path: root/Biz/Bild.hs

{-# LANGUAGE DeriveAnyClass #-}
{-# LANGUAGE DeriveGeneric #-}
{-# LANGUAGE LambdaCase #-}
{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE QuasiQuotes #-}
{-# LANGUAGE RecordWildCards #-}
{-# LANGUAGE TupleSections #-}
{-# LANGUAGE NoImplicitPrelude #-}

-- | A general purpose build tool.
--
-- : out bild
--
-- == Design constraints
--
--    * only input is one or more a namespaces. no subcommands, no packages
--
--    * no need to write specific build rules, one rule for hs, one for rs, one
--    for scm, and so on
--
--    * no need to distinguish between exe and lib, just have a single output,
--    or figure it out automatically
--
--    * never concerned with deployment/packaging - leave that to another tool
--      (scp? tar?)
--
--    * local dev builds should be preserved, while remote nix builds are used
--    for the final package
--
-- == Features
--
--    * namespace maps to filesystem
--
--        * no need for `bild -l` for listing available targets.
--          Use `ls` or `tree`
--
--        * you build namespaces, not files/modules/packages/etc
--
--    * namespace maps to language modules
--
--        * build settings can be set in the file comments
--
--    * pwd is always considered the the source directory,
--      no `src` vs `doc` etc.
--
--    * build methods automaticatly detected with file extensions
--
--    * flags modify the way to interact with the build, some ideas:
--
--        * -p = turn on profiling
--
--        * -o = optimize level
--
--    * the build is planned out with an analysis, which can be viewed
--      beforehand with `--json`. The analysis includes compiler flags, which
--      can be used in `repl` for testing compilation locally.
--
-- == Example Commands
--
-- > bild [opts] <target..>
--
-- The general scheme is to build the things described by the targets. A target
-- is a namespace. You can list as many as you want, but you must list at least
-- one. It could just be `:!bild %` in vim to build whatever you're working on,
-- or `bild **/*` to build everything, or `fd .hs -X bild` to build all Haskell
-- files.
--
-- Build outputs will go into the `_` directory in the root of the project.
--
-- > bild A/B.hs
--
-- This will build the file at ./A/B.hs, which translates to something like
-- `ghc --make A.B`.
--
-- == Build Metadata
--
-- Metadata is set in the comments with a special syntax. For system-level deps,
-- we list the deps in comments in the target file, like:
--
-- > -- : sys cmark
--
-- The name is used to lookup the package in `nixpkgs.pkgs.<name>`.
-- Language-level deps can automatically determined by passing parsed import
-- statements to a package database, eg `ghc-pkg find-module`.
--
-- The output executable is named with:
--
-- > -- : out my-program
--
-- or
--
-- > -- : out my-app.js
--
-- When multiple compilers are possible we use the @out@ extension to determine
-- target platform. If @out@ does not have an extension, each build type falls
-- back to a default, namely an executable binary.
--
-- This method of setting metadata in the module comments works pretty well,
-- and really only needs to be done in the entrypoint module anyway.
--
-- Local module deps are included by just giving the repo root to the underlying
-- compiler for the target, and the compiler does work of walking the source
-- tree.
module Biz.Bild where

import Alpha hiding (sym, (<.>))
import qualified Biz.Cli as Cli
import qualified Biz.Log as Log
import Biz.Namespace (Namespace (..))
import qualified Biz.Namespace as Namespace
import qualified Biz.Test as Test
import qualified Control.Concurrent.Async as Async
import qualified Data.Aeson as Aeson
import qualified Data.ByteString.Char8 as Char8
import qualified Data.ByteString.Lazy as ByteString
import qualified Data.Char as Char
import Data.Conduit ((.|))
import qualified Data.Conduit as Conduit
import qualified Data.Conduit.List as Conduit
import qualified Data.Conduit.Process as Conduit
import qualified Data.Map as Map
import qualified Data.Set as Set
import qualified Data.String as String
import qualified Data.Text as Text
import qualified Data.Text.IO as Text.IO
import qualified System.Directory as Dir
import qualified System.Environment as Env
import qualified System.Exit as Exit
import System.FilePath (replaceExtension, (</>))
import qualified System.IO as IO
import qualified System.Process as Process
import qualified Text.Regex.Applicative as Regex

