From 6f46531c5985bde0f368f59d8532929c1aa9fdc2 Mon Sep 17 00:00:00 2001
From: John Bintz <john@coswellproductions.com>
Date: Wed, 15 Oct 2008 22:35:16 -0400
Subject: [PATCH] a bunch of cleanup, reworking xml parsing and ui changes

---
 build/javascript.hxml             |   3 +-
 build/tests.hxml                  |   3 +-
 htdocs/index.html                 |   3 +-
 htdocs/style.css                  |  28 +++++++-
 src/CodeParser.hx                 |  27 +++-----
 src/ConstantTokenProcessor.hx     |   2 -
 src/FunctionTokenProcessor.hx     |  54 +++++++++++++--
 src/JavaScriptTarget.hx           |  20 ++++--
 src/MyTests.hx                    |   1 +
 src/RegenerateDataFiles.hx        |  16 +----
 src/TestCodeParser.hx             |   4 +-
 src/TestFunctionTokenProcessor.hx |  18 -----
 src/TestTokenProcessor.hx         |  35 ++++++++++
 src/TokenProcessor.hx             | 108 +++++++++++++++++++++++-------
 14 files changed, 227 insertions(+), 95 deletions(-)
 create mode 100644 src/TestTokenProcessor.hx

diff --git a/build/javascript.hxml b/build/javascript.hxml
index 9881998..965984a 100755
--- a/build/javascript.hxml
+++ b/build/javascript.hxml
@@ -1,5 +1,4 @@
 -js ../htdocs/harmoniouscode.js
 -main JavaScriptTarget
--resource ../data/functions_tokens_cache.hxd
--resource ../data/constant_tokens_cache.hxd
+-resource ../data/all_tokens.hxd
 -cp ../src
diff --git a/build/tests.hxml b/build/tests.hxml
index f9b7c18..49aa12f 100644
--- a/build/tests.hxml
+++ b/build/tests.hxml
@@ -1,5 +1,4 @@
 -main MyTests
 -neko ../neko/my_tests.n
--resource ../data/functions_tokens_cache.hxd
--resource ../data/constant_tokens_cache.hxd
+-resource ../data/all_tokens.hxd
 -cp ../src
diff --git a/htdocs/index.html b/htdocs/index.html
index 089ace3..9466adf 100644
--- a/htdocs/index.html
+++ b/htdocs/index.html
@@ -12,10 +12,11 @@
     <div id="form" style="display: none">
       <div id="form-holder">
         <form action="" method="post" onsubmit="return false;">
-          <textarea name="source" id="source" rows="15" cols="80"></textarea><br />
+          <textarea name="source" id="source"></textarea><br />
           <input id="analyze-code-button" type="button" value="Analyze Code" onclick="JavaScriptTarget.do_analysis(this.form.elements.source)" />
         </form>
       </div>
+      <div id="processing"></div>
       <div id="output"></div>
       <div id="permanent-ignore" style="display: none">
         To permanently ignore tokens (and globally ignore modules), do one of the following:
diff --git a/htdocs/style.css b/htdocs/style.css
index fea3b5e..3e77233 100644
--- a/htdocs/style.css
+++ b/htdocs/style.css
@@ -7,13 +7,18 @@ body, td, div, li, p, span {
   font-size: 12px;
 }
 
+h1 {
+  text-align: center
+}
+
 div#form-holder {
   text-align: center
 }
 
 input#analyze-code-button {
-  width: 200px;
+  width: 750px;
   padding: 5px;
+  margin-top: 5px;
 }
 
 div#loading {
@@ -63,13 +68,32 @@ th.is-filtering {
   color: blue;
 }
 
+textarea#source {
+  width: 750px;
+  height: 250px;
+}
+
 div#footer {
   text-align: center;
   font-size: 10px;
   color: #555;
-  font-style: oblique
+  font-style: oblique;
+  border-top: solid #555 1px;
+  padding-top: 5px;
+  margin-top: 5px
 }
 
