Lead a readme's relative links into the tree

A browser resolved them against the about page, where nothing lives
now that the page answers 404 below itself. They resolve the way
GitHub resolves a readme's links instead, from the readme's directory
and on the branch it was read from.

1 file changed, 55 insertions, 6 deletions

diff --git a/custom/extensions/about-render.lua b/custom/extensions/about-render.luaindex ecb2abe..a3932af 100644--- a/custom/extensions/about-render.lua+++ b/custom/extensions/about-render.lua@@ -30,7 +30,7 @@ local max_bytes = 512 * 1024 -- markers cannot drive the recursive parser into a stack overflow. local max_depth = 24 -local filename, chunks = "", {}+local filename, tree_base, tree_query, chunks = "", "", "", {}   -- Lua patterns backtrack, so "%s+$" costs the square of a run of spaces in@@ -108,11 +108,56 @@ local function safe_url(url) 	return url end +-- A readme percent-encodes what a link path cannot carry, and the path is+-- encoded again on its way out, so it is decoded first. An encoded NUL stays+-- encoded, since cgit's sinks would stop at a real one.+local function unescape_path(path)+	return (path:gsub("%%(%x%x)", function(hex)+		local byte = tonumber(hex, 16)+		if byte > 0 then+			return string.char(byte)+		end+	end))+end++-- The path in the tree that a relative link names. Like GitHub, it resolves+-- against the readme's own directory, or against the root when it starts+-- with a slash, and dot segments fold without climbing above the root.+local function tree_path(path)+	if path:sub(1, 1) ~= "/" then+		path = (filename:match("^(.*/)") or "") .. path+	end+	local segments = {}+	for segment in path:gmatch("[^/]+") do+		if segment == ".." then+			segments[#segments] = nil+		elseif segment ~= "." then+			segments[#segments + 1] = segment+		end+	end+	return table.concat(segments, "/")+end++-- Writes a link target that safe_url passed. A relative one leads into the+-- tree at the ref the readme came from, keeping its fragment but not a+-- query, which the tree page would read as its own. A readme with no tree+-- behind it keeps its links as written.+local function emit_href(url)+	if tree_base == "" or url:find("^#") or url:find("^%a[%w%+%.%-]*:") then+		html_attr(url)+		return+	end+	html_attr(tree_base)+	html_url_path(tree_path(unescape_path(url:match("^[^?#]*"))))+	html_attr(tree_query)+	html_attr(url:match("#.*$") or "")+end+  -- The page renders untrusted repository content. Every run of text goes out--- through cgit's html_txt, every attribute value through html_attr, and--- every link or image target through safe_url before html_attr. html() only--- ever gets literal tag scaffolding.+-- through cgit's html_txt, every attribute value through html_attr or, for a+-- path in the tree, html_url_path, and every link or image target through+-- safe_url first. html() only ever gets literal tag scaffolding.  local emit_inline @@ -130,7 +175,7 @@ emit_inline = function(list) 		elseif kind == "link" then 			local target = safe_url(node.url) 			if target then-				html("<a href='"); html_attr(target); html("'>")+				html("<a href='"); emit_href(target); html("'>") 				emit_inline(node.kids) 				html("</a>") 			else@@ -632,8 +677,12 @@ for _, ext in ipairs({ "1", "2", "3", "4", "5", "6", "7", "8", "9", "man" }) do 	handlers[ext] = render_man end -function filter_open(name)+-- A readme tracked in the repository comes with the url of the tree page at+-- its ref and the query that follows a path there, both escaped for a url.+function filter_open(name, base, query) 	filename = name or ""+	tree_base = base or ""+	tree_query = query or "" 	chunks = {} end 
diff --git a/custom/extensions/about-render.lua b/custom/extensions/about-render.lua
index ecb2abe..a3932af 100644
--- a/custom/extensions/about-render.lua
+++ b/custom/extensions/about-render.lua
@@ -30,7 +30,7 @@ local max_bytes = 512 * 1024
 -- markers cannot drive the recursive parser into a stack overflow.
 local max_depth = 24
 
-local filename, chunks = "", {}
+local filename, tree_base, tree_query, chunks = "", "", "", {}
 
 
 -- Lua patterns backtrack, so "%s+$" costs the square of a run of spaces in
@@ -108,11 +108,56 @@ local function safe_url(url)
 	return url
 end
 
+-- A readme percent-encodes what a link path cannot carry, and the path is
+-- encoded again on its way out, so it is decoded first. An encoded NUL stays
+-- encoded, since cgit's sinks would stop at a real one.
+local function unescape_path(path)
+	return (path:gsub("%%(%x%x)", function(hex)
+		local byte = tonumber(hex, 16)
+		if byte > 0 then
+			return string.char(byte)
+		end
+	end))
+end
+
+-- The path in the tree that a relative link names. Like GitHub, it resolves
+-- against the readme's own directory, or against the root when it starts
+-- with a slash, and dot segments fold without climbing above the root.
+local function tree_path(path)
+	if path:sub(1, 1) ~= "/" then
+		path = (filename:match("^(.*/)") or "") .. path
+	end
+	local segments = {}
+	for segment in path:gmatch("[^/]+") do
+		if segment == ".." then
+			segments[#segments] = nil
+		elseif segment ~= "." then
+			segments[#segments + 1] = segment
+		end
+	end
+	return table.concat(segments, "/")
+end
+
+-- Writes a link target that safe_url passed. A relative one leads into the
+-- tree at the ref the readme came from, keeping its fragment but not a
+-- query, which the tree page would read as its own. A readme with no tree
+-- behind it keeps its links as written.
+local function emit_href(url)
+	if tree_base == "" or url:find("^#") or url:find("^%a[%w%+%.%-]*:") then
+		html_attr(url)
+		return
+	end
+	html_attr(tree_base)
+	html_url_path(tree_path(unescape_path(url:match("^[^?#]*"))))
+	html_attr(tree_query)
+	html_attr(url:match("#.*$") or "")
+end
+
 
 -- The page renders untrusted repository content. Every run of text goes out
--- through cgit's html_txt, every attribute value through html_attr, and
--- every link or image target through safe_url before html_attr. html() only
--- ever gets literal tag scaffolding.
+-- through cgit's html_txt, every attribute value through html_attr or, for a
+-- path in the tree, html_url_path, and every link or image target through
+-- safe_url first. html() only ever gets literal tag scaffolding.
 
 local emit_inline
 
@@ -130,7 +175,7 @@ emit_inline = function(list)
 		elseif kind == "link" then
 			local target = safe_url(node.url)
 			if target then
-				html("<a href='"); html_attr(target); html("'>")
+				html("<a href='"); emit_href(target); html("'>")
 				emit_inline(node.kids)
 				html("</a>")
 			else
@@ -632,8 +677,12 @@ for _, ext in ipairs({ "1", "2", "3", "4", "5", "6", "7", "8", "9", "man" }) do
 	handlers[ext] = render_man
 end
 
-function filter_open(name)
+-- A readme tracked in the repository comes with the url of the tree page at
+-- its ref and the query that follows a path there, both escaped for a url.
+function filter_open(name, base, query)
 	filename = name or ""
+	tree_base = base or ""
+	tree_query = query or ""
 	chunks = {}
 end