main :: IO ()
main = Cli.main <| Cli.Plan help move test_ pure
  where
    test_ =
      Test.group
        "Biz.Bild"
        [ Test.unit "can bild bild" <| do
            root <- Env.getEnv "BIZ_ROOT"
            path <- Dir.makeAbsolute "Biz/Bild.hs"
            case Namespace.fromPath root path of
              Nothing -> Test.assertFailure "can't find ns for bild"
              Just ns ->
                analyze mempty ns
                  +> build False False
                  +> \case
                    [Exit.ExitFailure _] ->
                      Test.assertFailure "can't bild bild"
                    _ ->
                      pure ()
        ]

move :: Cli.Arguments -> IO ()
move args = do
  root <- Env.getEnv "BIZ_ROOT"
  IO.hSetBuffering stdout IO.NoBuffering
    >> pure (Cli.getAllArgs args (Cli.argument "target"))
    /> filter (not <. Namespace.isCab)
    +> filterM Dir.doesFileExist
    +> traverse Dir.makeAbsolute
    /> map (Namespace.fromPath root)
    /> catMaybes
    +> foldM analyze mempty
    /> Map.filter (namespace .> isBuildableNs)
    +> printOrBuild
    +> exitSummary
  where
    printOrBuild :: Analysis -> IO [ExitCode]
    printOrBuild targets
      | args `Cli.has` Cli.longOption "json" =
        Log.wipe >> putJSON targets >> pure [Exit.ExitSuccess]
      | otherwise = do
        root <- Env.getEnv "BIZ_ROOT"
        createHier root
        build isTest isLoud targets
    isTest = args `Cli.has` Cli.longOption "test"
    isLoud = args `Cli.has` Cli.longOption "loud"
    putJSON = Aeson.encode .> ByteString.toStrict .> Char8.putStrLn

nixStore :: String
nixStore = "/nix/store/00000000000000000000000000000000-"

help :: Cli.Docopt
help =
  [Cli.docopt|
bild

Usage:
  bild test
  bild [options] <target>...

Options:
  --test      Run tests on a target after building
  --loud      Show all output from compiler
  --json      Print the build plan as JSON, don't build
  -h, --help  Print this info
|]

exitSummary :: [Exit.ExitCode] -> IO ()
exitSummary exits =
  if failures > 0
    then Exit.die <| show failures
    else Exit.exitSuccess
  where
    failures = length <| filter isFailure exits

type Dep = String

data Out = Lib String | Bin String | None
  deriving (Show, Eq)

instance Aeson.ToJSON Out where
  toJSON =
    Aeson.String <. Text.pack <. \case
      Bin a -> a
      Lib a -> a
      None -> ""

data Compiler
  = Copy
  | Gcc
  | GhcLib
  | GhcExe
  | Guile
  | NixBuild
  | Rustc
  | Sbcl
  deriving (Eq, Show, Generic)

instance Aeson.ToJSON Compiler where
  toJSON =
    Aeson.String <. \case
      Copy -> "cp"
      Gcc -> "gcc"
      GhcLib -> "ghc"
      GhcExe -> "ghc"
      Guile -> "guile"
      NixBuild -> "nix-build"
      Rustc -> "rustc"
      Sbcl -> "sbcl"

data Target = Target
  { -- | Output name
    out :: Out,
    -- | Output path (into cabdir)
    outPath :: FilePath,
    -- | Fully qualified namespace partitioned by '.'
    namespace :: Namespace,
    -- | Absolute path to file
    path :: FilePath,
    -- | Language-specific dependencies
    langdeps :: Set Dep,
    -- | System-level dependencies
    sysdeps :: Set Dep,
    -- | Which compiler should we use?
    compiler :: Compiler,
    -- | Where is this machine being built? Schema: user@location
    builder :: Text,
    -- | Flags and arguments passed to 'Compiler' when building
    compilerFlags :: [Text]
  }
  deriving (Show, Generic, Aeson.ToJSON)

