lcpbit file creationi completed

felipelouza · Jan 11, 2019 · 9a8e9cc · 9a8e9cc
1 parent 9b0876f
commit 9a8e9cc
Show file tree

Hide file tree

Showing 6 changed files with 32 additions and 24 deletions.
diff --git a/config.h b/config.h
@@ -135,7 +135,7 @@ typedef struct {
   FILE *unsortedLcp;       // if !NULL file containing unsorted LCP values
   FILE *unsortedLcp_size;  // if !NULL file containing the size of sorted blocks in unsorted_Lcp
   bool smallAlpha;         // the alphabet is small
-  int extMem;              // run in external memory: if ==1 only keep BWTs in external memory 
+  bool extMem;             // if true run in external memory
   int algorithm;           // preferred algorithm to use: gap8 gap16 gap128 gap256, if!=8,16,128,256 then use gap  
   bool mmapZ;              // mmap Z arrays
   bool mmapB;              // mmap B array

diff --git a/gap.c b/gap.c
@@ -61,8 +61,8 @@ int main(int argc, char *argv[]) {
   bool hm = false;
   g.unsortedLcp = NULL;
   g.outPath = NULL;
-  g.extMem = g.algorithm = 0;
-  g.smallAlpha=g.mmapZ=g.mmapBWT=g.mmapB= g.lcpMerge = g.lcpCompute = false;
+  g.algorithm = 0;
+  g.extMem = g.smallAlpha=g.mmapZ=g.mmapBWT=g.mmapB= g.lcpMerge = g.lcpCompute = false;
   g.outputDA = 0;
   g.dbOrder = 0;           // order for deBruijn graph 
   int num_threads = 0;
@@ -103,8 +103,7 @@ int main(int argc, char *argv[]) {
         num_threads = atoi(optarg);     // number of consumer threads 
         break;       
       case 'E':
-        g.extMem++; 
-        break;                    // use external memory (see mergegap.c)
+        g.extMem=true; break;           // use external memory (see mergegap.c)
       case 'Z':
         g.mmapZ=true; break;      // mmap merge and newmerge
       case 'B':
@@ -117,6 +116,20 @@ int main(int argc, char *argv[]) {
         exit(EXIT_FAILURE);
       }
   }
+  if(g.dbOrder<0 || g.dbOrder==1) { // illegal order
+    printf("dbOrder must be al least 2\n");
+    exit(EXIT_FAILURE);
+  }
+  if(g.dbOrder>1) { // valid order !=0
+    if(!g.extMem) {
+      printf("Option -D forces option -E\n");
+      g.extMem = true;
+    }
+    if(g.algorithm!=128) {
+      printf("Option -D forces option -A 128\n");
+      g.algorithm = 128;
+    }
+  }
   if(num_threads <0) {
     printf("Invalid number of threads, must be non negative\n");
     exit(EXIT_FAILURE);

diff --git a/io.c b/io.c
@@ -181,12 +181,11 @@ void bitfile_flush(bitfile *b)
 // init a bitfile: opening file and filling it with size zero bits
 // if temp==true the file is anonymous and immediately deleted, otherwise 
 // the file has .bitfile extension and maintained after the end of the computation   
-void bitfile_create(bitfile *b, size_t size, char *path, bool temp) {
+void bitfile_create(bitfile *b, size_t size, char *path, int order) {
   // create local copy of template
-  char s[strlen(path)+11];
-  if(temp) { // if this is a temp file create unique name 
+  char s[Filename_size];
+  if(order==0) { // if this is a temp file create unique name 
     sprintf(s,"%s.bitXXXXXX",path);
-    assert(strlen(s)==strlen(path) + 10);
     // get file descriptor for tmp file fill it with 0s and delete file  
     b->fd = mkstemp(s);
     if(b->fd == -1) die("bitfile_create: Tempfile creation failed");
@@ -196,11 +195,10 @@ void bitfile_create(bitfile *b, size_t size, char *path, bool temp) {
     if(e!=0)        die("bitfile_create: Tempfile unlink failed");
     }
   else { // we keep this file (to compute the DB-graph) 
-    sprintf(s,"%s.bitfile1",path);
-    assert(strlen(s)==strlen(path) + 9);
+    sprintf(s,"%s.%d.lcpbit1",path,order);
     // get file descriptor for bitfile fill it with 0s  
     b->fd = open(s,O_RDWR|O_CREAT|O_TRUNC, 0666);
-    if(b->fd == -1) die("bitfile_create: bitfileq creation failed");
+    if(b->fd == -1) die("bitfile_create: bitfile creation failed");
     int e = ftruncate(b->fd,(size+7)/8); // fill with size 0 bits
     if(e!=0)        die("bitfile_create: bitfile ftruncate failed");
   }
@@ -281,12 +279,12 @@ off_t bitfile_tell(bitfile *b) {
 
