Refactored file type detection engine

This commit is contained in:
n1474335 2018-12-18 17:44:42 +00:00
parent d02124550b
commit e6fb0be1d0
12 changed files with 867 additions and 727 deletions

View file

@ -5,7 +5,7 @@
*/
import Operation from "../Operation";
import Magic from "../lib/Magic";
import {detectFileType} from "../lib/FileType";
/**
* Detect File Type operation
@ -34,17 +34,21 @@ class DetectFileType extends Operation {
*/
run(input, args) {
const data = new Uint8Array(input),
type = Magic.magicFileType(data);
types = detectFileType(data);
if (!type) {
if (!types.length) {
return "Unknown file type. Have you tried checking the entropy of this data to determine whether it might be encrypted or compressed?";
} else {
let output = "File extension: " + type.ext + "\n" +
"MIME type: " + type.mime;
let output;
if (type.desc && type.desc.length) {
output += "\nDescription: " + type.desc;
}
types.forEach(type => {
output += "File extension: " + type.extension + "\n" +
"MIME type: " + type.mime + "\n";
if (type.description && type.description.length) {
output += "\nDescription: " + type.description + "\n";
}
});
return output;
}

View file

@ -6,9 +6,8 @@
import Operation from "../Operation";
// import OperationError from "../errors/OperationError";
import Magic from "../lib/Magic";
import Utils from "../Utils";
import {extractFile} from "../lib/FileExtraction";
import {detectFileType, extractFile} from "../lib/FileType";
/**
* Extract Files operation
@ -40,13 +39,13 @@ class ExtractFiles extends Operation {
const bytes = new Uint8Array(input);
// Scan for embedded files
const fileDetails = scanForEmbeddedFiles(bytes);
const detectedFiles = scanForEmbeddedFiles(bytes);
// Extract each file that we support
const files = [];
fileDetails.forEach(fileDetail => {
detectedFiles.forEach(detectedFile => {
try {
files.push(extractFile(bytes, fileDetail));
files.push(extractFile(bytes, detectedFile.fileDetails, detectedFile.offset));
} catch (err) {}
});
@ -70,22 +69,21 @@ class ExtractFiles extends Operation {
* @param data
*/
function scanForEmbeddedFiles(data) {
let type;
const types = [];
const detectedFiles = [];
for (let i = 0; i < data.length; i++) {
type = Magic.magicFileType(data.slice(i));
if (type) {
types.push({
offset: i,
ext: type.ext,
mime: type.mime,
desc: type.desc
const fileDetails = detectFileType(data.slice(i));
if (fileDetails.length) {
fileDetails.forEach(match => {
detectedFiles.push({
offset: i,
fileDetails: match,
});
});
}
}
return types;
return detectedFiles;
}
export default ExtractFiles;

View file

@ -9,7 +9,7 @@ import { fromHex } from "../lib/Hex";
import Operation from "../Operation";
import OperationError from "../errors/OperationError";
import Utils from "../Utils";
import Magic from "../lib/Magic";
import {detectFileType} from "../lib/FileType";
/**
* Render Image operation
@ -72,8 +72,8 @@ class RenderImage extends Operation {
}
// Determine file type
const type = Magic.magicFileType(input);
if (!(type && type.mime.indexOf("image") === 0)) {
const types = detectFileType(input);
if (!(types.length && types[0].mime.indexOf("image") === 0)) {
throw new OperationError("Invalid file type");
}
@ -92,9 +92,9 @@ class RenderImage extends Operation {
let dataURI = "data:";
// Determine file type
const type = Magic.magicFileType(data);
if (type && type.mime.indexOf("image") === 0) {
dataURI += type.mime + ";";
const types = detectFileType(data);
if (types.length && types[0].mime.indexOf("image") === 0) {
dataURI += types[0].mime + ";";
} else {
throw new OperationError("Invalid file type");
}

View file

@ -6,7 +6,7 @@
import Operation from "../Operation";
import Utils from "../Utils";
import Magic from "../lib/Magic";
import {detectFileType} from "../lib/FileType";
/**
* Scan for Embedded Files operation
@ -41,7 +41,7 @@ class ScanForEmbeddedFiles extends Operation {
*/
run(input, args) {
let output = "Scanning data for 'magic bytes' which may indicate embedded files. The following results may be false positives and should not be treat as reliable. Any suffiently long file is likely to contain these magic bytes coincidentally.\n",
type,
types,
numFound = 0,
numCommonFound = 0;
const ignoreCommon = args[0],
@ -49,20 +49,23 @@ class ScanForEmbeddedFiles extends Operation {
data = new Uint8Array(input);
for (let i = 0; i < data.length; i++) {
type = Magic.magicFileType(data.slice(i));
if (type) {
if (ignoreCommon && commonExts.indexOf(type.ext) > -1) {
numCommonFound++;
continue;
}
numFound++;
output += "\nOffset " + i + " (0x" + Utils.hex(i) + "):\n" +
" File extension: " + type.ext + "\n" +
" MIME type: " + type.mime + "\n";
types = detectFileType(data.slice(i));
if (types.length) {
types.forEach(type => {
if (ignoreCommon && commonExts.indexOf(type.extension) > -1) {
numCommonFound++;
return;
}
if (type.desc && type.desc.length) {
output += " Description: " + type.desc + "\n";
}
numFound++;
output += "\nOffset " + i + " (0x" + Utils.hex(i) + "):\n" +
" File extension: " + type.extension + "\n" +
" MIME type: " + type.mime + "\n";
if (type.description && type.description.length) {
output += " Description: " + type.description + "\n";
}
});
}
}