gliss 0.1.2 → 0.1.3

data/VERSION

@@ -1 +1 @@
-0.1.2
+0.1.3

data/lib/gliss.rb

@@ -24,19 +24,25 @@ require 'set'
 module Gliss
   Gloss = Struct.new(:sha, :tag, :text)
   CommitMsg = Struct.new(:sha, :log)
-  GLOSS_RE=/^((?:[^\s]){3})(-?)(.*?)\1(.*)/
-  GLOSS_TAG=3
-  GLOSS_TEXT=4
-  GLOSS_STRIP=2
-  INDENT_RE=/^(\s+)(.*)$/
+  GLOSS_TAG_RE=/(([^\s])\3\3)(-?)(.*?)\2(.*)/
+  GLOSS_INDENTED_RE=/^(\s*)#{GLOSS_TAG_RE}/
+  GLOSS_MIDLINE_RE=/^(.*?)#{GLOSS_TAG_RE}/
+  GLOSS_RE=/^()#{GLOSS_TAG_RE}/
+  
+  GLOSS_TAG=5
+  GLOSS_TEXT=6
+  GLOSS_STRIP=4
+  GLOSS_BEFORE=1
+
+  BARE_INDENT_RE=/(\s+)(.*)/
   INDENT_AMOUNT=1
   INDENTED_TEXT=2
 
   attr_reader :filter
 
-  def self.glosses_between(repo, older, newer)
+  def self.glosses_between(repo, older, newer, allow_indented=false)
     commits_between(repo, older, newer).inject([]) do |acc, commit|
-      acc + glosses_in(commit.log, commit.sha)
+      acc + glosses_in(commit.log, commit.sha, allow_indented)
     end
   end
 
@@ -51,40 +57,79 @@ module Gliss
     end
   end
 
-  def self.glosses_in(message, sha=nil)
+  def self.glosses_in(message, sha=nil, allow_indented=false)
     result = []
     continuing = false
+    ws = ""
     indent_matcher = nil
 
     message.each_line do |line|
       line.chomp!
       match = nil
       if continuing
-        match = line.match((indent_matcher || INDENT_RE))
+        match = line.match((indent_matcher || /^(?:#{ws})#{BARE_INDENT_RE}$/))
         
         if match
-          indent_matcher ||= /^(#{match[INDENT_AMOUNT]})(.*)$/
+          indent_matcher ||= /^(?:#{ws})(#{match[INDENT_AMOUNT]})(.*)$/
           text = match[INDENTED_TEXT].strip
           result[-1].text << text
+          next
         else
           indent_matcher = nil
           continuing = false
         end
       end
-          
-      match = line.match(GLOSS_RE)
+      
+      match = begins_gloss(line, sha, allow_indented)
       if match
-        continuing = true
-        indent_matcher = nil
-        tag = match[GLOSS_TAG].strip
-        text = match[GLOSS_TEXT].strip
-        result << Gloss.new(sha, tag, [text])
+        continuing, indent_matcher, ws, new_gloss = match
+        result << new_gloss
       end
     end
 
     result.each {|g| g.text.reject! {|t| t == ''}; g.text = g.text.join(" ")}
   end
   
+  def self.begins_gloss(line, sha, allow_indented=false)
+    match = line.match(allow_indented ? GLOSS_INDENTED_RE : GLOSS_RE)
+    if match
+      ws = match[GLOSS_BEFORE]
+      tag = match[GLOSS_TAG].strip
+      text = match[GLOSS_TEXT].strip
+      return [true, nil, ws, Gloss.new(sha, tag, [text])]
+    end
+    nil
+  end
+  
+  def self.split_glosses(gloss, split_glosses=false)
+    if split_glosses
+      result_glosses = [gloss]
+      text = gloss.text
+      match = text.match(GLOSS_MIDLINE_RE)
+      while match
+        result_glosses[-1].text = match[GLOSS_BEFORE].strip if result_glosses[-1]
+        tag = match[GLOSS_TAG].strip
+        text = match[GLOSS_TEXT].strip
+        result_glosses << Gloss.new(gloss.sha, tag, text)
+        match = text.match(GLOSS_MIDLINE_RE)
+      end
+
+      if block_given?
+        result_glosses.each do |g|
+          yield g
+        end
+      else
+        result_glosses
+      end
+    else
+      if block_given?
+        yield gloss
+      else
+        return [gloss]
+      end
+    end
+  end
+
   class App
     def initialize(args=nil)
       @args = (args || ARGV.dup)
@@ -104,9 +149,11 @@ module Gliss
     
