feat: support reading of compressed trace files

Author: DavePearce

pkg/cmd/util.go

@@ -174,7 +174,7 @@ func ReadTraceFile(filename string) lt.TraceFile {
 		tracefile lt.TraceFile
 	)
 	// Read data file
-	data, err := os.ReadFile(filename)
+	filename, data, err := file.ReadAndUncompress(filename)
 	// Check success
 	if err == nil {
 		// Check file extension
@@ -220,7 +220,7 @@ func ReadTraceFile(filename string) lt.TraceFile {
 func ReadBatchedTraceFile(filename string) []lt.TraceFile {
 	var (
 		stats  = util.NewPerfStats()
-		lines  = file.ReadInputFile(filename)
+		lines  = file.ReadInputFileAsLines(filename)
 		traces = make([]lt.TraceFile, 0)
 	)
 	// Read constraints line by line

pkg/test/util/check_valid.go

@@ -378,7 +378,7 @@ func (p *traceId) String() string {
 // ReadTracesFile reads a file containing zero or more traces expressed as JSON, where
 // each trace is on a separate line.
 func ReadTracesFile(filename string) []lt.TraceFile {
-	lines := file.ReadInputFile(filename)
+	lines := file.ReadInputFileAsLines(filename)
 	traces := make([]lt.TraceFile, len(lines))
 	// Read constraints line by line
 	for i, line := range lines {

pkg/util/file/files.go

@@ -15,14 +15,58 @@ package file
 import (
 	"bufio"
 	"compress/bzip2"
+	"compress/gzip"
 	"errors"
 	"io"
 	"os"
 	"path"
+	"strings"
 )
 
-// ReadInputFile reads an input file as a sequence of lines.
-func ReadInputFile(filename string) []string {
+// ReadAndUncompress reads a given file and, if its extension indicates it is
+// compressed, then it decompresses it before returning the decompressed byte
+// and the underlying filename (e.g. if original filename was "file.lt.gz" then
+// it returns "file.lt").  Supported compression formats are "bz2" and "gz".
+func ReadAndUncompress(filename string) (string, []byte, error) {
+	var (
+		ext    = path.Ext(filename)
+		reader io.Reader
+		err    error
+		data   []byte
+	)
+	// Open file for reading
+	file, err := os.Open(filename)
+	// Setup file close behaviour
+	defer func() {
+		if err := file.Close(); err != nil {
+			panic(err)
+		}
+	}()
+	//
+	reader = file
+	// Check for error
+	if err == nil {
+		// check extension
+		switch ext {
+		case ".bz2":
+			reader = bzip2.NewReader(reader)
+			filename = strings.ReplaceAll(filename, ext, "")
+		case ".gz":
+			if reader, err = gzip.NewReader(reader); err != nil {
+				return filename, nil, err
+			}
+			//
+			filename = strings.ReplaceAll(filename, ext, "")
+		}
+		//
+		data, err = io.ReadAll(reader)
+	}
+	//
+	return filename, data, err
+}
+
+// ReadInputFileAsLines reads an input file as a sequence of lines.
+func ReadInputFileAsLines(filename string) []string {
 	file, err := os.Open(filename)
 	// Check whether file exists
 	if errors.Is(err, os.ErrNotExist) {