Compare commits

..

21 Commits

Author SHA1 Message Date
Solene Rapenne 12010338c2 we don't support -m from vger 2022-08-21 13:20:29 +02:00
Solene Rapenne 1e8d353859 todo 2022-08-21 13:20:19 +02:00
Solene Rapenne 872aeffe77 use Network.URI instead of a regex 2022-08-21 13:16:17 +02:00
Solene Rapenne 9144140aff bump licence 2022-08-19 18:22:09 +02:00
Solene Rapenne 9816876c62 remove shell.nix 2022-08-19 18:20:57 +02:00
Solene Rapenne 014b9efb64 mention hix 2022-08-19 18:19:20 +02:00
Solene Rapenne caaca49a54 integration tests should use the new vger path 2022-08-19 18:19:07 +02:00
Solene Rapenne 48ab9299d8 use hix project framework 2022-08-19 18:16:39 +02:00
Solene Rapenne 92f356e09f improve gemini protocol cases 2022-08-19 15:04:41 +02:00
Solene Rapenne 1809a0d13a move file 2022-08-19 14:59:54 +02:00
Solene Rapenne 95a2ef910e apply ormolu for formatting 2022-08-19 14:58:58 +02:00
Solene Rapenne 32bee65a20 support code 31 for directories 2022-08-17 12:48:25 +02:00
Solene Rapenne 74803a76ca support index 2022-08-17 12:35:44 +02:00
Solene Rapenne 9a145bb042 add getopt to support language 2022-08-13 21:12:22 +02:00
Solene Rapenne aa900ac405 unit tests for mime 2022-08-13 17:52:09 +02:00
Solene Rapenne d8c9d26e30 use uppercase for unit file 2022-08-13 14:03:10 +02:00
Solene Rapenne 56740f7f14 start cleaning up the README 2022-08-13 14:01:46 +02:00
Solene Rapenne aa3aff9854 remove C bits from Procfile 2022-08-13 14:00:41 +02:00
Solene Rapenne bb1de0e0cc more unit tests 2022-08-13 14:00:29 +02:00
Solene Rapenne 0b1d5d399d haskell mode 2022-08-13 13:56:09 +02:00
Solene Rapenne 7212128831 init version haskell 2022-08-12 14:49:24 +02:00
25 changed files with 628 additions and 1050 deletions

5
.gitignore vendored
View File

@ -1,2 +1,3 @@
*.o
vger
*.hi
Vger
unit

View File

@ -1,4 +1,4 @@
Copyright (c) 2020-2021 Solène Rapenne <solene@openbsd.org>
Copyright (c) 2020-2022 Solène Rapenne <solene@openbsd.org>
BSD 2-clause License

View File

@ -1,33 +0,0 @@
include config.mk
PREFIX?=/usr/local/
CFLAGS += -pedantic -Wall -Wextra -Wmissing-prototypes \
-Wstrict-prototypes -Wwrite-strings ${EXTRAFLAGS}
.SUFFIXES: .c .o
.c.o:
${CC} ${CFLAGS} -c $<
all: vger
clean:
find . \( -name vger -o \
-name unit_test -o \
-name "*.o" -o \
-name "*.core" \) \
-delete
vger: main.c vger.c mimes.o utils.o opts.h
${CC} ${CFLAGS} -o $@ main.c mimes.o utils.o
install: vger
install -o root -g wheel vger ${PREFIX}/bin/
install -o root -g wheel vger.8 ${PREFIX}/man/man8/
unit_test: tests.c vger.o
${CC} ${CFLAGS} -o $@ vger.o tests.c mimes.o utils.o
test: vger unit_test
./unit_test
cd tests && sh test.sh

View File

@ -31,12 +31,14 @@ For all supported OS, it's possible to run **Vger** in a chroot and drop privile
```
git clone https://tildegit.org/solene/vger.git
cd vger
./configure (only really useful for Linux)
make
doas make install
nix build
```
On GNU/Linux, make sure you installed `libbsd`, it has been reported that using clang was required too.
# Development
```
nix run .#ghcid -- . Main main tests
```
For NixOS/Nix users, there is a `shell.nix` listing the dependencies.

41
app/Vger.hs Normal file
View File

@ -0,0 +1,41 @@
{-# Language ApplicativeDo #-}
{-# Language RecordWildCards #-}
import Gemini
import Options.Applicative
data Sample = Sample
{ baseDir :: String
, language :: String
, virtualhost :: Bool
}
sample :: Parser Sample
sample = do
baseDir <- strOption
( long "baseDir"
<> short 'd'
<> help "base directory to serve files from"
<> value "/var/gemini/")
language <- strOption
( long "language"
<> short 'l'
<> help "language to use in the response for gemini files"
<> value "")
virtualhost <- switch
( long "virtualhost"
<> short 'v'
<> help "virtualhost support")
pure Sample{..}
opts :: ParserInfo Sample
opts = info (sample <**> helper)
(fullDesc)
main :: IO ()
main = do
options <- execParser opts
request <- get_request
answer <- create_answer request (language options) (baseDir options) (virtualhost options)
putStr (make_reply answer)
putStr (content answer)

View File

@ -1 +0,0 @@
EXTRAFLAGS=

14
configure vendored
View File

@ -1,14 +0,0 @@
#!/bin/sh
OS="$(uname -s)"
case "$OS" in
Linux)
EXTRAFLAGS=-lbsd
;;
*)
EXTRAFLAGS=""
;;
esac
echo "EXTRAFLAGS=${EXTRAFLAGS}" > config.mk

