extractbb.lua wrapper for safety

git-svn-id: svn://tug.org/texlive/trunk@71664 c570f23f-e606-0410-a88d-b1316a301751

21 files changed, 271 insertions, 16 deletions

diff --git a/Build/source/texk/kpathsea/ChangeLog b/Build/source/texk/kpathsea/ChangeLog
index b7e89628ec3..cbcee62f617 100644
--- a/Build/source/texk/kpathsea/ChangeLog
+++ b/Build/source/texk/kpathsea/ChangeLog
@@ -1,3 +1,8 @@
+2024-07-01  Karl Berry  <karl@tug.org>
+
+	* texmf.cnf (shell_escape_commands): restore extractbb,
+	now as a wrapper script that protects against the exploit.
+
 2024-06-29  Karl Berry  <karl@tug.org>
 
 	* texmf.cnf (shell_escape_commands): remove extractbb.
diff --git a/Build/source/texk/kpathsea/texmf.cnf b/Build/source/texk/kpathsea/texmf.cnf
index 333e6325b3d..5df6bd97f9c 100644
--- a/Build/source/texk/kpathsea/texmf.cnf
+++ b/Build/source/texk/kpathsea/texmf.cnf
@@ -649,6 +649,7 @@ shell_escape = p
 % 
 shell_escape_commands = \
 bibtex,bibtex8,\
+extractbb,\
 gregorio,\
 kpsewhich,\
 l3sys-query,\
@@ -660,7 +661,6 @@ r-mpost,\
 texosquery-jre8,\
 
 % other programs considered:
-% extractbb - allowed in the past, but can be exploited.
 % dvips - but external commands can be executed, need at least -R1.
 % epspdf, ps2pdf, pstopdf - need to respect openout_any,
 %   and gs -dSAFER must be used and check for shell injection with filenames.
diff --git a/Build/source/texk/texlive/linked_scripts/Makefile.am b/Build/source/texk/texlive/linked_scripts/Makefile.am
index ba4e7be48fd..0bd39fbab96 100644
--- a/Build/source/texk/texlive/linked_scripts/Makefile.am
+++ b/Build/source/texk/texlive/linked_scripts/Makefile.am
@@ -235,6 +235,7 @@ texmf_other_scripts = \
 	texfindpkg/texfindpkg.lua \
 	texdoc/texdoc.tlu \
 	texfot/texfot.pl \
+	texlive/extractbb.lua \
 	texlive/fmtutil-sys.sh \
 	texlive/fmtutil-user.sh \
 	texlive/fmtutil.pl \
diff --git a/Build/source/texk/texlive/linked_scripts/Makefile.in b/Build/source/texk/texlive/linked_scripts/Makefile.in
index 447f6518ae1..13b8d4046ad 100644
--- a/Build/source/texk/texlive/linked_scripts/Makefile.in
+++ b/Build/source/texk/texlive/linked_scripts/Makefile.in
@@ -452,6 +452,7 @@ texmf_other_scripts = \
 	texfindpkg/texfindpkg.lua \
 	texdoc/texdoc.tlu \
 	texfot/texfot.pl \
+	texlive/extractbb.lua \
 	texlive/fmtutil-sys.sh \
 	texlive/fmtutil-user.sh \
 	texlive/fmtutil.pl \
diff --git a/Build/source/texk/texlive/linked_scripts/scripts.lst b/Build/source/texk/texlive/linked_scripts/scripts.lst
index 479473cdf8c..04eacde6e68 100644
--- a/Build/source/texk/texlive/linked_scripts/scripts.lst
+++ b/Build/source/texk/texlive/linked_scripts/scripts.lst
@@ -176,6 +176,7 @@ texdirflatten/texdirflatten
 texfindpkg/texfindpkg.lua
 texdoc/texdoc.tlu
 texfot/texfot.pl
+texlive/extractbb.lua
 texlive/fmtutil-sys.sh
 texlive/fmtutil-user.sh
 texlive/fmtutil.pl
