From eb245de82b561193124a38f39f939e922364f65c Mon Sep 17 00:00:00 2001
From: WangFei <151241843@qq.com>
Date: Fri, 24 Apr 2020 17:48:38 +0800
Subject: [PATCH] improve the parse method and add the inverse parse
---
README.md | 31 ++++++++++++++-
examples/inverse_parse_com.php | 38 ++++++++++++++++++
examples/parse_dom_string.php | 39 ++++++++++++++++++
src/DomParser/Config.php | 22 +++++++---
src/DomParser/DomParser.php | 18 ++++++---
src/DomParser/Node.php | 73 ++++++++++++++++++++++++++++++++++
src/DomParser/Parser.php | 12 +++---
tests/DomParserTest.php | 13 +++---
8 files changed, 223 insertions(+), 23 deletions(-)
create mode 100644 examples/inverse_parse_com.php
create mode 100644 examples/parse_dom_string.php
diff --git a/README.md b/README.md
index 9df136f..559607f 100644
--- a/README.md
+++ b/README.md
@@ -1,12 +1,14 @@
# dom-parser
-A PHP library , use to parse normal and special DOM string like vue react . to DOM Tree .
+A PHP library , use to parse normal and special DOM string like vue react. to DOM Tree.
+By modifying config variables, can parse other HTML-like markup languages string too.
## Installation
composer require faitheir/dom-parser
## Base Usage
-
+#### Parse Method
+parse Dom string to DOM Tree (\Faitheir\DomParser\Node $dom).
```php
$html = <<< 'HTML'
@@ -37,4 +39,29 @@ $parser = new \Faitheir\DomParser\DomParser();
$dom = $parser->setConfig([])->parse($html);
print_r($dom);
+```
+
+#### Inverse Parse Method
+parse Dom Tree(\Faitheir\DomParser\Node $dom) to DOM string.
+```php
+$string = $parser->setConfig([])->invParse($dom);
+or
+Config::getInstance()->setConfig([
+ 'tag_indent' => ' '
+]);
+$string = (string) $dom;
+```
+
+## Config Method
+```php
+# demo
+# match string '{{php id="world"}}'
+
+$configs = [
+ 'start_tag_reg' => '/^\s*{{([^}\s\/!]+)/is'
+];
+# other confs see DomPaser/Config.php
+
+$parser = $parser->setConfig($configs);
+$dom = $parser->parse($html);
```
\ No newline at end of file
diff --git a/examples/inverse_parse_com.php b/examples/inverse_parse_com.php
new file mode 100644
index 0000000..f28976f
--- /dev/null
+++ b/examples/inverse_parse_com.php
@@ -0,0 +1,38 @@
+
+
+
+
+
+
+
+
+
+
+
+
+ Hello World !
+
+
+
+
+HTML;
+
+$dom = $parser->parse($html);
+
+$string = $parser->invParse($dom);
+echo $string;
\ No newline at end of file
diff --git a/examples/parse_dom_string.php b/examples/parse_dom_string.php
new file mode 100644
index 0000000..8c905f7
--- /dev/null
+++ b/examples/parse_dom_string.php
@@ -0,0 +1,39 @@
+
+
+
+
+
+
+
+
+
+
+
+
+ Hello World !
+
+
+
+
+HTML;
+
+
+echo '';
+$dom = $parser->parse($html);
+
+print_r($dom);
\ No newline at end of file
diff --git a/src/DomParser/Config.php b/src/DomParser/Config.php
index 8bad547..90e0995 100644
--- a/src/DomParser/Config.php
+++ b/src/DomParser/Config.php
@@ -4,25 +4,37 @@
class Config
{
+ static private $instance;
+ private function __clone(){}
# parser configs
private $_config = [
+ # parse
'start_tag_reg' => '/^\s*<([^>\s\/!]+)/is',
'start_end_tag_reg' => '/(^\s*>)|(^\s*\/>)/is',
'end_tag_reg' => '/^\s*<([^>\s\/]+)/is',
'content_reg' => '/^\s*([^<]+)|()/is',
'attrs_reg' => '/^\s*([^=>< ]+)="([^"]*)"|\s([^=><\s]+)(?=\s|>)/iU',
+ # inverse parse
+ 'tag_indent' => ' ',
+ 'hide_genid' => false,
];
- /**
- * Config constructor.
