raka 0.3.1 → 0.3.4
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/README.md +343 -119
- data/VERSION +1 -1
- data/bin/raka +37 -12
- data/lib/{compile.rb → raka/compile.rb} +10 -4
- data/lib/{interface.rbs → raka/interface.rbs} +0 -0
- data/lib/{lang → raka/lang}/psql/impl.rb +0 -0
- data/lib/{lang → raka/lang}/python/impl.rb +0 -0
- data/lib/raka/lang/r/impl.rb +23 -0
- data/lib/{lang → raka/lang}/shell/impl.rb +0 -0
- data/lib/{output_type.rb → raka/output_type.rb} +0 -0
- data/lib/{protocol.rb → raka/protocol.rb} +0 -0
- data/lib/{token.rb → raka/token.rb} +0 -0
- data/lib/raka.rb +9 -4
- metadata +13 -14
- data/lib/lang/r/impl.rb +0 -38
- data/lib/lang/r/io.R +0 -113
- data/lib/temp.json +0 -9167
data/bin/raka
CHANGED
@@ -17,8 +17,6 @@ def detect_main
|
|
17
17
|
return rakas[0] if rakas.length == 1
|
18
18
|
end
|
19
19
|
|
20
|
-
entry = detect_main
|
21
|
-
|
22
20
|
options = { rake: {}, raka_finished: false }
|
23
21
|
def set_option(opts, key, value)
|
24
22
|
if opts[:raka_finished]
|
@@ -28,17 +26,31 @@ def set_option(opts, key, value)
|
|
28
26
|
end
|
29
27
|
end
|
30
28
|
parser = OptionParser.new do |opts|
|
31
|
-
opts.banner = 'Usage: raka [options]
|
29
|
+
opts.banner = 'Usage: raka [options] <output> -- [rake options]'
|
32
30
|
|
33
31
|
opts.on('-v', '--[no-]verbose', 'Run verbosely') do |v|
|
34
32
|
set_option(options, :verbose, v)
|
35
33
|
end
|
34
|
+
|
35
|
+
opts.on('-f', '--file FILE', String, 'Run even when up to date') do |s|
|
36
|
+
set_option(options, :file, s)
|
37
|
+
end
|
38
|
+
|
39
|
+
opts.on('-j', '--jobs JOBS', Integer, 'Run in parallel') do |n|
|
40
|
+
set_option(options, :jobs, n)
|
41
|
+
end
|
36
42
|
end
|
37
43
|
|
44
|
+
if ARGV.empty?
|
45
|
+
puts parser.help
|
46
|
+
exit(1)
|
47
|
+
end
|
38
48
|
both_args = ARGV.join(' ').split(' -- ')
|
39
|
-
extra_args = both_args[1]
|
40
49
|
self_args = both_args[0].split(/\s+/)
|
41
50
|
parser.parse!(self_args)
|
51
|
+
extra_args = (both_args[1] || ' ').lstrip
|
52
|
+
|
53
|
+
entry = options[:file] || detect_main
|
42
54
|
|
43
55
|
env = if options[:verbose]
|
44
56
|
'LOG_LEVEL=0 '
|
@@ -46,13 +58,26 @@ env = if options[:verbose]
|
|
46
58
|
''
|
47
59
|
end
|
48
60
|
targets = self_args.join(' ')
|
49
|
-
cmd =
|
61
|
+
cmd = ''
|
62
|
+
opt_str = "-f #{entry}"
|
63
|
+
opt_str += " -m -j #{options[:jobs]}" if options.key?(:jobs)
|
64
|
+
cmd += "#{env}rake #{opt_str} #{extra_args} #{targets}"
|
50
65
|
puts cmd
|
51
|
-
output
|
52
|
-
|
53
|
-
|
54
|
-
|
55
|
-
|
56
|
-
puts 'Error: rake returns the following information:'
|
66
|
+
output = []
|
67
|
+
ro, out = IO.pipe
|
68
|
+
re, err = IO.pipe
|
69
|
+
pid = fork do
|
70
|
+
status = system(cmd, out: out, err: err)
|
71
|
+
puts 'Error: rake returns the following information:' unless status
|
72
|
+
exit($CHILD_STATUS.exitstatus)
|
57
73
|
end
|
58
|
-
|
74
|
+
out.close
|
75
|
+
err.close
|
76
|
+
ro.each_line do |l|