diff --git a/Master/bin/aarch64-linux/extractbb b/Master/bin/aarch64-linux/extractbb
index 619e6a2f6ab..b592ca9b806 120000
--- a/Master/bin/aarch64-linux/extractbb
+++ b/Master/bin/aarch64-linux/extractbb
@@ -1 +1 @@
-xdvipdfmx
\ No newline at end of file
+../../texmf-dist/scripts/texlive/extractbb.lua
\ No newline at end of file
diff --git a/Master/bin/amd64-freebsd/extractbb b/Master/bin/amd64-freebsd/extractbb
index 619e6a2f6ab..b592ca9b806 120000
--- a/Master/bin/amd64-freebsd/extractbb
+++ b/Master/bin/amd64-freebsd/extractbb
@@ -1 +1 @@
-xdvipdfmx
\ No newline at end of file
+../../texmf-dist/scripts/texlive/extractbb.lua
\ No newline at end of file
diff --git a/Master/bin/amd64-netbsd/extractbb b/Master/bin/amd64-netbsd/extractbb
index 619e6a2f6ab..b592ca9b806 120000
--- a/Master/bin/amd64-netbsd/extractbb
+++ b/Master/bin/amd64-netbsd/extractbb
@@ -1 +1 @@
-xdvipdfmx
\ No newline at end of file
+../../texmf-dist/scripts/texlive/extractbb.lua
\ No newline at end of file
diff --git a/Master/bin/armhf-linux/extractbb b/Master/bin/armhf-linux/extractbb
index 619e6a2f6ab..b592ca9b806 120000
--- a/Master/bin/armhf-linux/extractbb
+++ b/Master/bin/armhf-linux/extractbb
@@ -1 +1 @@
-xdvipdfmx
\ No newline at end of file
+../../texmf-dist/scripts/texlive/extractbb.lua
\ No newline at end of file
diff --git a/Master/bin/i386-freebsd/extractbb b/Master/bin/i386-freebsd/extractbb
index 619e6a2f6ab..b592ca9b806 120000
--- a/Master/bin/i386-freebsd/extractbb
+++ b/Master/bin/i386-freebsd/extractbb
@@ -1 +1 @@
-xdvipdfmx
\ No newline at end of file
+../../texmf-dist/scripts/texlive/extractbb.lua
\ No newline at end of file
diff --git a/Master/bin/i386-linux/extractbb b/Master/bin/i386-linux/extractbb
index 619e6a2f6ab..b592ca9b806 120000
--- a/Master/bin/i386-linux/extractbb
+++ b/Master/bin/i386-linux/extractbb
@@ -1 +1 @@
-xdvipdfmx
\ No newline at end of file
+../../texmf-dist/scripts/texlive/extractbb.lua
\ No newline at end of file
diff --git a/Master/bin/i386-netbsd/extractbb b/Master/bin/i386-netbsd/extractbb
index 619e6a2f6ab..b592ca9b806 120000
--- a/Master/bin/i386-netbsd/extractbb
+++ b/Master/bin/i386-netbsd/extractbb
@@ -1 +1 @@
-xdvipdfmx
\ No newline at end of file
+../../texmf-dist/scripts/texlive/extractbb.lua
\ No newline at end of file
diff --git a/Master/bin/i386-solaris/extractbb b/Master/bin/i386-solaris/extractbb
index 619e6a2f6ab..b592ca9b806 120000
--- a/Master/bin/i386-solaris/extractbb
+++ b/Master/bin/i386-solaris/extractbb
@@ -1 +1 @@
-xdvipdfmx
\ No newline at end of file
+../../texmf-dist/scripts/texlive/extractbb.lua
\ No newline at end of file
diff --git a/Master/bin/universal-darwin/extractbb b/Master/bin/universal-darwin/extractbb
index 619e6a2f6ab..b592ca9b806 120000
--- a/Master/bin/universal-darwin/extractbb
+++ b/Master/bin/universal-darwin/extractbb
@@ -1 +1 @@