248
flake.lock Normal file
View File

@ -0,0 +1,248 @@
{
"nodes": {
"easy-hls": {
"inputs": {
"nixpkgs": "nixpkgs"
},
"locked": {
"lastModified": 1646146030,
"narHash": "sha256-wkKDVVL2/rUGokxs4kqCE+ZzxljBqYDZvBuQGVGXYJM=",
"owner": "jkachmar",
"repo": "easy-hls-nix",
"rev": "ecb85ab6ba0aab0531fff32786dfc51feea19370",
"type": "github"
},
"original": {
"owner": "jkachmar",
"repo": "easy-hls-nix",
"type": "github"
}
},
"flake-utils": {
"locked": {
"lastModified": 1649676176,
"narHash": "sha256-OWKJratjt2RW151VUlJPRALb7OU2S5s+f0vLj4o1bHM=",
"owner": "numtide",
"repo": "flake-utils",
"rev": "a4b154ebbdc88c8498a5c7b01589addc9e9cb678",
"type": "github"
},
"original": {
"owner": "numtide",
"repo": "flake-utils",
"type": "github"
}
},
"hix": {
"inputs": {
"easy-hls": "easy-hls",
"flake-utils": "flake-utils",
"nixpkgs": "nixpkgs_2",
"nixpkgs_ghc8107": "nixpkgs_ghc8107",
"nixpkgs_ghc865": "nixpkgs_ghc865",
"nixpkgs_ghc884": "nixpkgs_ghc884",
"nixpkgs_ghc901": "nixpkgs_ghc901",
"nixpkgs_ghc902": "nixpkgs_ghc902",
"nixpkgs_ghc922": "nixpkgs_ghc922",
"nmd": "nmd",
"obelisk": "obelisk",
"thax": "thax"
},
"locked": {
"lastModified": 1658021306,
"narHash": "sha256-chR2NPBuJgI/aaN5IdH83DabILJQ8AEfjRsDlhkEQiM=",
"owner": "tek",
"repo": "hix",
"rev": "bedd71655b5fe8af681d28f197dff483edfc2f19",
"type": "github"
},
"original": {
"owner": "tek",
"repo": "hix",
"type": "github"
}
},
"nixpkgs": {
"locked": {
"lastModified": 1615055905,
"narHash": "sha256-Ig7CXgE5C3mwtTVBl8nSTessa1oMAm6Pm/p+T6iAJD8=",
"owner": "nixos",
"repo": "nixpkgs",
"rev": "f3fc074642a25ef5a1423412f09946149237e338",
"type": "github"
},
"original": {
"owner": "nixos",
"repo": "nixpkgs",
"type": "github"
}
},
"nixpkgs_2": {
"locked": {
"lastModified": 1651927509,
"narHash": "sha256-fGVGUdEsriuAL1vkUh29FlOQmEkPRnSfRGImWYaVjos=",
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "2fdb6f2e08e7989b03a2a1aa8538d99e3eeea881",
"type": "github"
},
"original": {
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "2fdb6f2e08e7989b03a2a1aa8538d99e3eeea881",
"type": "github"
}
},
"nixpkgs_ghc8107": {
"locked": {
"lastModified": 1642069818,
"narHash": "sha256-666w6j8wl/bojfgpp0k58/UJ5rbrdYFbI2RFT2BXbSQ=",
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "46821ea01c8f54d2a20f5a503809abfc605269d7",
"type": "github"
},
"original": {
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "46821ea01c8f54d2a20f5a503809abfc605269d7",
"type": "github"
}
},
"nixpkgs_ghc865": {
"locked": {
"lastModified": 1602473909,
"narHash": "sha256-tZfZhmjYNX4UrUySEjbosJO7Bj3eiCGnjfX5DkVClQw=",
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "cfed29bfcb28259376713005d176a6f82951014a",
"type": "github"
},
"original": {
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "cfed29bfcb28259376713005d176a6f82951014a",
"type": "github"
}
},
"nixpkgs_ghc884": {
"locked": {
"lastModified": 1616670887,
"narHash": "sha256-wn+l9qJfR5sj5Gq4DheJHAcBDfOs9K2p9seW2f35xzs=",
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "c0e881852006b132236cbf0301bd1939bb50867e",
"type": "github"
},
"original": {
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "c0e881852006b132236cbf0301bd1939bb50867e",
"type": "github"
}
},
"nixpkgs_ghc901": {
"locked": {
"lastModified": 1639713555,
"narHash": "sha256-w1TacWjnqhC19n+rheyOif3JxwvWMbyxfgqYCY0FLdQ=",
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "45a3f9d7725c7e21b252c223676cc56fb2ed5d6d",
"type": "github"
},
"original": {
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "45a3f9d7725c7e21b252c223676cc56fb2ed5d6d",
"type": "github"
}
},
"nixpkgs_ghc902": {
"locked": {
"lastModified": 1642069818,
"narHash": "sha256-666w6j8wl/bojfgpp0k58/UJ5rbrdYFbI2RFT2BXbSQ=",
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "46821ea01c8f54d2a20f5a503809abfc605269d7",
"type": "github"
},
"original": {
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "46821ea01c8f54d2a20f5a503809abfc605269d7",
"type": "github"
}
},
"nixpkgs_ghc922": {
"locked": {
"lastModified": 1651927509,
"narHash": "sha256-fGVGUdEsriuAL1vkUh29FlOQmEkPRnSfRGImWYaVjos=",
"owner": "nixos",
"repo": "nixpkgs",
"rev": "2fdb6f2e08e7989b03a2a1aa8538d99e3eeea881",
"type": "github"
},
"original": {
"owner": "nixos",
"repo": "nixpkgs",
"rev": "2fdb6f2e08e7989b03a2a1aa8538d99e3eeea881",
"type": "github"
}
},
"nmd": {
"flake": false,
"locked": {
"lastModified": 1648078774,
"narHash": "sha256-rWaMxlODw2uji55ANn+t0/20elreUGvvY9FaRNbBJMo=",
"ref": "master",
"rev": "de522bdd533350b3afb41e1ce9b3afb72922fba2",
"revCount": 31,
"type": "git",
"url": "https://gitlab.com/rycee/nmd"
},
"original": {
"type": "git",
"url": "https://gitlab.com/rycee/nmd"
}
},
"obelisk": {
"flake": false,
"locked": {
"lastModified": 1650800768,
"narHash": "sha256-bBr61MzVFyIMpsri8inOPxCHSL1ogFbpb3uGS7pZ+X0=",
"owner": "tek",
"repo": "obelisk",
"rev": "7991b4d86be0ae04e22bcbe91607d99eb113c2c2",
"type": "github"
},
"original": {
"owner": "tek",
"ref": "ghc9",
"repo": "obelisk",
"type": "github"
}
},
"root": {
"inputs": {
"hix": "hix"
}
},
"thax": {
"locked": {
"lastModified": 1650770034,
"narHash": "sha256-OI4dxtNKgXogHBwYnNAfDABhHEcmWiBkkboZ+cCfyzk=",
"owner": "tek",
"repo": "thax",
"rev": "dc0025eebaba4ad97fcf48b68c14c8901d90daa5",
"type": "github"
},
"original": {
"owner": "tek",
"repo": "thax",
"type": "github"
}
}
},
"root": "root",
"version": 7
}

60
flake.nix Normal file
View File

@ -0,0 +1,60 @@
{
description = "vger";
inputs.hix.url = github:tek/hix;
outputs = { hix, ... }:
hix.lib.flake {
ghci.extensions = hix.inputs.nixpkgs.lib.mkForce [];
ghci.args = hix.inputs.nixpkgs.lib.mkForce [];
base = ./.;
packages = { vger = ./.; };
hpack.packages.vger = {
name = "vger";
executables.vger = {
ghc-options = [
"-Wunused-packages"
];
main = "Vger.hs";
source-dirs = "app";
dependencies = [
"vger"
"base"
"optparse-applicative"
"network-uri"
];
};
library = {
ghc-options = [
"-Wunused-packages"
];
source-dirs = "lib";
dependencies = [
"base"
"regex-pcre"
"regex-compat"
"mime-types"
"utf8-string"
"directory"
"text"
"network-uri"
];
};
tests.test = {
ghc-options = [
"-Wunused-packages"
];
main = "Main.hs";
source-dirs = "tests";
dependencies = [
"vger"
"base"
"HUnit"
"network-uri"
];
};
};
};
}