|
77
|
+
puts l
|
78
|
+
output << l.chomp
|
79
|
+
end
|
80
|
+
re.each_line { |l| puts l; }
|
81
|
+
|
82
|
+
Process.wait(pid)
|
83
|
+
puts 'All targets are up to date' if output.empty? && $CHILD_STATUS.exitstatus == 0
|
@@ -28,8 +28,7 @@ class DSLCompiler
|
|
28
28
|
# Raka task structure, input task is rake's task pushed into blocks
|
29
29
|
def dsl_task(token, task)
|
30
30
|
name = task.name
|
31
|
-
deps = task.prerequisites
|
32
|
-
|
31
|
+
deps = task.prerequisites
|
33
32
|
output_info = token._parse_output_ name
|
34
33
|
task_info = {
|
35
34
|
name: name,
|
@@ -41,6 +40,10 @@ class DSLCompiler
|
|
41
40
|
OpenStruct.new(output_info.to_h.merge(task_info))
|
42
41
|
end
|
43
42
|
|
43
|
+
def stem(path)
|
44
|
+
File.basename(path, File.extname(path))
|
45
|
+
end
|
46
|
+
|
44
47
|
# resolve auto variables with only output info,
|
45
48
|
# useful when resolve extra deps (task is not available yet)
|
46
49
|
def resolve_by_output(target, output_info)
|
@@ -50,8 +53,10 @@ class DSLCompiler
|
|
50
53
|
.gsub('$(scope)', info.scope.nil? ? '' : info.scope)
|
51
54
|
.gsub('$(target_scope)', info.target_scope.nil? ? '' : info.target_scope)
|
52
55
|
.gsub('$(output)', info.output)
|
53
|
-
.gsub('$(output_stem)', info.stem)
|
56
|
+
.gsub('$(output_stem)', stem(info.stem))
|
54
57
|
.gsub('$(input_stem)', info.input_stem.nil? ? '' : info.input_stem)
|
58
|
+
.gsub('$(func)', info.func.nil? ? '' : info.func)
|
59
|
+
.gsub('$(ext)', info.ext)
|
55
60
|
.gsub('$@', info.name)
|
56
61
|
|
57
62
|
protect_percent_symbol text do |safe_text|
|
@@ -75,7 +80,8 @@ class DSLCompiler
|
|
75
80
|
|
76
81
|
protect_percent_symbol text do |safe_text|
|
77
82
|
# add numbered auto variables like $0, $2 referring to the first and third deps
|
78
|
-
safe_text.gsub(/\$\(dep(\d+)\)/, '%{\1}') % array_to_hash(task.deps)
|
83
|
+
safe_text = safe_text.gsub(/\$\(dep(\d+)\)/, '%{\1}') % array_to_hash(task.deps)
|
84
|
+
safe_text.gsub(/\$\(dep(\d+)_stem\)/, '%{\1}') % array_to_hash(task.deps.map {|d| stem(d)})
|
79
85
|
end
|
80
86
|
end
|
81
87
|
|
File without changes
|
File without changes
|
File without changes
|
@@ -0,0 +1,23 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require_relative '../../protocol'
|
4
|
+
|
5
|
+
# r language protocol
|
6
|
+
class R
|
7
|
+
def initialize(libs = [], **kwargs)
|
8
|
+
@libs = libs
|
9
|
+
super(**kwargs)
|
10
|
+
end
|
11
|
+
|
12
|
+
def build(code, _)
|
13
|
+
libraries = @libs.map { |name| "suppressPackageStartupMessages(library(#{name}))" }
|
14
|
+
|
15
|
+
[libraries, code].join "\n"
|
16
|
+
end
|
17
|
+
|
18
|
+
def run_script(env, fname, _task)
|
19
|
+
env.send :sh, "Rscript #{fname}"
|
20
|
+
end
|
21
|
+
end
|
22
|
+
|
23
|
+
creator :r, R
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
data/lib/raka.rb
CHANGED
@@ -2,13 +2,14 @@
|
|
2
2
|
|
3
3
|
require 'logger'
|
4
4
|
|
5
|
-
require_relative './compile'
|