-xdvipdfmx
\ No newline at end of file
+../../texmf-dist/scripts/texlive/extractbb.lua
\ No newline at end of file
diff --git a/Master/bin/x86_64-cygwin/extractbb b/Master/bin/x86_64-cygwin/extractbb
index b011eb0c845..b592ca9b806 120000
--- a/Master/bin/x86_64-cygwin/extractbb
+++ b/Master/bin/x86_64-cygwin/extractbb
@@ -1 +1 @@
-xdvipdfmx.exe
\ No newline at end of file
+../../texmf-dist/scripts/texlive/extractbb.lua
\ No newline at end of file
diff --git a/Master/bin/x86_64-darwinlegacy/extractbb b/Master/bin/x86_64-darwinlegacy/extractbb
index 619e6a2f6ab..b592ca9b806 120000
--- a/Master/bin/x86_64-darwinlegacy/extractbb
+++ b/Master/bin/x86_64-darwinlegacy/extractbb
@@ -1 +1 @@
-xdvipdfmx
\ No newline at end of file
+../../texmf-dist/scripts/texlive/extractbb.lua
\ No newline at end of file
diff --git a/Master/bin/x86_64-linux/extractbb b/Master/bin/x86_64-linux/extractbb
index 619e6a2f6ab..b592ca9b806 120000
--- a/Master/bin/x86_64-linux/extractbb
+++ b/Master/bin/x86_64-linux/extractbb
@@ -1 +1 @@
-xdvipdfmx
\ No newline at end of file
+../../texmf-dist/scripts/texlive/extractbb.lua
\ No newline at end of file
diff --git a/Master/bin/x86_64-linuxmusl/extractbb b/Master/bin/x86_64-linuxmusl/extractbb
index 619e6a2f6ab..b592ca9b806 120000
--- a/Master/bin/x86_64-linuxmusl/extractbb
+++ b/Master/bin/x86_64-linuxmusl/extractbb
@@ -1 +1 @@
-xdvipdfmx
\ No newline at end of file
+../../texmf-dist/scripts/texlive/extractbb.lua
\ No newline at end of file
diff --git a/Master/bin/x86_64-solaris/extractbb b/Master/bin/x86_64-solaris/extractbb
index 619e6a2f6ab..b592ca9b806 120000
--- a/Master/bin/x86_64-solaris/extractbb
+++ b/Master/bin/x86_64-solaris/extractbb
@@ -1 +1 @@
-xdvipdfmx
\ No newline at end of file
+../../texmf-dist/scripts/texlive/extractbb.lua
\ No newline at end of file
diff --git a/Master/texmf-dist/scripts/texlive/extractbb.lua b/Master/texmf-dist/scripts/texlive/extractbb.lua
new file mode 100755
index 00000000000..7584814ee80
--- /dev/null
+++ b/Master/texmf-dist/scripts/texlive/extractbb.lua
@@ -0,0 +1,247 @@
+#!/usr/bin/env texlua
+-- $Id$
+-- SPDX-License-Identifier: CC0-1.0
+-- SPDX-FileCopyrightText: 2024 Max Chernoff
+--
+-- A generic wrapper to make commands safe to run with restricted shell escape.
+-- 
+-- Originally created for extractbb, which is listed in shell_escape_commands,
+-- but can be run as dvipdfm(x), which in turn can run arbitrary commands
+-- using its -D option.
+-- 
+-- The idea is to exec "ebb --ebb <other args>", since only argv[1] is
+-- used by dvipdfmx to determine its behavior.
+--
+-- Note: This script can only adjust the paths and arguments of the target
+-- executable; it *CANNOT* make an arbitrary program safe to run with
+-- restricted shell escape.
+
+-- A shorter, less paranoid version.
+-- (Prepend a hyphen to the line below to enable).