 span.ignore-code-holder {
   font-family: monospace
+}
+
+div#processing, div#code-announcement {
+  text-align: center;
+  font-size: 14px;
+  font-weight: bold
+}
+
+div#code-announcement {
+  border-bottom: solid #555 1px;
+  margin-bottom: 5px
 }
\ No newline at end of file
diff --git a/src/CodeParser.hx b/src/CodeParser.hx
index c623e65..ead6e41 100644
--- a/src/CodeParser.hx
+++ b/src/CodeParser.hx
@@ -20,16 +20,9 @@ class CodeParser {
     /**
       Load all possible token processors from disk.
     **/
-    public function load_processors_from_disk() {
-      for (processor_type_name in processor_types) {
-        var processor : TokenProcessor = Type.createInstance(Type.resolveClass(processor_type_name), []);
-
-        if (!processor.load_from_cache()) {
-          processor.populate_from_file();
-          processor.save_to_cache();
-        }
-
-        this.token_processors.set(processor_type_name, processor);
+    public function load_all_processors_from_disk() {
+      for (processor in TokenProcessor.load_all_from_cache()) {
+        this.token_processors.set(Type.getClassName(Type.getClass(processor)), processor);
       }
     }
   #end
@@ -38,12 +31,8 @@ class CodeParser {
     Load all possible token processors form haXe Resources.
   **/
   public function load_processors_from_resources() {
-    for (processor_type_name in processor_types) {
-      var processor : TokenProcessor = Type.createInstance(Type.resolveClass(processor_type_name), []);
-
-      processor.load_from_resource();
-
-      this.token_processors.set(processor_type_name, processor);
+    for (processor in TokenProcessor.load_all_from_resource()) {
+      this.token_processors.set(Type.getClassName(Type.getClass(processor)), processor);
     }
   }
 
@@ -118,8 +107,10 @@ class CodeParser {
           if (!tokens_found.exists(token)) {
             if (!flattened_tokens.exists(token)) {
               for (token_processor in this.token_processors.iterator()) {
-                if (token_processor.tokenHash.exists(token)) {
-                  results.push(token_processor.tokenHash.get(token).toResult()); break;
+                if ((token_processor.get_default_token_type() == FunctionToken) == is_function) {
+                  if (token_processor.tokenHash.exists(token)) {
+                    results.push(token_processor.tokenHash.get(token).toResult()); break;
+                  }
                 }
               }
               tokens_found.set(token, true);
diff --git a/src/ConstantTokenProcessor.hx b/src/ConstantTokenProcessor.hx
index 09d6a69..9d3fba3 100644
--- a/src/ConstantTokenProcessor.hx
+++ b/src/ConstantTokenProcessor.hx
@@ -1,6 +1,4 @@
 class ConstantTokenProcessor extends TokenProcessor {
-  public static var cachePath : String = "../data/constant_tokens_cache.hxd";
-  override public function get_cache_path() { return ConstantTokenProcessor.cachePath; }
   override public function get_default_token_type() { return ConstantToken; }
 
   public static var source_path : String = "../data";
diff --git a/src/FunctionTokenProcessor.hx b/src/FunctionTokenProcessor.hx
index 0bd0eba..b29d00c 100644
--- a/src/FunctionTokenProcessor.hx
+++ b/src/FunctionTokenProcessor.hx
@@ -1,17 +1,25 @@
 class FunctionTokenProcessor extends TokenProcessor {
-  public static var cachePath : String = "../data/functions_tokens_cache.hxd";
-  override public function get_cache_path() { return FunctionTokenProcessor.cachePath; }
-  public static var sourcePath : String = "../data/phpdoc_function_versions.xml";
+  public static var source_path : String = "../data/phpdoc_function_versions.xml";
   override public function get_default_token_type() { return FunctionToken; }
 