6
|
-
require_relative './protocol'
|
7
|
-
require_relative './token'
|
5
|
+
require_relative './raka/compile'
|
6
|
+
require_relative './raka/protocol'
|
7
|
+
require_relative './raka/token'
|
8
8
|
|
9
9
|
# initialize raka
|
10
10
|
class Raka
|
11
11
|
Pattern = Pattern
|
12
|
+
P = Pattern
|
12
13
|
attr_reader :logger
|
13
14
|
|
14
15
|
def create_logger(level)
|
@@ -53,7 +54,7 @@ class Raka
|
|
53
54
|
@options.input_types |= @options.output_types # any output can be used as intermediate
|
54
55
|
# specify root of scopes in options, scopes will append to each root
|
55
56
|
@scopes = options.scopes.empty? ? [] : [options.scopes]
|
56
|
-
@options.lang.each { |path| load File::join(File::dirname(__FILE__), "
|
57
|
+
@options.lang.each { |path| load File::join(File::dirname(__FILE__), "raka/#{path}/impl.rb") }
|
57
58
|
@options.user_lang.each { |path| load path.to_s + '.rb' }
|
58
59
|
|
59
60
|
# These are where the dsl starts
|
@@ -67,4 +68,8 @@ class Raka
|
|
67
68
|
block.call
|
68
69
|
@scopes.pop
|
69
70
|
end
|
71
|
+
|
72
|
+
def stem(path)
|
73
|
+
File.basename(path, File.extname(path))
|
74
|
+
end
|
70
75
|
end
|
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: raka
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.3.
|
4
|
+
version: 0.3.4
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- yarray
|
8
8
|
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date: 2022-
|
11
|
+
date: 2022-05-09 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: rake
|
@@ -111,7 +111,8 @@ dependencies:
|
|
111
111
|
description: An extensible, concise and light weight DSL on Rake to automate data
|
112
112
|
processing tasks
|
113
113
|
email: '08to09@gmail.com'
|
114
|
-
executables:
|
114
|
+
executables:
|
115
|
+
- raka
|
115
116
|
extensions: []
|
116
117
|
extra_rdoc_files:
|
117
118
|
- LICENSE
|
@@ -121,18 +122,16 @@ files:
|
|
121
122
|
- README.md
|
122
123
|
- VERSION
|
123
124
|
- bin/raka
|
124
|
-
- lib/compile.rb
|
125
|
-
- lib/interface.rbs
|
126
|
-
- lib/lang/psql/impl.rb
|
127
|
-
- lib/lang/python/impl.rb
|
128
|
-
- lib/lang/r/impl.rb
|
129
|
-
- lib/lang/r/io.R
|
130
|
-
- lib/lang/shell/impl.rb
|
131
|
-
- lib/output_type.rb
|
132
|
-
- lib/protocol.rb
|
133
125
|
- lib/raka.rb
|
134
|
-
- lib/
|
135
|
-
- lib/
|
126
|
+
- lib/raka/compile.rb
|
127
|
+
- lib/raka/interface.rbs
|
128
|
+
- lib/raka/lang/psql/impl.rb
|
129
|
+
- lib/raka/lang/python/impl.rb
|
130
|
+
- lib/raka/lang/r/impl.rb
|
131
|
+
- lib/raka/lang/shell/impl.rb
|
132
|
+
- lib/raka/output_type.rb
|
133
|
+
- lib/raka/protocol.rb
|
134
|
+
- lib/raka/token.rb
|
136
135
|
homepage: http://github.com/yarray/raka
|
137
136
|
licenses:
|
138
137
|
- MIT
|
data/lib/lang/r/impl.rb
DELETED
@@ -1,38 +0,0 @@
|
|
1
|
-
# frozen_string_literal: true
|
2
|
-
|
3
|
-
require_relative '../protocol'
|
4
|
-
|
5
|
-
# r language protocol
|
6
|
-
class R
|
7
|
-
