First version of an Optical Character Recognition (OCR) module for MISP

misp_modules/modules/import/ocr.py

@@ -0,0 +1,59 @@
+import json
+import base64
+
+try:
+    import Image
+except ImportError:
+    from PIL import Image
+
+from pytesseract import image_to_string
+from io import BytesIO
+misperrors = {'error': 'Error'}
+userConfig = { };
+
+inputSource = ['file']
+
+moduleinfo = {'version': '0.1', 'author': 'Alexandre Dulaunoy',
+              'description': 'Optical Character Recognition (OCR) module for MISP',
+              'module-type': ['import']}
+
+moduleconfig = []
+
+
+def handler(q=False):
+    if q is False:
+        return False
+    r = {'results': []}
+    request = json.loads(q)
+    image = base64.b64decode(request["data"])
+    image_file = BytesIO(image)
+    ocrized = image_to_string(Image.open(image_file))
+    freetext = {}
+    freetext['values'] = ocrized
+    freetext['type'] = 'freetext'
+    r['results'].append(freetext)
+    return r
+
+
+def introspection():
+    modulesetup = {}
+    try:
+        userConfig
+        modulesetup['userConfig'] = userConfig
+    except NameError:
+        pass
+    try:
+        inputSource
+        modulesetup['inputSource'] = inputSource
+    except NameError:
+        pass
+    return modulesetup
+
+
+def version():
+    moduleinfo['config'] = moduleconfig
+    return moduleinfo
+
+if __name__ == '__main__':
+    x = open('test.json', 'r')
+    handler(q=x.read())