From 7e37713c95bcdb0630b7deab92142583977c817e Mon Sep 17 00:00:00 2001
From: j <j@mailb.org>
Date: Tue, 3 Nov 2015 23:36:19 +0100
Subject: [PATCH] better metadata lookup

---
 oml/meta/__init__.py     | 23 ++++++++++++++++-------
 oml/meta/abebooks.py     |  2 +-
 oml/meta/lookupbyisbn.py |  5 +++--
 3 files changed, 20 insertions(+), 10 deletions(-)

diff --git a/oml/meta/__init__.py b/oml/meta/__init__.py
index 97805cb..f42fa29 100644
--- a/oml/meta/__init__.py
+++ b/oml/meta/__init__.py
@@ -21,6 +21,7 @@ providers = [
     ('openlibrary', 'olid'),
     ('loc', 'lccn'),
     ('worldcat', 'oclc'),
+    ('worldcat', 'isbn'),
     ('lookupbyisbn', 'asin'),
     ('lookupbyisbn', 'isbn'),
     ('abebooks', 'isbn')
@@ -36,23 +37,31 @@ def find(query):
     '''
     return results
 
+def lookup_provider(arg):
+    provider, id, ids, key, value = arg
+    values = set()
+    for key, value in ids:
+        if key == id or provider in ('openlibrary', ):
+            for kv in globals()[provider].get_ids(key, value):
+                values.add(kv)
+    return values
+
 def lookup(key, value):
     if not isvalid_id(key, value):
         return {}
     data = {key: [value]}
-    ids = [(key, value)]
+    ids = set([(key, value)])
     provider_data = {}
     done = False
+
     while not done:
         done = True
         for provider, id in providers:
-            for key, value in ids:
-                for kv in globals()[provider].get_ids(key, value):
-                    if not kv in ids:
-                        ids.append(kv)
-                        done = False
+            result = lookup_provider((provider, id, ids, key, value))
+            done = not result - ids
+            ids.update(result)
     logger.debug('FIXME: sort ids')
-    ids.sort(key=lambda i: ox.sort_string(''.join(i)))
+    ids = sorted(ids, key=lambda i: ox.sort_string(''.join(i)))
     logger.debug('IDS %s', ids)
     for k, v in ids:
         for provider, id in providers:
diff --git a/oml/meta/abebooks.py b/oml/meta/abebooks.py
index ebb1352..b7054f6 100644
--- a/oml/meta/abebooks.py
+++ b/oml/meta/abebooks.py
@@ -15,7 +15,7 @@ base = 'http://www.abebooks.com'
 def get_ids(key, value):
     ids = []
     if key == 'isbn':
-        url = '%s/servlet/SearchResults?isbn=%s&sts=t' % (base, id)
+        url = '%s/servlet/SearchResults?isbn=%s&sts=t' % (base, value)
         data = read_url(url, unicode=True)
         urls = re.compile('href="(/servlet/BookDetailsPL[^"]+)"').findall(data)
         if urls:
diff --git a/oml/meta/lookupbyisbn.py b/oml/meta/lookupbyisbn.py
index 2eb5365..2a38174 100644
--- a/oml/meta/lookupbyisbn.py
+++ b/oml/meta/lookupbyisbn.py
@@ -31,7 +31,7 @@ def get_ids(key, value):
         m = re.compile('href="(/Lookup/Book/[^"]+?)"').findall(data)
         if m:
             asin = m[0].split('/')[-3]
-            if not stdnum.isbn.is_valid(asin):
+            if stdnum.isbn.to_isbn10(asin) or not stdnum.isbn.is_valid(asin):
                 ids.append(('asin', asin))
     if key == 'isbn':
         add_other_isbn(value)
@@ -89,5 +89,6 @@ def lookup(id):
     return r
 
 def amazon_lookup(asin):
-    html = read_url('http://www.amazon.com/dp/%s' % asin).decode('utf-8', 'ignore')
+    url = 'http://www.amazon.com/dp/%s' % asin
+    html = read_url(url, timeout=-1).decode('utf-8', 'ignore')
     return list(set(find_isbns(find_re(html, 'Formats</h3>.*?</table'))))