Convert mbox files to standalone html files

No thread or date structure, just one isolated file per message.
Only text/plain and some multipart formats

mbox2web

@@ -0,0 +1,113 @@
+#!/usr/bin/python3
+
+import jinja2
+import mailbox
+import os
+import re
+import sys
+
+basedir = "."
+
+jenv = jinja2.Environment(
+                loader=jinja2.FileSystemLoader(["templates"]),
+                autoescape=True,
+              )
+
+def get_message_id(msg):
+    """
+    Extract the message id from a message
+
+    Note that this assumes that there is (at least) one message id. If
+    this is not the case, it will raise an exception (currently an
+    IndexError, but we may use something more suitable in the future).
+    """
+    match = re.search(r'<(.*?)>', msg["Message-ID"])
+    return match.group(1)
+
+def encode_message_id(msgid):
+    encmsgid = re.sub('[^!"$-.0-9:=@-z|~]', lambda x: "{%02x}" % (ord(x.group(0))), msgid)
+    return encmsgid
+
+def render_message(msg):
+    msgtmpl = jenv.get_template("message2.html")
+    bodyhtml = render_body(msg)
+    context = {
+        "message_id": msg["Message-Id"],
+        "subject": msg["Subject"],
+        "from": msg["From"],
+        "date": msg["Date"],
+        "bodyhtml": bodyhtml,
+    }
+    msghtml = msgtmpl.render(context)
+    return jinja2.Markup(msghtml)
+
+def render_body(msg):
+    content_type = msg.get_content_type()
+    if content_type == "text/plain":
+        bodytmpl = jenv.get_template("body_text_plain.html")
+        context = {
+            "body": msg.get_payload()
+        }
+        bodyhtml = bodytmpl.render(context)
+        return jinja2.Markup(bodyhtml)
+    elif content_type == "multipart/mixed":
+        partshtml = []
+        for part in msg.get_payload():
+            partshtml.append(render_body(part))
+        bodytmpl = jenv.get_template("body_multipart_mixed.html")
+        context = {
+            "parts": partshtml
+        }
+        bodyhtml = bodytmpl.render(context)
+        return jinja2.Markup(bodyhtml)
+    elif content_type == "multipart/digest":
+        partshtml = []
+        for part in msg.get_payload():
+            partshtml.append(render_message(part))
+        bodytmpl = jenv.get_template("body_multipart_digest.html")
+        context = {
+            "parts": partshtml
+        }
+        bodyhtml = bodytmpl.render(context)
+        return jinja2.Markup(bodyhtml)
+    elif content_type == "message/rfc822":
+        partshtml = []
+        for part in msg.get_payload():
+            partshtml.append(render_message(part))
+        bodytmpl = jenv.get_template("body_message_rfc822.html")
+        context = {
+            "parts": partshtml
+        }
+        bodyhtml = bodytmpl.render(context)
+        return jinja2.Markup(bodyhtml)
+    else:
+        raise RuntimeError("Content-type " + content_type + " not implemented yet")
+
+
+def archive(msg):
+    mid = get_message_id(msg)
+    encmid = encode_message_id(mid)
+    msgdir = basedir + "/msg/" + encmid
+    os.makedirs(msgdir, exist_ok=True)
+    with open(msgdir + "/index.html", "w") as hfd:
+        msgtmpl = jenv.get_template("message.html")
+        bodyhtml = render_body(msg)
+        context = {
+            "list": "LUGA",
+            "message_id": mid,
+            "subject": msg["Subject"],
+            "from": msg["From"],
+            "date": msg["Date"],
+            "bodyhtml": bodyhtml,
+        }
+        msghtml = msgtmpl.render(context)
+        hfd.write(msghtml)
+
+
+for f in sys.argv[1:]:
+    print("F", f)
+    mb = mailbox.mbox(f)
+
+    for m in mb:
+        archive(m)
+