raka 0.3.1 → 0.3.4

data/bin/raka

@@ -17,8 +17,6 @@ def detect_main
   return rakas[0] if rakas.length == 1
 end
 
-entry = detect_main
-
 options = { rake: {}, raka_finished: false }
 def set_option(opts, key, value)
   if opts[:raka_finished]
@@ -28,17 +26,31 @@ def set_option(opts, key, value)
   end
 end
 parser = OptionParser.new do |opts|
-  opts.banner = 'Usage: raka [options] [raka file] -- [rake options]'
+  opts.banner = 'Usage: raka [options] <output> -- [rake options]'
 
   opts.on('-v', '--[no-]verbose', 'Run verbosely') do |v|
     set_option(options, :verbose, v)
   end
+
+  opts.on('-f', '--file FILE', String, 'Run even when up to date') do |s|
+    set_option(options, :file, s)
+  end
+
+  opts.on('-j', '--jobs JOBS', Integer, 'Run in parallel') do |n|
+    set_option(options, :jobs, n)
+  end
 end
 
+if ARGV.empty?
+  puts parser.help
+  exit(1)
+end
 both_args = ARGV.join(' ').split(' -- ')
-extra_args = both_args[1]
 self_args = both_args[0].split(/\s+/)
 parser.parse!(self_args)
+extra_args = (both_args[1] || ' ').lstrip
+
+entry = options[:file] || detect_main
 
 env = if options[:verbose]
         'LOG_LEVEL=0 '
@@ -46,13 +58,26 @@ env = if options[:verbose]
         ''
       end
 targets = self_args.join(' ')
-cmd = "#{env}rake -f #{entry}  #{extra_args} #{targets}"
+cmd = ''
+opt_str = "-f #{entry}"
+opt_str += " -m -j #{options[:jobs]}" if options.key?(:jobs)
+cmd += "#{env}rake #{opt_str} #{extra_args} #{targets}"
 puts cmd
-output, err, code = Open3.capture3(cmd)
-if code == 0
-  # TOOD: if empty, print all exist
-  puts output unless output.empty?
-else
-  puts 'Error: rake returns the following information:'
+output = []
+ro, out = IO.pipe
+re, err = IO.pipe
+pid = fork do
+  status = system(cmd, out: out, err: err)
+  puts 'Error: rake returns the following information:' unless status
+  exit($CHILD_STATUS.exitstatus)
 end
-puts err
+out.close
+err.close
+ro.each_line do |l|
+  puts l
+  output << l.chomp
+end
+re.each_line { |l| puts l; }
+
+Process.wait(pid)
+puts 'All targets are up to date' if output.empty? && $CHILD_STATUS.exitstatus == 0

data/lib/{compile.rb → raka/compile.rb}

@@ -28,8 +28,7 @@ class DSLCompiler
   # Raka task structure, input task is rake's task pushed into blocks
   def dsl_task(token, task)
     name = task.name
-    deps = task.prerequisites
-
+    deps = task.prerequisites 
     output_info = token._parse_output_ name
     task_info = {
       name: name,
@@ -41,6 +40,10 @@ class DSLCompiler
     OpenStruct.new(output_info.to_h.merge(task_info))
   end
 
+  def stem(path)
+    File.basename(path, File.extname(path))
+  end
+
   # resolve auto variables with only output info,
   # useful when resolve extra deps (task is not available yet)
   def resolve_by_output(target, output_info)
@@ -50,8 +53,10 @@ class DSLCompiler
       .gsub('$(scope)', info.scope.nil? ? '' : info.scope)
       .gsub('$(target_scope)', info.target_scope.nil? ? '' : info.target_scope)
       .gsub('$(output)', info.output)
-      .gsub('$(output_stem)', info.stem)
+      .gsub('$(output_stem)', stem(info.stem))
       .gsub('$(input_stem)', info.input_stem.nil? ? '' : info.input_stem)
+      .gsub('$(func)', info.func.nil? ? '' : info.func)
+      .gsub('$(ext)', info.ext)
       .gsub('$@', info.name)
 
     protect_percent_symbol text do |safe_text|
@@ -75,7 +80,8 @@ class DSLCompiler
 
     protect_percent_symbol text do |safe_text|
       # add numbered auto variables like $0, $2 referring to the first and third deps
-      safe_text.gsub(/\$\(dep(\d+)\)/, '%{\1}') % array_to_hash(task.deps)
+      safe_text = safe_text.gsub(/\$\(dep(\d+)\)/, '%{\1}') % array_to_hash(task.deps)
+      safe_text.gsub(/\$\(dep(\d+)_stem\)/, '%{\1}') % array_to_hash(task.deps.map {|d| stem(d)})
     end
   end
 

