revert all my stupid commits, we'll try this again later when we have a test server

[dak.git] / dak / process_unchecked.py

diff --git a/dak/process_unchecked.py b/dak/process_unchecked.py
index 123fd9f37e7e6f7316300f8b98c90452b12a086c..04afb7b30065648d3cd3cdf633778915477cb766 100755 (executable)
--- a/dak/process_unchecked.py
+++ b/dak/process_unchecked.py
@@ -630,11 +630,11 @@ def check_files():
 
             # Check the md5sum & size against existing files (if any)
             files[f]["pool name"] = utils.poolify (changes["source"], files[f]["component"])
-            files_id = database.get_files_id(files[f]["pool name"] + f, files[f]["size"], files[f]["md5sum"], files[f]["location id"])
+            files_id = database.get_files_id(files[f]["pool name"] + f, files[f]["size"], files[f]["md5sum"], files[f]["sha1sum"], files[f]["sha256sum"], files[f]["location id"])
             if files_id == -1:
                 reject("INTERNAL ERROR, get_files_id() returned multiple matches for %s." % (f))
             elif files_id == -2:
-                reject("md5sum and/or size mismatch on existing copy of %s." % (f))
+                reject("md5sum, sha1sum, sha256sum and/or size mismatch on existing copy of %s." % (f))
             files[f]["files id"] = files_id
 
             # Check for packages that have moved from one component to another
@@ -919,13 +919,111 @@ def check_urgency ():
 ################################################################################
 
 def check_hashes ():
-    utils.check_hash(".changes", files, "md5sum", apt_pkg.md5sum)
-    utils.check_hash(".dsc", dsc_files, "md5sum", apt_pkg.md5sum)
+    # Make sure we recognise the format of the Files: field
+    format = changes.get("format", "0.0").split(".",1)
+    if len(format) == 2:
+        format = int(format[0]), int(format[1])
+    else:
+        format = int(float(format[0])), 0
+
+    check_hash(".changes", files, "md5sum", apt_pkg.md5sum)
+    check_hash(".dsc", dsc_files, "md5sum", apt_pkg.md5sum)
+
+    # (hashname, function, originate)
+    # If originate is true, we have to calculate it because
+    # the changes file version is too early for it to be
+    # included
+    hashes = [("sha1", apt_pkg.sha1sum, False),
+              ("sha256", apt_pkg.sha256sum, False)]
+
+    if format <= (1,8):
+        hashes["sha1"] = True
+        hashes["sha256"] = True
+
+    for x in changes:
+        if x.startswith("checksum-"):
+            h = x.split("-",1)[1]
+            if h not in dict(hashes):
+                reject("Unsupported checksum field in .changes" % (h))
+
+    for x in dsc:
+        if x.startswith("checksum-"):
+            h = x.split("-",1)[1]
+            if h not in dict(hashes):
+                reject("Unsupported checksum field in .dsc" % (h))
+
+    for h,f,o in hashes:
+        try:
+            fs = utils.build_file_list(changes, 0, "checksums-%s" % h, h)
+            if o:
+                create_hash(fs, h, f, files)
+            else:
+                check_hash(".changes %s" % (h), fs, h, f, files)
+        except NoFilesFieldError:
+            reject("No Checksums-%s: field in .changes" % (h))
+        except UnknownFormatError, format:
+            reject("%s: unknown format of .changes" % (format))
+        except ParseChangesError, line:
+            reject("parse error for Checksums-%s in .changes, can't grok: %s." % (h, line))
+
+        if "source" not in changes["architecture"]: continue
+
+        try:
+            fs = utils.build_file_list(dsc, 1, "checksums-%s" % h, h)
+            if o:
+                create_hash(fs, h, f, dsc_files)
+            else:
+                check_hash(".dsc %s" % (h), fs, h, f, dsc_files)
+        except UnknownFormatError, format:
+            reject("%s: unknown format of .dsc" % (format))
+        except NoFilesFieldError:
+            reject("No Checksums-%s: field in .dsc" % (h))
+        except ParseChangesError, line:
+            reject("parse error for Checksums-%s in .dsc, can't grok: %s." % (h, line))
+
+################################################################################
+
+def create_hash (lfiles, key, testfn, basedict = None):
+    for f in lfiles.keys():
+        try:
+            file_handle = utils.open_file(f)
+        except CantOpenError:
+            continue
+
+        # Check hash
+        basedict[f]['%ssum' % key] = testfn(file_handle)
+        file_handle.close()
+
+
+################################################################################
+
+def check_hash (where, lfiles, key, testfn, basedict = None):
+    if basedict:
+        for f in basedict.keys():
+            if f not in lfiles:
+                reject("%s: no %s checksum" % (f, key))
+
+    for f in lfiles.keys():
+        if basedict and f not in basedict:
+            reject("%s: extraneous entry in %s checksums" % (f, key))
+
+        try:
+            file_handle = utils.open_file(f)
+        except CantOpenError:
+            continue
 
-    # This is stupid API, but it'll have to do for now until
-    # we actually have proper abstraction
-    for m in utils.ensure_hashes(changes, dsc, files, dsc_files):
-        reject(m)
+        # Check hash
+        if testfn(file_handle) != lfiles[f][key]:
+            reject("%s: %s check failed." % (f, key))
+        file_handle.close()
+        # Store the hashes for later use
+        basedict[f]['%ssum' % key] = lfiles[f][key]
+        # Check size
+        actual_size = os.stat(f)[stat.ST_SIZE]
+        size = int(lfiles[f]["size"])
+        if size != actual_size:
+            reject("%s: actual file size (%s) does not match size (%s) in %s"
+                   % (f, actual_size, size, where))
 
 ################################################################################