This commit is contained in:
Nick Lang
2019-05-10 09:16:33 -06:00
committed by GitHub
parent 01e62965a1
commit 206f5e2754
34 changed files with 1300 additions and 826 deletions
+66 -69
View File
@@ -6,95 +6,92 @@ from dateutil.parser import parse as parse_date
from elasticsearch import Elasticsearch
def print_search_stats(results):
print('=' * 80)
print('Total %d found in %dms' % (results['hits']['total'], results['took']))
print('-' * 80)
print("=" * 80)
print("Total %d found in %dms" % (results["hits"]["total"], results["took"]))
print("-" * 80)
def print_hits(results):
" Simple utility function to print results of a search query. "
print_search_stats(results)
for hit in results['hits']['hits']:
for hit in results["hits"]["hits"]:
# get created date for a repo and fallback to authored_date for a commit
created_at = parse_date(hit['_source'].get('created_at', hit['_source']['authored_date']))
print('/%s/%s/%s (%s): %s' % (
hit['_index'], hit['_type'], hit['_id'],
created_at.strftime('%Y-%m-%d'),
hit['_source']['description'].split('\n')[0]))
created_at = parse_date(
hit["_source"].get("created_at", hit["_source"]["authored_date"])
)
print(
"/%s/%s/%s (%s): %s"
% (
hit["_index"],
hit["_type"],
hit["_id"],
created_at.strftime("%Y-%m-%d"),
hit["_source"]["description"].split("\n")[0],
)
)
print('=' * 80)
print("=" * 80)
print()
# get trace logger and set level
tracer = logging.getLogger('elasticsearch.trace')
tracer = logging.getLogger("elasticsearch.trace")
tracer.setLevel(logging.INFO)
tracer.addHandler(logging.FileHandler('/tmp/es_trace.log'))
tracer.addHandler(logging.FileHandler("/tmp/es_trace.log"))
# instantiate es client, connects to localhost:9200 by default
es = Elasticsearch()
print('Empty search:')
print_hits(es.search(index='git'))
print("Empty search:")
print_hits(es.search(index="git"))
print('Find commits that says "fix" without touching tests:')
result = es.search(
index='git',
doc_type='doc',
index="git",
doc_type="doc",
body={
'query': {
'bool': {
'must': {
'match': {'description': 'fix'}
},
'must_not': {
'term': {'files': 'test_elasticsearch'}
}
}
}
}
)
print_hits(result)
print('Last 8 Commits for elasticsearch-py:')
result = es.search(
index='git',
doc_type='doc',
body={
'query': {
'term': {
'repository': 'elasticsearch-py'
}
},
'sort': [
{'committed_date': {'order': 'desc'}}
],
'size': 8
}
)
print_hits(result)
print('Stats for top 10 committers:')
result = es.search(
index='git',
doc_type='doc',
body={
'size': 0,
'aggs': {
'committers': {
'terms': {
'field': 'committer.name.keyword',
},
'aggs': {
'line_stats': {
'stats': {'field': 'stats.lines'}
"query": {
"bool": {
"must": {"match": {"description": "fix"}},
"must_not": {"term": {"files": "test_elasticsearch"}},
}
}
}
}
}
},
)
print_hits(result)
print("Last 8 Commits for elasticsearch-py:")
result = es.search(
index="git",
doc_type="doc",
body={
"query": {"term": {"repository": "elasticsearch-py"}},
"sort": [{"committed_date": {"order": "desc"}}],
"size": 8,
},
)
print_hits(result)
print("Stats for top 10 committers:")
result = es.search(
index="git",
doc_type="doc",
body={
"size": 0,
"aggs": {
"committers": {
"terms": {"field": "committer.name.keyword"},
"aggs": {"line_stats": {"stats": {"field": "stats.lines"}}},
}
},
},
)
print_search_stats(result)
for committer in result['aggregations']['committers']['buckets']:
print('%15s: %3d commits changing %6d lines' % (
committer['key'], committer['doc_count'], committer['line_stats']['sum']))
print('=' * 80)
for committer in result["aggregations"]["committers"]["buckets"]:
print(
"%15s: %3d commits changing %6d lines"
% (committer["key"], committer["doc_count"], committer["line_stats"]["sum"])
)
print("=" * 80)