Php parser array and comments trubble

[nikic]

Php parser array and comments trubble

[nikic]

Yes, comments are assigned to the node after them, so if there is nothing following the comment it won't be collected anywhere. There isn't really anything I can do against this, short of switching to a completely different model (where you get a parse tree instead of an abstract syntax tree). If you need to preserve all comments the PHP parser isn't a suitable tool, you'll need to work on the tokens directly.

[pscheit]

is this the same issue?

if (true) {
  // body comment
}

body comment is not stored in if? What about a fallback for comments to be attached to its parent (with line number?) when no element is coming afterwards?

Why are inline comments attached to the following element?

$myVar = 8; // 0-based
$otherVar = 9; // 1-based

0-based is assigned to $otherVar and // 1-based is dropped. Would you agree, that this is wrong for inline comments?

Is the parse Tree you name, the one were the comment would be an individual node?

[nikic]

Yep, those also have the same issue. The comment is assigned to the next node that follows, so for the if it is just dropped and for your second example it's assigned to the next assignment. I do agree that this is wrong, but I don't see an easy way to change this (with the current approach). Also it's often not really clear what statement a comment "belongs" to. E.g. consider the following code:

<?php

$someCodeIs;         // Here is a big block
$writtenInHere;      // of documentary text
$otherMultipleLines; // that isn't really related
                     // to the particular statements
                     // on the left. It's just a multiline
                     // comment formatted like that.

Here you can't really say that a comment on the right belongs to the statement left to it.

In any case, I'll try out whether there is some easy way to generate a parse trace which retains all information from the source code. I'm not sure whether this is possible with the data I have available right now.

[yasheena]

I tried to fix this problem by adding the file /node/Stmt/Comment.php to the parser with the following content:

<?php

/**
 * @property null|PHPParser_Node_Expr $num Number of loops to break
 */
class PHPParser_Node_Stmt_Comment extends PHPParser_Node_Stmt
{
    /**
     * Constructs a comment node (if there is no node to attach a comment to).
     *
     * @param array                    $attributes Additional attributes
     */
    public function __construct(array $attributes = array()) {
        parent::__construct(
            $attributes
        );
    }
}

I also added the folowing function to the file /PrettyPrinter/Default.php:

// For printing comments which couldn't be assigned to another node
public function pStmt_Comment($node) {
    return '';  
}

Now I searched for the right position in the file Parser.php to add a "Comment-Node", if during reducing the stack a still existing (unhandled) comment exists. Do you think this method would be useful to handle this comment problem? If yes, I need a hint: The stack handling of the parser is so complex so I have no idea where I have to insert the additional code.

[Ekstazi]

You shouldn't work with stack directly, think that you need to add a new rule for comments in parser.

[nikic]

Closing this in favor of #41.