sam/linkify
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Merge branch 'features/hostname_validation' into 'master'

Browse source 
Add .onion in the default TLDs, Validate IPv6, IDN compatibility in email and mentions

See merge request pleroma/elixir-libraries/linkify!29
This commit is contained in:
Haelwenn 2020-11-19 20:42:20 +00:00
commit 9b86e5e5e4
3 changed files with 187 additions and 68 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

87
lib/linkify/parser.ex
View file

@ -9,23 +9,13 @@ defmodule Linkify.Parser do
@match_url ~r{^(?:\W*)?(?<url>(?:https?:\/\/)?[\w.-]+(?:\.[\w\.-]+)+[\w\-\._~%:\/?#[\]@!\$&'\(\)\*\+,;=.]+$)}u
@match_hostname ~r{^\W*(?<scheme>https?:\/\/)?(?:[^@\n]+\\w@)?(?<host>[^:#~\/\n?]+)}u
@match_ip ~r"^(([0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])\.){3}([0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])$"
# @user
# @user@example.com
# credo:disable-for-next-line
@match_mention ~r/^(?:\W*)?(?<long>@[a-zA-Z\d_-]+@[a-zA-Z0-9_-](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?(?:\.[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?)*)|^(?:\W*)?(?<short>@[a-zA-Z\d_-]+)/u
# https://www.w3.org/TR/html5/forms.html#valid-e-mail-address
@match_email ~r"^[a-zA-Z0-9.!#$%&'*+\/=?^_`{|}~-]+@[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?(?:\.[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?)*$"u
@get_scheme_host ~r{^\W*(?<scheme>https?:\/\/)?(?:[^@\n]+\\w@)?(?<host>[^:#~\/\n?]+)}u
@match_hashtag ~r/^(?<tag>\#[[:word:]_]*[[:alpha:]_·][[:word:]_·\p{M}]*)/u
@match_skipped_tag ~r/^(?<tag>(a|code|pre)).*>*/
@delimiters ~r/[,.;:>]*$/
@delimiters ~r/[,.;:>?!]*$/
@prefix_extra [
"magnet:?",
@ -41,7 +31,11 @@ defmodule Linkify.Parser do
"ssb://"
]
@tlds "./priv/tlds.txt" |> File.read!() |> String.split("\n", trim: true) |> MapSet.new()
@tlds "./priv/tlds.txt"
|> File.read!()
|> String.split("\n", trim: true)
|> Enum.concat(["onion"])
|> MapSet.new()
@default_opts %{
url: true,
@ -59,7 +53,7 @@ defmodule Linkify.Parser do
~s{Check out <a href="http://google.com">google.com</a>}
"""
@types [:url, :email, :hashtag, :extra, :mention]
@types [:url, :hashtag, :extra, :mention, :email]
def parse(input, opts \\ %{})
def parse(input, opts) when is_binary(input), do: {input, %{}} |> parse(opts) |> elem(0)
@ -220,7 +214,11 @@ defmodule Linkify.Parser do
end
def email?(buffer, opts) do
valid_url?(buffer) && Regex.match?(@match_email, buffer) && valid_tld?(buffer, opts)
# Note: In reality the local part can only be checked by the remote server
case Regex.run(~r/^(?<user>.*)@(?<host>[^@]+)$/, buffer, capture: [:user, :host]) do
[_user, hostname] -> valid_hostname?(hostname) && valid_tld?(hostname, opts)
_ -> false
end
end
defp valid_url?(url), do: !Regex.match?(@invalid_url, url)
@ -233,7 +231,7 @@ defmodule Linkify.Parser do
Will skip validation and return `true` if `:validate_tld` set to `:no_scheme` and the url has a scheme.
"""
def valid_tld?(url, opts) do
[scheme, host] = Regex.run(@match_hostname, url, capture: [:scheme, :host])
[scheme, host] = Regex.run(@get_scheme_host, url, capture: [:scheme, :host])
cond do
opts[:validate_tld] == false ->
@ -247,18 +245,63 @@ defmodule Linkify.Parser do
true
true ->
tld = host |> String.trim_trailing(".") |> String.split(".") |> List.last()
tld = host |> strip_punctuation() |> String.split(".") |> List.last()
MapSet.member?(@tlds, tld)
end
end
def ip?(buffer), do: Regex.match?(@match_ip, buffer)
def safe_to_integer(string, base \\ 10) do
String.to_integer(string, base)
rescue
_ ->
nil
end
def ip?(buffer) do
v4 = String.split(buffer, ".")
v6 =
buffer
|> String.trim_leading("[")
|> String.trim_trailing("]")
|> String.split(":", trim: true)
cond do
length(v4) == 4 ->
!Enum.any?(v4, fn x -> safe_to_integer(x, 10) not in 0..255 end)
length(v6) in 1..8 ->
!Enum.any?(v4, fn x -> safe_to_integer(x, 16) not in 0..0xFFFF end)
false ->
false
end
end
# IDN-compatible, ported from musl-libc's is_valid_hostname()
def valid_hostname?(hostname) do
hostname
|> String.to_charlist()
|> Enum.any?(fn s ->
!(s >= 0x80 || s in 0x30..0x39 || s in 0x41..0x5A || s in 0x61..0x7A || s in '.-')
end)
|> Kernel.!()
end
def match_mention(buffer) do
case Regex.run(@match_mention, buffer, capture: [:long, :short]) do
[mention, ""] -> mention
["", mention] -> mention
_ -> nil
case Regex.run(~r/^@(?<user>[a-zA-Z\d_-]+)(@(?<host>[^@]+))?$/, buffer,
capture: [:user, :host]
) do
[user, ""] ->
"@" <> user
[user, hostname] ->
if valid_hostname?(hostname) && valid_tld?(hostname, []),
do: "@" <> user <> "@" <> hostname,
else: nil
_ ->
nil
end
end