 // save hi bit of name to bitfile0
 // used for extracting info for dbGraph
-void extract_bitfile(char *name, size_t size, char *outpath)  
+void extract_bitfile(char *name, size_t size, char *outpath, int order)  
 {
   FILE *f = fopen(name,"rb");
   if(f==NULL) die("extract_bitfile: unable to open input file");
-  char s[strlen(outpath) + 10];
-  sprintf(s,"%s.bitfile0",outpath);
+  char s[Filename_size];
+  sprintf(s,"%s.%d.lcpbit0",outpath,order);
   assert(strlen(s)==strlen(outpath) + 9);
   FILE *g = fopen(s,"wb");
   if(f==NULL) die("extract_bitfile: unable to open output file");

diff --git a/io.h b/io.h
@@ -37,15 +37,15 @@ typedef struct{
 } bitfile;
 
 void bitfile_flush(bitfile *b);
-void bitfile_create(bitfile *b, size_t size, char *path, bool);
+void bitfile_create(bitfile *b, size_t size, char *path, int);
 void bitfile_destroy(bitfile *b);
 void bitfile_rewind(bitfile *b);
 bool bitfile_read_or_write(bitfile *b, bool new);
 void bitfile_skip(bitfile *b, uint64_t s);
 off_t bitfile_tell(bitfile *b);
 
 // used for extract hi bit from mergefile for later dbGraph construction 
-void extract_bitfile(char *name, size_t size, char *outpath);
+void extract_bitfile(char *name, size_t size, char *outpath, int order);
 
 
 #endif
diff --git a/merge128ext.h b/merge128ext.h
@@ -205,8 +205,7 @@ void gap128ext(g_data *g, bool lastRound) {
   open_bw_files(g);
   // create 0 initialized bitfile (reading and writing using bitfile_* functions)
   bitfile b;
-  if(g->dbOrder>1) bitfile_create(&b,g->mergeLen,g->outPath,false);
-  else bitfile_create(&b,g->mergeLen,g->outPath,true);
+  bitfile_create(&b,g->mergeLen,g->outPath,g->dbOrder);
   // allocate Z (merge) Znew (reading only Z, writing only newZ) 
   alloc_merge_arrays(g);
 
@@ -248,7 +247,7 @@ void gap128ext(g_data *g, bool lastRound) {
     // update solid block files:
     rewind(ibList->fout);
     ibList->fin = ibList->fout;
-  } while(!merge_completed && (prefixLength!=g->dbOrder+1));  // end main loop
+  } while(!merge_completed && (prefixLength!=g->dbOrder));  // end main loop
   if(ibList->fin!=NULL) fclose(ibList->fin);
 
   if (g->verbose>0) {
@@ -270,7 +269,7 @@ void gap128ext(g_data *g, bool lastRound) {
 
   // for dbGraph info we need to extract the hi bit from the merge arrray
   if(g->dbOrder>1) 
-    extract_bitfile(g->merge_fname, g->mergeLen, g->outPath);
+    extract_bitfile(g->merge_fname, g->mergeLen, g->outPath, g->dbOrder);
 
   // computation complete, do the merging. 
   // The following call writes the merged BWT back to file g->bwfname

diff --git a/util.c b/util.c
@@ -331,7 +331,6 @@ void mergeBWT128ext(g_data *g, bool lastRound)
   if(sizeof(symbol)>sizeof(palette)) die("Sorry, alphabet is too large (mergeBWT128ext");
   assert(!g->lcpMerge && g->extMem && g->numBwt<=128);
   FILE *daOutFile=NULL;
-
   check_g_data(g);
   array_clear(g->inCnt,g->numBwt,0);
   if(g->outputDA && lastRound)
@@ -377,8 +376,7 @@ void mergeBWT128ext(g_data *g, bool lastRound)
     g->inCnt[currentColor]++; // one more char read from currentColor BWT
   }  
   close_bw_files(g);
-  if(g->outputDA)
-    close_da_files(g);
+  if(g->outputDA  && lastRound) close_da_files(g);
   // final check on the merging 
   for(int i=0;i<g->numBwt;i++) assert(g->inCnt[i]==g->bwtLen[i]);