toolcontext: add `keep_going` flag to possibly let the tool continue after errors

[nit.git] / src / toolcontext.nit

# This file is part of NIT ( http://www.nitlanguage.org ).
#
# Copyright 2006-2008 Floréal Morandat <morandat@lirmm.fr>
# Copyright 2008-2012 Jean Privat <jean@pryen.org>
# Copyright 2009 Jean-Sebastien Gelinas <calestar@gmail.com>
# Copyright 2014 Alexandre Terrasa <alexandre@moz-code.org>
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

# Common command-line tool infrastructure than handle options and error messages
module toolcontext

import opts
import location
import version
import template

# A warning or an error
class Message
        super Comparable
        redef type OTHER: Message

        # The origin of the message in the source code, if any.
        var location: nullable Location

        # The category of the message.
        #
        # Used by quality-control tool for statistics or to enable/disable things individually.
        var tag: nullable String

        # The human-readable description of the message.
        #
        # It should be short and fit on a single line.
        # It should also have meaningful information first in case
        # on truncation by an IDE for instance.
        var text: String

        # Comparisons are made on message locations.
        redef fun <(other: OTHER): Bool do
                if location == null then return true
                if other.location == null then return false

                return location.as(not null) < other.location.as(not null)
        end

        redef fun to_s: String
        do
                var l = location
                if l == null then
                        return text
                else
                        return "{l}: {text}"
                end
        end

        # A colored version of the message including the original source line
        fun to_color_string: String
        do
                var esc = 27.ascii
                #var red = "{esc}[0;31m"
                #var bred = "{esc}[1;31m"
                #var green = "{esc}[0;32m"
                var yellow = "{esc}[0;33m"
                var def = "{esc}[0m"

                var tag = tag
                if tag != null then
                        tag = " ({tag})"
                else
                        tag = ""
                end
                var l = location
                if l == null then
                        return "{text}{tag}"
                else if l.file == null then
                        return "{yellow}{l}{def}: {text}{tag}"
                else
                        return "{yellow}{l}{def}: {text}{tag}\n{l.colored_line("1;31")}"
                end
        end
end

# Global context for tools
class ToolContext
        # Number of errors
        var error_count: Int = 0

        # Number of warnings
        var warning_count: Int = 0

        # Directory where to generate log files
        var log_directory: String = "logs"

        # Messages
        private var messages = new Array[Message]
        private var message_sorter: Comparator = default_comparator

        # Does an error prevent the program to stop at `check_errors`?
        #
        # Default to false.
        # Set this value to `true` if you need to keep the program going in case of error.
        var keep_going = false is writable

        # Output all current stacked messages and display total error informations
        #
        # Return true if no errors occurred.
        #
        # If some errors occurred, the behavior depends on the value of `keep_going`.
        # If `keep_going` is false, then the program exits.
        # Else, the error count and the warning count are reset and false is returned.
        fun check_errors: Bool
        do
                if messages.length > 0 then
                        message_sorter.sort(messages)

                        for m in messages do
                                if opt_no_color.value then
                                        sys.stderr.write("{m}\n")
                                else
                                        sys.stderr.write("{m.to_color_string}\n")
                                end
                        end

                        messages.clear
                end

                if error_count > 0 then
                        errors_info
                        if not keep_going then exit(1)
                        return false
                end
                return true
        end

        # Display (and reset) total error informations
        fun errors_info
        do
                if error_count == 0 and warning_count == 0 then return
                if opt_no_color.value then return
                sys.stderr.write "Errors: {error_count}. Warnings: {warning_count}.\n"
                error_count = 0
                warning_count = 0
        end

        # Display an error
        fun error(l: nullable Location, s: String)
        do
                messages.add(new Message(l,null,s))
                error_count = error_count + 1
                if opt_stop_on_first_error.value then check_errors
        end

        # Add an error, show errors and quit
        fun fatal_error(l: nullable Location, s: String)
        do
                error(l,s)
                check_errors
        end

        # Display a first-level warning.
        #
        # First-level warnings are warnings that SHOULD be corrected,
        # and COULD usually be immediately corrected.
        #
        # * There is a simple correction
        # * There is no reason to let the code this way (no reasonable @supresswarning-like annotation)
        # * They always are real issues (no false positive)
        #
        # First-level warnings are displayed by default (except if option `-q` is given).
        fun warning(l: nullable Location, tag: String, text: String)
        do
                if opt_warning.value.has("no-{tag}") then return
                if not opt_warning.value.has(tag) and opt_warn.value == 0 then return
                messages.add(new Message(l, tag, text))
                warning_count = warning_count + 1
                if opt_stop_on_first_error.value then check_errors
        end