138
lib/Gemini.hs Normal file
View File

@ -0,0 +1,138 @@
module Gemini where
import Control.Exception
import Data.ByteString.UTF8
import Data.Text
import Network.Mime
import Network.URI
import System.Directory (doesDirectoryExist, doesFileExist)
import Text.Regex
import Text.Regex.PCRE
getFile :: FilePath -> IO (Maybe String)
getFile s = do
exists <- doesFileExist s
case exists of
True -> do
result <- try (readFile s) :: IO (Either SomeException String)
case result of
Left ex -> return Nothing
Right ex -> return (Just ex)
False -> return Nothing
-- remove any .. in the uri that could escape the location
sanitize_uri :: String -> String
sanitize_uri path =
subRegex (mkRegex "\\.\\.\\/") path ""
remove_crnlf :: String -> String
remove_crnlf uri =
subRegex (mkRegex "\r\n$") uri ""
-- return a file as a string
read_file :: String -> IO String
read_file path = do
text <- readFile path
pure text
-- read from stdin
get_request :: IO String
get_request = do
stdin <- getContents
pure stdin
data Answer = MkAnswer
{ code :: Int,
lang :: String,
content :: String,
mime :: String
}
get_mime :: String -> String
get_mime filename
| (fileNameExtensions (pack filename)) == [(pack "gmi" :: Extension)] = "text/gemini"
| (fileNameExtensions (pack filename)) == [(pack "gemini" :: Extension)] = "text/gemini"
| (fileNameExtensions (pack filename)) == [(pack "md" :: Extension)] = "text/markdown"
| otherwise = (toString (defaultMimeLookup (pack filename)))
geminiInvalidURI :: Answer
geminiInvalidURI =
MkAnswer
{ code = 50,
lang = "",
mime = "",
content = ""
}
create_answer :: String -> String -> String -> Bool -> IO Answer
create_answer request language baseDir vhost = do
-- parse the URI
case parseURI (remove_crnlf (sanitize_uri request)) of
Nothing -> return geminiInvalidURI
Just uri -> do
-- look for authority info
case uriAuthority uri of
Nothing -> pure geminiInvalidURI
-- request is valid from here
Just auth -> do
let domain = uriRegName auth
let file = uriPath uri
let baseDir' =
if vhost
then baseDir ++ "/" ++ domain
else baseDir
content <- getFile $ baseDir' ++ file
case content of
-- file has been found
Just x ->
return
( MkAnswer
{ code = 20,
lang = language,
mime = get_mime file,
content = x
}
)
Nothing -> do
-- no file
-- try /index.gmi
content' <- getFile $ baseDir' ++ file ++ "index.gmi"
case content' of
Just y ->
return
( MkAnswer
{ code = 20,
lang = language,
mime = get_mime $ file ++ "index.gmi",
content = y
}
)
Nothing -> do
isdir <- doesDirectoryExist $ baseDir' ++ file
if isdir
then
return
( MkAnswer
{ code = 31,
lang = "",
mime = file ++ "/",
content = ""
}
)
else
return
( MkAnswer
{ code = 52,
lang = language,
mime = "",
content = "can't find " ++ baseDir' ++ file
}
)
make_reply :: Answer -> String
make_reply (MkAnswer code lang content mime)
| code == 20 && mime == "text/gemini" && lang /= "" = "20 " ++ mime ++ "; lang=" ++ lang ++ " \r\n"
| code == 20 && mime == "text/gemini" && lang == "" = "20 " ++ mime ++ "; \r\n"
| code == 20 && mime /= "text/gemini" = "20 " ++ mime ++ "\r\n"
| code == 31 = "31 " ++ mime ++ "\r\n"
| otherwise = "55\r\n"

3
lib/TODO Normal file
View File

@ -0,0 +1,3 @@
-- before reading a file, check if it's a symbolic link and if so, read the target
-- cffi foreign import

97
main.c
View File

@ -1,97 +0,0 @@
#include "vger.c"
int
main(int argc, char **argv)
{
char request[GEMINI_REQUEST_MAX] = {'\0'};
char user[_SC_LOGIN_NAME_MAX] = {'\0'};
char hostname[GEMINI_REQUEST_MAX] = {'\0'};
char path[GEMINI_REQUEST_MAX] = {'\0'};
char query[GEMINI_REQUEST_MAX] = {'\0'};
char cgi_dir[PATH_MAX] = {'\0'};
char rel_cgi_dir[PATH_MAX] = {'\0'};
char chroot_dir[PATH_MAX] = DEFAULT_CHROOT;
char tmp[PATH_MAX] = {'\0'};
int option = 0;
int virtualhost = 0;
/*
* request : contain the whole request from client : gemini://...\r\n
* user : username, used in drop_privileges()
* hostname : extracted from hostname. used with virtualhosts and cgi SERVER_NAME
* query : after a ? in cgi : gemini://...?query
*/
while ((option = getopt(argc, argv, ":d:l:m:u:c:vi")) != -1) {
switch (option) {
case 'd':
esnprintf(chroot_dir, sizeof(chroot_dir), "%s", optarg);
break;
case 'l':
esnprintf(lang, sizeof(lang), "lang=%s", optarg);
break;
case 'm':
esnprintf(default_mime, sizeof(default_mime), "%s", optarg);
break;
case 'u':
esnprintf(user, sizeof(user), "%s", optarg);
break;
case 'c':
esnprintf(rel_cgi_dir, sizeof(rel_cgi_dir), "%s", optarg);
/* remove leading / */
while (*rel_cgi_dir == '/')
memmove(rel_cgi_dir, rel_cgi_dir+1,
strlen(rel_cgi_dir)); // strlen +1-1
break;
case 'v':
virtualhost = 1;
break;
case 'i':
doautoidx = 1;
break;
}
}
read_request(request);
split_request(request, hostname, path, query);
/* do chroot if an user is supplied */
if (*user)
drop_privileges(user);
/* set actual chroot_dir */
if (virtualhost) {
esnprintf(tmp, sizeof(tmp), "%s/%s", chroot_dir, hostname);
esnprintf(chroot_dir, sizeof(chroot_dir), "%s", tmp);
}
/* cgi_dir is in chroot_dir */
if (*rel_cgi_dir)
esnprintf(cgi_dir, sizeof(cgi_dir),
"%s/%s", chroot_dir, rel_cgi_dir);
set_rootdir(chroot_dir, cgi_dir, user);
if (strlen(path) == 0) { /* this is root dir */
esnprintf(path, sizeof(path), "./");
} else {
uridecode(path);
remove_double_dot(path);
}
uridecode(query);
/* is it cgi ? */
if (*cgi_dir)
if (do_cgi(rel_cgi_dir, path, hostname, query) == 0)
stop(EXIT_SUCCESS, NULL);
/* *** from here, cgi didn't run *** */
/* check if path available */
check_path(path, sizeof(path));
/* regular file to stdout */
display_file(path);
stop(EXIT_SUCCESS, NULL);
}