   #if neko
     public override function populate_from_file() {
-      this.populate_from_string(neko.io.File.getContent(sourcePath));
+      this.populate_from_string(neko.io.File.getContent(source_path));
     }
 
     public function populate_from_string(s : String) {
       this.tokenHash = new Hash<Token>();
-      for (child in Xml.parse(s).firstElement()) {
+      var tokens_parsed = 0;
+
+      //
+      // haXe XML parsing is slow, as it uses a custom XML parser.
+      // so I'll use a custom XML parser for this data.
+      //
+      /*var start = Date.now();
+      var first_element = Xml.parse(s).firstElement();
+      var end = Date.now();
+      trace(end.getTime() - start.getTime());
+      for (child in first_element) {
         if (child.nodeType == Xml.Element) {
           if (child.nodeName == "function") {
             var version = child.get("from");
@@ -19,9 +27,45 @@ class FunctionTokenProcessor extends TokenProcessor {
             version = ~/\:/.replace(version, " ");
             var token = child.get("name");
             this.tokenHash.set(token, new FunctionToken(child.get("name"), version));
+            tokens_parsed++;
           }
         }
+      }*/
+
+      var s_length = s.length;
+      var i = 0;
+
+      var version_regexp = ~/from=\'([^\']*)\'/i;
+      var token_regexp   = ~/name=\'([^\']*)\'/i;
+
+      while (i < s_length) {
+        var new_i = s.indexOf("<function", i);
+        if (new_i != -1) {
+          var tag_end = s.indexOf(">", new_i);
+          if (tag_end != -1) {
+            var tag = s.substr(new_i, tag_end - new_i + 1);
+
+            if (version_regexp.match(tag) && token_regexp.match(tag)) {
+              var version = version_regexp.matched(1);
+              var token   = token_regexp.matched(1);
+              version = ~/PECL /.replace(version, "");
+              version = ~/\:/.replace(version, " ");
+
+              this.tokenHash.set(token, new FunctionToken(token, version));
+              tokens_parsed++;
+              i = tag_end;
+            } else {
+              i++;
+            }
+          } else {
+            i++;
+          }
+        } else {
+          break;
+        }
       }
+      
+      trace("tokens parsed: " + tokens_parsed);
     }
   #end
 }
\ No newline at end of file
diff --git a/src/JavaScriptTarget.hx b/src/JavaScriptTarget.hx
index 157170e..6c2b793 100644
--- a/src/JavaScriptTarget.hx
+++ b/src/JavaScriptTarget.hx
@@ -91,14 +91,12 @@ class JavaScriptTarget {
     static public function display_version_information() {
       var version_info = new CodeVersionInformation(current_results, ignored_modules);
 
-      var output = "Your code in requires the following minimum PHP & PECL module versions:";
+      var output = "<div id=\"code-announcement\">Your code requires the following minimum PHP & PECL module versions:</div>";
 
       var minimum = version_info.final_versions.get("minimum");
 
       output += "<form action=\"\" onsubmit=\"return false\">";
 
-      output += "<ul>";
-
       var all_modules_hash = new Hash<Bool>();
 
       for (module in minimum.keys()) { all_modules_hash.set(module, true); }
@@ -291,8 +289,20 @@ class JavaScriptTarget {
     static public function do_analysis(textarea) {
       show_only_modules = new Hash<Bool>();
 
-      JavaScriptTarget.get_results(textarea.value);
-      JavaScriptTarget.display_version_information();
+      js.Lib.document.getElementById('processing').innerHTML = "Analyzing code...";
+      untyped {
+        js.Lib.document.getElementById('analyze-code-button').disabled = true;
+      }
+
+      haxe.Timer.delay(function() {
+        JavaScriptTarget.get_results(textarea.value);
+        JavaScriptTarget.display_version_information();
+
+        js.Lib.document.getElementById('processing').innerHTML = "";
+        untyped {
+          js.Lib.document.getElementById('analyze-code-button').disabled = false;
+        }
+      }, 100);
     }
 
