When specifying Git dependencies in Cargo.toml developers can set 3

additional keys: rev, branch, tag [1].  These are reflected in a
projects' Cargo.lock file as e.g.

git+https://github.com/servo/angle?branch=servo#a1371e8a160128677af863d1d73f150862ba42b2
git+https://github.com/rust-lang/libc?tag=0.2.26#288942e6858a4b2f8ee56338da5386263b9c4b82

Currently cargo-crates.awk generates the wrong output in these cases:

GH_TUPLE=	servo:angle?branch=servo:a1371e8a160128677af863d1d73f150862ba42b2:angle \
		rust-lang:libc?tag=0.2.26:288942e6858a4b2f8ee56338da5386263b9c4b82:libc

Fix cargo-crates.awk to ignore the query string (except in the tag
case) and generate

GH_TUPLE=	servo:angle:a1371e8a160128677af863d1d73f150862ba42b2:angle \
		rust-lang:libc:0.2.26:libc

instead.

[1] https://github.com/rust-lang/cargo/blob/master/src/doc/specifying-dependencies.md#specifying-dependencies-from-git-repositories

PR:		220548
Reported by:	jbeich
Reviewed by:	jbeich, mat
Differential Revision:	https://reviews.freebsd.org/D11571

1 files changed, 44 insertions, 9 deletions

diff --git a/Mk/Scripts/cargo-crates.awk b/Mk/Scripts/cargo-crates.awk
index cf4302204d00..8daaf05065eb 100644
--- a/Mk/Scripts/cargo-crates.awk
+++ b/Mk/Scripts/cargo-crates.awk
@@ -1,3 +1,6 @@
+# MAINTAINER: ports@FreeBSD.org
+# $FreeBSD$
+
 BEGIN {
 	gh_tuple_len = 0
 	crates_len = 0
@@ -17,18 +20,50 @@ BEGIN {
 	gsub("[^a-zA-Z_]", "", package_name)
 }
 
-/^source = "git\+(https|http|git):\/\/github.com\/.*#.*"/ {
-	split($3, url, "#")
+function split_url(s) {
+	# scheme:[//[user[:password]@]host[:port]][/path][?query][#fragment]
+	split(s, url_scheme, "://")
+	url["scheme"] = url_scheme[1]
 
-	gsub("^\"git\+", "", url[1])
-	split(url[1], repourl, "/")
-	account = repourl[4]
-	project = repourl[5]
-	gsub("\.git$", "", project)
+	split(url_scheme[2], url_fragment, "#")
+	url["fragment"] = url_fragment[2]
 
-	tag = url[2]
-	gsub("\"$", "", tag)
+	split(url_fragment[1], url_query, "?")
+	url["query"] = url_query[2]
 
+	split(url_query[1], url_authority, "/")
+	url["path"] = substr(url_query[1], length(url_authority[1]) + 1)
+
+	split(url_authority[1], url_auth, "@")
+
+	if (length(url_auth) == 2) {
+		split(url_auth[1], url_user, ":")
+		url["user"] = url_user[1]
+		url["password"] = url_user[2]
+		split(url_auth[2], url_host, ":")
+	} else {
+		url["user"] = ""
+		url["password"] = ""
+		split(url_auth[1], url_host, ":")
+	}
+	url["host"] = url_host[1]
+	url["port"] = url_host[2]
+}
+
+/^source = "git\+(https|http|git):\/\/github.com\/.*#.*"/ {
+	split_url(substr($3, 1, length($3) - 1))
+	
+	split(url["path"], path, "/")
+	account = path[2]
+	project = path[3]
+	gsub("\.git$", "", project)
+	
+	if (match(url["query"], "^tag=")) {
+		split(url["query"], tag_, "=")
+		tag = tag_[2]
+	} else {
+		tag = url["fragment"]
+	}
 	gh_tuple[gh_tuple_len++] = sprintf(\
 		"%s:%s:%s:%s", account, project, tag, package_name)
 }