From 4778137e61534f4f2507ecc2126a255cd2d51b19 Mon Sep 17 00:00:00 2001
From: jrobinso <933148+jrobinso@users.noreply.github.com>
Date: Wed, 9 Oct 2024 10:12:33 -0700
Subject: [PATCH] Extend file format determination code.

---
 .../org/broad/igv/track/FileFormatUtils.java  | 22 +++++++++++++++++++
 .../org/broad/igv/track/TrackProperties.java  | 12 ++++++++++
 .../java/org/broad/igv/util/ParsingUtils.java |  7 ++++--
 .../broad/igv/track/FileFormatUtilsTest.java  |  4 ++++
 4 files changed, 43 insertions(+), 2 deletions(-)

diff --git a/src/main/java/org/broad/igv/track/FileFormatUtils.java b/src/main/java/org/broad/igv/track/FileFormatUtils.java
index 4969624c83..6171ff27c9 100644
--- a/src/main/java/org/broad/igv/track/FileFormatUtils.java
+++ b/src/main/java/org/broad/igv/track/FileFormatUtils.java
@@ -4,6 +4,7 @@
 import htsjdk.samtools.util.BlockCompressedInputStream;
 import org.broad.igv.ucsc.twobit.UnsignedByteBuffer;
 import org.broad.igv.ucsc.twobit.UnsignedByteBufferImpl;
+import org.broad.igv.util.ParsingUtils;
 import org.broad.igv.util.stream.IGVSeekableStreamFactory;
 
 import java.io.*;
@@ -89,6 +90,27 @@ public static String determineFormat(String path) throws IOException {
             if (firstLine.startsWith("##gff-version")) {
                 return "gff";
             }
+            if(firstLine.startsWith("##fileformat=")) {
+                return firstLine.substring(13);   // Non standard extension of VCF convention
+            }
+
+            // Read maximum of first 100 lines searching for format indication.
+            int n = 0;
+            String nextLine;
+            while((nextLine = reader.readLine()) != null && n++ < 100) {
+                if(nextLine.startsWith("#")) continue;
+                if(nextLine.startsWith("track")) {
+                    TrackProperties properties = new TrackProperties();
+                    ParsingUtils.parseTrackLine(nextLine, properties);
+                    if(properties.getFormat() != null) {
+                        return properties.getFormat();
+                    }
+                }
+                if(nextLine.startsWith("fixedStep") || nextLine.startsWith("variableStep")) {
+                    return "wig";
+                }
+            }
+
             if (maybeSampleInfo(bytes)) {
                 return "sampleinfo";
             }
diff --git a/src/main/java/org/broad/igv/track/TrackProperties.java b/src/main/java/org/broad/igv/track/TrackProperties.java
index 16ffa2043f..96658053f7 100644
--- a/src/main/java/org/broad/igv/track/TrackProperties.java
+++ b/src/main/java/org/broad/igv/track/TrackProperties.java
@@ -143,6 +143,11 @@ public enum BaseCoord {
 
     private String coverageURL;
 
+    /**
+     * Non-standard track field to indicate file format
+     */
+    private String format;
+
     /**
      * Track attributes (meta data)
      */
@@ -153,6 +158,13 @@ public TrackProperties() {
 
     }
 
+    public String getFormat() {
+        return format;
+    }
+
+    public void setFormat(String format) {
+        this.format = format;
+    }
 
     public void setTrackLine(String trackLine) {
         this.trackLine = trackLine;
diff --git a/src/main/java/org/broad/igv/util/ParsingUtils.java b/src/main/java/org/broad/igv/util/ParsingUtils.java
index 06e4685aa3..68d72fe00d 100644
--- a/src/main/java/org/broad/igv/util/ParsingUtils.java
+++ b/src/main/java/org/broad/igv/util/ParsingUtils.java
@@ -386,7 +386,10 @@ public static boolean parseTrackLine(String nextLine, TrackProperties trackPrope
                     String key = kv.get(0).toLowerCase().trim();
                     String value = kv.get(1).replaceAll("\"", "");
 
-                    if (key.equals("coords")) {
+                    if(key.equals("format")) {
+                        trackProperties.setFormat(value);
+                    }
+                    else if (key.equals("coords")) {
                         if (value.equals("0")) {
                             trackProperties.setBaseCoord(TrackProperties.BaseCoord.ZERO);
                         } else if (value.equals("1")) {
@@ -394,7 +397,7 @@ public static boolean parseTrackLine(String nextLine, TrackProperties trackPrope
                         }
 
                     }
-                    if (key.equals("name")) {
+                    else if (key.equals("name")) {
                         trackProperties.setName(value);
                         //dhmay adding name check for TopHat junctions files. graphType is also checked.
                         if (value.equals("junctions")) {
diff --git a/src/test/java/org/broad/igv/track/FileFormatUtilsTest.java b/src/test/java/org/broad/igv/track/FileFormatUtilsTest.java
index 1010d9537d..eaf23587d4 100644
--- a/src/test/java/org/broad/igv/track/FileFormatUtilsTest.java
+++ b/src/test/java/org/broad/igv/track/FileFormatUtilsTest.java
@@ -42,5 +42,9 @@ public void testDetermineFormat() throws Exception {
         String sampleInfoFile = "http://igvdata.broadinstitute.org/data/hg18/tcga/gbm/gbmsubtypes/sampleTable.txt.gz";
         format = FileFormatUtils.determineFormat(sampleInfoFile);
         assertEquals("sampleinfo", format);
+
+        String wigFile = TestUtils.DATA_DIR + "wig/dm3_var_sample.wig";
+        format = FileFormatUtils.determineFormat(wigFile);
+        assertEquals("wig", format);
     }
 }
\ No newline at end of file