def initialize(src, libs = [], **kwargs)
|
8
|
-
@src = src
|
9
|
-
@libs = libs
|
10
|
-
super(**kwargs)
|
11
|
-
end
|
12
|
-
|
13
|
-
def build(code, _)
|
14
|
-
libraries = ([
|
15
|
-
:pipeR
|
16
|
-
] + @libs).map { |name| "suppressPackageStartupMessages(library(#{name}))" }
|
17
|
-
|
18
|
-
sources = ["source('#{File.dirname(__FILE__)}/io.R')"] +
|
19
|
-
(@src ? [@src] : []).map { |name| "source('#{SRC_DIR}/#{name}.R')" }
|
20
|
-
|
21
|
-
extra = [
|
22
|
-
'`|` <- `%>>%`',
|
23
|
-
"conn_args <- list(host='#{HOST}', user='#{USER}', dbname='#{DB}', port='#{PORT}')",
|
24
|
-
'args <- commandArgs(trailingOnly = T)',
|
25
|
-
'sql_input <- init_sql_input(conn_args, args[1])',
|
26
|
-
'table_input <- init_table_input(conn_args, args[1])',
|
27
|
-
'table_output <- init_table_output(conn_args, args[1])'
|
28
|
-
]
|
29
|
-
|
30
|
-
[libraries, sources, extra, code].join "\n"
|
31
|
-
end
|
32
|
-
|
33
|
-
def run_script(env, fname, task)
|
34
|
-
env.send :sh, "Rscript #{fname} '#{task.scope || 'public'}'"
|
35
|
-
end
|
36
|
-
end
|
37
|
-
|
38
|
-
creator :r, R
|
data/lib/lang/r/io.R
DELETED
@@ -1,113 +0,0 @@
|
|
1
|
-
suppressPackageStartupMessages(library(RPostgreSQL))
|
2
|
-
suppressPackageStartupMessages(library(ggplot2))
|
3
|
-
suppressPackageStartupMessages(library(gridExtra))
|
4
|
-
|
5
|
-
# input
|
6
|
-
# -----
|
7
|
-
file_input <- function(fileName, coltypes = NA) {
|
8
|
-
read.table(fileName, header = T, sep = ',', colClasses = coltypes)
|
9
|
-
}
|
10
|
-
|
11
|
-
join_file_input <- function(namesstr, key) {
|
12
|
-
files <- strsplit(namesstr, split = '\\+')[[1]]
|
13
|
-
dataList <- lapply(files, (function(f) fileInput(f)))
|
14
|
-
do.call(merge, c(dataList, by = key))
|
15
|
-
}
|
16
|
-
|
17
|
-
init_table_input <- function(conn_args, schema) {
|
18
|
-
function(name, cols = '*', where = 'true') {
|
19
|
-
colstr <- paste(cols, collapse = ',')
|
20
|
-
|
21
|
-
sql <- sprintf("SELECT %s FROM %s WHERE %s", colstr, name, where)
|
22
|
-
init_sql_input(conn_args, schema)(sql)
|
23
|
-
}
|
24
|
-
}
|
25
|
-
|
26
|
-
init_sql_input <- function(conn_args, schema) {
|
27
|
-
function(sql) {
|
28
|
-
conn <- do.call(dbConnect, c(dbDriver('PostgreSQL'), conn_args))
|
29
|
-
on.exit(dbDisconnect(conn))
|
30
|
-
dbGetQuery(conn, paste('SET search_path TO', paste(c(schema, 'public'), sep = ',')))
|
31
|
-
buffer <- dbSendQuery(conn, sql)
|
32
|
-
fetch(buffer, n=-1)
|
33
|
-
}
|
34
|
-
}
|
35
|
-
|
36
|
-
create_parents <- function(output) {
|
37
|
-
dir.create(dirname(output), showWarnings = FALSE, recursive = TRUE)
|
38
|
-
}
|
39
|
-
|
40
|
-
# Output
|
41
|
-
# ------
|
42
|
-
# TODO consider deprecating this, perhaps more specific ones like graph_output is still useful
|
43
|
-
auto_output <- function(report, ...) {
|
44
|
-
actual <- switch(class(report)[1],
|
45
|
-
gg = ggplot_output,
|
46
|
-
data.frame = csv_output,
|
47
|
-
print)
|
48
|
-
actual(report, ...)
