mirror of
https://github.com/dense-analysis/ale.git
synced 2025-12-24 13:01:28 +08:00
* Add option to open loclist/quicklist when there are errors I copied PR #137, and tries to complete it by correcting some issues and adding vader tests. About tests, first time with vader, can you give some feedback if there are what you expected in PR #137. * Remove old code + fix indent issue * add g:ale_keep_list_window_open option * Correct bug with keep open option * Add comment into vader file * Fix errors for Travis CI build
410 lines
14 KiB
VimL
410 lines
14 KiB
VimL
" Author: w0rp <devw0rp@gmail.com>
|
|
" Description: Backend execution and job management
|
|
" Executes linters in the background, using NeoVim or Vim 8 jobs
|
|
|
|
" Stores information for each job including:
|
|
"
|
|
" linter: The linter dictionary for the job.
|
|
" buffer: The buffer number for the job.
|
|
" output: The array of lines for the output of the job.
|
|
let s:job_info_map = {}
|
|
|
|
function! s:GetJobID(job) abort
|
|
if has('nvim')
|
|
"In NeoVim, job values are just IDs.
|
|
return a:job
|
|
endif
|
|
|
|
" In Vim 8, the job is a special variable, and we open a channel for each
|
|
" job. We'll use the ID of the channel instead as the job ID.
|
|
return ch_info(job_getchannel(a:job)).id
|
|
endfunction
|
|
|
|
function! ale#engine#InitBufferInfo(buffer) abort
|
|
if !has_key(g:ale_buffer_info, a:buffer)
|
|
" job_list will hold the list of jobs
|
|
" loclist holds the loclist items after all jobs have completed.
|
|
" new_loclist holds loclist items while jobs are being run.
|
|
let g:ale_buffer_info[a:buffer] = {
|
|
\ 'job_list': [],
|
|
\ 'loclist': [],
|
|
\ 'new_loclist': [],
|
|
\}
|
|
endif
|
|
endfunction
|
|
|
|
function! ale#engine#ClearJob(job) abort
|
|
let l:job_id = s:GetJobID(a:job)
|
|
let l:linter = s:job_info_map[l:job_id].linter
|
|
|
|
if has('nvim')
|
|
call jobstop(a:job)
|
|
else
|
|
" We must close the channel for reading the buffer if it is open
|
|
" when stopping a job. Otherwise, we will get errors in the status line.
|
|
if ch_status(job_getchannel(a:job)) ==# 'open'
|
|
call ch_close_in(job_getchannel(a:job))
|
|
endif
|
|
|
|
call job_stop(a:job)
|
|
endif
|
|
|
|
call remove(s:job_info_map, l:job_id)
|
|
endfunction
|
|
|
|
function! s:StopPreviousJobs(buffer, linter) abort
|
|
let l:new_job_list = []
|
|
|
|
for l:job in g:ale_buffer_info[a:buffer].job_list
|
|
let l:job_id = s:GetJobID(l:job)
|
|
|
|
if has_key(s:job_info_map, l:job_id)
|
|
\&& s:job_info_map[l:job_id].linter.name ==# a:linter.name
|
|
" Stop jobs which match the buffer and linter.
|
|
call ale#engine#ClearJob(l:job)
|
|
else
|
|
" Keep other jobs in the list.
|
|
call add(l:new_job_list, l:job)
|
|
endif
|
|
endfor
|
|
|
|
" Update the list, removing the previously run job.
|
|
let g:ale_buffer_info[a:buffer].job_list = l:new_job_list
|
|
endfunction
|
|
|
|
function! s:GatherOutputVim(channel, data) abort
|
|
let l:job_id = s:GetJobID(ch_getjob(a:channel))
|
|
|
|
if !has_key(s:job_info_map, l:job_id)
|
|
return
|
|
endif
|
|
|
|
call add(s:job_info_map[l:job_id].output, a:data)
|
|
endfunction
|
|
|
|
function! s:GatherOutputNeoVim(job, data, event) abort
|
|
let l:job_id = s:GetJobID(a:job)
|
|
|
|
if !has_key(s:job_info_map, l:job_id)
|
|
return
|
|
endif
|
|
|
|
" Join the lines passed to ale, because Neovim splits them up.
|
|
" a:data is a list of strings, where every item is a new line, except the
|
|
" first one, which is the continuation of the last item passed last time.
