From e8a75a42df688e575d6b7ea5d983dc0b790c1c32 Mon Sep 17 00:00:00 2001
From: brent s <bts@square-r00t.net>
Date: Thu, 21 Jan 2021 11:59:18 -0500
Subject: [PATCH] centralize getting the hash because duh

---
 _base.py     | 28 +++++++++++++++++++---------
 arch.py      | 14 +++++---------
 grml.py      | 14 +++++---------
 sysresccd.py | 16 ++++++----------
 4 files changed, 35 insertions(+), 37 deletions(-)

diff --git a/_base.py b/_base.py
index ccf4aa3..13f28e0 100644
--- a/_base.py
+++ b/_base.py
@@ -90,17 +90,16 @@ class BaseUpdater(object):
             return(None)
         if not self.iso_url:
             raise RuntimeError('iso_url attribute must be set first')
-        req = requests.get(self.iso_url, stream = True, headers = {'User-Agent': 'curl/7.74.0'})
-        if not req.ok:
-            raise RuntimeError('Received non-200/30x {0} for {1}'.format(req.status_code, self.iso_url))
+        req_chk = requests.head(self.iso_url, headers = {'User-Agent': 'curl/7.74.0'})
+        if not req_chk.ok:
+            raise RuntimeError('Received non-200/30x {0} for {1}'.format(req_chk.status_code, self.iso_url))
         os.makedirs(os.path.dirname(self.dest_iso), exist_ok = True)
-        with req as uri:
+        with requests.get(self.iso_url, stream = True, headers = {'User-Agent': 'curl/7.74.0'}) as req:
+            req.raise_for_status()
             with open(self.dest_iso, 'wb') as fh:
-                shutil.copyfileobj(uri.raw, fh)
-        hasher = hashlib.new(self.hash_type)
-        with open(self.dest_iso, 'rb') as fh:
-            hasher.update(fh.read())
-        realhash = hasher.hexdigest().lower()
+                for chunk in req.iter_content(chunk_size = 8192):
+                    fh.write(chunk)
+        realhash = self.getISOHash()
         if realhash != self.new_hash:
             raise RuntimeError('Hash mismatch: {0} (LOCAL), {1} (REMOTE)'.format(realhash, self.new_hash))
         self.updateVer()
@@ -110,6 +109,17 @@ class BaseUpdater(object):
         raise RuntimeError('BaseUpdater should be subclassed and its updateVer, getCurVer, and getNewVer methods '
                            'should be replaced.')
 
+    def getISOHash(self):
+        hasher = hashlib.new(self.hash_type)
+        # TODO: later on when python 3.8 is more prevalent, https://stackoverflow.com/a/1131238/733214
+        with open(self.dest_iso, 'rb') as fh:
+            while True:
+                chunk = fh.read(8192)
+                if not chunk:
+                    break
+                hasher.update(chunk)
+        return(hasher.hexdigest().lower())
+
     def getNewVer(self):
         raise RuntimeError('BaseUpdater should be subclassed and its updateVer, getCurVer, and getNewVer methods '
                            'should be replaced.')
diff --git a/arch.py b/arch.py
index be844a9..7219c5b 100755
--- a/arch.py
+++ b/arch.py
@@ -13,7 +13,6 @@
 
 import datetime
 import json
-import hashlib
 import os
 import re
 ##
@@ -133,17 +132,14 @@ class Updater(_base.BaseUpdater):
         # if ver_info.get('arch') != self.arch:
         #     self.do_update = True
         #     self.force_update = True
-        try:
-            hasher = hashlib.new(self.hash_type)
-            with open(self.dest_iso, 'rb') as fh:
-                hasher.update(fh.read())
-            if self.old_hash != hasher.hexdigest().lower():
-                self.do_update = True
-                self.force_update = True
-        except FileNotFoundError:
+        if not os.path.isfile(self.dest_iso):
             self.do_update = True
             self.force_update = True
             return(None)
+        realhash = self.getISOHash()
+        if self.old_hash != realhash:
+            self.do_update = True
+            self.force_update = True
         return(None)
 
     def getNet(self):
diff --git a/grml.py b/grml.py
index 5abf4ef..a932088 100755
--- a/grml.py
+++ b/grml.py
@@ -10,7 +10,6 @@
 
 import datetime
 import json
-import hashlib
 import os
 import re
 ##
@@ -94,17 +93,14 @@ class Updater(_base.BaseUpdater):
         if ver_info.get('arch') != self.arch:
             self.do_update = True
             self.force_update = True
-        try:
-            hasher = hashlib.new(self.hash_type)
-            with open(self.dest_iso, 'rb') as fh:
-                hasher.update(fh.read())
-            if self.old_hash != hasher.hexdigest().lower():
-                self.do_update = True
-                self.force_update = True
-        except FileNotFoundError:
+        if not os.path.isfile(self.dest_iso):
             self.do_update = True
             self.force_update = True
             return(None)
+        realhash = self.getISOHash()
+        if self.old_hash != realhash:
+            self.do_update = True
+            self.force_update = True
         return(None)
 
     def getNewVer(self):
diff --git a/sysresccd.py b/sysresccd.py
index e815d01..8002430 100755
--- a/sysresccd.py
+++ b/sysresccd.py
@@ -10,7 +10,6 @@
 
 import datetime
 import json
-import hashlib
 import os
 import re
 ##
@@ -78,17 +77,14 @@ class Updater(_base.BaseUpdater):
         if ver_info.get('arch') != self.arch:
             self.do_update = True
             self.force_update = True
-        try:
-            hasher = hashlib.new(self.hash_type)
-            with open(self.dest_iso, 'rb') as fh:
-                hasher.update(fh.read())
-            if self.old_hash != hasher.hexdigest().lower():
-                self.do_update = True
-                self.force_update = True
-        except FileNotFoundError:
+        if not os.path.isfile(self.dest_iso):
+            self.do_update = True
+            self.force_update = True
+            return (None)
+        realhash = self.getISOHash()
+        if self.old_hash != realhash:
             self.do_update = True
             self.force_update = True
-            return(None)
         return (None)
 
     def getNewVer(self):