     private
     def output_glosses(repo)
-      Gliss::glosses_between(repo, @from, @to).each do |gloss|
-        if gloss.tag =~ @filter
-          @output_proc.call(gloss)
+      Gliss::glosses_between(repo, @from, @to, @allow_indented_glosses).each do |gloss|
+        Gliss::split_glosses(gloss, @split_glosses) do |gl|
+          if gl.tag =~ @filter
+            @output_proc.call(gl)
+          end
         end
       end 
     end
@@ -147,11 +194,24 @@ module Gliss
           @repo = repo
         end
         
-        opts.on("-f REGEX", "--filter REGEX", "Output only messages with tags matching REGEX", "(default is all tags)") do |filter|
+        opts.on("-f REGEX", "--filter REGEX", "Output only messages with tags matching", "REGEX (default is all tags)") do |filter|
           new_filter = Regexp.new(filter)
           @filter = @filter ? Regexp.union(@filter, new_filter) : new_filter
         end
         
+        opts.on("--split-glosses", "Attempt to find multiple glosses in a line.", "Note that this option may return","spurious glosses") do
+          @split_glosses = true
+        end
+        
+        opts.on("--allow-indented-glosses", "Find glosses that don't begin at","the beginning of a line") do
+          @allow_indented_glosses = true
+        end
+        
+        opts.on("--permissive", "Find as many malformed glosses as possible.","Implies --allow-indented-glosses and", "--split-glosses") do
+          @split_glosses = true
+          @allow_indented_glosses = true
+        end
+        
         opts.on("-w", "--whole-commit", "Output entire commit messages that contain glosses") do
           @seen_messages = Set.new
           @output_proc = Proc.new do |gloss|

data/spec/gliss_spec.rb

@@ -17,15 +17,104 @@ This line does not contain a gloss.
   
   This is a gloss in paragraph format.
   This is a good format for longer comments.
+===UGH=== this is a gloss that has a ===MALFORMED=== gloss attached
+
+===YUCK=== this is a gloss that has two ===MALFORMED=== glosses attached ===TO=== it
+
+===BLEAH=== this is a gloss that has two ===MALFORMED=== glosses attached 
+   ===ONE=== of which is on the second line
+
+Foo
+
+  ===INDENTED=== this should only be a gloss if indenting is allowed
+  ===INDENTED2=== this should not be part of the preceding gloss
+    but these lines should be appended on
+
+abcCRUFTabc this should not be a gloss
+END
+  
+  SECOND_TEST = <<-END
+This is a test commit.
+  ===FOO=== this is a bogus gloss
+
+===BAR=== this is not a bogus gloss, but ===MIDLINE=== this is
+  ===MIDLINE2=== and so is this
+
+
 END
   