+--[=[
+arg[0] = arg[0]:gsub("extractbb", "ebb")
+table.insert(arg, 1, "ebb")
+table.insert(arg, 2, "--ebb")
+os.exec(arg)
+os.exit(1)
+--]=]
+
+---------------------
+--- Configuration ---
+---------------------
+
+-- The base name of this script. (Example: ``extractbb'')
+local SCRIPT_NAME = "extractbb"
+
+-- The extension of the script. Extensionless-names are also permitted.
+-- (Example: ``lua'')
+local SCRIPT_EXT = "lua"
+
+-- The base name of the path to the target program. (Example: ``xdvipdfmx'')
+local TARGET_PATH_NAME = "xdvipdfmx"
+
+-- The name to use when calling the target program. Equivalent to ``argv[0]''
+-- in C. (Example: ``extractbb'')
+local TARGET_EXEC_NAME = "ebb"
+
+-- Any extra arguments to be prepended to the target program, before any
+-- user-supplied arguments. Equivalent to ``argv[1], ...'' in C.
+-- (Example: ``--extractbb'')
+local TARGET_PREPEND_ARGS = { "--ebb" }
+
+-- Any extra arguments to be appended to the target program, after any
+-- user-supplied arguments. Equivalent to ``..., argv[argc]'' in C.
+local TARGET_APPEND_ARGS = {}
+
+-- Sets the value of ``openin_any'' to this value. If ``nil'', then the value
+-- will be left unchanged. (Example: ``r'')
+local READ_PERMS = "r"
+
+-- Sets the value of ``openout_any'' to this value. If ``nil'', then the value
+-- will be left unchanged. (Example: ``p'')
+local WRITE_PERMS = "p"
+
+-- The name of the Lua interpreter. (Example: ``texlua'')
+local INTERPRETER_NAME = "texlua"
+
+-- The extension of the interpreter. Extensionless-names are also permitted.
+-- (Example: ``exe'')
+local INTERPRETER_EXT = "exe"
+
+
+----------------------
+--- Initialization ---
+----------------------
+
+-- Save often-used globals for a slight speed boost.
+local insert = table.insert
+
+-- Set the kpathsea program name
+kpse.set_program_name(INTERPRETER_NAME, SCRIPT_NAME)
+
+-- Rename the input arguments so we don't get confused
+local script_args = arg
+
+
+----------------------------
+--- Function Definitions ---
+----------------------------
+
+-- Error messages
+local function error(message)
+    io.stderr:write("! ERROR (extractbb.lua): " .. message, "\n")
+    io.stderr:write(debug.traceback(nil, 2), "\n")
+    io.stderr:flush()
+    os.exit(1)
+end
+
+-- Get the directory, name, and extension from a full path. We'll split on
+-- either a forward or backward slash---Windows can use either, and we don't
+-- need to support Unix systems with TL installed to a directory with
+-- backslashes in its name.
+local split_dir_pattern = "^(.*)[/\\]([^/\\]-)$"
+local split_ext_pattern = "(.*)%.([^.]-)$"
+
+local function split_path(path)
+    -- Make sure that we were given a string
+    if type(path) ~= "string" then
+        return nil, nil, nil
+    end
+
+    -- Split the (directory) from the (name and extension)
+    local dir, name_ext = path:match(split_dir_pattern)
+
+    -- No directory
+    if not dir then
+        dir      = nil
+        name_ext = path
+
+    -- A bare directory (with a trailing slash)
+    elseif name_ext == "" then
+        return dir, nil, nil
+    end
+
+    -- Split the (name) from the (extension)
+    local name, ext = name_ext:match(split_ext_pattern)
+
+    -- No extension (or a dotfile)
+    if (not name) or (name == "") then
+        name = name_ext
+        ext  = nil
+    end
+
+    return dir, name, ext
+end
+
+-- See if a file exists
+local function file_exists(path)
+    local mode = lfs.attributes(path, "mode")
+    return (mode == "file") or (mode == "link")
+end
+
+
+---------------------
+--- Safety Checks ---
+---------------------
+
+-- Make sure that we're running unrestricted.
+if status.shell_escape ~= 1 then
+    error("Shell escape has been disabled.")