139
mimes.c
View File

@ -1,139 +0,0 @@
#include <string.h>
#include <unistd.h>
#include <string.h>
#include "mimes.h"
#include "opts.h"
/* extension to mimetype table */
static const struct {
const char *extension;
const char *type;
} database[] = {
{"gmi", "text/gemini"},
{"gemini", "text/gemini"},
{"7z", "application/x-7z-compressed"},
{"atom", "application/atom+xml"},
{"avi", "video/x-msvideo"},
{"bin", "application/octet-stream"},
{"bmp", "image/x-ms-bmp"},
{"cco", "application/x-cocoa"},
{"crt", "application/x-x509-ca-cert"},
{"css", "text/css"},
{"deb", "application/octet-stream"},
{"dll", "application/octet-stream"},
{"dmg", "application/octet-stream"},
{"doc", "application/msword"},
{"eot", "application/vnd.ms-fontobject"},
{"exe", "application/octet-stream"},
{"flv", "video/x-flv"},
{"fs", "application/octet-stream"},
{"gif", "image/gif"},
{"hqx", "application/mac-binhex40"},
{"htc", "text/x-component"},
{"html", "text/html"},
{"ico", "image/x-icon"},
{"img", "application/octet-stream"},
{"iso", "application/octet-stream"},
{"jad", "text/vnd.sun.j2me.app-descriptor"},
{"jar", "application/java-archive"},
{"jardiff", "application/x-java-archive-diff"},
{"jng", "image/x-jng"},
{"jnlp", "application/x-java-jnlp-file"},
{"jpeg", "image/jpeg"},
{"jpg", "image/jpeg"},
{"js", "application/javascript"},
{"json", "application/json"},
{"kml", "application/vnd.google-earth.kml+xml"},
{"kmz", "application/vnd.google-earth.kmz"},
{"m3u8", "application/vnd.apple.mpegurl"},
{"m4a", "audio/x-m4a"},
{"m4v", "video/x-m4v"},
{"md", "text/markdown"},
{"mid", "audio/midi"},
{"midi", "audio/midi"},
{"mkv", "video/x-matroska"},
{"mml", "text/mathml"},
{"mng", "video/x-mng"},
{"mov", "video/quicktime"},
{"mp3", "audio/mpeg"},
{"mp4", "video/mp4"},
{"mpeg", "video/mpeg"},
{"mpg", "video/mpeg"},
{"msi", "application/octet-stream"},
{"msm", "application/octet-stream"},
{"msp", "application/octet-stream"},
{"odb", "application/vnd.oasis.opendocument.database"},
{"odc", "application/vnd.oasis.opendocument.chart"},
{"odf", "application/vnd.oasis.opendocument.formula"},
{"odg", "application/vnd.oasis.opendocument.graphics"},
{"odi", "application/vnd.oasis.opendocument.image"},
{"odm", "application/vnd.oasis.opendocument.text-master"},
{"odp", "application/vnd.oasis.opendocument.presentation"},
{"ods", "application/vnd.oasis.opendocument.spreadsheet"},
{"odt", "application/vnd.oasis.opendocument.text"},
{"ogg", "audio/ogg"},
{"oth", "application/vnd.oasis.opendocument.text-web"},
{"otp", "application/vnd.oasis.opendocument.presentation-template"},
{"pac", "application/x-ns-proxy-autoconfig"},
{"pdf", "application/pdf"},
{"pem", "application/x-x509-ca-cert"},
{"pl", "application/x-perl"},
{"pm", "application/x-perl"},
{"png", "image/png"},
{"ppt", "application/vnd.ms-powerpoint"},
{"ps", "application/postscript"},
{"ra", "audio/x-realaudio"},
{"rar", "application/x-rar-compressed"},
{"rpm", "application/x-redhat-package-manager"},
{"rss", "application/rss+xml"},
{"rtf", "application/rtf"},
{"run", "application/x-makeself"},
{"sea", "application/x-sea"},
{"sit", "application/x-stuffit"},
{"svg", "image/svg+xml"},
{"svgz", "image/svg+xml"},
{"swf", "application/x-shockwave-flash"},
{"tcl", "application/x-tcl"},
{"tif", "image/tiff"},
{"tiff", "image/tiff"},
{"tk", "application/x-tcl"},
{"ts", "video/mp2t"},
{"txt", "text/plain"},
{"war", "application/java-archive"},
{"wbmp", "image/vnd.wap.wbmp"},
{"webm", "video/webm"},
{"webp", "image/webp"},
{"wml", "text/vnd.wap.wml"},
{"wmlc", "application/vnd.wap.wmlc"},
{"wmv", "video/x-ms-wmv"},
{"woff", "application/font-woff"},
{"xhtml", "application/xhtml+xml"},
{"xls", "application/vnd.ms-excel"},
{"xml", "text/xml"},
{"xpi", "application/x-xpinstall"},
{"zip", "application/zip"}
};
#ifndef nitems
#define nitems(_a) (sizeof((_a)) / sizeof((_a)[0]))
#endif
const char *
get_file_mime(const char *path, const char *default_mime)
{
size_t i;
char *extension;
/* search for extension after last '.' in path */
if ((extension = strrchr(path, '.')) != NULL) {
/* look for the MIME in the database */
for (i = 0; i < nitems(database); i++) {
if (strcmp(database[i].extension, extension + 1) == 0)
return (database[i].type);
}
}
/* no MIME found, set a default one */
return default_mime;
}

View File

@ -1 +0,0 @@
const char *get_file_mime(const char *, const char *);

16
opts.h
View File

@ -1,16 +0,0 @@
#include <limits.h> /* PATH_MAX */
/* Defaults values */
#define DEFAULT_MIME "application/octet-stream"
#define DEFAULT_LANG ""
#define DEFAULT_CHROOT "/var/gemini"
#define DEFAULT_INDEX "index.gmi"
#define DEFAULT_AUTOIDX 0
/*
* Options used later
*/
/* longest hardcoded mimetype is 56 long so 64 should be enough */
static char default_mime[64] = DEFAULT_MIME;
static char lang[16] = DEFAULT_LANG;
static unsigned int doautoidx = DEFAULT_AUTOIDX;

View File

@ -1,4 +0,0 @@
with (import <nixpkgs> {});
mkShell {
buildInputs = [ gcc libbsd gnumake ];
}

50
tests.c
View File

@ -1,50 +0,0 @@
#include <stdlib.h>
#include <string.h>
#include <stdio.h>
#include "vger.h"
// to test
void test_status(void);
void test_status_error(void);
void test_uridecode(char*, const int);
void
test_uridecode(char *str, const int result)
{
char reference[GEMINI_REQUEST_MAX] = {'\0'};
strlcpy(reference, str, sizeof(reference));
uridecode(str);
if (strncmp(reference, str, strlen(str)) != result)
{
printf("uridecode error\n");
printf("Strings should be %s\n", (result == 0) ? "identical" : "different");
printf("passed %s\n", reference);
printf("got %s\n", str);
exit(1);
}
}
void
test_status(void)
{
status(20, "text/gemini");
}
void
test_status_error(void)
{
status(51, "file not found");
status(50, "Forbidden path");
status(50, "Internal server error");
}
int
main(void)
{
test_status_error();
test_status();
//test_uridecode("host.name", 0);
//test_uridecode("host.name/percent%25-encode%3.gmi", 1);
return(0);
}

41
tests/Main.hs Normal file
View File

@ -0,0 +1,41 @@
module Main where
import Test.HUnit
import Gemini
regex_10 = TestCase (assertEqual
"ensure sanitization works"
"gemini://perso.pw/main.gmi/passwd"
(sanitize_uri "gemini://perso.pw/../../../main.gmi/../../passwd"))
mime_1 = TestCase (assertEqual
"gmi files should be text/gemini"
"text/gemini"
(get_mime "main.gmi"))
mime_2 = TestCase (assertEqual
"markdown file"
"text/plain"
(get_mime "file.txt"))
mime_3 = TestCase (assertEqual
"jpeg file"
"image/jpeg"
(get_mime "picture.jpg"))
mime_4 = TestCase (assertEqual
"png file"
"image/png"
(get_mime "picture.png"))
tests = TestList
[ regex_10
, mime_1
, mime_2
, mime_3
, mime_4
]
main :: IO Counts
main = do
runTestTT tests