     static public function toggle_module_and_redraw(module : String) {
diff --git a/src/MyTests.hx b/src/MyTests.hx
index 16ab0c4..0545254 100644
--- a/src/MyTests.hx
+++ b/src/MyTests.hx
@@ -2,6 +2,7 @@ class MyTests {
   static function main() {
     var r = new haxe.unit.TestRunner();
     r.add(new TestToken());
+    r.add(new TestTokenProcessor());
     r.add(new TestFunctionToken());
     r.add(new TestFunctionTokenProcessor());
     r.add(new TestConstantToken());
diff --git a/src/RegenerateDataFiles.hx b/src/RegenerateDataFiles.hx
index 843482a..53d60c0 100644
--- a/src/RegenerateDataFiles.hx
+++ b/src/RegenerateDataFiles.hx
@@ -1,17 +1,3 @@
 class RegenerateDataFiles {
-  public static function main() {
-    var functionProcessor = new FunctionTokenProcessor();
-    if (!functionProcessor.load_from_cache()) {
-      neko.Lib.print("Regenerating functions cache...\n");
-      functionProcessor.populate_from_file();
-      functionProcessor.save_to_cache();
-    }
-
-    var constantProcessor = new ConstantTokenProcessor();
-    if (!constantProcessor.load_from_cache()) {
-      neko.Lib.print("Regenerating constants cache...\n");
-      constantProcessor.populate_from_file();
-      constantProcessor.save_to_cache();
-    }
-  }
+  public static function main() { TokenProcessor.save_all_to_cache(); }
 }
\ No newline at end of file
diff --git a/src/TestCodeParser.hx b/src/TestCodeParser.hx
index 3d9c768..d96bca4 100644
--- a/src/TestCodeParser.hx
+++ b/src/TestCodeParser.hx
@@ -14,14 +14,14 @@ class TestCodeParser extends haxe.unit.TestCase {
   #if neko
     function testCodeParserLoadTokens() {
       var p = new CodeParser();
-      p.load_processors_from_disk();
+      p.load_all_processors_from_disk();
       assertTrue(p.token_processors.exists("FunctionTokenProcessor"));
       assertTrue(p.token_processors.exists("ConstantTokenProcessor"));
     }
 
     function testProcessCode() {
       var p = new CodeParser();
-      p.load_processors_from_disk();
+      p.load_all_processors_from_disk();
 