+end
+
+if status.safer_option ~= 0 then
+    error("The ``safer'' option has been enabled.")
+end
+
+-- Set the file permissions.
+if READ_PERMS then
+    os.setenv("openin_any", READ_PERMS)
+end
+
+if WRITE_PERMS then
+    os.setenv("openout_any", WRITE_PERMS)
+end
+
+-- Get the directory of the script and interpreter
+local script_path = debug.getinfo(1, "S").source:sub(2)
+local script_dir, script_name, script_ext = split_path(script_path)
+
+local interpreter_dir = kpse.var_value("SELFAUTOLOC")
+local _, interpreter_name, interpreter_ext = split_path(script_args[-1])
+if os.type == 'windows' then
+    interpreter_ext = INTERPRETER_EXT
+end
+-- Look up the script again with kpathsea
+local resolved_script_path = kpse.find_file(
+    script_name .. "." .. (script_ext or SCRIPT_EXT), "texmfscripts", true
+)
+
+-- Make sure that our paths are correct
+if not script_dir then
+    error("Empty script dir")
+end
+
+if not resolved_script_path then
+    error("Empty resolved script path")
+end
+
+if (script_dir ~= interpreter_dir) and (script_path ~= resolved_script_path) then
+    error("The script is in an incorrect location: " .. script_dir)
+end
+
+if script_name ~= SCRIPT_NAME then
+    error("Incorrect script name: " .. script_name)
+end
+
+if interpreter_name ~= INTERPRETER_NAME then
+    error("Incorrect interpreter name: " .. interpreter_name)
+end
+
+if (script_ext ~= SCRIPT_EXT) and (script_ext ~= nil) then
+    error("Incorrect script extension: " .. script_ext)
+end
+
+if (interpreter_ext ~= INTERPRETER_EXT) and (interpreter_ext ~= nil) then
+    error("Incorrect interpreter extension: " .. interpreter_ext)
+end
+
+-- Get the path to the target program
+local target_ext  = interpreter_ext and ("." .. interpreter_ext) or ""
+local target_path = interpreter_dir .. "/" .. TARGET_PATH_NAME .. target_ext
+
+-- Make sure that the target program exists
+if not file_exists(target_path) then
+    error("The target program does not exist: " .. target_path)
+end
+
+
+----------------------
+--- Run the target ---
+----------------------
+
+-- Generate the target arguments
+local target_args = {
+    [0] = target_path,      -- Path to the executable
+    [1] = TARGET_EXEC_NAME, -- argv[0]
+}
+
+-- argv[2] through argv[n]
+for _, arg in ipairs(TARGET_PREPEND_ARGS) do
+    insert(target_args, arg)
+end
+
+for i = 1, #script_args do
+    -- We use a numeric iterator here to avoid ``arg[-1]'' and ``arg[0]''.
+    insert(target_args, script_args[i])
+end
+
+for _, arg in ipairs(TARGET_APPEND_ARGS) do
+    insert(target_args, arg)
+end
+
+-- Run the target program, replacing the current process
+local _, err = os.exec(target_args)
+
+if err then
+    error("The target program failed to run.")
+end
diff --git a/Master/texmf-dist/web2c/texmf.cnf b/Master/texmf-dist/web2c/texmf.cnf
index 333e6325b3d..5df6bd97f9c 100644
--- a/Master/texmf-dist/web2c/texmf.cnf
+++ b/Master/texmf-dist/web2c/texmf.cnf
@@ -649,6 +649,7 @@ shell_escape = p
 % 
 shell_escape_commands = \
 bibtex,bibtex8,\
+extractbb,\
 gregorio,\
 kpsewhich,\
 l3sys-query,\
@@ -660,7 +661,6 @@ r-mpost,\
 texosquery-jre8,\
 
 % other programs considered:
-% extractbb - allowed in the past, but can be exploited.
 % dvips - but external commands can be executed, need at least -R1.
 % epspdf, ps2pdf, pstopdf - need to respect openout_any,
 %   and gs -dSAFER must be used and check for shell injection with filenames.