From eb0b41ccb03ee59ed78b0ddd9b50300fbe80e365 Mon Sep 17 00:00:00 2001
From: Dominic Della Valle <ddvpublic@gmail.com>
Date: Tue, 13 Feb 2018 11:28:00 -0500
Subject: [PATCH] Add gateway URL parser

Allows HTTP gateway urls as arguments
---
 main.go                     | 33 +++++++++++++++++++++++----------
 sharness/t0030-arguments.sh | 24 ++++++++++++++++++++++++
 2 files changed, 47 insertions(+), 10 deletions(-)
 create mode 100755 sharness/t0030-arguments.sh

diff --git a/main.go b/main.go
index c094b95..1887617 100644
--- a/main.go
+++ b/main.go
@@ -4,10 +4,11 @@ import (
 	"fmt"
 	"os"
 	"os/signal"
+	"regexp"
 	"strings"
 	"syscall"
 
-	path "gx/ipfs/QmT3rzed1ppXefourpmoZ7tyVQfsGPQZ1pHDngLmCvXxd3/go-path"
+	ipath "gx/ipfs/QmT3rzed1ppXefourpmoZ7tyVQfsGPQZ1pHDngLmCvXxd3/go-path"
 	fallback "gx/ipfs/QmaWDhoQaV6cDyy6NSKFgPaUAGRtb4SMiLpaDYEsxP7X8P/fallback-ipfs-shell"
 	cli "gx/ipfs/Qmc1AtgBdoUHP8oYSqU81NRYdzohmF45t5XNwVMvhCxsBA/cli"
 )
@@ -36,21 +37,19 @@ func main() {
 		}
 
 		outfile := c.String("output")
-		arg := c.Args().First()
+		inPath, err := parsePath(c.Args().First())
+		if err != nil {
+			fmt.Fprintf(os.Stderr, "Argument parse failure: %s\n", err)
+			os.Exit(1)
+		}
 
 		// Use the final segment of the object's path if no path was given.
 		if outfile == "" {
-			ipfsPath, err := path.ParsePath(arg)
-			if err != nil {
-				fmt.Fprintf(os.Stderr, "ParsePath failure: %s\n", err)
-				os.Exit(1)
-			}
-			segments := ipfsPath.Segments()
+			segments := inPath.Segments()
 			outfile = segments[len(segments)-1]
 		}
 
 		var shell fallback.Shell
-		var err error
 
 		if c.String("node") == "fallback" {
 			shell, err = fallback.NewShell()
@@ -76,7 +75,7 @@ func main() {
 			return nil
 		}
 
-		if err := shell.Get(arg, outfile); err != nil {
+		if err := shell.Get(inPath.String(), outfile); err != nil {
 			os.Remove(outfile)
 			fmt.Fprintf(os.Stderr, "ipget failed: %s\n", err)
 			os.Exit(2)
@@ -132,3 +131,17 @@ func movePostfixOptions(args []string) []string {
 	// append extracted arguments to the real args
 	return append(args, the_args...)
 }
+
+func parsePath(path string) (ipath.Path, error) {
+	ipfsPath, err := ipath.ParsePath(path)
+	if err == nil { // valid canonical path
+		return ipfsPath, nil
+	}
+	// allow HTTP gateway URLs as input as well
+	matches := regexp.MustCompile(`^https?.*(/(ipfs|ipns|ipld)/.+)$`).FindStringSubmatch(path)
+	if len(matches) < 1 {
+		return "", fmt.Errorf("%q does not appear to be a valid IPFS path or HTTP gateway URL", path)
+	}
+
+	return ipath.ParsePath(matches[1])
+}
diff --git a/sharness/t0030-arguments.sh b/sharness/t0030-arguments.sh
new file mode 100755
index 0000000..33f5841
--- /dev/null
+++ b/sharness/t0030-arguments.sh
@@ -0,0 +1,24 @@
+#!/bin/sh
+
+test_description="test the ipget argument parser"
+
+. ./lib/sharness/sharness.sh
+
+test_expect_success "retrieve a known popular single file with a gateway HTTP URL" "
+    ipget http://ipfs.io/ipfs/QmQ2r6iMNpky5f1m4cnm3Yqw8VSvjuKpTcK1X7dBR1LkJF/cat.gif &&
+    echo 'c5ea0d6cacf1e54635685803ec4edbe0d4fe8465' > expected &&
+    shasum cat.gif | cut -d ' ' -f 1 > actual &&
+    diff expected actual
+"
+
+test_expect_failure "don't allow non-gateway URLS" "
+    ipget ftp://ipfs.io/ipfs/QmQ2r6iMNpky5f1m4cnm3Yqw8VSvjuKpTcK1X7dBR1LkJF/cat.gif
+"
+
+test_expect_success "retrieve a directory" "
+    ipget --node=spawn QmQ2r6iMNpky5f1m4cnm3Yqw8VSvjuKpTcK1X7dBR1LkJF &&
+    ls QmQ2r6iMNpky5f1m4cnm3Yqw8VSvjuKpTcK1X7dBR1LkJF > /dev/null &&
+    ls QmQ2r6iMNpky5f1m4cnm3Yqw8VSvjuKpTcK1X7dBR1LkJF/cat.gif > /dev/null
+"
+
+test_done