|
49
|
-
}
|
50
|
-
|
51
|
-
rplot_output <- function(f, output, size = c(8, 8)) {
|
52
|
-
create_parents(output)
|
53
|
-
size <- ceiling(size / 2.54 * 300 / 96) # convert cm to inch
|
54
|
-
par(mar = c(0,0,0,0))
|
55
|
-
# use pdf since png can be very weird, e.g. radarplot
|
56
|
-
pdf(output, width = size[1], height = size[2], pointsize = 12)
|
57
|
-
f()
|
58
|
-
dev.off()
|
59
|
-
}
|
60
|
-
|
61
|
-
# use facet or latex subfigure, do not handle list anymore
|
62
|
-
ggplot_output <- function(report, output, size = c(15, 15), fontScale = 1, compact = F) {
|
63
|
-
create_parents(output)
|
64
|
-
if (compact) {
|
65
|
-
margin <- 1
|
66
|
-
} else {
|
67
|
-
margin <- 5
|
68
|
-
}
|
69
|
-
|
70
|
-
report <- report + theme_bw() +
|
71
|
-
theme(axis.title = element_text(size = 13 * fontScale),
|
72
|
-
axis.text = element_text(size = 11 * fontScale),
|
73
|
-
legend.title = element_text(size = 13 * fontScale),
|
74
|
-
legend.text = element_text(size = 11 * fontScale),
|
75
|
-
legend.key.size = unit(13 * fontScale, 'pt'),
|
76
|
-
legend.margin = unit(0, "cm"),
|
77
|
-
legend.background = element_rect(fill = alpha('white', 0)),
|
78
|
-
plot.margin = unit(c(margin, margin, 0, 0),"mm"))
|
79
|
-
|
80
|
-
width <- size[1]
|
81
|
-
height <- size[2]
|
82
|
-
|
83
|
-
ggsave(report, filename = output,
|
84
|
-
dpi = 300, units = 'cm', width = width, height = height, limitsize = FALSE)
|
85
|
-
}
|
86
|
-
|
87
|
-
csv_output <- function(report, output) {
|
88
|
-
create_parents(output)
|
89
|
-
write.table(format(report, digits = 4), file = toString(output), row.names = FALSE,
|
90
|
-
sep = ',', quote = FALSE)
|
91
|
-
}
|
92
|
-
|
93
|
-
txt_output <- function(report, output) {
|
94
|
-
create_parents(output)
|
95
|
-
write.table(format(report, digits = 4), file = toString(output), row.names = FALSE,
|
96
|
-
sep = ',', quote = FALSE)
|
97
|
-
}
|
98
|
-
|
99
|
-
init_table_output <- function(conn_args, schema) {
|
100
|
-
function(report, output, placeholder = NULL) {
|
101
|
-
conn <- do.call(dbConnect, c(dbDriver('PostgreSQL'), conn_args))
|
102
|
-
on.exit(dbDisconnect(conn))
|
103
|
-
dbGetQuery(conn, paste('SET search_path TO', paste(c(schema, 'public'), sep = ',')))
|
104
|
-
|
105
|
-
if (dbExistsTable(conn, output)) {
|
106
|
-
dbRemoveTable(conn, output)
|
107
|
-
}
|
108
|
-
dbWriteTable(conn, output, report, row.names = F)
|
109
|
-
if (!is.null(placeholder)) {
|
110
|
-
write.table(data.frame(), file = placeholder, col.names = FALSE)
|
111
|
-
}
|
112
|
-
}
|
113
|
-
}
|