   before(:each) do
     @glosses = Gliss.glosses_in(TEST_MESSAGE)
   end
   
+  it "correctly splits glosses when asked to" do
+    ugh = @glosses.select {|g| g.tag == "UGH"}
+    ugh.size.should == 1
+    splits = Gliss::split_glosses(ugh[0], true)
+    splits.size.should == 2
+    splits[0].tag.should == "UGH"
+    splits[0].text.should == "this is a gloss that has a"
+    splits[1].text.should == "gloss attached"
+    splits[1].tag.should == "MALFORMED"
+  end
+
+  it "correctly splits malformed glosses with more than one midline gloss when asked to" do
+    yuck = @glosses.select {|g| g.tag == "YUCK"}
+    yuck.size.should == 1
+    splits = Gliss::split_glosses(yuck[0], true)
+    splits.size.should == 3
+    splits.map{|g| g.tag}.should == %w{YUCK MALFORMED TO}
+    splits.map{|g| g.text}.should == ["this is a gloss that has two", "glosses attached", "it"]
+  end
+
+  it "correctly splits malformed glosses with more than one mid-multiline gloss gloss when asked to" do
+    bleah = @glosses.select {|g| g.tag == "BLEAH"}
+    bleah.size.should == 1
+    splits = Gliss::split_glosses(bleah[0], true)
+    splits.size.should == 3
+    splits.map{|g| g.tag}.should == %w{BLEAH MALFORMED ONE}
+    splits.map{|g| g.text}.should == ["this is a gloss that has two", "glosses attached", "of which is on the second line"]
+  end
+
+  it "correctly operates in permissive mode" do
+    @glosses = Gliss.glosses_in(SECOND_TEST, nil, true)
+    @glosses.size.should == 2
+    bar = @glosses.select {|g| g.tag == "BAR"}
+    bar.size.should == 1
+    splits = Gliss::split_glosses(bar[0], true)
+    splits.size.should == 3
+    splits.map{|g| g.tag}.should == %w{BAR MIDLINE MIDLINE2}
+    splits.map{|g| g.text}.should == ["this is not a bogus gloss, but", "this is", "and so is this"]
+  end
+  
   it "correctly identifies single-line glosses" do
     @glosses.select {|g| g.tag == "FOO" && g.text == "is an example gloss of type FOO"}.size.should == 1
   end
+
+  it "does not spuriously identify indented glosses" do
+    @glosses.select {|g| g.tag == "INDENTED"}.size.should == 0
+    @glosses.select {|g| g.tag == "INDENTED2"}.size.should == 0
+  end
+
+  it "correctly identifies indented glosses when told to" do
+    @glosses = Gliss.glosses_in(TEST_MESSAGE, nil, true)
+    @glosses.select {|g| g.tag == "INDENTED"}.size.should == 1
+  end
+  
+  it "correctly identifies consecutive indented glosses" do
+    @glosses = Gliss.glosses_in(TEST_MESSAGE, nil, true)
+    @glosses.select {|g| g.tag == "INDENTED"}.size.should == 1
+    @glosses.select {|g| g.tag == "INDENTED" && g.text == "this should only be a gloss if indenting is allowed"}.size.should == 1
+    @glosses.select {|g| g.tag == "INDENTED" && g.text =~ /INDENTED2/}.size.should == 0
+    @glosses.select {|g| g.tag == "INDENTED2"}.size.should == 1
+    @glosses.select {|g| g.tag == "INDENTED2" && g.text =~ /not be part of the preceding gloss/}.size.should == 1
+  end
+
+  it "correctly identifies indented multiline glosses when told to" do
+    @glosses = Gliss.glosses_in(TEST_MESSAGE, nil, true)
+    @glosses.select {|g| g.tag == "INDENTED2"}.size.should == 1
+    @glosses.select {|g| g.tag == "INDENTED2" && g.text =~ /but these lines should be appended on/}.size.should == 1
+  end
   
   it "correctly identifies multiple-line glosses" do
     @glosses.select {|g| g.tag == "BAR" && g.text == "is a gloss that spans several lines because each line has the same indent or at least that much"}.size.should == 1
@@ -46,7 +135,7 @@ END
   end
   
   it "finds multiple glosses in a message" do
-    @glosses.size.should == 5
+    @glosses.size.should == 8
   end
   
   it "does not create spurious glosses" do

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: gliss
 version: !ruby/object:Gem::Version 
-  hash: 31
+  hash: 29
   prerelease: false
   segments: 
   - 0
   - 1
-  - 2
-  version: 0.1.2
+  - 3
+  version: 0.1.3
 platform: ruby
 authors: 
 - Will Benton
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-06-28 00:00:00 -05:00
+date: 2011-07-20 00:00:00 -05:00
 default_executable: gliss
 dependencies: 
 - !ruby/object:Gem::Dependency