Revert one of my earlier changes - Heng was right, CIGAR P not sensible in a padded...

[samtools.git] / sam_header.c

diff --git a/sam_header.c b/sam_header.c
index 05d75deb2c4ced014c6ae5f8612260d05b34660a..d348d10e89b0aa26a1da127b83a1d86091f02763 100644 (file)
--- 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};
 
@@ -434,8 +434,14 @@ 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 +569,7 @@ 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 +578,7 @@ 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