Support for PLAIN BB-Code and CODE=rich (#36)

* had to reset my branch * remove unused files in parser-directory after reset * Fix typos/mistakes in code comments * Properly set up phpunit testsuite and extract helper methods to trait * Add phpunit to phan github action * simplify the parser and make sure text-nodes are buffered * stop parsing after unclosed bbcode there is no reason the check the text-buffer after a unclosed bbcode, because the unclosed bbcode consumes the rest of the input anyway. * fix wording in test * fix wording in test #2 * add comment about nested bbcodes in test * added parser-helpers trait again * used more descriptive variable-names in parser * used more descriptive variable-names in parser * hide phan progress-bar, because github cannot render it * (wip) added support for nested code-blocks * some fixes and adjusted node-export logic * tests added * fix code=rich export issue * small adjustments in architecture + saner api * test if pr is broken * yep, pr is broken * fix IteratorAggregate usage * make tests type-safe * added more tests and fixed some associated issues * fix typo in file-path * fixed some typos and added more tests * use expectException in tests Co-authored-by: JR Cologne <[email protected]>
dev-community-de · Jul 17, 2020 · 38ef1fe · 38ef1fe
1 parent 8a1a862
commit 38ef1fe
Show file tree

Hide file tree

Showing 17 changed files with 2,908 additions and 448 deletions.
diff --git a/.github/workflows/phan.yml → .github/workflows/phan-phpunit.yml b/.github/workflows/phan.yml → .github/workflows/phan-phpunit.yml
@@ -1,12 +1,12 @@
-name: Static analysis (phan)
+name: Static analysis (phan) & Unit Testing (phpunit)
 
 on:
   pull_request:
     paths:
       - "**.php"
 
 jobs:
-  phan:
+  phan-phpunit:
     runs-on: ubuntu-latest
     steps:
       - name: Install PHP and dependencies
@@ -25,4 +25,6 @@ jobs:
       - name: Install composer packages
         run: composer install
       - name: Run phan from vendor folder
-        run: vendor/bin/phan -S -k ./.phan/config.php
+        run: vendor/bin/phan -S -k ./.phan/config.php --no-progress-bar
+      - name: Run phpunit from vendor folder
+        run: vendor/bin/phpunit
diff --git a/app/CodeFormatterApp.php b/app/CodeFormatterApp.php
@@ -3,9 +3,10 @@
 namespace DevCommunityDE\CodeFormatter;
 
 use DevCommunityDE\CodeFormatter\CodeFormatter\CodeFormatter;
+use DevCommunityDE\CodeFormatter\Parser\ElemNode;
+use DevCommunityDE\CodeFormatter\Parser\Node;
 use DevCommunityDE\CodeFormatter\Parser\Parser;
-use DevCommunityDE\CodeFormatter\Parser\PregParser;
-use DevCommunityDE\CodeFormatter\Parser\Token;
+use DevCommunityDE\CodeFormatter\Parser\TextNode;
 
 /**
  * Class CodeFormatterApp.
@@ -22,7 +23,7 @@ class CodeFormatterApp
      */
     public function __construct(Parser $parser = null)
     {
-        $this->parser = $parser ?? new PregParser();
+        $this->parser = $parser ?? new Parser();
     }
 
     /**
@@ -32,59 +33,76 @@ public function __construct(Parser $parser = null)
      */
     public function run()
     {
-        $tokens = $this->parseInput();
+        $nodes = $this->parseInput();
 
-        foreach ($tokens as $token) {
-            echo $this->formatToken($token);
+        foreach ($nodes as $node) {
+            echo $this->formatNode($node);
         }
     }
 
     /**
      * parses the code from stdin.
      *
-     * @return iterable<Token>
+     * @return iterable<Node>
      */
     private function parseInput(): iterable
     {
         return $this->parser->parseFile('php://input');
     }
 
     /**
-     * formats a token based on its type and language.
+     * formats a node based on its type and language.
      *
-     * @param Token $token
+     * @param Node $node
      *
      * @return string
      */
-    private function formatToken(Token $token): string
+    private function formatNode(Node $node): string
     {
-        if ($token->isText()) {
-            return $token->getBody();
+        if ($node instanceof TextNode) {
+            return $this->exportNode($node, null);
         }
 
-        $language = $token->getAttribute('lang');
-        \assert(null !== $language);
+        \assert($node instanceof ElemNode);
 
+        if (!$node->isCode()) {
+            // export node as-is (PLAIN bbcode)
+            return $this->exportNode($node, null);
+        }
+
+        if ($node->isRich()) {
+            // iterate over all child-nodes in this case
+            $buffer = '';
+            foreach ($node->getBody() as $subNode) {
+                $buffer .= $this->formatNode($subNode);
+            }
+            return $this->exportNode($node, $buffer);
+        }
+
+        $language = $node->getLang() ?: 'text';
         $formatter = CodeFormatter::create($language);
+
         if (null === $formatter) {
-            // no formatter found, return token as is
-            return $this->exportToken($token, null);
+            // no formatter found, return node as-is
+            return $this->exportNode($node, null);
         }
 
-        $result = $formatter->exec($token->getBody());
-        return $this->exportToken($token, $result);
+        $nodeBody = $node->getBody();
+        \assert(\is_string($nodeBody));
+        $result = $formatter->exec($nodeBody);
+        return $this->exportNode($node, $result);
     }
 