162
test/linkify_test.exs
View file

@ -244,7 +244,7 @@ defmodule LinkifyTest do
end
expected =
~s(Hello again, <span class="h-card"><a href="#/user/user">@<span>@user</span></a></span>.&lt;script&gt;&lt;/script&gt;\nThis is on another :moominmamma: line. <a href="/tag/2hu" target="_blank">#2hu</a> <a href="/tag/epic" target="_blank">#epic</a> <a href="/tag/phantasmagoric" target="_blank">#phantasmagoric</a>)
~s(Hello again, @user.&lt;script&gt;&lt;/script&gt;\nThis is on another :moominmamma: line. <a href="/tag/2hu" target="_blank">#2hu</a> <a href="/tag/epic" target="_blank">#epic</a> <a href="/tag/phantasmagoric" target="_blank">#phantasmagoric</a>)
assert Linkify.link(text,
mention: true,
@ -385,22 +385,14 @@ defmodule LinkifyTest do
text = "That's @user@example.com's server"
expected =
"That's <a href=\"https://example.com/user/user@example.com\">@user@example.com</a>'s server"
assert Linkify.link(text,
mention: true,
mention_prefix: "https://example.com/user/"
) == expected
assert Linkify.link(text, mention: true, mention_prefix: "https://example.com/user/") ==
text
end
test "mentions with symbols before them" do
text = "@@example hey! >@@test@example.com"
test "mentions with no word-separation before them" do
text = "@@example hey! >@@test@example.com idolm@ster"
expected =
"@<a href=\"/users/example\">@example</a> hey! >@<a href=\"/users/test@example.com\">@test@example.com</a>"
assert Linkify.link(text, mention: true, mention_prefix: "/users/") == expected
assert Linkify.link(text, mention: true, mention_prefix: "/users/") == text
end
test "invalid mentions" do
@ -408,6 +400,29 @@ defmodule LinkifyTest do
assert Linkify.link(text, mention: true, mention_prefix: "/users/") == text
end
test "IDN domain" do
text = "hello @lain@我爱你.com"
expected = "hello <a href=\"/users/lain@我爱你.com\">@lain@我爱你.com</a>"
assert Linkify.link(text, mention: true, mention_prefix: "/users/") == expected
text = "hello @lain@xn--6qq986b3xl.com"
expected = "hello <a href=\"/users/lain@xn--6qq986b3xl.com\">@lain@xn--6qq986b3xl.com</a>"
assert Linkify.link(text, mention: true, mention_prefix: "/users/") == expected
end
test ".onion domain" do
text = "Hey @admin@vww6ybal4bd7szmgncyruucpgfkqahzddi37ktceo3ah7ngmcopnpyyd.onion"
expected =
"Hey <a href=\"/users/admin@vww6ybal4bd7szmgncyruucpgfkqahzddi37ktceo3ah7ngmcopnpyyd.onion\">@admin@vww6ybal4bd7szmgncyruucpgfkqahzddi37ktceo3ah7ngmcopnpyyd.onion</a>"
assert Linkify.link(text, mention: true, mention_prefix: "/users/") == expected
end
end
describe "hashtag links" do
@ -505,12 +520,52 @@ defmodule LinkifyTest do
test "turn urls with schema into urls" do
text = "📌https://google.com"
expected = "📌<a href=\"https://google.com\">https://google.com</a>"
assert Linkify.link(text, rel: false) == expected
text = "http://www.cs.vu.nl/~ast/intel/"
expected = "<a href=\"http://www.cs.vu.nl/~ast/intel/\">http://www.cs.vu.nl/~ast/intel/</a>"
assert Linkify.link(text) == expected
text = "https://forum.zdoom.org/viewtopic.php?f=44&t=57087"
expected =
"<a href=\"https://forum.zdoom.org/viewtopic.php?f=44&t=57087\">https://forum.zdoom.org/viewtopic.php?f=44&t=57087</a>"
assert Linkify.link(text) == expected
text = "https://en.wikipedia.org/wiki/Sophia_(Gnosticism)#Mythos_of_the_soul"
expected =
"<a href=\"https://en.wikipedia.org/wiki/Sophia_(Gnosticism)#Mythos_of_the_soul\">https://en.wikipedia.org/wiki/Sophia_(Gnosticism)#Mythos_of_the_soul</a>"
assert Linkify.link(text) == expected
text = "https://en.wikipedia.org/wiki/Duff's_device"
expected =
"<a href=\"https://en.wikipedia.org/wiki/Duff's_device\">https://en.wikipedia.org/wiki/Duff's_device</a>"
assert Linkify.link(text) == expected
text = "https://1.1.1.1/"
expected = "<a href=\"https://1.1.1.1/\">https://1.1.1.1/</a>"
