def testParse(self):
query_text = """
SELECT
'comment' AS type,
date_added, post.title AS title,
keyvalue.value || ': ' || comment.body AS body,
'?Post:' || comment.post_id || '#Comments' AS url
FROM
comment
LEFT JOIN json USING (json_id)
LEFT JOIN json AS json_content ON (json_content.directory = json.directory AND json_content.file_name='content.json')
LEFT JOIN keyvalue ON (keyvalue.json_id = json_content.json_id AND key = 'cert_user_id')
LEFT JOIN post ON (comment.post_id = post.post_id)
WHERE
post.date_added > 123
ORDER BY
date_added DESC
LIMIT 20
"""
query = DbQuery(query_text)
assert query.parts["LIMIT"] == "20"
assert query.fields["body"] == "keyvalue.value || ': ' || comment.body"
assert re.sub("[ \r\n]", "",
str(query)) == re.sub("[ \r\n]", "", query_text)
query.wheres.append("body LIKE '%hello%'")
assert "body LIKE '%hello%'" in str(query)
def actionFeedQuery(self, to, limit=10, day_limit=3):
from Site import SiteManager
rows = []
stats = []
total_s = time.time()
num_sites = 0
for address, site_data in list(self.user.sites.items()):
feeds = site_data.get("follow")
if not feeds:
continue
if type(feeds) is not dict:
self.log.debug("Invalid feed for site %s" % address)
continue
num_sites += 1
for name, query_set in feeds.items():
site = SiteManager.site_manager.get(address)
if not site or not site.storage.has_db:
continue
s = time.time()
try:
query_raw, params = query_set
query_parts = re.split(r"UNION(?:\s+ALL|)", query_raw)
for i, query_part in enumerate(query_parts):
db_query = DbQuery(query_part)
if day_limit:
where = " WHERE %s > strftime('%%s', 'now', '-%s day')" % (
db_query.fields.get("date_added",
"date_added"), day_limit)
if "WHERE" in query_part:
query_part = re.sub(
"WHERE (.*?)(?=$| GROUP BY)",
where + " AND (\\1)", query_part)
else:
query_part += where
query_parts[i] = query_part
query = " UNION ".join(query_parts)
if ":params" in query:
query_params = map(helper.sqlquote, params)
query = query.replace(":params",
",".join(query_params))
res = site.storage.query(
query + " ORDER BY date_added DESC LIMIT %s" % limit)
except Exception as err: # Log error
self.log.error("%s feed query %s error: %s" %
(address, name, Debug.formatException(err)))
stats.append({
"site": site.address,
"feed_name": name,
"error": str(err)
})
continue
for row in res:
row = dict(row)
if not isinstance(row["date_added"],
(int, float, complex)):
self.log.debug("Invalid date_added from site %s: %r" %
(address, row["date_added"]))
continue
if row["date_added"] > 1000000000000: # Formatted as millseconds
row["date_added"] = row["date_added"] / 1000
if "date_added" not in row or row[
"date_added"] > time.time() + 120:
self.log.debug(
"Newsfeed item from the future from from site %s" %
address)
continue # Feed item is in the future, skip it
row["site"] = address
row["feed_name"] = name
rows.append(row)
stats.append({
"site": site.address,
"feed_name": name,
"taken": round(time.time() - s, 3)
})
time.sleep(0.001)
return self.response(
to, {
"rows": rows,
"stats": stats,
"num": len(rows),
"sites": num_sites,
"taken": round(time.time() - total_s, 3)
})
def actionFeedSearch(self, to, search, limit=30, day_limit=30):
if "ADMIN" not in self.site.settings["permissions"]:
return self.response(to, "FeedSearch not allowed")
from Site import SiteManager
rows = []
stats = []
num_sites = 0
total_s = time.time()
search_text, filters = self.parseSearch(search)
for address, site in SiteManager.site_manager.list().items():
if not site.storage.has_db:
continue
if "site" in filters:
if filters["site"].lower() not in [
site.address,
site.content_manager.contents["content.json"].get(
"title").lower()
]:
continue
if site.storage.db: # Database loaded
feeds = site.storage.db.schema.get("feeds")
else:
try:
feeds = site.storage.loadJson("dbschema.json").get("feeds")
except:
continue
if not feeds:
continue
num_sites += 1
for name, query in feeds.items():
s = time.time()
try:
db_query = DbQuery(query)
params = []
# Filters
if search_text:
db_query.wheres.append("(%s LIKE ? OR %s LIKE ?)" %
(db_query.fields["body"],
db_query.fields["title"]))
search_like = "%" + search_text.replace(" ", "%") + "%"
params.append(search_like)
params.append(search_like)
if filters.get("type") and filters["type"] not in query:
continue
if day_limit:
db_query.wheres.append(
"%s > strftime('%%s', 'now', '-%s day')" %
(db_query.fields.get("date_added",
"date_added"), day_limit))
# Order
db_query.parts["ORDER BY"] = "date_added DESC"
db_query.parts["LIMIT"] = str(limit)
res = site.storage.query(str(db_query), params)
except Exception as err:
self.log.error("%s feed query %s error: %s" %
(address, name, Debug.formatException(err)))
stats.append({
"site": site.address,
"feed_name": name,
"error": str(err),
"query": query
})
continue
for row in res:
row = dict(row)
if not row["date_added"] or row["date_added"] > time.time(
) + 120:
continue # Feed item is in the future, skip it
row["site"] = address
row["feed_name"] = name
rows.append(row)
stats.append({
"site": site.address,
"feed_name": name,
"taken": round(time.time() - s, 3)
})
return self.response(
to, {
"rows": rows,
"num": len(rows),
"sites": num_sites,
"taken": round(time.time() - total_s, 3),
"stats": stats
})