From 91e63c7e5f92e2064f8195d413ee4f1b94ea2a28 Mon Sep 17 00:00:00 2001
From: Martin Zimmermann <info@posativ.org>
Date: Wed, 11 Jun 2014 09:52:20 +0200
Subject: [PATCH] simplify import format detection

---
 isso/migrate.py              | 60 +++++++++++++++++-------------------
 isso/tests/test_migration.py |  9 +++---
 isso/tests/wordpress.xml     |  2 +-
 3 files changed, 34 insertions(+), 37 deletions(-)

diff --git a/isso/migrate.py b/isso/migrate.py
index 7afd736..c3f0500 100644
--- a/isso/migrate.py
+++ b/isso/migrate.py
@@ -6,6 +6,7 @@ import sys
 import os
 import io
 import re
+import logging
 import textwrap
 
 from time import mktime, strptime, time
@@ -26,6 +27,7 @@ except ImportError:
 
 from xml.etree import ElementTree
 
+logger = logging.getLogger("isso")
 
 def strip(val):
     if isinstance(val, string_types):
@@ -146,14 +148,6 @@ class Disqus(object):
                                     initial_indent="  ", subsequent_indent="  "))
                 print("")
 
-    @classmethod
-    def detect(cls, peek):
-
-        if 'xmlns="http://disqus.com' in peek:
-            return "http://disqus.com"
-
-        return None
-
 
 class WordPress(object):
 
@@ -164,11 +158,13 @@ class WordPress(object):
         self.xmlfile = xmlfile
         self.count = 0
 
-        with io.open(xmlfile) as fp:
-            ns = WordPress.detect(fp.read(io.DEFAULT_BUFFER_SIZE))
-
-        if ns:
-            self.ns = "{" + ns + "}"
+        for line in io.open(xmlfile):
+            m = WordPress.detect(line)
+            if m:
+                self.ns = WordPress.ns.replace("1.0", m.group(1))
+                break
+        else:
+            logger.warn("No WXR namespace found, assuming 1.0")
 
     def insert(self, thread):
 
@@ -242,12 +238,19 @@ class WordPress(object):
 
     @classmethod
     def detect(cls, peek):
+        return re.compile("http://wordpress.org/export/(1\.\d)/").search(peek)
+
 
-        m = re.search("http://wordpress.org/export/1\.\d/", peek)
-        if m:
-            return m.group(0)
+def autodetect(peek):
 
-        return None
+    if 'xmlns="http://disqus.com' in peek:
+        return Disqus
+
+    m = WordPress.detect(peek)
+    if m:
+        return WordPress
+
+    return None
 
 
 def dispatch(type, db, dump):
@@ -255,20 +258,15 @@ def dispatch(type, db, dump):
             if input("Isso DB is not empty! Continue? [y/N]: ") not in ("y", "Y"):
                 raise SystemExit("Abort.")
 
-        if type is None:
-
+        if type == "disqus":
+            cls = Disqus
+        elif type == "wordpress":
+            cls = WordPress
+        else:
             with io.open(dump) as fp:
-                peek = fp.read(io.DEFAULT_BUFFER_SIZE)
+                cls = autodetect(fp.read(io.DEFAULT_BUFFER_SIZE))
 
-            if WordPress.detect(peek):
-                type = "wordpress"
-
-            if Disqus.detect(peek):
-                type = "disqus"
-
-        if type == "wordpress":
-            WordPress(db, dump).migrate()
-        elif type == "disqus":
-            Disqus(db, dump).migrate()
-        else:
+        if cls is None:
             raise SystemExit("Unknown format, abort.")
+
+        cls(db, dump).migrate()
diff --git a/isso/tests/test_migration.py b/isso/tests/test_migration.py
index 857eff1..4b08b94 100644
--- a/isso/tests/test_migration.py
+++ b/isso/tests/test_migration.py
@@ -13,7 +13,7 @@ from os.path import join, dirname
 from isso.core import Config
 
 from isso.db import SQLite3
-from isso.migrate import Disqus, WordPress
+from isso.migrate import Disqus, WordPress, autodetect
 
 
 class TestMigration(unittest.TestCase):
@@ -82,14 +82,13 @@ class TestMigration(unittest.TestCase):
                     xmlns:dc="http://purl.org/dc/elements/1.1/"
                     xmlns:wp="http://wordpress.org/export/%s/">"""
 
-        self.assertEqual(WordPress.detect(wp % "invalid"), None)
+        self.assertEqual(autodetect(wp % "foo"), None)
 
         for version in ("1.0", "1.1", "1.2", "1.3"):
-            self.assertEqual(WordPress.detect(wp % version),
-                             "http://wordpress.org/export/%s/" % version)
+            self.assertEqual(autodetect(wp % version), WordPress)
 
         dq = '''\
         <?xml version="1.0"?>
         <disqus xmlns="http://disqus.com"
                 xmlns:dsq="http://disqus.com/disqus-internals"'''
-        self.assertIsNotNone(Disqus.detect(dq))
+        self.assertEqual(autodetect(dq), Disqus)
diff --git a/isso/tests/wordpress.xml b/isso/tests/wordpress.xml
index 3d62987..7560aef 100644
--- a/isso/tests/wordpress.xml
+++ b/isso/tests/wordpress.xml
@@ -141,4 +141,4 @@
             <link>http://example.tld/?p=6</link>
         </item>
     </channel>
-</rss>
\ No newline at end of file
+</rss>