assert Linkify.link(text) == expected
text = "https://1.1.1.1:8080/"
expected = "<a href=\"https://1.1.1.1:8080/\">https://1.1.1.1:8080/</a>"
assert Linkify.link(text) == expected
end
test "skip prefix" do
test "strip prefix" do
assert Linkify.link("http://google.com", strip_prefix: true) ==
"<a href=\"http://google.com\">google.com</a>"
@ -541,35 +596,10 @@ defmodule LinkifyTest do
assert Linkify.link(text, new_window: true) == expected
text = "@username"
expected = "@username"
assert Linkify.link(text, new_window: true) == expected
text = "http://www.cs.vu.nl/~ast/intel/"
expected = "<a href=\"http://www.cs.vu.nl/~ast/intel/\">http://www.cs.vu.nl/~ast/intel/</a>"
assert Linkify.link(text) == expected
text = "https://forum.zdoom.org/viewtopic.php?f=44&t=57087"
expected =
"<a href=\"https://forum.zdoom.org/viewtopic.php?f=44&t=57087\">https://forum.zdoom.org/viewtopic.php?f=44&t=57087</a>"
assert Linkify.link(text) == expected
text = "https://en.wikipedia.org/wiki/Sophia_(Gnosticism)#Mythos_of_the_soul"
expected =
"<a href=\"https://en.wikipedia.org/wiki/Sophia_(Gnosticism)#Mythos_of_the_soul\">https://en.wikipedia.org/wiki/Sophia_(Gnosticism)#Mythos_of_the_soul</a>"
assert Linkify.link(text) == expected
text = "https://en.wikipedia.org/wiki/Duff's_device"
expected =
"<a href=\"https://en.wikipedia.org/wiki/Duff's_device\">https://en.wikipedia.org/wiki/Duff's_device</a>"
assert Linkify.link(text) == expected
end
end
@ -676,13 +706,59 @@ defmodule LinkifyTest do
assert Linkify.link(text) == expected
end
test "Does not link trailing punctuation" do
test "Do not link trailing punctuation" do
text = "You can find more info at https://pleroma.social."
expected =
"You can find more info at <a href=\"https://pleroma.social\">https://pleroma.social</a>."
assert Linkify.link(text) == expected
text = "Of course it was google.com!!"
expected = "Of course it was <a href=\"http://google.com\">google.com</a>!!"
assert Linkify.link(text) == expected
text =
"First I had to login to hotmail.com, then I had to delete emails because my 15MB quota was full."
expected =
"First I had to login to <a href=\"http://hotmail.com\">hotmail.com</a>, then I had to delete emails because my 15MB quota was full."
assert Linkify.link(text) == expected
text = "I looked at theonion.com; it was no longer funny."
expected =
"I looked at <a href=\"http://theonion.com\">theonion.com</a>; it was no longer funny."
assert Linkify.link(text) == expected
end
test "IDN and punycode domain" do
text = "FrauBücher.com says Neiiighhh!"
expected = "<a href=\"http://FrauBücher.com\">FrauBücher.com</a> says Neiiighhh!"
assert Linkify.link(text) == expected
text = "xn--fraubcher-u9a.com says Neiiighhh!"
expected =
"<a href=\"http://xn--fraubcher-u9a.com\">xn--fraubcher-u9a.com</a> says Neiiighhh!"
assert Linkify.link(text) == expected
end
test ".onion domain" do
text =
"The riseup.net hidden service is at vww6ybal4bd7szmgncyruucpgfkqahzddi37ktceo3ah7ngmcopnpyyd.onion"
expected =
"The <a href=\"http://riseup.net\">riseup.net</a> hidden service is at <a href=\"http://vww6ybal4bd7szmgncyruucpgfkqahzddi37ktceo3ah7ngmcopnpyyd.onion\">vww6ybal4bd7szmgncyruucpgfkqahzddi37ktceo3ah7ngmcopnpyyd.onion</a>"
assert Linkify.link(text) == expected
end
end
end

6
test/parser_test.exs
View file

@ -304,7 +304,7 @@ defmodule Linkify.ParserTest do
"misskey.loki"
]
def valid_emails, do: ["rms@ai.mit.edu", "vc@cock.li"]
def invalid_emails, do: ["rms[at]ai.mit.edu", "vc@cock", "xmpp:lain@trashserver.net"]
def valid_custom_tld_emails, do: ["guardian@33y6fjyhs3phzfjj.onion", "hi@company.null"]
def valid_emails, do: ["rms@ai.mit.edu", "vc@cock.li", "guardian@33y6fjyhs3phzfjj.onion"]
def invalid_emails, do: ["rms[at]ai.mit.edu", "vc@cock"]
def valid_custom_tld_emails, do: ["hi@company.null"]
end