Aggiunta correzioni per verificare il MIME-type dei file downloadati.

pyproject.toml

@@ -15,10 +15,11 @@ license = "MIT"
 requires-python = ">= 3.6"
 dependencies = [
     "click",
+    "lddcollect",
     "lxml",
+    "python-magic",
     "pyyaml",
-    "requests",
+    "requests"
-    "lddcollect"
 ]
 classifiers = [
     "Development Status :: 5 - Production/Stable",

src/loaih/build.py

@@ -3,6 +3,7 @@
 """Classes and functions to build an AppImage."""
 
 import os
+import datetime
 import glob
 import subprocess
 import shutil
@@ -11,6 +12,7 @@ import shlex
 import tempfile
 import hashlib
 import requests
+import magic
 import loaih
 
 
@@ -27,6 +29,7 @@ class Collection(list):
         arch = [ x for x in arch if version.urls[x] != '-' ]
         self.extend([ Build(version, ar) for ar in arch ])
 
+class BuildException(Exception): pass
 
 class Build():
     """Builds a single version."""
@@ -237,7 +240,7 @@ class Build():
 
             # Download the archive
             try:
-                self.__download_archive__(archive)
+                self.__download_archive_debug__(archive)
             except Exception as error:
                 print(f"Failed to download {archive}: {error}.")
 
@@ -453,9 +456,33 @@ class Build():
         with open(f"{file}.md5", 'w', encoding='utf-8') as checkfile:
             checkfile.write(f"{retval.hexdigest()}  {os.path.basename(file)}")
 
-    def __download_archive__(self, archive):
+    def __download_archive__(self, archive) -> str:
         return self.__download__(self.url, archive)
 
+    def __download_archive_debug__(self, archive) -> str:
+        """Analyses the downloaded archive to prevent issues with unpacking."""
+        break_control = 0
+        testedfilename = ""
+        while break_control <= 5:
+            timenow = datetime.datetime.now()
+            testedfilename, resource = self.__download_debug__(self.url, archive)
+            mime = magic.Magic(mime=True)
+            mimetype = mime.from_file(testedfilename)
+            if mimetype == 'application/gzip':
+                return testedfilename
+
+            # On the contrary, we will dump a logfile, remove the download and
+            # redo the download.
+            with open(os.path.join(self.downloadpath, 'downloadfailure.log'), 'wa') as logfile:
+                logfile.write(f"{timenow.isoformat()};{resource.url};{mimetype}")
+
+            file.unlink(testedfilename)
+            break_control += 1
+
+        # If it arrives here, 5 attempts to download the archive have failed.
+        raise BuildException(f"All downloads failed for {archive}. Exiting.")
+
+
     def __download__(self, url: str, filename: str):
         basename = filename
         if '/' in filename:
@@ -473,6 +500,23 @@ class Build():
                     file.write(chunk)
         return filename
 
+    def __download_debug__(self, url: str, filename: str) -> (str, requests.Response):
+        basename = filename
+        if '/' in filename:
+            basename = filename.split('/')[-1]
+
+        full_url = url
+        if url.endswith('/'):
+            # URL has to be completed with basename of filename
+            full_url = url + basename
+
+        with requests.get(full_url, stream=True, timeout=10) as resource:
+            resource.raise_for_status()
+            with open(filename, 'wb') as file:
+                for chunk in resource.iter_content(chunk_size=8192):
+                    file.write(chunk)
+        return (filename, resource)
+
     def __select_tarballs__(self):
         retval = [ self.tarballs[0] ]