|
|
call ale#engine#JoinNeovimOutput(s:job_info_map[l:job_id].output, a:data)
|
|
endfunction
|
|
|
|
function! ale#engine#JoinNeovimOutput(output, data) abort
|
|
if empty(a:output)
|
|
call extend(a:output, a:data)
|
|
else
|
|
" Extend the previous line, which can be continued.
|
|
let a:output[-1] .= get(a:data, 0, '')
|
|
|
|
" Add the new lines.
|
|
call extend(a:output, a:data[1:])
|
|
endif
|
|
endfunction
|
|
|
|
function! s:HandleExit(job) abort
|
|
if a:job ==# 'no process'
|
|
" Stop right away when the job is not valid in Vim 8.
|
|
return
|
|
endif
|
|
|
|
let l:job_id = s:GetJobID(a:job)
|
|
|
|
if !has_key(s:job_info_map, l:job_id)
|
|
return
|
|
endif
|
|
|
|
let l:job_info = s:job_info_map[l:job_id]
|
|
let l:linter = l:job_info.linter
|
|
let l:output = l:job_info.output
|
|
let l:buffer = l:job_info.buffer
|
|
let l:next_chain_index = l:job_info.next_chain_index
|
|
|
|
" Call the same function for stopping jobs again to clean up the job
|
|
" which just closed.
|
|
call s:StopPreviousJobs(l:buffer, l:linter)
|
|
|
|
if l:next_chain_index < len(get(l:linter, 'command_chain', []))
|
|
call s:InvokeChain(l:buffer, l:linter, l:next_chain_index, l:output)
|
|
return
|
|
endif
|
|
|
|
let l:linter_loclist = ale#util#GetFunction(l:linter.callback)(l:buffer, l:output)
|
|
|
|
" Make some adjustments to the loclists to fix common problems.
|
|
call s:FixLocList(l:buffer, l:linter_loclist)
|
|
|
|
for l:item in l:linter_loclist
|
|
let l:item.linter_name = l:linter.name
|
|
endfor
|
|
|
|
" Add the loclist items from the linter.
|
|
call extend(g:ale_buffer_info[l:buffer].new_loclist, l:linter_loclist)
|
|
|
|
if !empty(g:ale_buffer_info[l:buffer].job_list)
|
|
" Wait for all jobs to complete before doing anything else.
|
|
return
|
|
endif
|
|
|
|
" Sort the loclist again.
|
|
" We need a sorted list so we can run a binary search against it
|
|
" for efficient lookup of the messages in the cursor handler.
|
|
call sort(g:ale_buffer_info[l:buffer].new_loclist, 'ale#util#LocItemCompare')
|
|
|
|
" Now swap the old and new loclists, after we have collected everything
|
|
" and sorted the list again.
|
|
let g:ale_buffer_info[l:buffer].loclist = g:ale_buffer_info[l:buffer].new_loclist
|
|
let g:ale_buffer_info[l:buffer].new_loclist = []
|
|
|
|
if g:ale_set_quickfix || g:ale_set_loclist
|
|
call ale#list#SetLists(g:ale_buffer_info[l:buffer].loclist)
|
|
endif
|
|
|
|
if g:ale_set_signs
|
|
call ale#sign#SetSigns(l:buffer, g:ale_buffer_info[l:buffer].loclist)
|
|
endif
|
|
|
|
if exists('*ale#statusline#Update')
|
|
" Don't load/run if not already loaded.
|
|
call ale#statusline#Update(l:buffer, g:ale_buffer_info[l:buffer].loclist)
|
|
endif
|
|
|
|
" Call user autocommands. This allows users to hook into ALE's lint cycle.
|
|
silent doautocmd User ALELint
|
|
|
|
" Mark line 200, column 17 with a squiggly line or something
|
|
" matchadd('ALEError', '\%200l\%17v')
|
|
endfunction
|
|
|
|
function! s:HandleExitNeoVim(job, data, event) abort
|
|
call s:HandleExit(a:job)
|
|
endfunction
|
|
|
|
function! s:HandleExitVim(channel) abort
|
|
call s:HandleExit(ch_getjob(a:channel))
|
|
endfunction
|
|
|
|
function! s:FixLocList(buffer, loclist) abort
|
|
" Some errors have line numbers beyond the end of the file,
|
|
" so we need to adjust them so they set the error at the last line
|
|
" of the file instead.
|
|
let l:last_line_number = ale#util#GetLineCount(a:buffer)
|
|
|
|
for l:item in a:loclist
|
|
if l:item.lnum == 0
|
|
" When errors appear at line 0, put them at line 1 instead.
