (i==num_mem-1)
mem_status = FERA$_OVER_RUN;
}
total_data += num_pts[i];
}
}
if (total_data % num_chan != 0) {
status = FERA$_PHASE_LOST;
return status;
}
buffer = malloc(total_data*sizeof(short));
for (i=0,ind=0; i< num_mem; ind+=num_pts[i],i++) {
if (num_pts[i]) {
char *name = ArrayRef(setup->head_nid+FERA_N_MEM_NAME, i);
if (name) {
return_on_error(DevCamChk(CamQstopw(name, 0, 2, num_pts[i], &buffer[ind], 16, 0), &one, &one), status)
}
}
}
status = Unpack(&buffer, num_pts, num_mem, num_chan, total_data/num_chan);
if ((status&1) || (status == FERA$_CONFUSED) || (status == FERA$_OVERFLOW))
put_status = Put(buffer, total_data/num_chan, num_chan, setup->head_nid + FERA_N_EXT_CLOCK, setup->head_nid +FERA_N_OUTPUT);
free(buffer);
return (put_status &1) ? ((status&1) ? mem_status : status) : put_status;
}