removed the crippling cocaine addiction from regexp pattern matching

2023-04-19 11:30:24 +04:00 · 2023-04-19 11:30:24 +04:00 · 3b8632f8a7
parent 6dcf46f40a
commit 3b8632f8a7
8 changed files with 110 additions and 14 deletions
--- a/hyde.rb
+++ b/hyde.rb
@ -2,6 +2,12 @@ require 'mime-types'
 require 'webrick'

 module Hyde
+  # Branding and version
+  VERSION = "0.1"
+  attr_reader :VERSION
+  VLINE = "<ADDRESS>\n      Hyde/#{Hyde::VERSION} on WEBrick/#{WEBrick::VERSION} (Ruby/#{RUBY_VERSION}/#{RUBY_RELEASE_DATE})\n    </ADDRESS>"
+  attr_reader :VLINE
+  
  class Server < WEBrick::HTTPServer
    def initialize(config={},&setup)
      super(config)
@ -11,6 +17,7 @@ module Hyde
      end
    end
  end
+
  module ErrorPages
    # 404 text
    def error404(request, filepath)
@ -20,17 +27,18 @@ module Hyde
          filepath
        ) 
      else
-        request.response.body = "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.0//EN\">\n<HTML>\n  <HEAD><TITLE>File not found</TITLE></HEAD>\n  <BODY>\n    <H1>File not found</H1>\n    #{filepath}\n    <HR>\n    <ADDRESS>\n      Hyde on WEBrick/#{WEBrick::VERSION} (Ruby/#{RUBY_VERSION}/#{RUBY_RELEASE_DATE})\n    </ADDRESS>\n  </BODY>\n</HTML>"      
+        request.response.body = "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.0//EN\">\n<HTML>\n  <HEAD><TITLE>File not found</TITLE></HEAD>\n  <BODY>\n    <H1>File not found</H1>\n    #{filepath}\n    <HR>\n    #{Hyde::VLINE}\n  </BODY>\n</HTML>"      
      end
      request.response["Content-Type"] = 'text/html'
    end
    module_function :error404
  end
-  # Interchangeable pattern matching
+  
+  # Interchangeable glob/regex/string pattern matching
  module PatternMatching
-    def prep_path(path,safe_regex: true)
+    def prep_path(path,safe_regexp: true)
      raise Exception, "Not permitted" if @lock_methods
-      @safe_regex = safe_regex
+      @safe_regexp = safe_regexp
      @path = normalize(path) if path.kind_of? String
      @path = path if path.kind_of? Regexp
    end
@ -40,9 +48,18 @@ module Hyde
      return true if @path == ""
      split_path = path.split("/").filter { |x| x != "" }
      if @path.kind_of? Regexp then
-        # "safe" as in "will not result in path matching anomalies"
-        return @path.match normalize_input(path) unless @safe_regex
-        return Regexp.new("^"+@path.to_s+"$").match split_path[0]
+        # this chunk of fuck is needed to force regexp into 3 rules:
+        # 1) unsafe regexp means match the whole (remaining) line.
+        # 3) safe regexp means match only the part before the next slash
+        # 2) a .match? only returns when there are no leftovers
+        # this forces the matching to work somewhat consistently
+        test = @path.match normalize_input(path) unless @safe_regexp
+        test = @path.match split_path[0] if @safe_regexp
+        if test and (test.pre_match == "") and (test.post_match == "") then
+          return true
+        else
+          return false
+        end
      else
        # algorithm to match path segments until no more left in @path
        @path.split("/").filter { |x| x != "" }
@ -91,12 +108,14 @@ module Hyde
      @request = request
      @response = response
      @handles = {}
+      @indexlist = []
    end
    attr_reader :request
    attr_reader :response
    attr_accessor :filepath
    attr_accessor :path
    attr_accessor :handles
+    attr_accessor :indexlist
  end

  # Context object with safe path encapsulation
@ -111,14 +130,15 @@ module Hyde
    undef :path=
    undef :filepath=
    undef :handles=
+    undef :indexlist=
  end