        # Display a second-level warning.
        #
        # Second-level warnings are warnings that should require investigation,
        # but cannot always be immediately corrected.
        #
        # * The correction could be complex. e.g. require a refactorisation or an API change.
        # * The correction cannot be done. e.g. Code that use a deprecated API for some compatibility reason.
        # * There is not a real issue (false positive). Note that this should be unlikely.
        # * Transitional: While a real warning, it fires a lot in current code, so a transition is needed
        #   in order to fix them before promoting the advice to a warning.
        #
        # In order to prevent warning inflation à la Java, second-level warnings are not displayed by
        # default and require an additional option `-W`.
        fun advice(l: nullable Location, tag: String, text: String)
        do
                if opt_warning.value.has("no-{tag}") then return
                if not opt_warning.value.has(tag) and opt_warn.value <= 1 then return
                messages.add(new Message(l, tag, text))
                warning_count = warning_count + 1
                if opt_stop_on_first_error.value then check_errors
        end

        # Display an info
        fun info(s: String, level: Int)
        do
                if level <= verbose_level then
                        print "{s}"
                end
        end

        # Executes a program while checking if it's available and if the execution ended correctly
        #
        # Stops execution and prints errors if the program isn't available or didn't end correctly
        fun exec_and_check(args: Array[String], error: String)
        do
                var prog = args.first
                args.remove_at 0

                # Is the wanted program available?
                var proc_which = new IProcess.from_a("which", [prog])
                proc_which.wait
                var res = proc_which.status
                if res != 0 then
                        print "{error}: executable \"{prog}\" not found"
                        exit 1
                end

                # Execute the wanted program
                var proc = new Process.from_a(prog, args)
                proc.wait
                res = proc.status
                if res != 0 then
                        print "{error}: execution of \"{prog} {args.join(" ")}\" failed"
                        exit 1
                end
        end

        # Global OptionContext
        var option_context = new OptionContext

        # Option --warn
        var opt_warn = new OptionCount("Show more warnings", "-W", "--warn")

        # Option --warning
        var opt_warning = new OptionArray("Show/hide a specific warning", "-w", "--warning")

        # Option --quiet
        var opt_quiet = new OptionBool("Do not show warnings", "-q", "--quiet")

        # Option --log
        var opt_log = new OptionBool("Generate various log files", "--log")

        # Option --log-dir
        var opt_log_dir = new OptionString("Directory where to generate log files", "--log-dir")

        # Option --nit-dir
        var opt_nit_dir = new OptionString("Base directory of the Nit installation", "--nit-dir")

        # Option --help
        var opt_help = new OptionBool("Show Help (This screen)", "-h", "-?", "--help")

        # Option --version
        var opt_version = new OptionBool("Show version and exit", "--version")

        # Option --set-dummy-tool
        var opt_set_dummy_tool = new OptionBool("Set toolname and version to DUMMY. Useful for testing", "--set-dummy-tool")

        # Option --verbose
        var opt_verbose = new OptionCount("Verbose", "-v", "--verbose")

        # Option --stop-on-first-error
        var opt_stop_on_first_error = new OptionBool("Stop on first error", "--stop-on-first-error")

        # Option --no-color
        var opt_no_color = new OptionBool("Do not use color to display errors and warnings", "--no-color")

        # Option --bash-completion
        var opt_bash_completion = new OptionBool("Generate bash_completion file for this program", "--bash-completion")

        # Option --stub-man
        var opt_stub_man = new OptionBool("Generate a stub manpage in pandoc markdown format", "--stub-man")

        # Verbose level
        var verbose_level: Int = 0

        init
        do
                option_context.add_option(opt_warn, opt_warning, opt_quiet, opt_stop_on_first_error, opt_no_color, opt_log, opt_log_dir, opt_nit_dir, opt_help, opt_version, opt_set_dummy_tool, opt_verbose, opt_bash_completion, opt_stub_man)

                # Hide some internal options
                opt_stub_man.hidden = true
                opt_bash_completion.hidden = true
                opt_set_dummy_tool.hidden = true
        end

        # Name, usage and synopsis of the tool.
        # It is mainly used in `usage`.
        # Should be correctly set by the client before calling `process_options`
        # A multi-line string is recommmended.
        #
        # eg. `"Usage: tool [OPTION]... [FILE]...\nDo some things."`
        var tooldescription: String = "Usage: [OPTION]... [ARG]..." is writable

        # Does `process_options` should accept an empty sequence of arguments.
        # ie. nothing except options.
        # Is `false` by default.
        #
        # If required, if should be set by the client before calling `process_options`
        var accept_no_arguments = false is writable

        # print the full usage of the tool.
        # Is called by `process_option` on `--help`.
        # It also could be called by the client.
        fun usage
        do
                print tooldescription
                option_context.usage
        end

        # Parse and process the options given on the command line
        fun process_options(args: Sequence[String])
        do
                self.opt_warn.value = 1

                # init options
                option_context.parse(args)

                if opt_help.value then
                        usage
                        exit 0
                end

                if opt_version.value then
                        print version
                        exit 0
                end

                if opt_bash_completion.value then
                        var bash_completion = new BashCompletion(self)
                        bash_completion.write_to(sys.stdout)
                        exit 0
                end

                if opt_stub_man.value then
                        print """
% {{{toolname.to_upper}}}(1)

# NAME

{{{tooldescription.split("\n")[1]}}}

# SYNOPSYS

{{{toolname}}} [*options*]...

