man_parser 0.1.0 → 0.1.1

data/README.markdown

@@ -2,7 +2,7 @@ Parse man source
 
 Install
 =======
-    sudo gem install grosser-man_parser -s http://gems.github.com/
+    sudo gem install man_parser -s http://gemcutter.org
 
 Usage
 =====

data/VERSION

@@ -1 +1 @@
-0.1.0
+0.1.1

data/lib/man_parser.rb

@@ -9,52 +9,67 @@ class ManParser
 
   def self.parse(cmd)
     sections = sections(source(cmd))
-    description, options = parse_description(sections['DESCRIPTION'])
-    options = options.map{|option| parse_option(option*' ') }
+    description, options = find_options(sections['OPTIONS']||sections['DESCRIPTION'])
+    description ||= sections['DESCRIPTION']
+    options = parse_options(options)
     {:description => description.map{|l|l.strip}.join(''), :options=>options, :sections=>sections}
   end
 
   private
 
+  # remove common prefix from options
+  def self.parse_options(options)
+    options = options.map{|option| parse_option(option*' ') }.reject{|o| o.empty?}
+    common = common_prefix(options.map{|o| o[:description]})
+    options.each{|o| o[:description] = o[:description].split(//)[common..-1].to_s}
+    options
+  end
+
   def self.root
     '/usr/share/man/man1'
   end
 
   def self.parse_option(option)
-    option = option.gsub(/\\f[IB](.*?)\\fR/,"\\1").gsub('\\','')
-
-    found =  if option =~ /^-(\w+), --([-\w]+)(=(\w+))?(.*)/
-      {:alias=>$1, :name=>$2, :argument=>$4, :description=>$5}
-    elsif option =~ /^--([-\w]+)(=(\w+))?(.*)/
-      {:name=>$1, :argument=>$3, :description=>$4}
-    elsif option =~ /^-([-\w]+)(.*)/
-      {:alias=>$1, :description=>$2}
-    end
-
+    option = without_markup(option)
+    found = option_parts(option)
     if not found
-      puts "#{option} <-> nil !"
-      return
+      puts "#{option} <-> nothing found !"
+      return {}
     end
 
-    found[:description] = found[:description].to_s.strip.sub(/\s*\.TP$/,'')
+    #remove ending .TP and any unnecessary whitespace
+    found[:description] = found[:description].to_s.strip.sub(/\s*\.TP$/,'').gsub(/\s{2,}/,' ')
     found.delete(:argument) unless found[:argument]
 
     found
   end
 
+  def self.common_prefix(texts)
+    shortest = texts.map{|t| t.size}.min || 0
+    shortest.downto(1) do |i|
+      common = texts.map{|t| t[0...i]}.uniq
+      next if common.size != 1
+      next if common.first =~ /^\w+$/ # ['\d hello world','\d hell is hot'] -> '\d '
+      return i
+    end
+    return 0
+  end
+
   # description can be split like "description, options, descriptions"
-  # so we remove the options part, and combind the 2 descriptions parts
-  def self.parse_description(text)
+  # so we remove the options part, and combine the 2 descriptions parts
+  def self.find_options(text)
     in_option = false
     already_switched = false
     options = []
+    known_names = []
     description = []
 
-    text.split("\n")[1..-1].each do |line|
+    text.split("\n").each do |line|
 
-      if start_of_option?(line) and not already_switched
+      if is_unknown_option?(line, known_names) and not already_switched
         in_option = true
         options << [] #new option
+        known_names << parse_option(line)[:name]
       elsif line =~ /^\.PP/ and in_option
         already_switched = true
         in_option = false
@@ -74,22 +89,48 @@ class ManParser
 
   # split into sections according to "SectionHead" aka .SH
   def self.sections(text)
+    lines = text.split("\n")+[".SH END"] 
     name = 'OUT_OF_SECTION'
-    sections = Hash.new([])
+    sections = {}
+    temp = []
 
-    text.split("\n").each do |line|
+    lines.each do |line|
       if line =~ /^\.SH (.*)$/
-        name = $1
+        sections[name] = temp * "\n"
+        temp = []
+        name = $1.gsub('"','').strip
       else
-        sections[name] += [line]
+        temp << line
       end
     end
 
