Detection of scripting-language-based exploits using parse tree transformation
Systems and methods for protecting client computers are described. One exemplary system includes an initial filter that gathers scripting-language-data from webpage data and a signature database of signatures of known scripting-language-data exploits. In addition, a parser generates a parse tree from the scripting-language-data, and a normalization and signature matching component reduces a complexity of the parse tree, generates a representation of at least a portion of a structure of the parse tree, and prevents the scripting-language-data from reaching an intended recipient if the representation matches one of the signatures of known scripting-language-data exploits.
1. A method for protecting client computers, comprising:
receiving webpage data at a proxy from a webpage before the data reaches
an intended recipient, wherein the proxy is disposed as an intermediary between a server serving up the webpage and the intended recipient;
gathering scripting-language-data from the webpage data;
identifying characteristics of the scripting-language-data;
scoring the characteristics of the scripting-language-data based upon a likelihood that the characteristics of the scripting-language-data are associated with malicious scripting-language-data;
generating a parse tree from the gathered scripting-language-data;
normalizing string-splitting constructs to reduce a level of string splitting in the parse tree;
replacing variable names of the parse tree with standard names;
generating a representation of at least a portion of the standard names;
comparing the representation with signatures of known scripting language exploits; and
determining whether to prevent the scripting-language-data from reaching the intended recipient based upon one or more of:
(i) a score of characteristics of a normalized scripting-language-data;
(ii) a score of characteristics of inspection data based on the scripting-language-data; and
(iii) when the representation matches a signature of the one of the known scripting language exploits.
2. The method of claim 1 , wherein the representation is a hash of at least a portion of the standard names.
3. The method of claim 1 , including:
disassembling a string that is generated from the scripting-language-data;
pseudo-executing the disassembled string; and
determining whether the scripting-language-data includes shellcode.
4. The method of claim 3 , wherein the disassembling is responsive to a size of the string exceeding a threshold.
5. The method of claim 1 , wherein identifying characteristics of the normalized scripting language-data includes identifying a presence of Eval functions, identifying statements exceeding a threshold-string-size, and identifying the existence of a DOM object, and wherein the inspection-data collected from the inspection points includes an indication whether a hidden Iframe is present, an indication whether a script writes another dynamic script, and an indication whether a script error matches a known anti-emulator method.
6. The method of claim 1 , wherein the normalized scripting-language-data comprises consolidated portions of scripting-language-data that are normalized into tokens; and
wherein the inspection data is collected during the emulated execution of data suspected of being scripting-language-based.
7. The method of claim 6 , wherein the normalized scripting-language-data indicates functions that the scripting-language-data would execute when executed in a web browser.
8. A system executed by a processor for protecting client computers; comprising:
an initial filter that gathers scripting-language-data from webpage data;
a signature database of signatures of known scripting-language-data exploits;
a parser that generates a parse tree from the scripting-language-data;
a normalization and signature matching component that reduces a complexity of the parse tree, generates a representation of at least a portion of a structure of the parse tree, and prevents the scripting-language-data from reaching an intended recipient when the representation matches one of the signatures of known scripting-language-data exploits; and
an analysis component that identifies characteristics of the scripting-language-data, scores the characteristics of scripting-language-data based upon a likelihood that the characteristics of the scripting-language-data are associated with malicious scripting-language data, and determines whether to block the scripting-language-data from reaching the intended recipient based upon a score of characteristics of a normalized scripting-language-data and a score of characteristics of inspection data based on the scripting language-data.
9. The system of claim 8 , wherein the normalization and signature matching component generates a hash representation of at least a portion of the structure of the parse tree.
10. The system of claim 8 , including:
a shellcode module that disassembles one or more strings that are generated during an emulated execution of the scripting-language-data so as to generate disassembled code and pseudo-executes at least a portion of the disassembled code to determine whether to block the scripting-language-data from reaching the intended recipient based upon pseudo-execution-data.
11. The system of claim 8 , wherein the shellcode module does not disassemble the one or more strings unless the one or mores strings exceeds a size threshold.
12. The system of claim 8 , wherein the normalized scripting-language-data comprises consolidated portions of scripting-language-data that are normalized into tokens; and
wherein the inspection data is collected during the emulated execution of data suspected of being scripting-language-based.
13. The system of claim 12 , wherein the normalized scripting-language-data indicates functions that the scripting-language-data would execute when executed in a web browser.
14. A non-transitory, tangible computer readable storage medium, encoded with processor readable instructions to perform a method for protecting client computers, the method comprising:
receiving webpage data at a proxy from a webpage before the data reaches an intended recipient, wherein the proxy is disposed as an intermediary between a server serving up the webpage and the intended recipient;
gathering scripting-language-data from the webpage data;
generating a parse tree from the gathered scripting-language-data;
identifying characteristics of the scripting-language-data;
scoring the characteristics of the scripting-language-data based upon a likelihood that the characteristics of the scripting-language-data are associated with malicious scripting-language-data;
normalizing string-splitting constructs to reduce a level of string splitting in the parse tree;
replacing variable names of the parse tree with standard names;
generating a representation of at least a portion of the standard names;
comparing the representation with signatures of known scripting language exploits; and
determining whether to prevent the scripting-language-data from reaching the intended recipient based upon one or more of:
(i) a score of characteristics of a normalized scripting-language-data;
(ii) a score of characteristics of inspection data based on the scripting-language-data; and
(iii) when the representation matches a signature of the one of the known scripting language exploits.
15. The non-transitory, tangible computer readable storage medium of claim 14 , wherein the representation is a hash of at least a portion of the standard names.
16. The non-transitory, tangible computer readable storage medium of claim 14 , the method including:
disassembling a string that is generated from the scripting-language-data;
pseudo-executing the disassembled string; and
determining whether the scripting-language-data includes shellcode.
17. The non-transitory, tangible computer readable storage medium of claim 16 , wherein the disassembling is responsive to a size of the string exceeding a threshold.
18. The non-transitory, tangible computer readable storage medium of claim 14 , wherein identifying characteristics of the normalized scripting-language-data includes identifying a presence of Eval functions, identifying statements exceeding a threshold-string-size, and identifying the existence of a DOM object, and wherein the inspection-data collected from the inspection points includes an indication whether a hidden Iframe is present, an indication whether a script writes another dynamic script, and an indication whether a script error matches a known anti-emulator method.
19. The non-transitory, tangible computer readable storage medium of claim 14 , wherein the normalized scripting-language-data comprises consolidated portions of scripting-language-data that are normalized into tokens; and
wherein the inspection data is collected during the emulated execution of data suspected of being scripting-language-based.
20. The non-transitory, tangible computer readable storage medium of claim 19 , wherein the normalized scripting-language-data indicates functions that the scripting-language-data would execute when executed in a web browser.