-- | We can't build everything yet...
isBuildableNs :: Namespace -> Bool
isBuildableNs = \case
  (Namespace _ Namespace.C) -> True
  (Namespace _ Namespace.Css) -> False
  (Namespace _ Namespace.Hs) -> True
  (Namespace _ Namespace.Json) -> False
  (Namespace _ Namespace.Keys) -> False
  (Namespace _ Namespace.Lisp) -> True
  (Namespace _ Namespace.Md) -> False
  (Namespace _ Namespace.None) -> False
  (Namespace _ Namespace.Py) -> False
  (Namespace _ Namespace.Sh) -> False
  (Namespace _ Namespace.Scm) -> True
  (Namespace _ Namespace.Rs) -> True
  (Namespace path Namespace.Nix)
    | path `elem` nixTargets -> True
    | otherwise -> False
  where
    nixTargets =
      [ ["Biz", "Pie"],
        ["Biz", "Que"],
        ["Biz", "Cloud"],
        ["Biz", "Dev"],
        ["Biz", "Dragons", "Analysis"]
      ]

-- | Emulate the *nix hierarchy in the cabdir.
outToPath :: Out -> FilePath
outToPath = \case
  Bin o -> "_/bin" </> o
  Lib o -> "_/lib" </> o
  None -> mempty

intdir, nixdir, vardir :: String
intdir = "_/int"
nixdir = "_/nix"
vardir = "_/var"

createHier :: String -> IO ()
createHier root =
  traverse_
    (Dir.createDirectoryIfMissing True)
    [ root </> (outToPath <| Bin ""),
      root </> (outToPath <| Lib ""),
      root </> intdir,
      root </> nixdir,
      root </> vardir
    ]

-- >>> removeVersion "array-0.5.4.0-DFLKGIjfsadi"
-- "array"
removeVersion :: String -> String
removeVersion = takeWhile (/= '.') .> butlast2
  where
    butlast2 s = take (length s - 2) s

