suppprt gzipped BED and PA5 input

asset.c

@@ -33,11 +33,19 @@
 #include <assert.h>
 #include <ctype.h>
 #include <unistd.h>
+#include <zlib.h>
 #include <sys/resource.h>
 #include <sys/time.h>
 
+#include "kseq.h"
+
 #include "asset.h"
 
+KSTREAM_INIT(gzFile, gzread, gzseek, BUFF_SIZE)
+
+void *kopen(const char *fn, int *_fd);
+int kclose(void *a);
+
 #define ARRAY_SIZE(a) (sizeof(a) / sizeof((a)[0]))
 
 double cputime(void)
@@ -77,7 +85,7 @@ double realtime(void)
     return tp.tv_sec + tp.tv_usec * 1e-6;
 }
 
-#if defined CTL_HW && defined HW_USERMEM
+#if (defined __APPLE__ && defined __MACH__) || defined __FreeBSD__ || defined __NetBSD__ || defined __OpenBSD__
 #include <sys/sysctl.h>
 #endif
 
@@ -235,3 +243,90 @@ void positive_or_die(int num)
         exit(EXIT_FAILURE);
     }
 }
+
+int is_empty_line(char *line)
+{
+    while (isspace(*line))
+        line++;
+    if(*line == '\0')
+        return 1;
+    return 0;
+}
+
+static kstring_t *kstring_init(int size)
+{
+    kstring_t *str;
+    str = (kstring_t *) calloc(1, sizeof(kstring_t));
+    str->m = size;
+    str->s = (char *) malloc(sizeof(char) * size);
+    return str;
+}
+
+static void kstring_destroy(kstring_t *str)
+{
+    if (!str) return;
+    free(str->s);
+    free(str);
+}
+
+iostream_t *iostream_open(const char *spath)
+{
+    iostream_t *iostream;
+    iostream = (iostream_t *) calloc(1, sizeof(iostream_t));
+    if (iostream == NULL)
+        return NULL;
+    iostream->koaux = kopen(spath, &iostream->fd);
+    if (iostream->koaux == 0) {
+        free(iostream);
+        return NULL;
+    }
+    iostream->fp = gzdopen(iostream->fd, "r");
+    if (iostream->fp == Z_NULL) {
+        kclose(iostream->koaux);
+        free(iostream);
+        return NULL;
+    }
+    iostream->stream = ks_init(iostream->fp);
+    iostream->buffer = kstring_init(BUFF_SIZE);
+
+    return iostream;
+}
+
+void iostream_close(iostream_t *iostream)
+{
+    if (!iostream) return;
+    ks_destroy(iostream->stream);
+    kstring_destroy(iostream->buffer);
+    gzclose(iostream->fp);
+    kclose(iostream->koaux);
+    free(iostream);
+}
+
+char *iostream_getline(iostream_t *iostream)
+{
+    int len;
+    kstring_t *buffer;
+    buffer = (kstring_t *) (iostream->buffer);
+    len = ks_getuntil(iostream->stream, KS_SEP_LINE, buffer, 0);
+    if (len >= 0) {
+        iostream->nline += 1;
+        return buffer->s;
+    }
+    return NULL;
+}
+
+#ifdef IOSTREAM_MAIN
+int main(int argc, char *argv[])
+{
+    iostream_t *iostream = iostream_open(argv[1]);
+    if (iostream == NULL) {
+        fprintf(stdout, "[E::%s] cannot open file %s to read\n", __func__, argv[1]);
+        exit (1);
+    }
+    char *line;
+    while ((line = iostream_getline(iostream)) != NULL)
+        fprintf(stdout, "[I::%s] Line %9ld: %s\n", __func__, iostream->nline, line);
+    iostream_close(iostream);
+}
+
+#endif

asset.h

@@ -32,6 +32,7 @@
 
 #include <stdint.h>
 #include <stdio.h>
+#include <zlib.h>
 
 #define SWAP(T, x, y) {T tmp = x; x = y; y = tmp;}
 #define MAX(x, y) (((x) > (y)) ? (x) : (y))
@@ -42,6 +43,15 @@
 #define BIN_V 0x2
 #define strcasecmp(s1, s2) strcmp_case_insensitive(s1, s2)
 
