feedwarrior

commit 24ed2be5033acac9b4c80f5da7a670c0e813fdff
parent 43ebb25273728f2dd0aa06ee90f68f40315221a2
Author: nolash <dev@holbrook.no>
Date:   Sun, 28 Jun 2020 17:43:12 +0200

Add entry object, calculate uuid from payload contents

Diffstat:
M
src/feedwarrior/__init__.py
 | 
1
+
M
src/feedwarrior/cmd/entry.py
 | 
27
++++++++++++++++++---------
M
src/feedwarrior/common.py
 | 
3
+++
M
src/feedwarrior/config.py
 | 
2
++
A
src/feedwarrior/entry.py
 | 
70
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

5 files changed, 94 insertions(+), 9 deletions(-)
diff --git a/src/feedwarrior/__init__.py b/src/feedwarrior/__init__.py
@@ -1,2 +1,3 @@
 from .feed import feed
+from . import entry
 from .config import load_config
diff --git a/src/feedwarrior/cmd/entry.py b/src/feedwarrior/cmd/entry.py
@@ -1,5 +1,14 @@
 # standard imports
 import email
+import logging
+import uuid
+import json
+
+# local imports
+import feedwarrior
+
+logg = logging.getLogger()
+
 
 def parse_args(argparser):
     argparser.add_argument('-l', required=True, help='log to add entry to')
@@ -11,14 +20,14 @@ def check_args(args):
     pass
 
 
-def process_as_multipart_file(config, feed, filename):
-    f = open(filename, 'r')
-    m = email.message_from_file(f)
-    f.close()
-    if not m.is_multipart():
-        raise ValueError('{} is not a MIME multipart message'.format(filename))
-    pass
+    return feedwarrior.entry(uu, m.get_payload())
+
 
 def execute(config, feed, args):
-    process_as_mime(config, feed, args.path)
-    pass
+    entry = feedwarrior.entry.from_multipart_file(args.path)
+    entry_serialized = entry.serialize()
+    uu = str(entry.uuid)
+    logg.debug('adding entry {}'.format(uu))
+    f = open(os.path.join(config.entries_dir, uu)
+    json.dump(f, entry_serialized)
+    f.close()
diff --git a/src/feedwarrior/common.py b/src/feedwarrior/common.py
@@ -1,5 +1,8 @@
 # standard imports
 import uuid
+import hashlib
+
+defaulthasher = hashlib.sha256
 
 def parse_uuid(uu):
     if type(uu).__name__ == 'str':
diff --git a/src/feedwarrior/config.py b/src/feedwarrior/config.py
@@ -1,6 +1,7 @@
 # standard imports
 import os
 import configparser
+import hashlib
 
 class config:
 
@@ -10,6 +11,7 @@ class config:
         self.data_dir = cp['FEEDWARRIOR']['datadir']
         self.feeds_dir = os.path.join(cp['FEEDWARRIOR']['datadir'], 'feeds')
         self.entries_dir = os.path.join(cp['FEEDWARRIOR']['datadir'], 'entries')
+        self.hasher = hashlib.sha256
 
 def load_config(filename):
     return config(filename)
diff --git a/src/feedwarrior/entry.py b/src/feedwarrior/entry.py
@@ -0,0 +1,70 @@
+# standard imports
+import email
+import uuid
+import logging
+import base64
+
+# local imports
+from .common import defaulthasher
+
+logg = logging.getLogger()
+
+
+class entry:
+
+    def __init__(self, uu, payload):
+        self.uuid = uu
+        self.payload = payload
+
+
+    def serialize(self):
+        return {
+            'uuid': str(self.uuid),
+            'payload': self.payload,
+                }
+
+   
+
+def from_multipart_file(filename, hasher=defaulthasher):
+#def process_as_multipart_file(config, feed, filename):
+    f = open(filename, 'r')
+    m = email.message_from_file(f)
+    f.close()
+    if not m.is_multipart():
+        raise ValueError('{} is not a MIME multipart message'.format(filename))
+
+    # the hasher calculates a uuid from the canonical order of the message contents
+    # TODO: currently the canonical order is the order of items in the message. this should
+    # rather be the lexiographical order of the hash integer values of the items.
+    htop = hasher()
+
+    # TODO: this is a naïve parser. If presumes that the message stucture will
+    # always be a multipart container on top. Therefore it will always discard the top item
+    subject = None
+    i = 0
+    for p in m.walk():
+        if i == 0:
+            subject = p.get('Subject')
+            i += 1
+            continue
+        if p.get_content_maintype() == 'multipart':
+            logg.warn('recursive multipart is not implemented, skipping part {}'.format(i))
+
+        hpart = hasher()
+        hpart.update(p.get_payload(decode=True))
+        psum = hpart.digest()
+        htop.update(psum)
+
+        i += 1
+
+    msum = htop.digest()
+    uu = uuid.UUID(bytes=msum[:16])
+    m.add_header('X-FEEDWARRIOR-HASH', htop.name)
+    m.add_header('X-FEEDWARRIOR-DIGEST', base64.encodebytes(msum).decode('ascii'))
+
+    if subject == None:
+        subject = str(uu)
+        logg.info('subject not specified, using uuid {}'.format(subject))
+
+    return entry(uu, m.as_bytes())
+