|
|
let l:item.lnum = 1
|
|
elseif l:item.lnum > l:last_line_number
|
|
let l:item.lnum = l:last_line_number
|
|
endif
|
|
endfor
|
|
endfunction
|
|
|
|
function! s:RunJob(command, generic_job_options) abort
|
|
let l:buffer = a:generic_job_options.buffer
|
|
let l:linter = a:generic_job_options.linter
|
|
let l:output_stream = a:generic_job_options.output_stream
|
|
let l:next_chain_index = a:generic_job_options.next_chain_index
|
|
let l:command = a:command
|
|
|
|
if l:command =~# '%s'
|
|
" If there is a '%s' in the command string, replace it with the name
|
|
" of the file.
|
|
let l:command = printf(l:command, shellescape(fnamemodify(bufname(l:buffer), ':p')))
|
|
endif
|
|
|
|
if has('nvim')
|
|
if l:output_stream ==# 'stderr'
|
|
" Read from stderr instead of stdout.
|
|
let l:job = jobstart(l:command, {
|
|
\ 'on_stderr': function('s:GatherOutputNeoVim'),
|
|
\ 'on_exit': function('s:HandleExitNeoVim'),
|
|
\})
|
|
elseif l:output_stream ==# 'both'
|
|
let l:job = jobstart(l:command, {
|
|
\ 'on_stdout': function('s:GatherOutputNeoVim'),
|
|
\ 'on_stderr': function('s:GatherOutputNeoVim'),
|
|
\ 'on_exit': function('s:HandleExitNeoVim'),
|
|
\})
|
|
else
|
|
let l:job = jobstart(l:command, {
|
|
\ 'on_stdout': function('s:GatherOutputNeoVim'),
|
|
\ 'on_exit': function('s:HandleExitNeoVim'),
|
|
\})
|
|
endif
|
|
else
|
|
let l:job_options = {
|
|
\ 'in_mode': 'nl',
|
|
\ 'out_mode': 'nl',
|
|
\ 'err_mode': 'nl',
|
|
\ 'close_cb': function('s:HandleExitVim'),
|
|
\}
|
|
|
|
if l:output_stream ==# 'stderr'
|
|
" Read from stderr instead of stdout.
|
|
let l:job_options.err_cb = function('s:GatherOutputVim')
|
|
elseif l:output_stream ==# 'both'
|
|
" Read from both streams.
|
|
let l:job_options.out_cb = function('s:GatherOutputVim')
|
|
let l:job_options.err_cb = function('s:GatherOutputVim')
|
|
else
|
|
let l:job_options.out_cb = function('s:GatherOutputVim')
|
|
endif
|
|
|
|
if has('win32')
|
|
" job_start commands on Windows have to be run with cmd /c,
|
|
" othwerwise %PATHTEXT% will not be used to programs ending int
|
|
" .cmd, .bat, .exe, etc.
|
|
let l:command = 'cmd /c ' . l:command
|
|
else
|
|
" Execute the command with the shell, to fix escaping issues.
|
|
let l:command = split(&shell) + split(&shellcmdflag) + [l:command]
|
|
|
|
" On Unix machines, we can send the Vim buffer directly.
|
|
" This is faster than reading the lines ourselves.
|
|
let l:job_options.in_io = 'buffer'
|
|
let l:job_options.in_buf = l:buffer
|
|
endif
|
|
|
|
" Vim 8 will read the stdin from the file's buffer.
|
|
let l:job = job_start(l:command, l:job_options)
|
|
endif
|
|
|
|
" Only proceed if the job is being run.
|
|
if has('nvim') || (l:job !=# 'no process' && job_status(l:job) ==# 'run')
|
|
" Add the job to the list of jobs, so we can track them.
|
|
call add(g:ale_buffer_info[l:buffer].job_list, l:job)
|
|
|
|
" Store the ID for the job in the map to read back again.
|
|
let s:job_info_map[s:GetJobID(l:job)] = {
|
|
\ 'linter': l:linter,
|
|
\ 'buffer': l:buffer,
|
|
\ 'output': [],
|
|
\ 'next_chain_index': l:next_chain_index,
|
|
\}
|
|
|
|
if has('nvim')
|
|
" In NeoVim, we have to send the buffer lines ourselves.
|
|
let l:input = join(getbufline(l:buffer, 1, '$'), "\n") . "\n"
|
|
|
|
call jobsend(l:job, l:input)
|
|
call jobclose(l:job, 'stdin')
|
|
elseif has('win32')
|
|
" On some Vim versions, we have to send the buffer data ourselves.
