From 982ca39c8ca49ded28bafcea9338c99d68a022bc Mon Sep 17 00:00:00 2001
From: Dominic Della Valle <ddvpublic@gmail.com>
Date: Mon, 26 Mar 2018 19:54:08 -0400
Subject: [PATCH] squashme: more rules on reserved names

---
 extractor.go        |  2 +-
 sanitize_windows.go | 33 ++++++++++++++++++++-------------
 2 files changed, 21 insertions(+), 14 deletions(-)

diff --git a/extractor.go b/extractor.go
index dac06a0..4a0d4ac 100644
--- a/extractor.go
+++ b/extractor.go
@@ -65,7 +65,7 @@ func (te *Extractor) Extract(reader io.Reader) error {
 func (te *Extractor) outputPath(tarPath string) string {
 	elems := strings.Split(tarPath, "/")  // break into elems
 	elems = elems[1:]                     // remove IPFS root
-	safePath := platformSanitize(elems)   // assure that our output path is platform legal
+	safePath := platformSanitize(elems)   // sanitize IPFS path to be platform legal
 	safePath = fp.Join(te.Path, safePath) // rebase on to extraction target root
 	return safePath
 }
diff --git a/sanitize_windows.go b/sanitize_windows.go
index c96923a..ab2f3ac 100644
--- a/sanitize_windows.go
+++ b/sanitize_windows.go
@@ -1,6 +1,7 @@
 package tar
 
 import (
+	"fmt"
 	"net/url"
 	"path/filepath"
 	"regexp"
@@ -8,38 +9,44 @@ import (
 )
 
 //https://msdn.microsoft.com/en-us/library/windows/desktop/aa365247(v=vs.85).aspx
+const reservedNamesRegexFmt = `(?i)^(%s)(?: *\.)?[^\w ]` //Case insensitive match reserved name, with or without whitespace between possible dot, but reserved word + non whitespace and non-dot is allowed
 var reservedNames = [...]string{"CON", "PRN", "AUX", "NUL", "COM1", "COM2", "COM3", "COM4", "COM5", "COM6", "COM7", "COM8", "COM9", "LPT1", "LPT2", "LPT3", "LPT4", "LPT5", "LPT6", "LPT7", "LPT8", "LPT9"}
+var reservedCharsRegex *regexp.Regexp
 
-const reservedCharsRegex = `[<>:"\\|?*]` //NOTE: `/` is not included, files with this in the name will cause problems
+func init() {
+	reservedCharsRegex = regexp.MustCompile(`[<>:"\\|?*]`) //NOTE: `/` is not included, files with this in the name are illegal in IPFS itself
+}
 
 func platformSanitize(pathElements []string) string {
-	// first pass: scan and prefix reserved names CON -> _CON
-	for i, pe := range pathElements {
+	//first pass: strip illegal tail & prefix reserved names `CON .` -> `_CON`
+	for pi := range pathElements {
+		pathElements[pi] = strings.TrimRight(pathElements[pi], ". ") //MSDN: Do not end a file or directory name with a space or a period
+
 		for _, rn := range reservedNames {
-			if pe == rn {
-				pathElements[i] = "_" + rn
+			re, _ := regexp.Compile(fmt.Sprintf(reservedNamesRegexFmt, rn)) //no err, regex is a constant with guaranteed constant argument
+			if matched := re.MatchString(pathElements[pi]); matched {
+				pathElements[pi] = "_" + pathElements[pi]
 				break
 			}
 		}
-		pathElements[i] = strings.TrimRight(pe, ". ") //MSDN: Do not end a file or directory name with a space or a period
 	}
+
 	//second pass: scan and encode reserved characters ? -> %3F
-	res := strings.Join(pathElements, `/`) // intentionally avoiding [file]path.Clean(), we want `\`'s intact
-	re := regexp.MustCompile(reservedCharsRegex)
-	illegalIndices := re.FindAllStringIndex(res, -1)
+	res := strings.Join(pathElements, `/`) //intentionally avoiding [file]path.Clean() being called with Join(); we do our own filtering first
+	illegalIndices := reservedCharsRegex.FindAllStringIndex(res, -1)
 
 	if illegalIndices != nil {
 		var lastIndex int
 		var builder strings.Builder
-		allocAssist := (len(res) - len(illegalIndices)) + (len(illegalIndices) * 3) // 3 = encoded length
+		allocAssist := (len(res) - len(illegalIndices)) + (len(illegalIndices) * 3) //3 = encoded length
 		builder.Grow(allocAssist)
 
 		for _, si := range illegalIndices {
-			builder.WriteString(res[lastIndex:si[0]])              // append up to problem char
-			builder.WriteString(url.QueryEscape(res[si[0]:si[1]])) // escape and append problem char
+			builder.WriteString(res[lastIndex:si[0]])              //append up to problem char
+			builder.WriteString(url.QueryEscape(res[si[0]:si[1]])) //escape and append problem char
 			lastIndex = si[1]
 		}
-		builder.WriteString(res[lastIndex:]) // append remainder
+		builder.WriteString(res[lastIndex:]) //append remainder
 		res = builder.String()
 	}