# OPTIONS
"""
                        for o in option_context.options do
                                var first = true
                                for n in o.names do
                                        if first then first = false else printn ", "
                                        printn "`{n}`"
                                end
                                print ""
                                print ":   {o.helptext}"
                                print ""
                        end
                        print """
# SEE ALSO

The Nit language documentation and the source code of its tools and libraries may be downloaded from <http://nitlanguage.org>"""
                        exit 0
                end

                var errors = option_context.get_errors
                if not errors.is_empty then
                        for e in errors do print "Error: {e}"
                        print tooldescription
                        print "Use --help for help"
                        exit 1
                end

                nit_dir = compute_nit_dir

                if option_context.rest.is_empty and not accept_no_arguments then
                        print tooldescription
                        print "Use --help for help"
                        exit 1
                end

                # Set verbose level
                verbose_level = opt_verbose.value

                if self.opt_quiet.value then self.opt_warn.value = 0

                if opt_log_dir.value != null then log_directory = opt_log_dir.value.as(not null)
                if opt_log.value then
                        # Make sure the output directory exists
                        log_directory.mkdir
                end

        end

        # Get the current `nit_version` or "DUMMY_VERSION" if `--set-dummy-tool` is set.
        fun version: String do
                if opt_set_dummy_tool.value then
                        return "DUMMY_VERSION"
                end
                return nit_version
        end

        # Get the name of the tool or "DUMMY_TOOL" id `--set-dummy-tool` is set.
        fun toolname: String do
                if opt_set_dummy_tool.value then
                        return "DUMMY_TOOL"
                end
                return sys.program_name.basename("")
        end

        # The identified root directory of the Nit project
        var nit_dir: String is noinit

        private fun compute_nit_dir: String
        do
                # the option has precedence
                var res = opt_nit_dir.value
                if res != null then
                        if not check_nit_dir(res) then
                                fatal_error(null, "Fatal Error: the value of --nit-dir does not seem to be a valid base Nit directory: {res}")
                        end
                        return res
                end

                # then the environ variable has precedence
                res = "NIT_DIR".environ
                if not res.is_empty then
                        if not check_nit_dir(res) then
                                fatal_error(null, "Fatal Error: the value of NIT_DIR does not seem to be a valid base Nit directory: {res}")
                        end
                        return res
                end

                # find the runpath of the program from argv[0]
                res = "{sys.program_name.dirname}/.."
                if check_nit_dir(res) then return res.simplify_path

                # find the runpath of the process from /proc
                var exe = "/proc/self/exe"
                if exe.file_exists then
                        res = exe.realpath
                        res = res.dirname.join_path("..")
                        if check_nit_dir(res) then return res.simplify_path
                end

                # search in the PATH
                var ps = "PATH".environ.split(":")
                for p in ps do
                        res = p/".."
                        if check_nit_dir(res) then return res.simplify_path
                end

                fatal_error(null, "Fatal Error: Cannot locate a valid base nit directory. It is quite unexpected. Try to set the environment variable `NIT_DIR` or to use the `--nit-dir` option.")
                abort
        end

        private fun check_nit_dir(res: String): Bool
        do
                return res.file_exists and "{res}/src/nit.nit".file_exists
        end
end

# This class generates a compatible `bash_completion` script file.
#
# On some Linux systems `bash_completion` allow the program to control command line behaviour.
#
#       $ nitls [TAB][TAB]
#       file1.nit              file2.nit              file3.nit
#
#       $ nitls --[TAB][TAB]
#       --bash-toolname        --keep                 --path                 --tree
#       --depends              --log                  --project              --verbose
#       --disable-phase        --log-dir              --quiet                --version
#       --gen-bash-completion  --no-color             --recursive            --warn
#       --help                 --only-metamodel       --source
#       --ignore-visibility    --only-parse           --stop-on-first-error
#
# Generated file can be placed in system bash_completion directory `/etc/bash_completion.d/`
# or source it in `~/.bash_completion`.
class BashCompletion
        super Template

        var toolcontext: ToolContext

        private fun extract_options_names: Array[String] do
                var names = new Array[String]
                for option in toolcontext.option_context.options do
                        for name in option.names do
                                if name.has_prefix("--") then names.add name
                        end
                end
                return names
        end

        redef fun rendering do
                var name = toolcontext.toolname
                var option_names = extract_options_names
                addn "# generated bash completion file for {name} {toolcontext.version}"
                addn "_{name}()"
                addn "\{"
                addn "  local cur prev opts"
                addn "  COMPREPLY=()"
                addn "  cur=\"$\{COMP_WORDS[COMP_CWORD]\}\""
                addn "  prev=\"$\{COMP_WORDS[COMP_CWORD-1]\}\""
                if not option_names.is_empty then
                        addn "  opts=\"{option_names.join(" ")}\""
                        addn "  if [[ $\{cur\} == -* ]] ; then"
                        addn "          COMPREPLY=( $(compgen -W \"$\{opts\}\" -- $\{cur\}) )"
                        addn "          return 0"
                        addn "  fi"
                end
                addn "\} &&"
                addn "complete -o default -F _{name} {name}"
        end
end