From de19711b8596b3293d2343954dccedcc9ace2446 Mon Sep 17 00:00:00 2001
From: pd3 <pd3@sanger.ac.uk>
Date: Mon, 5 Jun 2023 15:19:04 +0200
Subject: [PATCH] Add support for non-standard chromosome names containing [:-]
 characters

Note hts_parse_region() cannot be used because it requires the header
and without the header the caller does not learn the contig name.

Resolves #1620
---
 htslib/synced_bcf_reader.h |  2 ++
 synced_bcf_reader.c        | 32 +++++++++++++++++++++++++++-----
 2 files changed, 29 insertions(+), 5 deletions(-)

diff --git a/htslib/synced_bcf_reader.h b/htslib/synced_bcf_reader.h
index bbe5ea2ba..9a6b48438 100644
--- a/htslib/synced_bcf_reader.h
+++ b/htslib/synced_bcf_reader.h
@@ -338,6 +338,8 @@ int bcf_sr_set_regions(bcf_srs_t *readers, const char *regions, int is_file);
  *              supply 'from' in place of 'to'. When 'to' is negative, first
  *              abs(to) will be attempted and if that fails, 'from' will be used
  *              instead.
+ *              If chromosome name contains the characters ':' or '-', it should
+ *              be put in curly brackets, for example as "{weird-chr-name:1-2}:1000-2000"
  *
  *  The bcf_sr_regions_t struct returned by a successful call should be freed
  *  via bcf_sr_regions_destroy() when it is no longer needed.
diff --git a/synced_bcf_reader.c b/synced_bcf_reader.c
index 702f260ee..d24e5f444 100644
--- a/synced_bcf_reader.c
+++ b/synced_bcf_reader.c
@@ -1032,6 +1032,9 @@ void _regions_sort_and_merge(bcf_sr_regions_t *reg)
 }
 
 // File name or a list of genomic locations. If file name, NULL is returned.
+// Recognises regions in the form chr, chr:pos, chr:beg-end, chr:beg-, {weird-chr-name}:pos.
+// Cannot use hts_parse_region() as that requires the header and if header is not present,
+// wouldn't learn the chromosome name.
 static bcf_sr_regions_t *_regions_init_string(const char *str)
 {
     bcf_sr_regions_t *reg = (bcf_sr_regions_t *) calloc(1, sizeof(bcf_sr_regions_t));
@@ -1043,9 +1046,23 @@ static bcf_sr_regions_t *_regions_init_string(const char *str)
     hts_pos_t from, to;
     while ( 1 )
     {
-        while ( *ep && *ep!=',' && *ep!=':' ) ep++;
         tmp.l = 0;
-        kputsn(sp,ep-sp,&tmp);
+        if ( *ep=='{' )
+        {
+            while ( *ep && *ep!='}' ) ep++;
+            if ( !*ep )
+            {
+                hts_log_error("Could not parse the region, mismatching braces in: \"%s\"", str);
+                goto exit_nicely;
+            }
+            ep++;
+            kputsn(sp+1,ep-sp-2,&tmp);
+        }
+        else
+        {
+            while ( *ep && *ep!=',' && *ep!=':' ) ep++;
+            kputsn(sp,ep-sp,&tmp);
+        }
         if ( *ep==':' )
         {
             sp = ep+1;
@@ -1053,7 +1070,7 @@ static bcf_sr_regions_t *_regions_init_string(const char *str)
             if ( sp==ep )
             {
                 hts_log_error("Could not parse the region(s): %s", str);
-                free(reg); free(tmp.s); return NULL;
+                goto exit_nicely;
             }
             if ( !*ep || *ep==',' )
             {
@@ -1064,7 +1081,7 @@ static bcf_sr_regions_t *_regions_init_string(const char *str)
             if ( *ep!='-' )
             {
                 hts_log_error("Could not parse the region(s): %s", str);
-                free(reg); free(tmp.s); return NULL;
+                goto exit_nicely;
             }
             ep++;
             sp = ep;
@@ -1072,7 +1089,7 @@ static bcf_sr_regions_t *_regions_init_string(const char *str)
             if ( *ep && *ep!=',' )
             {
                 hts_log_error("Could not parse the region(s): %s", str);
-                free(reg); free(tmp.s); return NULL;
+                goto exit_nicely;
             }
             if ( sp==ep ) to = MAX_CSI_COOR-1;
             _regions_add(reg, tmp.s, from, to);
@@ -1088,6 +1105,11 @@ static bcf_sr_regions_t *_regions_init_string(const char *str)
     }
     free(tmp.s);
     return reg;
+
+exit_nicely:
+    bcf_sr_regions_destroy(reg);
+    free(tmp.s);
+    return NULL;
 }
 
 // ichr,ifrom,ito are 0-based;