- * @param array $confs
- */
- public function __construct($confs = [])
+ private function __construct($confs = [])
{
if (!empty($confs))
$this->setConfig($confs);
}
+ static public function getInstance($confs = [])
+ {
+ if (!self::$instance instanceof self) {
+ self::$instance = new self($confs);
+ }
+ if (!empty($confs))
+ self::$instance->setConfig($confs);
+
+ return self::$instance;
+ }
/**
* dynamic settings
diff --git a/src/DomParser/DomParser.php b/src/DomParser/DomParser.php
index 01935df..5b45cb2 100644
--- a/src/DomParser/DomParser.php
+++ b/src/DomParser/DomParser.php
@@ -1,5 +1,4 @@
config = new Config($config);
+ $this->config = Config::getInstance($config);
$this->parser = new Parser();
if ($domString)
@@ -50,7 +49,16 @@ public function parse($domString = '')
if ($domString)
$this->orgDomString = $domString;
- return $this->parser->config($this->config)->parse($this->orgDomString);
+ return $this->parser->parse($this->orgDomString);
}
+ /**
+ * inverse method
+ * @param $domTree
+ * @return string
+ */
+ public function invParse(Node $domTree)
+ {
+ return (string) $domTree;
+ }
}
\ No newline at end of file
diff --git a/src/DomParser/Node.php b/src/DomParser/Node.php
index 5755dc1..973f03b 100644
--- a/src/DomParser/Node.php
+++ b/src/DomParser/Node.php
@@ -137,5 +137,78 @@ public static function generContentNode($content = '')
return (new self('content'))->setNodeId()->setContent($content);
}
+ /**
+ * @return string
+ */
+ public function __toString()
+ {
+ # config
+ $config = Config::getInstance();
+ $tagIndent = $config->get('tag_indent');
+ $hideGenid = $config->get('hide_genid');
+ $tagIndentString = str_repeat($tagIndent, $this->level);
+
+ # content
+ if (in_array($this->nodeName, ['content'])) {
+ return $tagIndentString . $this->content . PHP_EOL;
+ }
+ # root
+ if (in_array($this->nodeName, ['root'])) {
+ return $this->_invParseChilds($this->childs);
+ }
+ if (empty($this->nodeName))
+ return '';
+
+ # normal
+ $string = $tagIndentString . '<' . $this->nodeName . ' ';
+ # id
+ if (!empty($this->domId))
+ $string .= ' id="' . $this->domId . '" ';
+ # gener id
+ if ($hideGenid == false)
+ $string .= ' data-genid="' . $this->nodeId . '" ';
+ # class
+ if (!empty($this->domAttrs['class'])) {
+ $string .= ' class="' . implode(' ', $this->domAttrs['class']) . '" ';
+ unset($this->domAttrs['class']);
+ }
+ # style
+ if (!empty($this->domAttrs['style'])) {
+// $string .= implode(' ', $this->domAttrs['class']);
+ unset($this->domAttrs['style']);
+ }
+ # attrs
+ if (!empty($this->domAttrs)) {
+ foreach ($this->domAttrs as $k => $v) {
+ $string .= ' ' . $k . '="' . $v . '" ';
+ }
+ }
+ # is single
+ if ($this->isSingle == true)
+ return $string . ' />' . PHP_EOL;
+
+ # start tag end
+ $string .= ' >' . PHP_EOL ;
+
+ if (!empty($this->childs))
+ $string .= $this->_invParseChilds($this->childs);
+
+ # end
+ $string .= $tagIndentString . 'nodeName . '>' . PHP_EOL;
+
+ return $string;
+ }
+
+ private function _invParseChilds($childs)
+ {
+ if (empty($childs))
+ return '';
+
+ $string = '';
+ foreach ($childs as $child) {
+ $string .= (string) $child;
+ }
+ return $string;
+ }
}
\ No newline at end of file
diff --git a/src/DomParser/Parser.php b/src/DomParser/Parser.php
index c5447c0..0c6f36b 100644
--- a/src/DomParser/Parser.php
+++ b/src/DomParser/Parser.php
@@ -8,11 +8,10 @@ class Parser
# config obj
private $_config;
- # set config obj
- public function config($config = null)
+ #
+ private function _getConfig()
{
- $this->_config = $config;
-
+ $this->_config = Config::getInstance();
#
$this->_startTagReg = $this->_config->get('start_tag_reg');
$this->_endTagReg = $this->_config->get('end_tag_reg');
@@ -30,6 +29,9 @@ public function config($config = null)
*/
public function parse($domString = '')
{
+ # init parse config
+ $this->_getConfig();
+
$root = Node::generRootNode();
$domString = trim($domString);
@@ -53,13 +55,11 @@ private function _parseChildNodes($domString = '', $parent = null)
$stringLen = strlen($domString);
# match start tad
list($domString, $node) = $this->_parseStartTag($domString, $parent);
-// $domString = $this->_parseStartTag($domString, $parent);
# match content
$domString = $this->_parseContent($domString, $parent);
# match end tag
-// $domString = $this->_parseEndTag($domString, $parent);
$domString = $this->_parseEndTag($domString, $node);
if ($stringLen == strlen($domString))
diff --git a/tests/DomParserTest.php b/tests/DomParserTest.php
index 8422859..e9e03d2 100644
--- a/tests/DomParserTest.php
+++ b/tests/DomParserTest.php
@@ -8,10 +8,12 @@
$html = <<< 'HTML'
-
-
+
+
-
+
@@ -26,7 +28,8 @@
HTML;
-echo '';
-$dom = $parser->setConfig([])->parse($html);
+//echo '';
+$dom = $parser->setConfig([
+])->parse($html);
print_r($dom);
\ No newline at end of file