Amend domain matcher with returning array of all matches #94
Conversation
| ips := []net.Address{} | ||
| for _, id := range indices { | ||
| ips = append(ips, h.ips[id]...) | ||
| } |
There was a problem hiding this comment.
这里DNS的static hosts现在若匹配到多个条目，会将IP整合到一起返回，不知是否合适。
| idx := s.domainMatcher.Match(domain) | ||
| if idx > 0 { | ||
| indices := s.domainMatcher.Match(domain) | ||
| for _, idx := range indices { |
There was a problem hiding this comment.
这里由于s.domainMatcher会返回多个匹配结果了，因此现在的Priority Matching可以顺序匹配多次，而非匹配一次失败后就直接Fallback到默认的轮询。
| }, | ||
| { | ||
| Domain: "x.y.com", | ||
| Result: []uint32{4, 6}, |
There was a problem hiding this comment.
DomainMatcher新增了"x.y.com"匹配到两个结果（DNS服务器）的用例。
| }, | ||
| { | ||
| Domain: "x.y.com", | ||
| Result: []uint32{4, 6}, |
There was a problem hiding this comment.
FullMatcher同样新增了"x.y.com"匹配到两个结果（DNS服务器）的用例。
| result = append(result, g.fullMatcher.Match(pattern)...) | ||
| result = append(result, g.domainMatcher.Match(pattern)...) | ||
| for _, e := range g.otherMatchers { | ||
| if e.m.Match(pattern) { | ||
| return e.id | ||
| result = append(result, e.id) | ||
| } | ||
| } |
There was a problem hiding this comment.
MatcherGroup现在会将fullMatcher, domainMatcher, otherMatchers全部尝试匹配一遍，并将结果整合在一起返回。返回的数组的优先级仍是 完整匹配 -> 子域名匹配 -> 正则匹配等。
由于domainMatcher设置时，fullMatcher与otherMatchers常常为空，反之亦然，因此全部尝试匹配对性能的影响不是太大。
| { // Will match server 1,2 and server 1 returns expected ip | ||
| ips, err := client.LookupIP("google.com") | ||
| if err != nil { | ||
| t.Fatal("unexpected error: ", err) | ||
| } | ||
|
|
||
| if r := cmp.Diff(ips, []net.IP{{8, 8, 8, 8}}); r != "" { | ||
| t.Fatal(r) | ||
| } | ||
| } | ||
|
|
||
| { // Will match server 1,2 and server 1 returns unexpected ip, then server 2 returns expected one | ||
| clientv4 := client.(feature_dns.IPv4Lookup) | ||
| ips, err := clientv4.LookupIPv4("ipv6.google.com") | ||
| if err != nil { | ||
| t.Fatal("unexpected error: ", err) | ||
| } | ||
|
|
||
| if r := cmp.Diff(ips, []net.IP{{8, 8, 8, 7}}); r != "" { | ||
| t.Fatal(r) | ||
| } | ||
| } | ||
|
|
||
| { // Will match server 1,2,3 and server 1,2 returns unexpected ip, then server 3 returns expected one | ||
| ips, err := client.LookupIP("api.google.com") | ||
| if err != nil { | ||
| t.Fatal("unexpected error: ", err) | ||
| } | ||
|
|
||
| if r := cmp.Diff(ips, []net.IP{{8, 8, 7, 7}}); r != "" { | ||
| t.Fatal(r) | ||
| } | ||
| } |
There was a problem hiding this comment.
DNS测试添加了三个用例：
- 查询google.com，会匹配DNS 1和2，DNS 1返回的IP是Expected的，直接返回；
- 查询ipv6.google.com，会匹配DNS 1和2，DNS 1返回的IP不是Expected的，这时并不直接Fallback到默认轮训，而是继续查询DNS 2，收到Expected IP并返回。
- 查询api.google.com，
会匹配DNS 1, 2和3，DNS 1和2返回的IP均不是Expected的，一直以Priority Matching查询到DNS 3，得到Expected IP并返回。由于Full Matcher优先级更高，会直接匹配DNS 3
| current.value = value | ||
| current.values = append(current.values, value) |
There was a problem hiding this comment.
DomainMatcher的Trie树的结点不再以新值覆盖旧值，而是添加进值数组里。旧值在前，代表更高的优先级。
| return m.matchers.Match(domain) > 0 | ||
| return len(m.matchers.Match(domain)) > 0 |
There was a problem hiding this comment.
路由规则现在变为查询该规则的Trie树后，返回的数组不为空即匹配。
Closes #92
采用了#92中的方案二：
MatcherGroup返回所有匹配的记录，DomainMatcher保留所有插入的记录。