|
|
let l:input = join(getbufline(l:buffer, 1, '$'), "\n") . "\n"
|
|
let l:channel = job_getchannel(l:job)
|
|
|
|
if ch_status(l:channel) ==# 'open'
|
|
call ch_sendraw(l:channel, l:input)
|
|
call ch_close_in(l:channel)
|
|
endif
|
|
endif
|
|
endif
|
|
endfunction
|
|
|
|
function! s:InvokeChain(buffer, linter, chain_index, input) abort
|
|
let l:output_stream = get(a:linter, 'output_stream', 'stdout')
|
|
|
|
if has_key(a:linter, 'command_chain')
|
|
" Run a chain of commands, one asychronous command after the other,
|
|
" so that many programs can be run in a sequence.
|
|
let l:chain_item = a:linter.command_chain[a:chain_index]
|
|
|
|
" The chain item can override the output_stream option.
|
|
if has_key(l:chain_item)
|
|
let l:output_stream = l:chain_item.output_stream
|
|
endif
|
|
|
|
let l:callback = ale#util#GetFunction(a:linter.callback)
|
|
|
|
if a:chain_index == 0
|
|
" The first callback in the chain takes only a buffer number.
|
|
let l:command = l:callback(a:buffer)
|
|
else
|
|
" The second callback in the chain takes some input too.
|
|
let l:command = l:callback(a:buffer, a:input)
|
|
endif
|
|
elseif has_key(a:linter, 'command_callback')
|
|
" If there is a callback for generating a command, call that instead.
|
|
let l:command = ale#util#GetFunction(a:linter.command_callback)(a:buffer)
|
|
else
|
|
let l:command = a:linter.command
|
|
endif
|
|
|
|
call s:RunJob(l:command, {
|
|
\ 'buffer': a:buffer,
|
|
\ 'linter': a:linter,
|
|
\ 'output_stream': l:output_stream,
|
|
\ 'next_chain_index': a:chain_index + 1,
|
|
\})
|
|
endfunction
|
|
|
|
function! ale#engine#Invoke(buffer, linter) abort
|
|
" Stop previous jobs for the same linter.
|
|
call s:StopPreviousJobs(a:buffer, a:linter)
|
|
call s:InvokeChain(a:buffer, a:linter, 0, [])
|
|
endfunction
|
|
|
|
" Given a buffer number, return the warnings and errors for a given buffer.
|
|
function! ale#engine#GetLoclist(buffer) abort
|
|
if !has_key(g:ale_buffer_info, a:buffer)
|
|
return []
|
|
endif
|
|
|
|
return g:ale_buffer_info[a:buffer].loclist
|
|
endfunction
|
|
|
|
" This function can be called with a timeout to wait for all jobs to finish.
|
|
" If the jobs to not finish in the given number of milliseconds,
|
|
" an exception will be thrown.
|
|
"
|
|
" The time taken will be a very rough approximation, and more time may be
|
|
" permitted than is specified.
|
|
function! ale#engine#WaitForJobs(deadline) abort
|
|
let l:start_time = system('date +%s%3N') + 0
|
|
|
|
if l:start_time == 0
|
|
throw 'Failed to read milliseconds from the clock!'
|
|
endif
|
|
|
|
let l:job_list = []
|
|
|
|
" Gather all of the jobs from every buffer.
|
|
for l:info in values(g:ale_buffer_info)
|
|
call extend(l:job_list, l:info.job_list)
|
|
endfor
|
|
|
|
let l:should_wait_more = 1
|
|
|
|
while l:should_wait_more
|
|
let l:should_wait_more = 0
|
|
|
|
for l:job in l:job_list
|
|
if job_status(l:job) ==# 'run'
|
|
let l:now = system('date +%s%3N') + 0
|
|
|
|
if l:now - l:start_time > a:deadline
|
|
" Stop waiting after a timeout, so we don't wait forever.
|
|
throw 'Jobs did not complete on time!'
|
|
endif
|
|
|
|
" Wait another 10 milliseconds
|
|
let l:should_wait_more = 1
|
|
sleep 10ms
|
|
break
|
|
endif
|
|
endfor
|
|
endwhile
|
|
|
|
" Sleep for a small amount of time after all jobs finish.
|
|
" This seems to be enough to let handlers after jobs end run, and
|
|
" prevents the occasional failure where this function exits after jobs
|
|
" end, but before handlers are run.
|
|
sleep 10ms
|
|
endfunction
|