View File

@ -12,113 +12,109 @@ MD5()
}
# serving a file
OUT=$(printf "gemini://host.name/main.gmi\r\n" | ../vger -d var/gemini/ | tee /dev/stderr | MD5)
OUT=$(printf "gemini://host.name/main.gmi\r\n" | ../result/bin/vger -d var/gemini/ | tee /dev/stderr | MD5)
if ! [ $OUT = "c7e352d6aae4ee7e7604548f7874fb9d" ] ; then echo "error" ; exit 1 ; fi
# default index.gmi file
OUT=$(printf "gemini://host.name\r\n" | ../vger -d var/gemini/ | tee /dev/stderr | MD5)
OUT=$(printf "gemini://host.name\r\n" | ../result/bin/vger -d var/gemini/ | tee /dev/stderr | MD5)
if ! [ $OUT = "fcc5a293f316e01f7b3103f97eca26b1" ] ; then echo "error" ; exit 1 ; fi
# default index.gmi file when using a trailing slash
OUT=$(printf "gemini://host.name/\r\n" | ../vger -d var/gemini/ | tee /dev/stderr | MD5)
OUT=$(printf "gemini://host.name/\r\n" | ../result/bin/vger -d var/gemini/ | tee /dev/stderr | MD5)
if ! [ $OUT = "fcc5a293f316e01f7b3103f97eca26b1" ] ; then echo "error" ; exit 1 ; fi
# default index.gmi file when client specify port
OUT=$(printf "gemini://host.name:1965\r\n" | ../vger -d var/gemini/ | tee /dev/stderr | MD5)
OUT=$(printf "gemini://host.name:1965\r\n" | ../result/bin/vger -d var/gemini/ | tee /dev/stderr | MD5)
if ! [ $OUT = "fcc5a293f316e01f7b3103f97eca26b1" ] ; then echo "error" ; exit 1 ; fi
# redirect to uri with trailing / if directory
OUT=$(printf "gemini://host.name/subdir\r\n" | ../vger -d var/gemini/ | tee /dev/stderr | MD5)
OUT=$(printf "gemini://host.name/subdir\r\n" | ../result/bin/vger -d var/gemini/ | tee /dev/stderr | MD5)
if ! [ $OUT = "84e5e7bb3eee0dfcc8db14865dc83e77" ] ; then echo "error" ; exit 1 ; fi
# redirect to uri with trailing / if directory and vhost enabled
OUT=$(printf "gemini://perso.pw/cgi-bin\r\n" | ../vger -vd var/gemini | tee /dev/stderr | MD5)
OUT=$(printf "gemini://perso.pw/cgi-bin\r\n" | ../result/bin/vger -v -d var/gemini | tee /dev/stderr | MD5)
if ! [ $OUT = "e0eb3a8e31bdb30c89d92d1d2b0a1fa1" ] ; then echo "error" ; exit 1 ; fi
# file from local directory with lang=fr and markdown MIME type
OUT=$(printf "gemini://perso.pw/file.md\r\n" | ../vger -d var/gemini/ -l fr | tee /dev/stderr | MD5)
OUT=$(printf "gemini://perso.pw/file.md\r\n" | ../result/bin/vger -d var/gemini/ -l fr | tee /dev/stderr | MD5)
if ! [ $OUT = "09c82ffe243ce3b3cfb04c2bc4a91acb" ] ; then echo "error" ; exit 1 ; fi
# file from local directory with lang=fr and unknown MIME type (default to application/octet-stream)
OUT=$(printf "gemini://perso.pw/foobar.unknown\r\n" | ../vger -d var/gemini/ -l fr | tee /dev/stderr | MD5)
OUT=$(printf "gemini://perso.pw/foobar.unknown\r\n" | ../result/bin/vger -d var/gemini/ -l fr | tee /dev/stderr | MD5)
if ! [ $OUT = "2c73bfb33dd2d12be322ebb85e03c015" ] ; then echo "error" ; exit 1 ; fi
# file from local directory and unknown MIME type, default forced to text/plain
OUT=$(printf "gemini://perso.pw/foobar.unknown\r\n" | ../vger -d var/gemini/ -m text/plain | tee /dev/stderr | MD5)
if ! [ $OUT = "8169f43fbb2032f4054b153c38fe61d6" ] ; then echo "error" ; exit 1 ; fi
# redirect file
OUT=$(printf "gemini://perso.pw/old_location\r\n" | ../vger -d var/gemini/ | tee /dev/stderr | MD5)
OUT=$(printf "gemini://perso.pw/old_location\r\n" | ../result/bin/vger -d var/gemini/ | tee /dev/stderr | MD5)
if ! [ $OUT = "cb4597b6fcc82cbc366ac9002fb60dac" ] ; then echo "error" ; exit 1 ; fi
# file from local directory using virtualhosts
OUT=$(printf "gemini://perso.pw/index.gmi\r\n" | ../vger -v -d var/gemini/ | tee /dev/stderr | MD5)
OUT=$(printf "gemini://perso.pw/index.gmi\r\n" | ../result/bin/vger -v -d var/gemini/ | tee /dev/stderr | MD5)
if ! [ $OUT = "5e5fca557e79f4521b21d4b81dc964c6" ] ; then echo "error" ; exit 1 ; fi
# file from local directory using virtualhosts without specifying a file
OUT=$(printf "gemini://perso.pw\r\n" | ../vger -v -d var/gemini/ | tee /dev/stderr | MD5)
OUT=$(printf "gemini://perso.pw\r\n" | ../result/bin/vger -v -d var/gemini/ | tee /dev/stderr | MD5)
if ! [ $OUT = "5e5fca557e79f4521b21d4b81dc964c6" ] ; then echo "error" ; exit 1 ; fi
# file from local directory using virtualhosts without specifying a file using lang = fr
OUT=$(printf "gemini://perso.pw/\r\n" | ../vger -v -d var/gemini/ -l fr | tee /dev/stderr | MD5)
OUT=$(printf "gemini://perso.pw/\r\n" | ../result/bin/vger -v -d var/gemini/ -l fr | tee /dev/stderr | MD5)
if ! [ $OUT = "7db981ce93fee268f29324912800f00d" ] ; then echo "error" ; exit 1 ; fi
# file from local directory using virtualhosts and IRI
OUT=$(printf "gemini://virtualhoßt/é è.gmi\r\n" | ../vger -v -d var/gemini/ | tee /dev/stderr | MD5)
OUT=$(printf "gemini://virtualhoßt/é è.gmi\r\n" | ../result/bin/vger -v -d var/gemini/ | tee /dev/stderr | MD5)
if ! [ $OUT = "282cee071d3bd20dbb6e6af38f217a29" ] ; then echo "error" ; exit 1 ; fi
# file from local directory using virtualhosts and IRI both with emojis
OUT=$(printf "gemini://⛴//❤️.gmi\r\n" | ../vger -v -d var/gemini/ | tee /dev/stderr | MD5)
OUT=$(printf "gemini://⛴//❤️.gmi\r\n" | ../result/bin/vger -v -d var/gemini/ | tee /dev/stderr | MD5)
if ! [ $OUT = "e354a1a29ea8273faaf0cdc29c1d8583" ] ; then echo "error" ; exit 1 ; fi
# auto index in directory without index.gmi must redirect
OUT=$(printf "gemini://host.name/autoidx\r\n" | ../vger -d var/gemini/ -i | tee /dev/stderr | MD5)
OUT=$(printf "gemini://host.name/autoidx\r\n" | ../result/bin/vger -d var/gemini/ -i | tee /dev/stderr | MD5)
if ! [ $OUT = "874f5e1af67eff6b93bedf8ac8033066" ] ; then echo "error" ; exit 1 ; fi
# auto index in directory
OUT=$(printf "gemini://host.name/autoidx/\r\n" | ../vger -d var/gemini/ -i | tee /dev/stderr | MD5)
OUT=$(printf "gemini://host.name/autoidx/\r\n" | ../result/bin/vger -d var/gemini/ -i | tee /dev/stderr | MD5)
if ! [ $OUT = "765bbbe2add810be8eb191bbde59e258" ] ; then echo "error" ; exit 1 ; fi
# cgi simple script
OUT=$(printf "gemini://host.name/cgi-bin/test.cgi\r\n" | ../vger -d var/gemini/ -c cgi-bin | tee /dev/stderr | MD5)
OUT=$(printf "gemini://host.name/cgi-bin/test.cgi\r\n" | ../result/bin/vger -d var/gemini/ -c var/gemini/cgi-bin | tee /dev/stderr | MD5)
if ! [ $OUT = "666e48200f90018b5e96c2cf974882dc" ] ; then echo "error" ; exit 1 ; fi
# cgi with use of variables
OUT=$(printf "gemini://host.name/cgi-bin/who.cgi?user=jean-mi\r\n" | ../vger -d var/gemini/ -c cgi-bin | tee /dev/stderr | MD5)
OUT=$(printf "gemini://host.name/cgi-bin/who.cgi?user=jean-mi\r\n" | ../result/bin/vger -d var/gemini/ -c var/gemini/cgi-bin | tee /dev/stderr | MD5)
if ! [ $OUT = "fa065a67d1f7c973501d4a9e3ca2ea57" ] ; then echo "error" ; exit 1 ; fi
# cgi with error
OUT=$(printf "gemini://host.name/cgi-bin/nope\r\n" | ../vger -d var/gemini/ -c cgi-bin | tee /dev/stderr | MD5)
OUT=$(printf "gemini://host.name/cgi-bin/nope\r\n" | ../result/bin/vger -d var/gemini/ -c var/gemini/cgi-bin | tee /dev/stderr | MD5)
if ! [ $OUT = "31b98e160402a073298c12f763d5db64" ] ; then echo "error" ; exit 1 ; fi
# cgi with PATH_INFO
OUT=$(printf "gemini://host.name/cgi-bin/test.cgi/path/info\r\n" | ../vger -d var/gemini -c cgi-bin | tee /dev/stderr | MD5)
OUT=$(printf "gemini://host.name/cgi-bin/test.cgi/path/info\r\n" | ../result/bin/vger -d var/gemini -c var/gemini/cgi-bin | tee /dev/stderr | MD5)
if ! [ $OUT = "ec64da76dc578ffb479fbfb23e3a7a5b" ] ; then echo "error" ; exit 1 ; fi
# virtualhost + cgi
OUT=$(printf "gemini://perso.pw/cgi-bin/test.cgi\r\n" | ../vger -v -d var/gemini/ -c cgi-bin | tee /dev/stderr | MD5)
OUT=$(printf "gemini://perso.pw/cgi-bin/test.cgi\r\n" | ../result/bin/vger -v -d var/gemini/ -c var/gemini/perso.pw/cgi-bin | tee /dev/stderr | MD5)
if ! [ $OUT = "666e48200f90018b5e96c2cf974882dc" ] ; then echo "error" ; exit 1 ; fi
# percent-decoding
OUT=$(printf "%s\r\n" "gemini://host.name/percent%25-encode%3f.gmi" | ../vger -d var/gemini/ | tee /dev/stderr | MD5)
OUT=$(printf "%s\r\n" "gemini://host.name/percent%25-encode%3f.gmi" | ../result/bin/vger -d var/gemini/ | tee /dev/stderr | MD5)
if ! [ $OUT = "83d59cca9ed7040145ac6df1992f5daf" ] ; then echo "error" ; exit 1 ; fi
# percent-decoding failing
OUT=$(printf "%s\r\n" "gemini://host.name/percent%25-encode%3.gmi" | ../vger -d var/gemini/ | tee /dev/stderr | MD5)
OUT=$(printf "%s\r\n" "gemini://host.name/percent%25-encode%3.gmi" | ../result/bin/vger -d var/gemini/ | tee /dev/stderr | MD5)
if ! [ $OUT = "c782da4173898f57033a0804b8e96fc3" ] ; then echo "error" ; exit 1 ; fi
# must fail only on OpenBSD !
# try to escape from unveil
if [ -f /bsd ]
then
OUT=$(printf "gemini://fail_on_openbsd/../../test.sh\r\n" | ../vger -d var/gemini/ -l fr | tee /dev/stderr | MD5)
OUT=$(printf "gemini://fail_on_openbsd/../../test.sh\r\n" | ../result/bin/vger -d var/gemini/ -l fr | tee /dev/stderr | MD5)
if [ $OUT = "$( ( printf '20 text/gemini; lang=fr\r\n' ; cat $0) | MD5)" ] ; then echo "error" ; exit 1 ; fi
fi
#type doas 2>/dev/null
#if [ $? -eq 0 ]; then
# # file from local directory chroot
# OUT=$(printf "gemini://perso.pw\r\n" | doas ../vger -v -d var/gemini/ -u solene -l fr | tee /dev/stderr | MD5)
# OUT=$(printf "gemini://perso.pw\r\n" | doas ../result/bin/vger -v -d var/gemini/ -u solene -l fr | tee /dev/stderr | MD5)
# if ! [ $OUT = "7db981ce93fee268f29324912800f00d" ] ; then echo "error" ; exit 1 ; fi
#fi

