From 92e451ad68d33919623a2dfaf4d306ae1fb30786 Mon Sep 17 00:00:00 2001
From: Ryan Hitchman <hitchmanr@gmail.com>
Date: Thu, 14 May 2009 22:00:48 -0600
Subject: [PATCH] google search plugin (same module as google image search)

---
 plugins/gis.py    | 23 -------------------
 plugins/google.py | 56 +++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 56 insertions(+), 23 deletions(-)
 delete mode 100644 plugins/gis.py
 create mode 100644 plugins/google.py

diff --git a/plugins/gis.py b/plugins/gis.py
deleted file mode 100644
index 3a1c0fc..0000000
--- a/plugins/gis.py
+++ /dev/null
@@ -1,23 +0,0 @@
-import yaml
-import urllib
-
-import hook
-
-
-@hook.command
-def gis(inp):
-    '''.gis <term> -- returns first google image result (safesearch off)'''
-    if not inp:
-        return gis.__doc__
-
-    req_url = 'http://ajax.googleapis.com/ajax/services/search/images?' \
-            'v=1.0&safe=off&q='
-    url = req_url + urllib.quote(inp, safe='')
-    json = urllib.urlopen(url).read()
-    parsed = yaml.load(json)
-    if not 200 <= parsed['responseStatus'] < 300:
-        raise IOError('error searching for images: %d: %s' % (
-                parsed['responseStatus'], ''))
-    if not parsed['responseData']['results']:
-        return 'no images found'
-    return parsed['responseData']['results'][0]['unescapedUrl']
diff --git a/plugins/google.py b/plugins/google.py
new file mode 100644
index 0000000..37c03bb
--- /dev/null
+++ b/plugins/google.py
@@ -0,0 +1,56 @@
+import yaml
+import urllib
+from lxml import html
+
+import hook
+
+
+def api_get(kind, query):
+    req_url = 'http://ajax.googleapis.com/ajax/services/search/%s?' \
+            'v=1.0&safe=off&q=%s'
+    url = req_url % (kind, urllib.quote(query, safe=''))
+    json = urllib.urlopen(url).read()
+    return yaml.load(json)
+
+
+@hook.command
+def gis(inp):
+    '''.gis <term> -- returns first google image result (safesearch off)'''
+    if not inp:
+        return gis.__doc__
+
+    parsed = api_get('images', inp)
+    if not 200 <= parsed['responseStatus'] < 300:
+        raise IOError('error searching for images: %d: %s' % (
+                parsed['responseStatus'], ''))
+    if not parsed['responseData']['results']:
+        return 'no images found'
+    return parsed['responseData']['results'][0]['unescapedUrl']
+
+@hook.command
+@hook.command('g')
+@hook.command('goog')
+def google(inp):
+    '''.g/.goog/.google <query> -- returns first google search result'''
+    if not inp:
+        return google.__doc__
+
+    parsed = api_get('web', inp)
+    if not 200 <= parsed['responseStatus'] < 300:
+        raise IOError('error searching for pages: %d: %s' % (
+                parsed['responseStatus'], ''))
+    if not parsed['responseData']['results']:
+        return 'no results found'
+
+    result = parsed['responseData']['results'][0]
+
+    title, content = map(lambda x: html.fromstring(x).text_content(),
+            (result['titleNoFormatting'], result['content']))
+    
+    out = '%s -- \x02%s\x02: "%s"' % (result['unescapedUrl'], title, content)
+    out = ' '.join(out.split())
+
+    if len(out) > 300:
+        out = out[:out.rfind(' ')] + '..."'
+
+    return out