     /**
-     * exports a token.
+     * exports a node.
      *
-     * @param Token       $token
+     * @param Node        $node
      * @param string|null $body
      *
      * @return string
      */
-    private function exportToken(Token $token, ?string $body): string
+    private function exportNode(Node $node, ?string $body): string
     {
-        return $this->parser->exportToken($token, $body);
+        return $this->parser->exportNode($node, $body);
     }
 }
diff --git a/app/Parser/ElemAttrs.php b/app/Parser/ElemAttrs.php
@@ -0,0 +1,63 @@
+<?php
+
+declare(strict_types=1);
+namespace DevCommunityDE\CodeFormatter\Parser;
+
+/**
+ * this class represents element attributes in a
+ * raw (match) and a tokenized (pairs) form.
+ *
+ * @internal
+ */
+final class ElemAttrs
+{
+    /** @var string */
+    private $match;
+
+    /** @var array */
+    private $pairs;
+
+    /**
+     * constructor.
+     *
+     * @param string               $match
+     * @param array<string,string> $pairs
+     */
+    public function __construct(string $match, array $pairs)
+    {
+        $this->match = $match;
+        $this->pairs = $pairs;
+    }
+
+    /**
+     * returns a value.
+     *
+     * @param string $name
+     *
+     * @return string|null
+     */
+    public function getValue(string $name): ?string
+    {
+        return $this->pairs[$name] ?? null;
+    }
+
+    /**
+     * returns the matched attribute string.
+     *
+     * @return string
+     */
+    public function getMatch(): string
+    {
+        return $this->match;
+    }
+
+    /**
+     * checks if some attributes are set.
+     *
+     * @return bool
+     */
+    public function hasMatch(): bool
+    {
+        return !empty($this->match);
+    }
+}
diff --git a/app/Parser/ElemNode.php b/app/Parser/ElemNode.php
@@ -0,0 +1,124 @@
+<?php
+
+declare(strict_types=1);
+namespace DevCommunityDE\CodeFormatter\Parser;
+
+/**
+ * this class represents a bbcode.
+ */
+final class ElemNode extends Node
+{
+    /** @var string */
+    private $name;
+
+    /** @var ElemAttrs */
+    private $attrs;
+
+    /** @var NodeList|string */
+    private $body;
+
+    /**
+     * constructs a new code-token.
+     *
+     * @param string          $name
+     * @param ElemAttrs       $attrs
+     * @param NodeList|string $body
+     */
+    public function __construct(string $name, ElemAttrs $attrs, $body)
+    {
+        \assert(\is_string($body) || $body instanceof NodeList);
+        parent::__construct(Node::KIND_ELEM);
+        $this->name = $name;
+        $this->attrs = $attrs;
+        $this->body = $body;
+    }
+
+    /**
+     * checks if this node represents a [CODE] bbcode.
+     *
+     * @return bool
+     */
+    public function isCode(): bool
+    {
+        return 0 === strcasecmp($this->name, 'code');
+    }
+
+    /**
+     * checks if this node represents a [CODE=rich] bbcode.
+     *
+     * @return bool
+     */
+    public function isRich(): bool
+    {
+        if (0 !== strcasecmp($this->name, 'code')) {
+            return false;
+        }
+
+        $langAttr = $this->getLang();
+
+        if (empty($langAttr)) {
+            /* defaults to "text" */
+            return false;
+        }
+
+        return 0 === strcasecmp($langAttr, 'rich');
+    }
+
+    /**
+     * utility method to retrieve a lang-attribute.
+     *
+     * @return string|null
+     */
+    public function getLang(): ?string
+    {
+        $langAttr = $this->getAttr('lang');
+
+        if (null === $langAttr && $this->isCode()) {
+            $langAttr = $this->getAttr('@value');
+        }
+
+        return $langAttr;
+    }
+
+    /**
+     * returns the element name.
+     *
+     * @return string
+     */
+    public function getName(): string
+    {
+        return $this->name;
+    }
+
+    /**
+     * returns a single node attribute.
+     *
+     * @param string $name
+     *
+     * @return string|null
+     */
+    public function getAttr(string $name): ?string
+    {
+        return $this->attrs->getValue($name);
+    }
+
+    /**
+     * returns the node attributes.
+     *
+     * @return string
+     */
+    public function getAttrMatch(): string
+    {
+        return $this->attrs->getMatch();
+    }
+
+    /**
+     * returns the node-body.
+     *
+     * @return NodeList|string
+     */
+    public function getBody()
+    {
+        return $this->body;
+    }
+}
diff --git a/app/Parser/Node.php b/app/Parser/Node.php
@@ -0,0 +1,37 @@
+<?php
+
+declare(strict_types=1);
+
+namespace DevCommunityDE\CodeFormatter\Parser;
+
+abstract class Node
+{
+    /** text node */
+    public const KIND_TEXT = 1;
+
+    /** element node (bbcode) */
+    public const KIND_ELEM = 2;
+
+    /** @var int */
+    private $kind;
+
+    /**
+     * constructs a new token.
+     *
+     * @param int $kind
+     */
+    public function __construct(int $kind)
+    {
+        $this->kind = $kind;
+    }
+
+    /**
+     * returns the node kind.
+     *
+     * @return int
+     */
+    public function getKind(): int
+    {
+        return $this->kind;
+    }
+}