       for (code in test_code) {
         var result = p.parse(code[0]);
diff --git a/src/TestFunctionTokenProcessor.hx b/src/TestFunctionTokenProcessor.hx
index 7bb3913..7a213db 100644
--- a/src/TestFunctionTokenProcessor.hx
+++ b/src/TestFunctionTokenProcessor.hx
@@ -12,22 +12,4 @@ class TestFunctionTokenProcessor extends haxe.unit.TestCase {
   public function testGenerateSampleToken() {
     assertTrue(token_processor.tokenHash.exists(function_name));
   }
-
-  public function testSerializeInfo() {
-    var test_xml = "<versions> <function name='one' from='PHP 4, PHP 5' /> <function name='two' from='PHP 4, PHP 5' /> </versions>";
-    token_processor.populate_from_string(test_xml);
-
-    var target_token_hash = "{one => { version => PHP 4, PHP 5, token => one }, two => { version => PHP 4, PHP 5, token => two }}";
-
-    assertEquals(target_token_hash, token_processor.tokenHash.toString());
-
-    var unwound_tokens = token_processor.unwind_tokens();
-
-    assertTrue(unwound_tokens.toString().length < target_token_hash.length);
-
-    token_processor = new FunctionTokenProcessor();
-    token_processor.populate_from_unwound_tokens(unwound_tokens);
-
-    assertEquals(target_token_hash, token_processor.tokenHash.toString());
-  }
 }
\ No newline at end of file
diff --git a/src/TestTokenProcessor.hx b/src/TestTokenProcessor.hx
new file mode 100644
index 0000000..ea200c6
--- /dev/null
+++ b/src/TestTokenProcessor.hx
@@ -0,0 +1,35 @@
+class TestTokenProcessor extends haxe.unit.TestCase {
+  function testSerializeMultipleProcessors() {
+    var token_processor_one = new TokenProcessor();
+    token_processor_one.tokenHash.set("one", new Token("one", "version one"));
+    token_processor_one.tokenHash.set("two", new Token("two", "version one"));
+    token_processor_one.tokenHash.set("three", new Token("three", "version two"));
+
+    var token_processor_two = new TokenProcessor();
+    token_processor_two.tokenHash.set("four", new Token("four", "version one"));
+    token_processor_two.tokenHash.set("five", new Token("five", "version two"));
+    token_processor_two.tokenHash.set("six", new Token("six", "version three"));
+
+    var normalized_data = TokenProcessor.normalize_processors([token_processor_one, token_processor_two]);
+
+    assertTrue(normalized_data.exists("types"));
+    assertEquals("{0 => TokenProcessor, 1 => TokenProcessor}", normalized_data.get("types").toString());
+
+    assertTrue(normalized_data.exists("versions"));
+    assertEquals("{version one => 0, version two => 1, version three => 2}".length, normalized_data.get("versions").toString().length);
+
+    assertTrue(normalized_data.exists("processor-0"));
+    assertTrue(normalized_data.exists("processor-1"));
+
+    var trap_invalid = true;
+    try {
+      TokenProcessor.unnormalize_processors(new Hash<Hash<Dynamic>>());
+      trap_invalid = false;
+    } catch (e : Dynamic) {}
+    assertTrue(trap_invalid);
+
+    var unnormalized_processors = TokenProcessor.unnormalize_processors(normalized_data);
+
+    assertTrue(unnormalized_processors.length == 2);
+  }
+}
\ No newline at end of file
diff --git a/src/TokenProcessor.hx b/src/TokenProcessor.hx
index ec69202..e5baba4 100644
--- a/src/TokenProcessor.hx
+++ b/src/TokenProcessor.hx
@@ -1,47 +1,109 @@
+import FunctionTokenProcessor;
+import ConstantTokenProcessor;
+
 class TokenProcessor {
   public var tokenHash : Hash<Token>;
-  public static var cachePath : String = null;
+  public static var cache_path : String = "../data/all_tokens.hxd";
 
   public function new() { this.tokenHash = new Hash<Token>(); }
-  public function get_cache_path() { return TokenProcessor.cachePath; }
   public function get_default_token_type() { return Token; }
 
+  public static var all_token_processors = [ "FunctionTokenProcessor", "ConstantTokenProcessor" ];
+
   #if neko
-    public function load_from_cache() : Bool {
-      if (neko.FileSystem.exists(this.get_cache_path())) {
-        this.populate_from_unwound_tokens(haxe.Unserializer.run(neko.io.File.getContent(this.get_cache_path())));
-        return true;
+    public static function load_all_from_cache() : Array<TokenProcessor> {
+      if (neko.FileSystem.exists(cache_path)) {
+        return unnormalize_processors(haxe.Unserializer.run(neko.io.File.getContent(cache_path)));
       } else {
-        return false;
+        return null;
       }
     }
 
