From 173e7a758569796a91120379853b94f35dcc4a81 Mon Sep 17 00:00:00 2001
From: Justin Bedo <cu@cua0.org>
Date: Tue, 25 Jan 2022 12:03:45 +1100
Subject: add bgz support and fix nested compression support

---
 lib/types.nix              |  1 +
 tools/compression-gzip.nix | 13 +++++++++----
 tools/compression.nix      | 28 ++++++++++++++++++++++++++--
 3 files changed, 36 insertions(+), 6 deletions(-)

diff --git a/lib/types.nix b/lib/types.nix
index 7e58ce0..11f4bf4 100644
--- a/lib/types.nix
+++ b/lib/types.nix
@@ -33,6 +33,7 @@ rec {
     vcf = { ref = any; };
     bed = { ref = any; };
     gz = filetype;
+    bgz = filetype;
     bz2 = filetype;
   };
 
diff --git a/tools/compression-gzip.nix b/tools/compression-gzip.nix
index 0229b94..80fb9f4 100644
--- a/tools/compression-gzip.nix
+++ b/tools/compression-gzip.nix
@@ -1,12 +1,17 @@
-{ bionix }:
+{ bionix, block ? false, flags ? "" }:
 
 with bionix;
+with types;
 
 input:
 
 stage {
   name = "gzip";
-  buildInputs = with pkgs; [ gzip ];
-  buildCommand = "gzip < ${input} > $out";
-  passthru.filetype = filetype.gzip input.filetype;
+  buildInputs = with pkgs; [ (if block then htslib else gzip) ];
+  buildCommand =
+    if block then
+      "bgzip -c ${flags} ${input} > $out"
+    else
+      "gzip ${flags} < ${input} > $out";
+  passthru.filetype = (if block then filetype.bgz else filetype.gz) input.filetype;
 }
diff --git a/tools/compression.nix b/tools/compression.nix
index c9c8777..b32d057 100644
--- a/tools/compression.nix
+++ b/tools/compression.nix
@@ -20,6 +20,7 @@ in
       vcf = _: f;
       bed = _: f;
       gz = _: gunzip attrs f;
+      bgz = _: gunzip attrs f;
       bz2 = _: bunzip2 attrs f;
     }
     f;
@@ -36,10 +37,31 @@ in
         cram = _: gz;
         vcf = _: gz;
         bed = _: gz;
-        gz = x: x;
+        gz = _: f;
+        bgz = _: f;
+        bz2 = _: with compression; gzip { } (uncompress { } f);
       }
       f;
 
+  bgzip = attrs: f:
+    let gz = gzip (attrs // { block = true; }) f;
+    in
+    types.matchFiletype "compressed"
+      {
+        fa = _: gz;
+        fq = _: gz;
+        bam = _: gz;
+        sam = _: gz;
+        cram = _: gz;
+        vcf = _: gz;
+        bed = _: gz;
+        bgz = _: f;
+        gz = _: with compression; bgzip { } (uncompress { } f);
+        bz2 = _: with compression; bgzip { } (uncompress { } f);
+      }
+      f;
+
+
   bzip2 = attrs: f:
     let bz2 = bzip2 attrs f;
     in
@@ -52,7 +74,9 @@ in
         cram = _: gz;
         vcf = _: gz;
         bed = _: gz;
-        bz2 = x: x;
+        bz2 = _: f;
+        bgz = _: with compression; bzip2 { } (uncompress { } f);
+        gz = _: with compression; bzip2 { } (uncompress { } f);
       }
       f;
 }
-- 
cgit v1.2.3