  # Handler classes
  class Probe
    include Hyde::PatternMatching
    include Hyde::PublicContextControlMethods
-    def initialize (path, safe_regex: true, &block_optional) 
-      prep_path path, safe_regex: safe_regex
+    def initialize (path, safe_regexp: true, &block_optional) 
+      prep_path path, safe_regexp: safe_regexp
      @block = block_optional 
    end
    def match(request)
@ -145,7 +165,7 @@ module Hyde
          request.response.body = data
          request.response["Content-Type"] = mimetype
        rescue Errno::ENOENT
-          Hyde::ErrorPages::error404 filepath
+          Hyde::ErrorPages::error404 request, request.request.path
        end
      end
    end
@ -181,8 +201,8 @@ module Hyde
  class Pathspec
    include Hyde::PatternMatching
    include Hyde::Handlers
-    def initialize (path, root_path: nil, safe_regex: true, &block)
-      prep_path path, safe_regex: safe_regex
+    def initialize (path, root_path: nil, safe_regexp: true, &block)
+      prep_path path, safe_regexp: safe_regexp
      @chain = []
      @handles = {}
      @root_override = root_path
@ -211,6 +231,9 @@ module Hyde
      @lock_methods = true
      @remap = true
    end
+    def index(list)
+      @indexlist = list if list.kind_of? Array
+    end
    def handle(code, &block)
      @handles[code] = block
    end
@ -228,13 +251,25 @@ module Hyde
        else
          request.filepath = request.filepath+next_path+"/"
        end
-        # error handle overlaying
+        # parameter overrides overlaying
        @handles.each_pair { |k,v| request.handles[k] = v }
+        request.indexlist = @indexlist if @indexlist
+        # do directory indexing
+        if cut_path.match /^\/?$/ then
+          request.indexlist.each do |x| 
+            try_index = @chain.find { |y| y.match? x }
+            if try_index then
+              request.path = x
+              return try_index.match request
+            end
+          end
+        end
        # passthrough to the next path object
        next_pathspec = @chain.find { |x| x.match? cut_path }
+        puts next_pathspec.inspect
        next_pathspec.match request if next_pathspec
-        # throw 404 if nowhere to go
        unless next_pathspec then
+          # throw 404 if nowhere to go
          Hyde::ErrorPages::error404 request, request.request.path
        end
      end
--- a/index.html
+++ b/index.html
@ -0,0 +1,10 @@
+<!DOCTYPE HTML>
+<html> 
+    <head>
+        <title> Unforseen consequences </title>
+    </head>
+    <body>
+        <p> Cleverly done, mister Freeman, but you are not supposed to be here </p>
+        <p> <b> Get back where you <a href="/index.html">belong</a> </b> </p>
+    </body>
+</html>
--- a/test/uploads/01-blog/index.html
+++ b/test/uploads/01-blog/index.html
@ -0,0 +1,11 @@
+<!DOCTYPE html>
+<html>
+    <head>
+        <title> test </title>
+    </head>
+    <body>
+        <h1> This is a test </h1>
+        <hr>
+        <address> yes </address>
+    </body>
+</html>
--- a/test/uploads/01-blog/test.md
+++ b/test/uploads/01-blog/test.md
@ -0,0 +1,7 @@
+# Welcome to the INFINITE HYPERNET
+
+---
+- THE HYPE IS REAL
+- SCENE IS DEAD
+- BLOOD OF LAMERS IS FUEL
+
--- a/test/uploads/02-rules/index.html
+++ b/test/uploads/02-rules/index.html
@ -0,0 +1,11 @@
+<!DOCTYPE html>
+<html>
+    <head>
+        <title> very test </title>
+    </head>
+    <body>
+        <h1> This is a very test </h1>
+        <hr>
+        <address> yes 2 </address>
+    </body>
+</html>
--- a/test/uploads/02-rules/megafuck.md
+++ b/test/uploads/02-rules/megafuck.md
@ -0,0 +1 @@
+# YES
--- a/test/uploads/index.html
+++ b/test/uploads/index.html
@ -0,0 +1,15 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0//EN">
+<HTML>
+  <HEAD><TITLE>File listing</TITLE></HEAD>
+  <BODY>
+    <H1>File listing:</H1>
+        <ul>
+            <li><a href="/uploads/01-blog/test.md">this</a></li>
+            <li><a href="/uploads/02-rules/megafuck.md">that</a></li>
+        </ul>
+    <HR>
+    <ADDRESS>
+        welcum to this crap
+    </ADDRESS>
+  </BODY>
+</HTML>
--- a/test_combined.rb
+++ b/test_combined.rb
@ -3,6 +3,7 @@ require_relative 'hyde'
 server = Hyde::Server.new Port: 8000 do
  root "/home/yessiest/hyde/test/"
  serve "index.html"
+  index ["index.html"]
  path "about" do
    serve "webrick" do |ctx|
      ctx.response.body = "WEBrick is a modular http server stack"
@ -13,6 +14,11 @@ server = Hyde::Server.new Port: 8000 do
      ctx.response['Content-Type'] = "text/plain"
    end
  end
+  path "uploads" do
+    index ["index.html"]
+    serve "**/*.md", safe_regex: false
+    serve ["*.html","**/*.html"], safe_regex: false
+  end
 end

 server.start