data/lib/raka/lang/r/impl.rb

@@ -0,0 +1,23 @@
+# frozen_string_literal: true
+
+require_relative '../../protocol'
+
+# r language protocol
+class R
+  def initialize(libs = [], **kwargs)
+    @libs = libs
+    super(**kwargs)
+  end
+
+  def build(code, _)
+    libraries = @libs.map { |name| "suppressPackageStartupMessages(library(#{name}))" }
+
+    [libraries, code].join "\n"
+  end
+
+  def run_script(env, fname, _task)
+    env.send :sh, "Rscript #{fname}"
+  end
+end
+
+creator :r, R

data/lib/raka.rb

@@ -2,13 +2,14 @@
 
 require 'logger'
 
-require_relative './compile'
-require_relative './protocol'
-require_relative './token'
+require_relative './raka/compile'
+require_relative './raka/protocol'
+require_relative './raka/token'
 
 # initialize raka
 class Raka
   Pattern = Pattern
+  P = Pattern
   attr_reader :logger
 
   def create_logger(level)
@@ -53,7 +54,7 @@ class Raka
     @options.input_types |= @options.output_types # any output can be used as intermediate
     # specify root of scopes in options, scopes will append to each root
     @scopes = options.scopes.empty? ? [] : [options.scopes]
-    @options.lang.each { |path| load File::join(File::dirname(__FILE__), "#{path}/impl.rb") }
+    @options.lang.each { |path| load File::join(File::dirname(__FILE__), "raka/#{path}/impl.rb") }
     @options.user_lang.each { |path| load path.to_s + '.rb' }
 
     # These are where the dsl starts
@@ -67,4 +68,8 @@ class Raka
     block.call
     @scopes.pop
   end
+
+  def stem(path)
+    File.basename(path, File.extname(path))
+  end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: raka
 version: !ruby/object:Gem::Version
-  version: 0.3.1
+  version: 0.3.4
 platform: ruby
 authors:
 - yarray
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2022-02-03 00:00:00.000000000 Z
+date: 2022-05-09 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake
@@ -111,7 +111,8 @@ dependencies:
 description: An extensible, concise and light weight DSL on Rake to automate data
   processing tasks
 email: '08to09@gmail.com'
-executables: []
+executables:
+- raka
 extensions: []
 extra_rdoc_files:
 - LICENSE
@@ -121,18 +122,16 @@ files:
 - README.md
 - VERSION
 - bin/raka
-- lib/compile.rb
-- lib/interface.rbs
-- lib/lang/psql/impl.rb
-- lib/lang/python/impl.rb
-- lib/lang/r/impl.rb
-- lib/lang/r/io.R
-- lib/lang/shell/impl.rb
-- lib/output_type.rb
-- lib/protocol.rb
 - lib/raka.rb
-- lib/temp.json
-- lib/token.rb
+- lib/raka/compile.rb
+- lib/raka/interface.rbs
+- lib/raka/lang/psql/impl.rb
+- lib/raka/lang/python/impl.rb
+- lib/raka/lang/r/impl.rb
+- lib/raka/lang/shell/impl.rb
+- lib/raka/output_type.rb
+- lib/raka/protocol.rb
+- lib/raka/token.rb
 homepage: http://github.com/yarray/raka
 licenses:
 - MIT

data/lib/lang/r/impl.rb

@@ -1,38 +0,0 @@
-# frozen_string_literal: true
-
-require_relative '../protocol'
-
-# r language protocol
-class R
-  def initialize(src, libs = [], **kwargs)
-    @src = src
-    @libs = libs
-    super(**kwargs)
-  end
-
-  def build(code, _)
-    libraries = ([
-      :pipeR
-    ] + @libs).map { |name| "suppressPackageStartupMessages(library(#{name}))" }
-
-    sources = ["source('#{File.dirname(__FILE__)}/io.R')"] +
-              (@src ? [@src] : []).map { |name| "source('#{SRC_DIR}/#{name}.R')" }
-
-    extra = [
-      '`|` <- `%>>%`',
-      "conn_args <- list(host='#{HOST}', user='#{USER}', dbname='#{DB}', port='#{PORT}')",
-      'args <- commandArgs(trailingOnly = T)',
-      'sql_input    <- init_sql_input(conn_args, args[1])',
-      'table_input  <- init_table_input(conn_args, args[1])',
-      'table_output <- init_table_output(conn_args, args[1])'
-    ]
-
-    [libraries, sources, extra, code].join "\n"
-  end
-
-  def run_script(env, fname, task)
-    env.send :sh, "Rscript #{fname} '#{task.scope || 'public'}'"
-  end
-end
-
-creator :r, R

