Numereous stability fixes: recover from HTTP and FeedParser errors

planet/feedparser.py

@@ -342,6 +342,10 @@ _cp1252 = {
 _urifixer = re.compile('^([A-Za-z][A-Za-z0-9+-.]*://)(/*)(.*?)')
 def _urljoin(base, uri):
     uri = _urifixer.sub(r'\1\3', uri)
+    try:
+        return urlparse.urljoin(base, uri)
+    except:
+        uri = urlparse.urlunparse([urllib.quote(part) for part in urlparse.urlparse(uri)])
         return urlparse.urljoin(base, uri)
 
 class _FeedParserMixin:

planet/reconstitute.py

@@ -188,14 +188,14 @@ def source(xsource, source, bozo, format):
 
     date(xsource, 'updated', source.get('updated_parsed',time.gmtime()))
 
+    if format: source['planet_format'] = format
+    if not bozo == None: source['planet_bozo'] = bozo and 'true' or 'false'
+
     # propagate planet inserted information
     for key, value in source.items():
         if key.startswith('planet_'):
             createTextElement(xsource, key.replace('_',':',1), value)
 
-    createTextElement(xsource, 'planet:bozo', bozo and 'true' or 'false')
-    createTextElement(xsource, 'planet:format', format)
-
 def reconstitute(feed, entry):
     """ create an entry document from a parsed feed """
     xdoc=minidom.parseString('<entry xmlns="http://www.w3.org/2005/Atom"/>\n')

planet/spider.py

@@ -99,6 +99,7 @@ def scrub(feed, data):
 
 def spiderFeed(feed):
     """ Spider (fetch) a single feed """
+    log = planet.logger
 
     # read cached feed info
     sources = config.cache_sources_directory()
@@ -125,8 +126,10 @@ def spiderFeed(feed):
         else:
             data.status = 500
 
+    activity_horizon = \
+        time.gmtime(time.time()-86400*config.activity_threshold(feed))
+
     # process based on the HTTP status code
-    log = planet.logger
     if data.status == 200 and data.has_key("url"):
         data.feed['planet_http_location'] = data.url
     elif data.status == 301 and data.has_key("entries") and len(data.entries)>0:
@@ -134,8 +137,17 @@ def spiderFeed(feed):
         data.feed['planet_http_location'] = data.url
     elif data.status == 304:
         log.info("Feed %s unchanged", feed)
-        if not feed_info.feed.has_key('planet_message'): return
+
+        if not feed_info.feed.has_key('planet_message'):
+            if feed_info.feed.has_key('planet_updated'):
+                updated = feed_info.feed.planet_updated
+                if feedparser._parse_date_iso8601(updated) >= activity_horizon:
+                    return
+        else:
+            if feed_info.feed.planet_message.startswith("no activity in"):
+               return
             del feed_info.feed['planet_message']
+
     elif data.status == 410:
         log.info("Feed %s gone", feed)
     elif data.status == 408:
@@ -146,7 +158,9 @@ def spiderFeed(feed):
         log.info("Updating feed %s", feed)
 
     # if read failed, retain cached information
-    if not data.version and feed_info.version: data.feed = feed_info.feed
+    if not data.version and feed_info.version:
+        data.feed = feed_info.feed
+        data.bozo = feed_info.feed.get('planet_bozo','true') == 'true'
     data.feed['planet_http_status'] = str(data.status)
 
     # capture etag and last-modified information
@@ -212,8 +226,6 @@ def spiderFeed(feed):
     
     # identify inactive feeds
     if config.activity_threshold(feed):
-        activity_horizon = \
-            time.gmtime(time.time()-86400*config.activity_threshold(feed))
         updated = [entry.updated_parsed for entry in data.entries
             if entry.has_key('updated_parsed')]
         updated.sort()
@@ -230,7 +242,9 @@ def spiderFeed(feed):
             data.feed['planet_message'] = msg
 
     # report channel level errors
-    if data.status == 403:
+    if data.status == 226:
+        if data.feed.has_key('planet_message'): del data.feed['planet_message']
+    elif data.status == 403:
         data.feed['planet_message'] = "403: forbidden"
     elif data.status == 404:
         data.feed['planet_message'] = "404: not found"
@@ -257,4 +271,12 @@ def spiderPlanet():
     planet.setTimeout(config.feed_timeout())
 
     for feed in config.subscriptions():
+        try:
             spiderFeed(feed)
+        except Exception,e:
+            import sys, traceback
+            type, value, tb = sys.exc_info()
+            log.error('Error processing %s', feed)
+            for line in (traceback.format_exception_only(type, value) +
+                traceback.format_tb(tb)):
+                log.error(line.rstrip())

tests/test_filters.py

@@ -78,11 +78,19 @@ class FilterTests(unittest.TestCase):
 
 try:
     from subprocess import Popen, PIPE
+
     sed=Popen(['sed','--version'],stdout=PIPE,stderr=PIPE)
     sed.communicate()
     if sed.returncode != 0:
         logger.warn("sed is not available => can't test stripAd_yahoo")
         del FilterTests.test_stripAd_yahoo
+
+    try:
+        import libxml2
+    except:
+        logger.warn("libxml2 is not available => can't test xpath_sifter")
+        del FilterTests.test_xpath_filter
+
 except ImportError:
     logger.warn("Popen is not available => can't test filters")
     for method in dir(FilterTests):