type Analysis = Map Namespace Target

analyze :: Analysis -> Namespace -> IO Analysis
analyze hmap ns = case Map.lookup ns hmap of
  Nothing -> do
    mTarget <- analyzeOne ns
    pure <| maybe hmap (\t -> Map.insert ns t hmap) mTarget
  Just _ -> pure hmap
  where
    analyzeOne :: Namespace -> IO (Maybe Target)
    analyzeOne namespace@(Namespace _ ext) = do
      let path = Namespace.toPath namespace
      content <-
        withFile path ReadMode <| \h ->
          IO.hSetEncoding h IO.utf8_bom
            >> Text.IO.hGetContents h
      let contentLines = Text.lines content
      root <- Env.getEnv "BIZ_ROOT"
      absPath <- Dir.makeAbsolute path
      user <- Env.getEnv "USER" /> Text.pack
      host <- Text.pack </ fromMaybe "interactive" </ Env.lookupEnv "HOSTNAME"
      Log.info ["bild", "analyze", str path]
      case ext of
        -- basically we don't support building these
        Namespace.Css -> pure Nothing
        Namespace.Json -> pure Nothing
        Namespace.Keys -> pure Nothing
        Namespace.Md -> pure Nothing
        Namespace.None -> pure Nothing
        Namespace.Py -> pure Nothing
        Namespace.Sh -> pure Nothing
        Namespace.C -> do
          let out =
                contentLines
                  /> Text.unpack
                  /> Regex.match (metaOut "//" <|> metaLib "//")
                  |> catMaybes
                  |> head
                  |> fromMaybe None
          let sysdeps =
                contentLines
                  /> Text.unpack
                  /> Regex.match (metaSys "//")
                  |> catMaybes
                  |> Set.fromList
          guileFlags <-
            if "guile_3_0" `elem` sysdeps
              then
                Process.readProcess "guile-config" ["compile"] ""
                  /> String.words
                  /> (++ ["-shared", "-fPIC"])
                  /> map Text.pack
              else pure mempty
          Just
            </ pure
              Target
                { langdeps = Set.empty, -- c has no lang deps...?
                  compiler = Gcc,
                  builder = user <> "@localhost",
                  compilerFlags =
                    concat
                      [ [o, dir, Text.pack absPath] ++ guileFlags
                        | let outable = out /= None,
                          o <- outable ?: (["-o"], []),
                          dir <- outable ?: ([Text.pack <| root </> outToPath out], [])
                      ],
                  outPath = outToPath out,
                  ..
                }
        Namespace.Hs -> do
          langdeps <- detectHaskellImports contentLines
          let out =
                contentLines
                  /> Text.unpack
                  /> Regex.match (metaOut "--")
                  |> catMaybes
                  |> head
                  |> fromMaybe None
          Just
            </ pure
              Target
                { builder = user <> "@localhost",
                  compiler = detectGhcCompiler out,
                  compilerFlags =
                    map
                      Text.pack
                      [ "-Werror",
                        "-i" <> root,
                        "-odir",
                        root </> intdir,
                        "-hidir",
                        root </> intdir,
                        "--make",
                        path
                      ]
                      ++ (out /= None)
                      ?: ( map
                             Text.pack
                             [ "-main-is",
                               Namespace.toHaskellModule namespace,
                               "-o",
                               root </> outToPath out
                             ],
                           []
                         ),
                  sysdeps =
                    contentLines
                      /> Text.unpack
                      /> Regex.match (metaSys "--")
                      |> catMaybes
                      |> Set.fromList,
                  outPath = outToPath out,
                  ..
                }
        Namespace.Lisp -> do
          let out =
                contentLines
                  /> Text.unpack
                  /> Regex.match (metaOut ";;")
                  |> catMaybes
                  |> head
                  |> fromMaybe None
          langdeps <- detectLispImports contentLines
          Just
            </ pure
              Target
                { sysdeps = Set.empty,
                  compiler = Sbcl,
                  compilerFlags =
                    map
                      Text.pack
                      [ "--load",
                        path,
                        "--eval",
                        "(require :asdf)",
                        "--eval",
                        "(sb-ext:save-lisp-and-die #p\"" <> (root </> outToPath out) <> "\" :toplevel #'main :executable t)"
                      ],
                  builder = user <> "@localhost",
                  outPath = outToPath out,
                  ..
                }
        Namespace.Nix -> do
          let builder =
                (host == "lithium")
                  ?: ( "local",
                       Text.concat
                         [ "ssh://",
                           user,
                           "@dev.simatime.com?ssh-key=/home/",
                           user,
                           "/.ssh/id_rsa"
                         ]
                     )
          Just
            </ pure
              Target
                { langdeps = Set.empty,
                  sysdeps = Set.empty,
                  compiler = NixBuild,
                  compilerFlags =
                    map
                      Text.pack
                      [ path,
                        "--out-link",
                        root </> nixdir </> Namespace.toPath namespace,
                        "--builders",
                        Text.unpack builder
                      ],
                  out = None,
                  outPath = outToPath None,
                  ..
                }
        Namespace.Scm -> do
          let out =
                contentLines
                  /> Text.unpack
                  /> Regex.match (metaOut ";;")
                  |> catMaybes
                  |> head
                  |> fromMaybe None
          Just
            </ pure
              Target
                { langdeps = Set.empty,
                  sysdeps = Set.empty,
                  compiler = Guile,
                  compilerFlags =
                    map
                      Text.pack
                      [ "compile",
                        "--r7rs",
                        "--load-path=" ++ root,
                        "--output=" ++ root </> intdir </> replaceExtension path ".scm.go",
                        path
                      ],
                  builder = user <> "@localhost",
                  outPath = outToPath out,
                  ..
                }
        Namespace.Rs -> do
          let out =
                contentLines
                  /> Text.unpack
                  /> Regex.match (metaOut "//")
                  |> catMaybes
                  |> head
                  |> fromMaybe None
          Just
            </ pure
              Target
                { langdeps = Set.empty,
                  sysdeps = Set.empty,
                  compiler = Rustc,
                  compilerFlags = map Text.pack [path, "-o", root </> outToPath out],
                  builder = user <> "@localhost",
                  outPath = outToPath out,
                  ..
                }
    detectHaskellImports :: [Text] -> IO (Set Dep)
    detectHaskellImports contentLines = do
      let imports =
            contentLines
              /> Text.unpack
              /> Regex.match haskellImports
              |> catMaybes
      pkgs <- foldM ghcPkgFindModule Set.empty imports
      root <- Env.getEnv "BIZ_ROOT"
      transitivePkgs <-
        imports
          |> map (Namespace.fromHaskellModule .> Namespace.toPath)
          |> traverse Dir.makeAbsolute
          +> filterM Dir.doesFileExist
          /> map (Namespace.fromPath root)
          /> catMaybes
          -- this is still an inefficiency, because this recurses before the
          -- hmap is updated by the fold, transitive imports will be
          -- re-visited. you can see this with `TERM=dumb bild`. to fix this i
          -- need shared state instead of a fold, or figure out how to do a
          -- breadth-first search instead of depth-first.
          +> foldM analyze hmap
          /> Map.elems
          /> map langdeps
          /> mconcat
      pure <| pkgs <> transitivePkgs
    detectLispImports contentLines = do
      let requires = contentLines /> Text.unpack /> Regex.match lispRequires |> catMaybes
      pure <| Set.fromList requires