+typedef struct {
+    int fd;
+    gzFile fp;
+    void *stream;
+    void *buffer;
+    void *koaux;
+    int64_t nline;
+} iostream_t;
+
 #ifdef __cplusplus
 extern "C" {
 #endif
@@ -59,6 +69,10 @@ void write_bin_header(FILE *fo);
 int is_valid_bin_header(int64_t magic_number);
 int strcmp_case_insensitive(const char *s1, const char *s2);
 void positive_or_die(int num);
+int is_empty_line(char *line);
+iostream_t *iostream_open(const char *spath);
+void iostream_close(iostream_t *iostream);
+char *iostream_getline(iostream_t *iostream);
 #ifdef __cplusplus
 }
 #endif

cov.c

@@ -38,9 +38,6 @@
 #include "cov.h"
 #include "asset.h"
 
-void *kopen(const char *fn, int *_fd);
-int kclose(void *a);
-
 KHASH_SET_INIT_STR(str)
 
 // read unmapped (0x4)
@@ -238,20 +235,15 @@ cov_t *bed_cstats(const char *bed, sdict_t *sdict)
     uint32_t i, s, e;
     uint64_t n, m, max_m, n_recs;
     char *line, cname[4096];
-    size_t ln = 0;
-    ssize_t read;
-    FILE *fp;
-    int fd;
-    void *fh;
+    iostream_t *fp;
     cov_t *covs;
 
     khash_t(str) *hmseq; // for absent sequences
     khint_t k;
     int absent;
     hmseq = kh_init(str);
 
-    fh = kopen(bed, &fd);
-    fp = fdopen(fd, "r");
+    fp = iostream_open(bed);
     if (fp == NULL) {
         fprintf(stderr, "[E::%s] cannot open file %s for reading\n", __func__, bed);
         exit(EXIT_FAILURE);
@@ -262,7 +254,9 @@ cov_t *bed_cstats(const char *bed, sdict_t *sdict)
     n_recs = 0;
     n = 0;
     line = 0;
-    while ((read = getline(&line, &ln, fp)) != -1) {
+    while ((line = iostream_getline(fp)) != NULL) {
+        if (is_empty_line(line))
+            continue;
         ++n_recs;
         if (n_recs % 1000000 == 0)
             fprintf(stderr, "[I::%s] %lu million records processed\n", __func__, n_recs / 1000000);
@@ -300,10 +294,7 @@ cov_t *bed_cstats(const char *bed, sdict_t *sdict)
         free((char *) kh_key(hmseq, k));
     kh_destroy(str, hmseq);
 
-    if (line)
-        free(line);
-    fclose(fp);
-    kclose(fh);
+    iostream_close(fp);
 
     return covs;
 }

graph.c

@@ -153,33 +153,29 @@ void graph_arc_index(graph_t *g)
 
 graph_t *read_graph_from_gfa(char *gfa)
 {
-    FILE *fp;
-    char *line = NULL;
-    size_t ln = 0;
-    ssize_t read;
+    iostream_t *fp;
+    char *line;
     char c0[1024], c1[1024], s0[4], s1[4], wts[1024];
     double wt;
 
     graph_t *g;
     g = graph_init();
     g->sdict = make_asm_dict_from_sdict(make_sdict_from_gfa(gfa, 0));
 
-    fp = fopen(gfa, "r");
+    fp = iostream_open(gfa);
     if (fp == NULL) {
         fprintf(stderr, "[E::%s] cannot open file %s for reading\n", __func__, gfa);
         exit(EXIT_FAILURE);
     }
 
-    while ((read = getline(&line, &ln, fp)) != -1) {
+    while ((line = iostream_getline(fp)) != NULL) {
         if (line[0] == 'L') {
             sscanf(line, "%*s %s %s %s %s %*s %s", c0, s0, c1, s1, wts);
             wt = strtof(wts + 5, NULL);
             graph_add_arc(g, (asm_sd_get(g->sdict, c0)<<1) | (s0[0]=='-'), (asm_sd_get(g->sdict, c1)<<1) | (s1[0]=='-'), -1, 0, wt);
         }
     }
-    if (line)
-        free(line);
-    fclose(fp);
+    iostream_close(fp);
 
     graph_arc_sort(g);
     graph_arc_index(g);