X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=sam_header.c;h=88b6a1c856bbf4e6b073c7c8d6e394c1a305094e;hb=28391e5898804ce6b805016d8c676fdf61442eb3;hp=05d75deb2c4ced014c6ae5f8612260d05b34660a;hpb=bef0453b4f1fefd6de0fad72abbcbbc379140a23;p=samtools.git

diff --git a/sam_header.c b/sam_header.c
index 05d75de..88b6a1c 100644
--- a/sam_header.c
+++ b/sam_header.c
@@ -38,7 +38,7 @@ const char *o_sq_tags[] = {"AS","M5","UR","SP",NULL};
 const char *r_sq_tags[] = {"SN","LN",NULL};
 const char *u_sq_tags[] = {"SN",NULL};
 
-const char *o_rg_tags[] = {"LB","DS","PU","PI","CN","DT","PL",NULL};
+const char *o_rg_tags[] = {"CN","DS","DT","FO","KS","LB","PG","PI","PL","PU","SM",NULL};
 const char *r_rg_tags[] = {"ID",NULL};
 const char *u_rg_tags[] = {"ID",NULL};
 
@@ -366,6 +366,7 @@ static HeaderLine *sam_header_line_parse(const char *headerLine)
     while (*to && *to=='\t') to++;
     if ( to-from != 1 ) {
         debug("[sam_header_line_parse] multiple tabs on line [%s] (%d)\n", headerLine,(int)(to-from));
+        free(hline);
 		return 0;
 	}
     from = to;
@@ -434,8 +435,14 @@ static int sam_header_line_validate(HeaderLine *hline)
         tag = tags->data;
         if ( !tag_exists(tag->key,required_tags[itype]) && !tag_exists(tag->key,optional_tags[itype]) )
         {
-            debug("Unknown tag [%c%c] for [%c%c].\n", tag->key[0],tag->key[1], hline->type[0],hline->type[1]);
-            return 0;
+            // Lower case tags are user-defined values.
+            if( !(islower(tag->key[0]) || islower(tag->key[1])) )
+            {
+                // Neither is lower case, but tag was not recognized.
+                debug("Unknown tag [%c%c] for [%c%c].\n", tag->key[0],tag->key[1], hline->type[0],hline->type[1]);
+                // return 0; // Even unknown tags are allowed - for forward compatibility with new attributes
+            }
+            // else - allow user defined tag
         }
         tags = tags->next;
     }
@@ -563,6 +570,7 @@ void *sam_header_parse2(const char *headerText)
     const char *text;
     char *buf=NULL;
     size_t nbuf = 0;
+	int tovalidate = 0;
 
     if ( !headerText )
 		return 0;
@@ -571,7 +579,7 @@ void *sam_header_parse2(const char *headerText)
     while ( (text=nextline(&buf, &nbuf, text)) )
     {
         hline = sam_header_line_parse(buf);
-        if ( hline && sam_header_line_validate(hline) )
+        if ( hline && (!tovalidate || sam_header_line_validate(hline)) )
             // With too many (~250,000) reference sequences the header parsing was too slow with list_append.
             hlines = list_append_to_end(hlines, hline);
         else
@@ -662,6 +670,36 @@ char **sam_header2list(const void *_dict, char type[2], char key_tag[2], int *_n
     return ret;
 }
 
+void *sam_header2key_val(void *iter, const char type[2], const char key_tag[2], const char value_tag[2], const char **_key, const char **_value)
+{
+    list_t *l = iter;
+    if ( !l ) return NULL;
+
+    while (l)
+    {
+        HeaderLine *hline = l->data;
+        if ( hline->type[0]!=type[0] || hline->type[1]!=type[1] )
+        {
+            l = l->next;
+            continue;
+        }
+
+        HeaderTag *key, *value;
+        key   = header_line_has_tag(hline,key_tag);
+        value = header_line_has_tag(hline,value_tag);
+        if ( !key && !value ) 
+        {
+            l = l->next;
+            continue;
+        }
+
+        *_key = key->value;
+        *_value = value->value;
+        return l->next;
+    }
+    return l;
+}
+
 const char *sam_tbl_get(void *h, const char *key)
 {
 	khash_t(str) *tbl = (khash_t(str)*)h;
@@ -732,4 +770,41 @@ void *sam_header_merge(int n, const void **_dicts)
     return out_dict;
 }
 
+char **sam_header2tbl_n(const void *dict, const char type[2], const char *tags[], int *n)
+{
+    int nout = 0;
+    char **out = NULL;
+
+    *n = 0;
+    list_t *l = (list_t *)dict;
+    if ( !l ) return NULL;
+
+    int i, ntags = 0;
+    while ( tags[ntags] ) ntags++;
+
+    while (l)
+    {
+        HeaderLine *hline = l->data;
+        if ( hline->type[0]!=type[0] || hline->type[1]!=type[1] )
+        {
+            l = l->next;
+            continue;
+        }
+        out = (char**) realloc(out, sizeof(char*)*(nout+1)*ntags);
+        for (i=0; i<ntags; i++)
+        {
+            HeaderTag *key = header_line_has_tag(hline, tags[i]);
+            if ( !key ) 
+            {
+                out[nout*ntags+i] = NULL;
+                continue;
+            }
+            out[nout*ntags+i] = key->value;
+        }
+        nout++;
+        l = l->next;
+    }
+    *n = nout;
+    return out;
+}