positivetechnologies / pt.pm Goto Github PK

Translate it after.

Even if IsIncludeIntermediateResult disabled. It's required frequently for linear to line-column text span conversion.

I mean: PT.PM.JavaAstConversion -> PT.PM.JavaUstConversion and other.

Parsers

https://github.com/zsoltmester/simple-abap-compiler

Samples:

• https://github.com/mauriciolauffer/ABAP
• https://github.com/abramsba/AbapExamples
• https://github.com/sulonen/abap

Or ImmutableArray extension.

Maybe use PatternDataStructure with patterns optimized information (FSM).

Move LastParseTree and LastUst to the returned object in Process method.
WorflowResult should have the following properties:

• List of ParseTree
• List of Ust
• List of MatchingResultDto
• Maybe list of read files SourceCode

Pattern: <["http"]>
Source: <?php $a = 'http'
Expected: http
Actual: 'htt

See this build for example: https://travis-ci.org/PositiveTechnologies/PT.PM/builds/210686796
Also see this PR: #14

• Use throw new ShouldNotBeVisitedException() instead of return VisitShouldNotBeVisited(context); everywhere.
• Remove VisitShouldNotBeVisited.

Translate not left-recursive rules to left-recursive ones due to the performance reason.
Check performance of rewritten grammar.

Use MIT license for that and upload it to NuGet.

Some patterns require certain file or directory paths.

Some php rules written as not left-recursive due to the performance issue, see the Java- and Java8 grammars paragraph in "Theory and Practice of Source Code Parsing with ANTLR and Roslyn" article.

But since ANTLR of version greater than or equal to 4.6 that is not an issue.

FileNode a bit confusing in visitors and other stuff. It should be removed.

Maybe use SourceCodeFile instead of simple string.

Add ability to load patterns from Json file, not only from string.

CompareTo returns a simple integer. It's not enough to an actual needs.

In some cases, it's necessary to return relative of absolute text spans except the result of the comparison (see IRelativeLocationPattern and IAbsoluteLocationPattern) and side effect text location used. It is necessary to get rid of them for all UstNode.

It will be used for patterns with private functions

Scan files with the following extensions:

• pks
• pkb
• tps
• vw

Implied console using instead of project assembly reference.

Ignore big JavaScript and other files with linguist-vendored attribute in .gitattributes file.

There are too many documentation files and articles in this repository.
Moving them to a separate repository will make maintaining more easier.
I suppose to name new repository as something like PT.PM.Doc.

• Integration of language parser.
  • ANTLR grammar writing if such parser does not exist.
• Developing of Converter to UST.
• Patterns writing and testing.
• DSL improving if required.

It helps to resolve performance problems for deep recursive expressions: InacBaseServiceApiDO.JAVA and ManyStringsConcat.java.

Sample:

<script type="text/javascript">
    var common_query = '<?php echo PMA_escapeJsString(PMA_generate_common_url('', '', '&'));?>';
</script>

I suggest to find <?php ?> with regex, process and remove them. After that parse JS code fragment with convenient JS parser.

hi，
I found cfg and pdg modul in this projecte,but they are not implemented。how can I implement such as simple defect analysis:

$a=_GET["taint"];
$b=$a;
echo $b;

or taint flow about interprocedural dependence analysis？any suggestion will be usefull！thanks a lot.

Implement an additional PT.PM.Cli's operation mode --dump-uast which allows to serialize and dump UAST without any matching against patterns. Dump should be saved to the named pipe or stdout (console) and contain a set of pairs Full file name -> UAST serialized to JSON for all processed sources.

<["\""]> should be recognized as a double quote instead of backslash + double quote.

Project: https://github.com/telerik/kendo-ui-core/tree/2017.1.223

Files with parsing errors:

• gulpfile.js
• build\gulp\gather-amd.js
• build\gulp\kendo-version.js
• build\gulp\license.js
• build\gulp\uglify.js
• build\gulp\wrap-umd.js
• build\gulp\tasks\karma.js
• docs_assets\javascripts\prettify.js
• docs-aspnet-core_assets\javascripts\prettify.js
• docs-aspnet-mvc_assets\javascripts\prettify.js
• src\angular.min.js
• src\jquery.min.js
• tests\core\class.js
• tests\data\datareader.js
• tests\data\xmldatareader.js
• tests\dropdownlist\knockout-min.js
• tests\model\initialization.js
• tests\mvvm\changing.js
• tests\window\templates-fixture.html

A basic class for all *Literal classes (StringLiteral, IntLiteral, BooleanLiteral, etc.).

Serialized BinaryOperatorLiteral node doesn't have BinaryOperator item.

Code sample:

<?php
$conn = mysql_connect('localhost', 'root', '');
mysql_select_db('test');
mysql_query("SELECT * FROM ci WHERE id=" . $_GET['a']);
?>

See attached screeshot.

Thanks.

At present time ParsingException is used for both.

Fill this property on Preprocess step.

Filter comments during scanning.