ghcPkgFindModule :: Set String -> String -> IO (Set String)
ghcPkgFindModule acc m = do
  packageDb <- Env.getEnv "GHC_PACKAGE_PATH"
  Process.readProcess
    "ghc-pkg"
    ["--package-db", packageDb, "--names-only", "--simple-output", "find-module", m]
    ""
    /> String.lines
    /> Set.fromList
    /> Set.union acc

-- | Some rules for detecting the how to compile a ghc module. If there is an
-- out, then we know it's some Exe, otherwise it's a Lib.
detectGhcCompiler :: Out -> Compiler
detectGhcCompiler = \case
  Bin _ -> GhcExe
  Lib _ -> GhcLib
  None -> GhcLib

isFailure :: Exit.ExitCode -> Bool
isFailure (Exit.ExitFailure _) = True
isFailure Exit.ExitSuccess = False

isSuccess :: Exit.ExitCode -> Bool
isSuccess Exit.ExitSuccess = True
isSuccess _ = False

test :: Bool -> Target -> IO Exit.ExitCode
test loud Target {..} = case compiler of
  GhcExe -> do
    root <- Env.getEnv "BIZ_ROOT"
    run
      <| Proc
        { loud = loud,
          cmd = root </> outToPath out,
          args = ["test"],
          ns = namespace,
          onFailure = Log.fail ["test", nschunk namespace] >> Log.br,
          onSuccess = Log.pass ["test", nschunk namespace] >> Log.br
        }
  _ ->
    Log.warn ["test", nschunk namespace, "unavailable"]
      >> Log.br
      >> pure Exit.ExitSuccess

build :: Bool -> Bool -> Analysis -> IO [Exit.ExitCode]
build andTest loud analysis = do
  root <- Env.getEnv "BIZ_ROOT"
  forM (Map.elems analysis) <| \target@Target {..} -> do
    case compiler of
      Gcc ->
        Log.info ["bild", label, "gcc", nschunk namespace]
          >> proc loud namespace "gcc" compilerFlags
        where
          label = case out of
            Bin _ -> "bin"
            _ -> "lib"
      GhcExe -> do
        Log.info ["bild", "dev", "ghc-exe", nschunk namespace]
        exitcode <- proc loud namespace "ghc" compilerFlags
        if andTest && isSuccess exitcode
          then test loud target
          else pure exitcode
      GhcLib -> do
        Log.info ["bild", "dev", "ghc-lib", nschunk namespace]
        proc
          loud
          namespace
          "ghc"
          compilerFlags
      Guile -> do
        Log.info ["bild", "dev", "guile", nschunk namespace]
        _ <- proc loud namespace "guild" compilerFlags
        when (out /= None) <| do
          writeFile
            (root </> outToPath out)
            <| Text.pack
            <| joinWith
              "\n"
              [ "#!/usr/bin/env bash",
                "guile -C \""
                  <> root </> intdir
                  <> "\" -e main "
                  <> "-s "
                  <> Namespace.toPath namespace
                  <> " \"$@\""
              ]
          p <- Dir.getPermissions <| root </> outToPath out
          Dir.setPermissions (root </> outToPath out) (Dir.setOwnerExecutable True p)
        pure Exit.ExitSuccess
      NixBuild -> do
        Log.info ["bild", "nix", builder, nschunk namespace]
        proc loud namespace "nix-build" compilerFlags
      Copy -> do
        Log.warn ["bild", "copy", "TODO", nschunk namespace]
        pure Exit.ExitSuccess
      Rustc -> do
        Log.info ["bild", "dev", "rust", nschunk namespace]
        proc loud namespace "rustc" compilerFlags
      Sbcl -> do
        Log.info ["bild", "dev", "lisp", nschunk namespace]
        proc loud namespace "sbcl" compilerFlags

