From 1ae519bb40c088f5f30162da40b5b4175e12aeee Mon Sep 17 00:00:00 2001
From: brent s <bts@square-r00t.net>
Date: Wed, 11 Dec 2019 06:33:24 -0500
Subject: [PATCH] hashing is done

---
 aif/constants_fallback.py |  5 +++-
 aif/utils/file_handler.py | 29 ++++++++++++++++-------
 aif/utils/hash_handler.py | 48 +++++++++++++++++++++++++++++++++++++++
 3 files changed, 73 insertions(+), 9 deletions(-)

diff --git a/aif/constants_fallback.py b/aif/constants_fallback.py
index fcb1cf1..12de2c0 100644
--- a/aif/constants_fallback.py
+++ b/aif/constants_fallback.py
@@ -1,4 +1,4 @@
-import os
+import hashlib
 import re
 import subprocess  # I wish there was a better way to get the supported LUKS ciphers.
 import uuid
@@ -287,3 +287,6 @@ MDADM_SUPPORTED_LAYOUTS = {5: (re.compile(r'^((left|right)-a?symmetric|[lr][as]|
                                 None)}
 # glibc doesn't support bcrypt/blowfish nor des (nor any of the others, like e.g. scrypt)
 CRYPT_SUPPORTED_HASHTYPES = ('sha512', 'sha256', 'md5')
+HASH_BUILTIN_SUPPORTED_TYPES = tuple(sorted(list(hashlib.algorithms_available)))
+HASH_EXTRA_SUPPORTED_TYPES = set(('adler32', 'crc32'))
+HASH_SUPPORTED_TYPES = tuple(sorted(list(hashlib.algorithms_available.union(HASH_EXTRA_SUPPORTED_TYPES))))
diff --git a/aif/utils/file_handler.py b/aif/utils/file_handler.py
index d10c699..4bfa239 100644
--- a/aif/utils/file_handler.py
+++ b/aif/utils/file_handler.py
@@ -9,6 +9,9 @@ class File(object):
         self.path_rel = pathlib.PurePosixPath(self.orig_path)
         self.path_full = pathlib.PurePosixPath(self.fullpath)
 
+    def __str(self):
+        return(self.fullpath)
+
 
 class Directory(object):
     def __init__(self, dir_path):
@@ -19,25 +22,35 @@ class Directory(object):
         self.files = []
         self.dirs = []
 
-    def populateFilesDirs(self, recursive = False):
+    def __str__(self):
+        return(self.fullpath)
+
+    def populateFilesDirs(self, recursive = False, native = False):
         if not recursive:
             for i in os.listdir(self.fullpath):
                 if os.path.isdir(os.path.join(self.fullpath, i)):
                     self.dirs.append(i)
                 elif os.path.isfile(os.path.join(self.fullpath, i)):
-                    self.files.append(i)
+                    if not native:
+                        self.files.append(i)
+                    else:
+                        self.files.append(File(i))
         else:
             for root, dirs, files in os.walk(self.fullpath):
                 for f in files:
                     fpath = os.path.join(root, f)
-                    relfpath = pathlib.PurePosixPath(fpath).relative_to(self.path_full)
-                    self.files.append(relfpath)
+                    relfpath = str(pathlib.PurePosixPath(fpath).relative_to(self.path_full))
+                    if not native:
+                        self.files.append(relfpath)
+                    else:
+                        self.files.append(relfpath)
                 for d in dirs:
                     dpath = os.path.join(root, d)
-                    reldpath = pathlib.PurePosixPath(dpath).relative_to(self.path_full)
+                    reldpath = str(pathlib.PurePosixPath(dpath).relative_to(self.path_full))
                     self.dirs.append(reldpath)
                 if root not in self.dirs:
-                    self.dirs.append(dirs)
-        self.dirs.sort()
-        self.files.sort()
+                    self.dirs.append(root)
+        if not native:
+            self.dirs.sort()
+            self.files.sort()
         return(None)
diff --git a/aif/utils/hash_handler.py b/aif/utils/hash_handler.py
index e69de29..260569f 100644
--- a/aif/utils/hash_handler.py
+++ b/aif/utils/hash_handler.py
@@ -0,0 +1,48 @@
+import hashlib
+import pathlib
+import zlib
+##
+import aif.constants_fallback
+from . import file_handler
+
+
+class Hash(object):
+    def __init__(self, file_path):
+        self.hashers = None
+
+    def configure(self, hashalgo = None):
+        self.hashers = {}
+        if hashalgo:
+            if not isinstance(hashalgo, list):
+                hashalgo = [hashalgo]
+        else:
+            hashalgo = list(aif.constants_fallback.HASH_SUPPORTED_TYPES)
+        for h in hashalgo:
+            if h not in aif.constants_fallback.HASH_SUPPORTED_TYPES:
+                raise ValueError('Hash algorithm not supported')
+            if h not in aif.constants_fallback.HASH_EXTRA_SUPPORTED_TYPES:
+                hasher = hashlib.new(h)
+            else:  # adler32 and crc32
+                hasher = getattr(zlib, h)
+            self.hashers[h] = hasher
+        return()
+
+    def hashData(self, data):
+        results = {}
+        if not self.hashers:
+            self.configure()
+        for hashtype, hasher in self.hashers.items():
+            if hashtype in aif.constants_fallback.HASH_EXTRA_SUPPORTED_TYPES:
+                results[hashtype] = hasher(data)
+            else:
+                rslt = hasher.update(data)
+                results[hashtype] = rslt.hexdigest()
+        return(results)
+
+    def hashFile(self, file_path):
+        if not isinstance(file_path, (str, file_handler.File, pathlib.Path, pathlib.PurePath)):
+            raise ValueError('file_path must be a path expression')
+        file_path = str(file_path)
+        with open(file_path, 'rb') as fh:
+            results = self.hashData(fh.read())
+        return(results)