-    sections.each{|k,v| sections[k] = v*"\n"}
     sections
   end
 
-  def self.start_of_option?(line)
-    !!( line =~ /^\\fB\\-/)
+  def self.without_markup(text)
+    text = text.gsub(/\\f[IB](.*?)\\f[RP]/,"\\1").gsub('\\','') # remove styles
+    text = text.gsub(/-\^-(\w)/,"--\\1") # remove weird ^ in e.g. grep options
+  end
+
+  # duplicate prevention, e.g. many lines in grep start with --mmap
+  def self.is_unknown_option?(line, known_names)
+    return if not is_option?(line)
+    name = parse_option(line)[:name]
+    name == nil or not known_names.include?(name)
+  end
+
+  def self.is_option?(line)
+    !! option_parts(line)
+  end
+
+  def self.option_parts(text)
+    text = without_markup(text).sub(/\.[A-Z]+ (")?/,'') # remove any initial markup
+    if text =~ /^-(\w+)[,"| ]+--(\w[-\w]*)(=(\w+)| <(\w+)>)?(.*)/
+      {:alias=>$1, :name=>$2, :argument=>$4||$5, :description=>$6}
+    elsif text =~ /^--(\w[-\w]*)(=(\w+))?(.*)/
+      {:name=>$1, :argument=>$3, :description=>$4}
+    elsif text =~ /^-(\w[-\w]*)(.*)/
+      {:alias=>$1, :description=>$2}
+    end
   end
 end

data/man_parser.gemspec

@@ -5,7 +5,7 @@
 
 Gem::Specification.new do |s|
   s.name = %q{man_parser}
-  s.version = "0.1.0"
+  s.version = "0.1.1"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Michael Grosser"]
@@ -20,6 +20,7 @@ Gem::Specification.new do |s|
      "VERSION",
      "lib/man_parser.rb",
      "man_parser.gemspec",
+     "spec/call",
      "spec/man_parser_spec.rb",
      "spec/spec_helper.rb"
   ]

data/spec/call

@@ -0,0 +1,4 @@
+#!/usr/bin/env ruby
+$LOAD_PATH << 'lib'
+require 'man_parser'
+print ManParser.send(ARGV[0], ARGV[1])

data/spec/man_parser_spec.rb

@@ -4,20 +4,20 @@ describe ManParser do
   describe :parse do
     it "finds the description" do
       d = ManParser.parse('ls')[:description]
-      d.should =~ /^\.PPList information about the FILEs \(the current/
+      d.should =~ /^\.\\\" Add any additional description here\.PPList infor/
       d.should =~ /problems, 2 if serious trouble\.$/
       d.should_not include('\\-\\-all')
       d.should_not include('do not ignore entries starting with')
       d.should_not include("      ")
     end
 
-    describe 'options' do
+    describe 'options in description' do
       def options
         ManParser.parse('ls')[:options]
       end
 
       it "finds all options" do
-        options.size.should == 58
+        options.size.should == 57
       end
 
       it "extracts the name" do
@@ -36,6 +36,34 @@ describe ManParser do
         options[2].should == {:name=>'author', :description=>'with -l, print the author of each file'}
       end
     end
+
+    describe 'options in OPTIONS section' do
+      def options
+        ManParser.parse('acpi')[:options]
+      end
+
+      it "finds all options x" do
+        options.size.should == 18
+      end
+
+      it "extracts the name" do
+        options.first[:name].should == 'battery'
+      end
+
+      it "extracts the alias" do
+        options.first[:alias].should == 'b'
+      end
+
+      it "extracts the description" do
+        options.first[:description].should == 'show battery information'
+      end
+    end
+
+    describe 'for grep' do
+      it "finds some options" do
+        ManParser.parse('grep')[:options].size.should == 70
+      end
+    end
   end
 
   describe :source do
@@ -80,15 +108,22 @@ describe ManParser do
       x.should == {:alias=>"T", :name => 'tabsize', :argument=>'COLS', :description=>"assume tab stops at each COLS instead of 8"}
     end
 
+    it "parses - and -- with <dir>" do
+      x = parse('.IP "\fB-d | --directory <dir>\fP " 10 bla bla')
+      x.should == {:alias=>'d', :name=>'directory', :argument=>'dir', :description=>'" 10 bla bla'}
+    end
+
     it "does not parse random stuff" do
       ManParser.stub!(:puts)
       x = parse('as we say: \fB\-T\fR, \fB\-\-tabsize\fR=\fICOLS\fR assume tab stops at each COLS instead of 8')
-      x.should == nil
+      x.should == {}
     end
   end
 
-  describe :start_of_option? do
+  describe :is_option? do
     {
+      '.BR \-P ", " \-\^\-perl\-regexp' => true,
+      '.IP "\fB-c | --cooling\fP " 10' => true,
       '\fB\-\-version\fR'=>true,
       '\fB\-1\fR'=>true,
       '\fB\-\-color\fR=\fIauto\fR'=>true,
@@ -98,7 +133,7 @@ describe ManParser do
       '   asdadas'=>false
     }.each do |line, success|
       it "recognises #{line} -- #{success}" do
-        ManParser.send(:start_of_option?, line).should == success
+        ManParser.send(:is_option?, line).should == success
       end
     end
   end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: man_parser
 version: !ruby/object:Gem::Version 
-  version: 0.1.0
+  version: 0.1.1
 platform: ruby
 authors: 
 - Michael Grosser
@@ -27,6 +27,7 @@ files:
 - VERSION
 - lib/man_parser.rb
 - man_parser.gemspec
+- spec/call
 - spec/man_parser_spec.rb
 - spec/spec_helper.rb
 has_rdoc: true