data Proc = Proc
  { loud :: Bool,
    cmd :: String,
    args :: [String],
    ns :: Namespace,
    onFailure :: IO (),
    onSuccess :: IO ()
  }

-- | Run a subprocess, streaming output if --loud is set.
run :: Proc -> IO Exit.ExitCode
run Proc {..} = do
  (Conduit.Inherited, stdout_, stderr_, cph) <- Conduit.streamingProcess <| Conduit.proc cmd args
  exitcode <-
    if loud
      then
        Async.runConcurrently
          <| Async.Concurrently (puts stdout_)
          *> (Async.Concurrently <| Conduit.waitForStreamingProcess cph)
      else
        Async.runConcurrently
          <| Async.Concurrently
          <| Conduit.waitForStreamingProcess cph
  if isFailure exitcode
    then puts stderr_ >> onFailure >> pure exitcode
    else onSuccess >> pure exitcode

-- | Helper for running a standard bild subprocess.
proc :: Bool -> Namespace -> String -> [Text] -> IO Exit.ExitCode
proc loud namespace cmd args =
  run
    <| Proc
      { loud = loud,
        ns = namespace,
        cmd = cmd,
        args = map Text.unpack args,
        onFailure = Log.fail ["bild", nschunk namespace] >> Log.br,
        onSuccess = Log.good ["bild", nschunk namespace] >> Log.br
      }

-- | Helper for printing during a subprocess
puts :: Conduit.ConduitM () ByteString IO () -> IO ()
puts thing = Conduit.runConduit <| thing .| Conduit.mapM_ putStr

nschunk :: Namespace -> Text
nschunk = Namespace.toPath .> Text.pack

metaDep :: Regex.RE Char Dep
metaDep =
  Regex.string "-- : dep "
    *> Regex.many (Regex.psym Char.isAlpha)

metaSys :: [Char] -> Regex.RE Char Dep
metaSys comment =
  Regex.string (comment ++ " : sys ")
    *> Regex.many (Regex.psym (not <. Char.isSpace))

metaOut :: [Char] -> Regex.RE Char Out
metaOut comment =
  Regex.string (comment ++ " : out ")
    *> Regex.many (Regex.psym (/= ' '))
    /> Bin

metaLib :: [Char] -> Regex.RE Char Out
metaLib comment =
  Regex.string (comment ++ " : lib ")
    *> Regex.many (Regex.psym (/= ' '))
    /> Lib

haskellImports :: Regex.RE Char String
haskellImports =
  Regex.string "import"
    *> Regex.some (Regex.psym Char.isSpace)
    *> Regex.many (Regex.psym Char.isLower)
    *> Regex.many (Regex.psym Char.isSpace)
    *> Regex.some (Regex.psym isModuleChar)
    <* Regex.many Regex.anySym

isModuleChar :: Char -> Bool
isModuleChar c =
  elem c <| concat [['A' .. 'Z'], ['a' .. 'z'], ['.', '_'], ['0' .. '9']]

-- Matches on `(require :package)` forms and returns `package`. The `require`
-- function is technically deprecated in Common Lisp, but no new spec has been
-- published with a replacement, and I don't wanna use asdf, so this is what we
-- use for Lisp imports.
lispRequires :: Regex.RE Char String
lispRequires =
  Regex.string "(require"
    *> Regex.some (Regex.psym Char.isSpace)
    *> Regex.many (Regex.psym isQuote)
    *> Regex.many (Regex.psym isModuleChar)
    <* Regex.many (Regex.psym (== ')'))
  where
    isQuote :: Char -> Bool
    isQuote c = c `elem` ['\'', ':']