From 136e5b6a9469e3fa9bb6ee64f5b3a20d2eeb4fc4 Mon Sep 17 00:00:00 2001
From: Alexander Kiryukhin <alexander.kiryukhin@tatar.ru>
Date: Tue, 28 Aug 2012 18:03:50 +0400
Subject: [PATCH] Alpha of new version

---
 nyaml.php  | 219 +++++++++++++++++++++++++++++++----------------------
 test.nyaml |   3 +
 test.php   |   8 ++
 3 files changed, 138 insertions(+), 92 deletions(-)
 create mode 100644 test.nyaml
 create mode 100644 test.php

diff --git a/nyaml.php b/nyaml.php
index 2caf399..6ba2716 100644
--- a/nyaml.php
+++ b/nyaml.php
@@ -1,106 +1,141 @@
 <?php
 
-class nyaml {
-    private $lines;
+namespace NXP;
 
-    public function file($fileName) {
+class nyaml
+{
+    /**
+     * Состояния
+     */
+    const PARSE_KEY = 0;
+    const PARSE_VALUE = 1;
+    const PARSE_INDENT = 2;
+    const PARSE_KEY_IN_QUOTE = 3;
+    const PARSE_KEY_IN_DQUOTE = 4;
+    const PARSE_VALUE_IN_QUOTE = 5;
+    const PARSE_VALUE_IN_DQUOTE = 6;
+
+    /**
+     * Термы
+     */
+    const T_SPACE = 0;
+    const T_NEWLINE = 1;
+    const T_CHAR = 2;
+    const T_QUOTE = 3;
+    const T_DQUOTE = 4;
+
+    private $state = 0, $result = array();
+    /**
+     * Термы
+     * @var array
+     */
+    private $terms = array(
+        self::T_SPACE => '/[\s|\t]/',
+        self::T_NEWLINE => '/[\n]/',
+        self::T_CHAR => '/[a-zA-Zа-яА-Я0-9]/',
+        self::T_QUOTE => '/[\']/',
+        self::T_DQUOTE => '/[\"]/'
+    );
+
+    /**
+     * Таблица переходов состояния
+     * @var array
+     */
+    private $actions = array(
+        array(self::PARSE_KEY,             self::T_CHAR,      self::PARSE_KEY),
+        array(self::PARSE_KEY,             self::T_SPACE,     self::PARSE_VALUE),
+        array(self::PARSE_KEY,             self::T_NEWLINE,   self::PARSE_VALUE),
+        array(self::PARSE_VALUE,           self::T_NEWLINE,   self::PARSE_KEY),
+        array(self::PARSE_VALUE,           self::T_SPACE,     self::PARSE_VALUE),
+        array(self::PARSE_VALUE,           self::T_CHAR,      self::PARSE_VALUE),
+        array(self::PARSE_KEY,             self::T_QUOTE,     self::PARSE_KEY_IN_QUOTE),
+        array(self::PARSE_KEY,             self::T_DQUOTE,    self::PARSE_KEY_IN_DQUOTE),
+        array(self::PARSE_VALUE,           self::T_QUOTE,     self::PARSE_VALUE_IN_QUOTE),
+        array(self::PARSE_VALUE,           self::T_DQUOTE,    self::PARSE_VALUE_IN_DQUOTE),
+        array(self::PARSE_KEY_IN_QUOTE,    self::T_QUOTE,     self::PARSE_KEY),
+        array(self::PARSE_KEY_IN_DQUOTE,   self::T_DQUOTE,    self::PARSE_KEY),
+        array(self::PARSE_VALUE_IN_QUOTE,  self::T_QUOTE,     self::PARSE_VALUE),
+        array(self::PARSE_VALUE_IN_DQUOTE, self::T_DQUOTE,    self::PARSE_VALUE),
+        array(self::PARSE_KEY_IN_QUOTE,    self::T_CHAR,      self::PARSE_KEY_IN_QUOTE),
+        array(self::PARSE_KEY_IN_DQUOTE,   self::T_CHAR,      self::PARSE_KEY_IN_DQUOTE),
+        array(self::PARSE_VALUE_IN_QUOTE,  self::T_CHAR,      self::PARSE_VALUE_IN_QUOTE),
+        array(self::PARSE_VALUE_IN_DQUOTE, self::T_CHAR,      self::PARSE_VALUE_IN_DQUOTE),
+        array(self::PARSE_KEY_IN_QUOTE,    self::T_SPACE,     self::PARSE_KEY_IN_QUOTE),
+        array(self::PARSE_KEY_IN_DQUOTE,   self::T_SPACE,     self::PARSE_KEY_IN_DQUOTE),
+        array(self::PARSE_VALUE_IN_QUOTE,  self::T_SPACE,     self::PARSE_VALUE_IN_QUOTE),
+        array(self::PARSE_VALUE_IN_DQUOTE, self::T_SPACE,     self::PARSE_VALUE_IN_DQUOTE),
+        array(self::PARSE_KEY_IN_QUOTE,    self::T_NEWLINE,   self::PARSE_KEY_IN_QUOTE),
+        array(self::PARSE_KEY_IN_DQUOTE,   self::T_NEWLINE,   self::PARSE_KEY_IN_DQUOTE),
+        array(self::PARSE_VALUE_IN_QUOTE,  self::T_NEWLINE,   self::PARSE_VALUE_IN_QUOTE),
+        array(self::PARSE_VALUE_IN_DQUOTE, self::T_NEWLINE,   self::PARSE_VALUE_IN_DQUOTE),
+    );
+
+    public function file($fileName)
+    {
         //Parse file
-        return $this->string(file_get_contents($fileName)); 
+        return $this->string(file_get_contents($fileName));
     }
 