124
utils.c
View File

@ -1,124 +0,0 @@
#include <sys/types.h>
#include <err.h>
#include <errno.h>
#include <regex.h>
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <syslog.h>
#include <unistd.h>
#include "utils.h"
#include "vger.h"
#if defined(__OpenBSD__) || defined(__FreeBSD__) || defined( __NetBSD__) || defined(__DragonFly__)
#include <string.h>
#else
#include <bsd/string.h>
#endif
/* e*foo() functions are the equivalent of foo() but handle errors.
* In case an error happens:
* The error is printed to stdout
* return 1
*/
#ifdef __OpenBSD__
void
eunveil(const char *path, const char *permissions)
{
if (unveil(path, permissions) == -1) {
status(41, "Error when unveil(), see logs");
stop(EXIT_FAILURE, "unveil on %s failed", path);
}
}
void
epledge(const char *promises, const char *execpromises)
{
if (pledge(promises, execpromises) == -1) {
status(41, "Error when pledge(), see logs");
stop(EXIT_FAILURE, "pledge failed for: %s", promises);
}
}
#endif
size_t
esnprintf(char *str, size_t size, const char *format, ...)
{
/* usage : esnprintf(str, sizeof(str), "%s ... %s", arg1, arg2); */
va_list ap;
size_t ret = 0;
va_start(ap, format);
ret = vsnprintf(str, size, format, ap);
va_end(ap);
if (ret < 0 || ret >= size) {
status(41, "vnsprintf failed: Output trunkated");
stop(EXIT_FAILURE, "vsnprintf: Output trunkated");
}
return ret;
}
int
esetenv(const char *name, const char *value, int overwrite)
{
int ret = 0;
ret = setenv(name, value, overwrite);
if (ret != 0) {
status(41, "setenv() failed, see logs");
stop(EXIT_FAILURE, "setenv() %s:%s", name, value);
}
return ret;
}
void
echdir(const char *path)
{
if (chdir(path) == -1) {
switch (errno) {
case ENOTDIR: /* FALLTHROUGH */
case ENOENT:
status(51, "file not found");
break;
case EACCES:
status(50, "Forbidden path");
break;
default:
status(50, "Internal server error");
break;
}
stop(EXIT_FAILURE, "chdir(%s) failed", path);
}
}
/* read the file fd byte after byte in buffer and write it to stdout
* return number of bytes read
*/
size_t
print_file(FILE *fd)
{
ssize_t nread = 0;
ssize_t datasent = 0;
char *buffer[BUFSIZ];
while ((nread = fread(buffer, 1, sizeof(buffer), fd)) != 0)
datasent += fwrite(buffer, 1, nread, stdout);
return datasent;
}
void
getsubexp(const char *str, regmatch_t m, char *dst)
{
size_t len = 0;
if ((len = m.rm_eo - m.rm_so) > 0) { /* skip empty substring */
len = m.rm_eo - m.rm_so;
memcpy(dst, str + m.rm_so, len);
dst[len] = '\0';
}
}

View File

@ -1,8 +0,0 @@
void getsubexp(const char *, regmatch_t, char *);
void echdir(const char *);
void epledge(const char *, const char *);
void eunveil(const char *, const char *);
int esetenv(const char *, const char *, int);
size_t esnprintf(char *, size_t, const char *, ...);
size_t print_file(FILE *fd);
void set_errmsg(const char *, ...);

5
vger.8
View File

@ -45,10 +45,11 @@ will read the file /var/gemini/hostname.example/file.gmi
Enable CGI support.
.Ar cgi_path
files will be executed as a cgi script instead of returning their content.
.Ar cgi_path should be relative to chroot so cgi can be called for different virtualhosts.
.Ar cgi_path must not end with '/'.
If using virtualhost, you must insert the virtualhost directory in the cgi path.
As example, for a request gemini://hostname.example/cgi-bin/hello.cgi, one must set:
.Bd -literal -offset indent
vger -c cgi-bin
vger -c /var/gemini/hostname.example/cgi-bin/hello.cgi
.Ed
.Pp
In this case,