data/lib/lang/r/io.R

@@ -1,113 +0,0 @@
-suppressPackageStartupMessages(library(RPostgreSQL))
-suppressPackageStartupMessages(library(ggplot2))
-suppressPackageStartupMessages(library(gridExtra))
-
-# input
-# -----
-file_input <- function(fileName, coltypes = NA) {
-    read.table(fileName, header = T, sep = ',', colClasses = coltypes)
-}
-
-join_file_input <- function(namesstr, key) {
-    files <- strsplit(namesstr, split = '\\+')[[1]]
-    dataList <- lapply(files, (function(f) fileInput(f)))
-    do.call(merge, c(dataList, by = key))
-}
-
-init_table_input <- function(conn_args, schema) {
-    function(name, cols = '*', where = 'true') {
-        colstr <- paste(cols, collapse = ',')
-
-        sql <- sprintf("SELECT %s FROM %s WHERE %s", colstr, name, where)
-        init_sql_input(conn_args, schema)(sql)
-    }
-}
-
-init_sql_input <- function(conn_args, schema) {
-    function(sql) {
-        conn <- do.call(dbConnect, c(dbDriver('PostgreSQL'), conn_args))
-        on.exit(dbDisconnect(conn))
-        dbGetQuery(conn, paste('SET search_path TO', paste(c(schema, 'public'), sep = ',')))
-        buffer <- dbSendQuery(conn, sql)
-        fetch(buffer, n=-1)
-    }
-}
-
-create_parents <- function(output) {
-    dir.create(dirname(output), showWarnings = FALSE, recursive = TRUE)
-}
-
-# Output
-# ------
-# TODO consider deprecating this, perhaps more specific ones like graph_output is still useful
-auto_output <- function(report, ...) {
-    actual <- switch(class(report)[1],
-                     gg = ggplot_output,
-                     data.frame = csv_output,
-                     print)
-    actual(report, ...)
-}
-
-rplot_output <- function(f, output, size = c(8, 8)) {
-    create_parents(output)
-    size <- ceiling(size / 2.54 * 300 / 96) # convert cm to inch
-    par(mar = c(0,0,0,0))
-    # use pdf since png can be very weird, e.g. radarplot
-    pdf(output, width = size[1], height = size[2], pointsize = 12)
-    f()
-    dev.off()
-}
-
-# use facet or latex subfigure, do not handle list anymore
-ggplot_output <- function(report, output, size = c(15, 15), fontScale = 1, compact = F) {
-    create_parents(output)
-    if (compact) {
-        margin <- 1
-    } else {
-        margin <- 5
-    }
-
-    report <- report + theme_bw() +
-        theme(axis.title = element_text(size = 13 * fontScale),
-              axis.text = element_text(size = 11 * fontScale),
-              legend.title = element_text(size = 13 * fontScale),
-              legend.text = element_text(size = 11 * fontScale),
-              legend.key.size = unit(13 * fontScale, 'pt'),
-              legend.margin = unit(0, "cm"),
-              legend.background = element_rect(fill = alpha('white', 0)),
-              plot.margin = unit(c(margin, margin, 0, 0),"mm"))
-
-    width <- size[1]
-    height <- size[2]
-
-    ggsave(report, filename = output,
-           dpi = 300, units = 'cm', width = width, height = height, limitsize = FALSE)
-}
-
-csv_output <- function(report, output) {
-    create_parents(output)
-    write.table(format(report, digits = 4), file = toString(output), row.names = FALSE,
-                sep = ',', quote = FALSE)
-}
-
-txt_output <- function(report, output) {
-    create_parents(output)
-    write.table(format(report, digits = 4), file = toString(output), row.names = FALSE,
-                sep = ',', quote = FALSE)
-}
-
-init_table_output <- function(conn_args, schema) {
-    function(report, output, placeholder = NULL) {
-        conn <- do.call(dbConnect, c(dbDriver('PostgreSQL'), conn_args))
-        on.exit(dbDisconnect(conn))
-        dbGetQuery(conn, paste('SET search_path TO', paste(c(schema, 'public'), sep = ',')))
-
-        if (dbExistsTable(conn, output)) {
-            dbRemoveTable(conn, output)
-        }
-        dbWriteTable(conn, output, report, row.names = F)
-        if (!is.null(placeholder)) {
-            write.table(data.frame(), file = placeholder, col.names = FALSE)
-        }
-    }
-}