-    public function save_to_cache() {
-      var fh = neko.io.File.write(this.get_cache_path(), true);
-      fh.writeString(haxe.Serializer.run(this.unwind_tokens()));
-      fh.close();
+    public static function save_all_to_cache() {
+      if (!neko.FileSystem.exists(cache_path)) {
+        var all_processors = new Array<TokenProcessor>();
+        for (processor_class in all_token_processors) {
+          var processor : TokenProcessor = Type.createInstance(Type.resolveClass(processor_class), []);
+          processor.populate_from_file();
+          all_processors.push(processor);
+        }
+
+        var fh = neko.io.File.write(cache_path, true);
+        fh.writeString(haxe.Serializer.run(normalize_processors(all_processors)));
+        fh.close();
+      }
     }
 
     public function populate_from_file() {}
   #end
 
-  public function load_from_resource() {
-    this.populate_from_unwound_tokens(haxe.Unserializer.run(haxe.Resource.getString(this.get_cache_path())));
+  public static function load_all_from_resource() {
+    return unnormalize_processors(haxe.Unserializer.run(haxe.Resource.getString(cache_path)));
   }
 
-  public function unwind_tokens() : Hash<String> {
-    var unwound_tokens = new Hash<String>();
-    for (token in this.tokenHash.keys()) {
-      unwound_tokens.set(token, this.tokenHash.get(token).version);
+  public static function normalize_processors(processors : Array<TokenProcessor>) : Hash<Hash<Dynamic>> {
+    if (processors.length == 0) { throw "no processors specified"; }
+    var normalized_data = new Hash<Hash<Dynamic>>();
+
+    var types = new Hash<String>();
+    var all_versions_with_index = new Hash<Int>();
+
+    var version_index = 0;
+    for (i in 0...processors.length) {
+      var i_string = Std.string(i);
+      var tokens_with_version_index = new Hash<Int>();
+      types.set(i_string, Type.getClassName(Type.getClass(processors[i])));
+      for (token in processors[i].tokenHash.keys()) {
+        var version = processors[i].tokenHash.get(token).version;
+        if (!all_versions_with_index.exists(version)) {
+          all_versions_with_index.set(version, version_index);
+          version_index++;
+        }
+        tokens_with_version_index.set(token, all_versions_with_index.get(version));
+      }
+
+      normalized_data.set("processor-" + i_string, tokens_with_version_index);
     }
-    return unwound_tokens;
+
+    trace("Unique version strings: " + version_index);
+
+    var flipped_versions = new Hash<String>();
+    for (version in all_versions_with_index.keys()) {
+      flipped_versions.set(Std.string(all_versions_with_index.get(version)), version);
+    }
+
+    normalized_data.set("versions", flipped_versions);
+    normalized_data.set("types", types);
+
+    return normalized_data;
   }
 
-  public function populate_from_unwound_tokens(unwound_tokens : Hash<String>) {
-    this.tokenHash = new Hash<Token>();
-    var token_type = get_default_token_type();
-    for (token in unwound_tokens.keys()) {
-      this.tokenHash.set(token, Type.createInstance(token_type, [ token, unwound_tokens.get(token) ]));
+  public static function unnormalize_processors(normalized_data : Hash<Hash<Dynamic>>) : Array<TokenProcessor> {
+    var unnormalized_processors = new Array<TokenProcessor>();
+
+    if (!normalized_data.exists("versions")) { throw "versions not defined"; }
+    if (!normalized_data.exists("types")) { throw "types not defined"; }
+
+    var versions = normalized_data.get("versions");
+    var types = normalized_data.get("types");
+
+    for (type_key in types.keys()) {
+      var i = Std.parseInt(type_key);
+      var processor : TokenProcessor = Type.createInstance(Type.resolveClass(types.get(type_key)), []);
+
+      var processor_key = "processor-" + type_key;
+      if (!normalized_data.exists(processor_key)) { throw "processor " + type_key + " not defined"; }
+
+      var processor_tokens = normalized_data.get(processor_key);
+      var token_type = processor.get_default_token_type();
+      for (token in processor_tokens.keys()) {
+        var version_lookup = Std.string(processor_tokens.get(token));
+        processor.tokenHash.set(token, Type.createInstance(token_type, [token, versions.get(version_lookup)]));
+      }
+
+      unnormalized_processors.push(processor);
     }
+
+    return unnormalized_processors;
   }
 }
\ No newline at end of file