468
vger.c
View File

@ -1,468 +0,0 @@
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/wait.h>
#include <ctype.h>
#include <dirent.h>
#include <err.h>
#include <errno.h>
#include <fcntl.h>
#include <limits.h>
#include <pwd.h>
#include <regex.h>
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <syslog.h>
#include <unistd.h>
#include "mimes.h"
#include "opts.h"
#include "utils.h"
#include "vger.h"
void
stop(const int r, const char *fmt, ...)
{
va_list ap, ap2;
fflush(stdout); /* ensure all data is sent */
/* log the request and retcode */
syslog(LOG_DAEMON|LOG_INFO, "\"%s\" %i %zd", _request, _retcode, _datasiz);
if (r != EXIT_SUCCESS) {
/* log and print error */
va_copy(ap2, ap);
va_start(ap, fmt);
vsyslog(LOG_ERR, fmt, ap);
va_end(ap);
va_start(ap2, fmt);
vfprintf(stderr, fmt, ap2);
va_end(ap2);
}
exit(r);
}
void
status(const int code, const char *fmt, ...)
{
va_list ap;
_datasiz += fprintf(stdout, "%i ", code);
va_start(ap, fmt);
_datasiz += vfprintf(stdout, fmt, ap);
va_end(ap);
_datasiz += fprintf(stdout, "\r\n"); /* make sure status end correctly */
_retcode = code; /* store return code for logs */
}
int
uridecode(char *uri)
{
int n = 0;
char c = '\0';
long l = 0;
char *pos = NULL;
if ((pos = strchr(uri, '%')) == NULL)
return n;
while ((pos = strchr(pos, '%')) != NULL) {
if (strlen(pos) < 3)
return n;
char hex[3] = {'\0'};
for (size_t i = 0; i < 2; i++)
hex[i] = tolower(pos[i + 1]);
errno = 0;
l = strtol(hex, 0, 16);
if (errno == ERANGE && (l == LONG_MAX || l == LONG_MIN))
continue; /* conversion failed */
c = (char)l;
pos[0] = c;
/* rewind of two char to remove %hex */
memmove(pos + 1, pos + 3, strlen(pos + 3) + 1); /* +1 for \0 */
n++;
pos++; /* avoid infinite loop */
}
return n;
}
void
drop_privileges(const char *user)
{
struct passwd *pw;
/*
* use chroot() if an user is specified requires root user to be
* running the program to run chroot() and then drop privileges
*/
/* is root? */
if (getuid() != 0) {
status(41, "privileges issue, see logs");
stop(EXIT_FAILURE, "%s",
"chroot requires program to be run as root");
}
/* search user uid from name */
if ((pw = getpwnam(user)) == NULL) {
status(41, "privileges issue, see logs");
stop(EXIT_FAILURE,
"the user %s can't be found on the system", user);
}
/* drop privileges */
if (setgroups(1, &pw->pw_gid) ||
setresgid(pw->pw_gid, pw->pw_gid, pw->pw_gid) ||
setresuid(pw->pw_uid, pw->pw_uid, pw->pw_uid)) {
status(41, "privileges issue, see logs");
stop(EXIT_FAILURE,
"dropping privileges to user %s (uid=%i) failed", \
user, pw->pw_uid);
}
}
void
set_rootdir(const char *chroot_dir, const char *cgi_dir, const char *user)
{
char capsule_dir[PATH_MAX] = {'\0'};
if (*user) {
/* chroot worked? */
if (chroot(chroot_dir) != 0) {
status(41, "privileges issue, see logs");
stop(EXIT_FAILURE,
"the chroot_dir %s can't be used for chroot", chroot_dir);
}
/* now chroot_dir is / */
esnprintf(capsule_dir, sizeof(capsule_dir), "%s", "/");
} else {
esnprintf(capsule_dir, sizeof(capsule_dir), "%s", chroot_dir);
}
#ifdef __OpenBSD__
/*
* prevent access to files other than the one in chroot_dir
*/
eunveil(capsule_dir, "r");
/* permission to execute what's inside cgi_dir */
if (*cgi_dir)
eunveil(cgi_dir, "rx");
eunveil(NULL, NULL); /* no more call to unveil() */
/* promise permissions */
if (*cgi_dir)
epledge("stdio rpath exec proc", NULL);
else
epledge("stdio rpath", NULL);
#endif
echdir(capsule_dir); /* move to the gemini data directory */
}
ssize_t
display_file(const char *path)
{
FILE *fd = NULL;
const char *file_mime;
/*
* special case : path ends with "/". The user requested a dir
*/
if ((path[strlen(path)-1] == '/') && (doautoidx)) {
/* no index.gmi, so display autoindex if enabled */
_datasiz += autoindex(path);
return _datasiz;
}
/* open the file requested */
if ((fd = fopen(path, "r")) != NULL) {
file_mime = get_file_mime(path, default_mime);
if (strcmp(file_mime, "text/gemini") == 0)
status(20, "%s; %s", file_mime, lang);
else
status(20, "%s", file_mime);
_datasiz += print_file(fd);
fclose(fd); /* close file descriptor */
} else {
/* return an error code and no content.
* seems unlikely to happen unless the file vanished
* since we checked with stat() if it exists
*/
status(51, "%s", "file not found and may have vanished");
}
return _datasiz;
}
int
do_cgi(const char *rel_cgi_dir, const char *path, const char *hostname, const char *query)
{
struct stat sbcgi = {0};
struct stat sbpath = {0};
char cgifp[PATH_MAX] = {'\0'}; /* cgi file to execute */
char path_dir[PATH_MAX] = {'\0'};
char *path_info = NULL;
/* get beginning of path */
/* path_dir is initialized so there is an \0 at the end */
memcpy(path_dir, path, strlen(rel_cgi_dir));
if (stat(rel_cgi_dir, &sbcgi) + stat(path_dir, &sbpath) != 0)
goto nocgi;
/* compare inodes */
if (sbcgi.st_ino != sbpath.st_ino)
goto nocgi; /* not in cgi_dir, go to display_file */
/* set env variables for CGI
* see
* https://lists.orbitalfox.eu/archives/gemini/2020/000315.html
*/
esetenv("GATEWAY_INTERFACE", "CGI/1.1", 1);
esetenv("SERVER_PROTOCOL", "GEMINI", 1);
esetenv("SERVER_SOFTWARE", "vger/1", 1);
if (*query)
esetenv("QUERY_STRING", query, 1);
/*
* if in cgi_dir, only the first file after cgi_dir/FILE
* is to be executed
* the rest is PATH_INFO
*/
/* find next item after cgi_dir in path:
* path + strlen(rel_cgi_dir) + 1 (skip '/')
*/
/* cgi file to execute */
esnprintf(cgifp, sizeof(cgifp), "%s", path + strlen(rel_cgi_dir) + 1);
if (!(*cgifp)) /* problem with cgi file, abort */
goto nocgi;
/* check if there is something after cgi file for PATH_INFO */
path_info = strchr(cgifp, '/');
if (path_info != NULL) {
esetenv("PATH_INFO", path_info, 1);
*path_info = '\0'; /* stop cgifp before PATH_INFO */
}
esetenv("SCRIPT_NAME", cgifp, 1);
esetenv("SERVER_NAME", hostname, 1);
echdir(rel_cgi_dir);
cgi(cgifp);
return 0;
nocgi:
return 1;
}
ssize_t
autoindex(const char *path)
{
/* display list of files in path + a link to parent (..) */
int n = 0;
struct dirent **namelist; /* this must be freed at last */
size_t bs = 0;
/* use alphasort to always have the same order on every system */
if ((n = scandir(path, &namelist, NULL, alphasort)) < 0) {
status(50, "Can't scan %s", path);
} else {
status(20, "text/gemini");
bs += fprintf(stdout, "=> .. ../\n"); /* display link to parent */
for (int j = 0; j < n; j++) {
/* skip self and parent */
if ((strcmp(namelist[j]->d_name, ".") == 0) ||
(strcmp(namelist[j]->d_name, "..") == 0)) {
continue;
}
/* add "/" at the end of a directory path */
if (namelist[j]->d_type == DT_DIR) {
bs += fprintf(stdout, "=> ./%s/ %s/\n",
namelist[j]->d_name, namelist[j]->d_name);
} else {
bs += fprintf(stdout, "=> ./%s %s\n",
namelist[j]->d_name, namelist[j]->d_name);
}
free(namelist[j]);
}
free(namelist);
}
return bs;
}
void
cgi(const char *cgicmd)
{
int fildes[2] = {0};
int retcode = 0;
pid_t pid = 0;
FILE *output = NULL;
if (pipe(fildes) != 0)
goto cgierr;
if ((pid = fork()) < 0)
goto cgierr;
if (pid > 0) { /* parent */
close(fildes[1]); /* make sure entry is closed to get EOF */
if ((output = fdopen(fildes[0], "r")) == NULL)
goto cgierr;
_datasiz += print_file(output);
close(fildes[0]);
fclose(output);
waitpid(pid, &retcode, 0);
stop(EXIT_SUCCESS, "cgi ran with exit code %d", status);
} else { /* child */
/* set pipe output equal to stdout & stderr */
dup2(fildes[1], STDOUT_FILENO);
close(fildes[1]); /* no longer required */
execl(cgicmd, cgicmd, NULL);
}
cgierr:
/* if execl is ok, this will never be reached */
close(fildes[0]);
close(fildes[1]);
status(42, "error when trying run cgi");
stop(EXIT_FAILURE, "error when trying to execl %s", cgicmd);
}
char *
read_request(char *request)
{
/* read 1024 +1 chars from stdin to get the request (1024 + \0) */
if (fgets(request, GEMINI_REQUEST_MAX, stdin) == NULL) {
/* EOF reached before reading anything */
if (feof(stdin)) {
status(59, "%s", "request is too short and probably empty");
stop(EXIT_FAILURE, "%s", "request is too short and probably empty");
/* error before reading anything */
} else if (ferror(stdin)) {
status(59, "Error while reading request: %s", request);
stop(EXIT_FAILURE, "Error while reading request: %s", request);
}
}
/* check if string ends with '\n', or to long */
if (request[strnlen(request, GEMINI_REQUEST_MAX) - 1] != '\n') {
status(59, "request is too long (1024 max): %s", request);
stop(EXIT_FAILURE, "request is too long (1024 max): %s", request);
}
/* remove \r\n at the end of string */
request[strcspn(request, "\r\n")] = '\0';
/* save request for logs */
esnprintf(_request, sizeof(_request), "%s", request);
return request;
}
void
remove_double_dot(char *request)
{
char *pos = NULL;
/* remove all "/.." for safety reasons */
while ((pos = strstr(request, "/..")) != NULL)
memmove(request, pos + 3, strlen(pos) + 1 - 3); /* "/.." = 3 */
}
void
check_path(char *path, size_t pathsiz)
{
struct stat sb = {0};
char tmp[PATH_MAX] = {'\0'};
if (stat(path, &sb) == -1) {
if (lstat(path, &sb) != -1 && S_ISLNK(sb.st_mode) == 1) {
if (readlink(path, tmp, sizeof(tmp)) > 0) {
status(30, "%s", tmp);
stop(EXIT_SUCCESS, NULL);
}
}
status(51, "%s", "file not found");
stop(EXIT_SUCCESS, NULL);
}
if (S_ISDIR(sb.st_mode)) {
/* check if dir path end with "/" */
if (path[strlen(path) - 1] != '/') {
esnprintf(tmp, sizeof(tmp), "/%s/", path);
status(31, "%s", tmp);
stop(EXIT_SUCCESS, NULL);
}
/* check if DEFAULT_INDEX exists in directory */
esnprintf(tmp, sizeof(tmp), "%s/%s", path, DEFAULT_INDEX);
if (stat(tmp, &sb) == 0)
esnprintf(path, pathsiz, "%s", tmp);
}
}
void
split_request(const char *request, char *hostname, char *path, char *query)
{
size_t nmatch = SE_MAX; /* 3 "()" + 1 for whole match */
char buf[BUFSIZ] = {'\0'}; /* to handle error messages */
int ret = 0;
regex_t greg; /* compiled gemini regex */
regmatch_t match[SE_MAX]; /* matches founds */
ret = regcomp(&greg, _gemini_regex, REG_EXTENDED);
if (ret != 0) {
regerror(ret, &greg, buf, sizeof(buf));
regfree(&greg);
status(50, "Internal server error");
stop(EXIT_FAILURE, "%s", buf);
}
ret = regexec(&greg, request, nmatch, match, 0);
if (ret != 0) {
regerror(ret, &greg, buf, sizeof(buf));
regfree(&greg);
status(59, "Malformed request");
stop(EXIT_FAILURE, "Malformed request, error:%s", buf);
}
/* one may want to check the return of getsubexp
* and change memcpy to strlcpy
* to make sure we didn't try to copy too long
* and that string isn't trunkated.
* It is unlikely to happen since dest string are as long as request
*/
getsubexp(request, match[1], hostname);
getsubexp(request, match[2], path);
getsubexp(request, match[3], query);
regfree(&greg);
}