-    public function string($nyaml) {
-        //Parse string
-
-        //Preprocessing
-        $nyaml = preg_replace("/(\n|^)\s*\#.*?\n/", "\n", $nyaml); //Cleaning from one-line comments
-        $nyaml = str_replace("\r", "", $nyaml); //Fix new lines
-        $nyaml = str_replace("\n\n", "\n", $nyaml); //Cleaning clear lines 
-        $nyaml = trim($nyaml,"\n");
-
-        $this->lines = explode("\n", $nyaml);
-        //Parsing
-        return $this->parse(0);
-     }
-
-    private function parse($level) {
-        $return=array();
-        $linelevel = $level;
-
-        while (count($this->lines)>0) {
-
-            $line = array_shift($this->lines);
-            $linelevel = $this->countLevel($line);
-
-            if ($linelevel<$level) { //If next line from root node stop parsing
-                array_unshift($this->lines, $line);
-                return $return;
-            }
-            if (substr($line,-1,1)==":") { //If new node
-                $key = substr($line, 0, strlen($line)-1);
-                $value = $this->parse($linelevel+1);
-                $return[trim($key)] = $value;
-            } elseif (strpos($line,":")!==false) { //If key-value pair
-                list($key, $value) = explode(":", $line, 2);
-                $return[trim($key)] = $this->parseValue($value);
-            } else { //If just value
-                $return[] = trim($line);
-            }
-        }
-        return $return;
-    }
-
-    private function parseValue($value) {
-        $value = trim($value);
-        switch (substr($value,0,1)) {
-            case "\"":
-                return substr($value,1,(strlen($value)-2));
-            break;
-            case "'":
-                return substr($value,1,(strlen($value)-2));
-            break;
-            case "[":
-                $result = $this->explode(",", substr($value,1,(strlen($value)-2)));
-                $result = array_map(array($this, "parseValue"), $result);
-                return $result;
-            break;
-            default:
-                return $value;
-        }
-    }
-    
-    private function explode($letter, $string) {
-        //smart exploding string
-        $quotes = array("\"","'","[","]");
-        $quote = '';
+    public function string($nyaml)
+    {
         $result = array();
-        $token = "";
-        for ($i = 0; $i<strlen($string); $i++) {
-            $symbol = substr($string,$i,1);
-            if ((in_array($symbol, $quotes)) && ($quote == '')) {
-                if ($symbol == '[') {
-                    $symbol = ']';
-                }
-                $quote = $symbol;
-            } elseif ((in_array($symbol, $quotes)) && ($quote == $symbol)) {
-                if ($quote == ']') {
-                    $token = "[$token]";
-                }
-                $quote = '';
-            } elseif (($symbol == $letter) && ($quote=='')) {
-                $result[] = $token;
-                $token = '';
-            } else {
-                $token .= $symbol;
+        $key = '';
+        $value = '';
+        for ($i = 0; $i < strlen($nyaml); $i++) {
+
+            $char = substr($nyaml, $i, 1);
+            $oldState = $this->state;
+            $newState = $this->getState($char);
+            $this->state = $newState;
+            if (($newState == self::PARSE_KEY) && ($oldState == self::PARSE_VALUE)) {
+                $result[$key] = $value;
+                $key = '';
+                $value = '';
+
+            }
+            if (($newState == self::PARSE_KEY) && ($oldState == self::PARSE_KEY) && ($oldState != self::PARSE_VALUE_IN_QUOTE) && ($oldState != self::PARSE_VALUE_IN_DQUOTE)) {
+                $key.=$char;
+            }
+            if ( ($newState == self::PARSE_VALUE)  && ($oldState != self::PARSE_VALUE_IN_QUOTE) && ($oldState != self::PARSE_VALUE_IN_DQUOTE) ) {
+                $value.=$char;
+            }
+
+            if (($newState == self::PARSE_KEY_IN_QUOTE) && ($oldState != self::PARSE_KEY)) {
+                $key.=$char;
+            }
+            if (($newState == self::PARSE_KEY_IN_DQUOTE) && ($oldState != self::PARSE_KEY)) {
+                $key.=$char;
+            }
+
+            if (($newState == self::PARSE_VALUE_IN_QUOTE) && ($oldState != self::PARSE_VALUE)) {
+                $value.=$char;
+            }
+            if (($newState == self::PARSE_VALUE_IN_DQUOTE) && ($oldState != self::PARSE_VALUE)) {
+                $value.=$char;
             }
         }
-        if ($token!='') {
-            $result[] = $token; 
-        }
+
+        $result[$key] = $value;
         return $result;
     }
 
-    private function countLevel($string) {
-        //Counting level by indenting
-        return (strlen($string) - strlen(ltrim($string,"\t ")));
+    public function getState($char)
+    {
+        $currentTerm = null;
+        foreach ($this->terms as $key => $term) {
+            if (preg_match($term, $char)) {
+                $currentTerm = $key;
+            }
+        }
+        if ($currentTerm === null) {
+            throw new Exception('Parse error');
+        }
+        foreach ($this->actions as $action) {
+            if (( $action[0] == $this->state ) && ( $action[1] == $currentTerm )) {
+
+                return $action[2];
+            }
+        }
     }
+
 }
diff --git a/test.nyaml b/test.nyaml
new file mode 100644
index 0000000..ac55a4a
--- /dev/null
+++ b/test.nyaml
@@ -0,0 +1,3 @@
+node1 value 1
+'node 2' value 2
+'node 3' "value 3"
\ No newline at end of file
diff --git a/test.php b/test.php
new file mode 100644
index 0000000..b57b514
--- /dev/null
+++ b/test.php
@@ -0,0 +1,8 @@
+<?php
+
+include 'nyaml.php';
+
+$nyaml = new NXP\nyaml();
+
+print_r($nyaml->file('test.nyaml'));
+