59
vger.cabal Normal file
View File

@ -0,0 +1,59 @@
cabal-version: 1.12
-- This file has been generated from package.yaml by hpack version 0.34.6.
--
-- see: https://github.com/sol/hpack
name: vger
version: 0.0.0
license: BSD2
license-file: LICENSE
build-type: Simple
library
exposed-modules:
Gemini
other-modules:
Paths_vger
hs-source-dirs:
lib
ghc-options: -Wunused-packages
build-depends:
base
, directory
, mime-types
, network-uri
, regex-compat
, regex-pcre
, text
, utf8-string
default-language: Haskell2010
executable vger
main-is: Vger.hs
hs-source-dirs:
app
ghc-options: -Wunused-packages
other-modules:
Paths_vger
build-depends:
base
, network-uri
, optparse-applicative
, vger
default-language: Haskell2010
test-suite test
type: exitcode-stdio-1.0
main-is: Main.hs
other-modules:
Paths_vger
hs-source-dirs:
tests
ghc-options: -Wunused-packages
build-depends:
HUnit
, base
, network-uri
, vger
default-language: Haskell2010

57
vger.h
View File

@ -1,57 +0,0 @@
#ifndef vger_h_INCLUDED
#define vger_h_INCLUDED
/* length of "gemini://" */
#define GEMINI_PART 9
/*
* number of bytes to read with fgets() : 2014 + 1.
* fgets() reads at most size-1 (1024 here).
* See https://gemini.circumlunar.space/docs/specification.html.
*/
#define GEMINI_REQUEST_MAX 1025
/* max subexpression in regex : 3 + 1 */
#define SE_MAX 4
/* gemini_regex:
* =============
* ^gemini://+ : in case of gemini:///
* 1: hostname
* ([^/|^\?|^:]*) :
* catch everything, stop when /, ? or : is found
* don't catch :port
* [:[0-9]*]? : skip :1234 (port number) if any
* / * : skip "/" if any
* 2: path
* ([^\?]*) :
* catch everything and stop at ? if any
* 3 : query
* [\?]?(.*)$:
* catch everything after ? if any
*/
static const char *_gemini_regex =
"^gemini://+([^/|^\?|^:]*)[:[0-9]*]?/*([^\?]*)[\?]?(.*)$";
/* global vars */
static int _retcode = 0;
static ssize_t _datasiz = 0;
static char _request[GEMINI_REQUEST_MAX] = {'\0'};
/* functions */
ssize_t autoindex(const char *);
void cgi(const char *);
char * read_request(char *);
void check_path(char *, size_t);
ssize_t display_file(const char *);
int do_cgi(const char *, const char *, const char *, const char *);
void drop_privileges(const char *);
void set_rootdir(const char *, const char *, const char *);
void remove_double_dot(char *);
void split_request(const char *, char *, char *, char *);
void status(const int, const char *, ...);
void stop(const int, const char *, ...);
int uridecode (char *);
#endif // vger_h_INCLUDED