cherry pick from nebula : full text (guotai junan) (vesoft-inc#2825)

* Ft search (vesoft-inc#5567) * Add optimizer rule for pushing limit into the fulltext index scan (vesoft-inc#5575) * Merge limit and full text index scan rule (vesoft-inc#5577) * Support to return score from es query (vesoft-inc#5580) * modify ft search (vesoft-inc#5584) * modify ft search * fix bug * address some comment * fix bug * fix conflict --------- Co-authored-by: hs.zhang <22708345+cangfengzhs@users.noreply.github.com> Co-authored-by: Yee <2520865+yixinglu@users.noreply.github.com>
yixinglu · Jun 12, 2023 · 4dff80e · 4dff80e
1 parent a832d57
commit 4dff80e
Show file tree

Hide file tree

Showing 58 changed files with 1,145 additions and 1,608 deletions.
diff --git a/src/clients/meta/MetaClient.cpp b/src/clients/meta/MetaClient.cpp
@@ -1370,7 +1370,7 @@ Status MetaClient::handleResponse(const RESP& resp) {
  case nebula::cpp2::ErrorCode::E_HOST_CAN_NOT_BE_ADDED:
  return Status::Error("Could not add a host, which is not a storage and not expired either");
  default:
- return Status::Error("Unknown error!");
+ return Status::Error("Unknown error %d!", static_cast<int>(resp.get_code()));
  }
 }
 

diff --git a/src/common/expression/Expression.cpp b/src/common/expression/Expression.cpp
@@ -512,10 +512,7 @@ Expression* Expression::decode(ObjectPool* pool, Expression::Decoder& decoder) {
  exp->resetFrom(decoder);
  return exp;
  }
- case Expression::Kind::kTSPrefix:
- case Expression::Kind::kTSWildcard:
- case Expression::Kind::kTSRegexp:
- case Expression::Kind::kTSFuzzy: {
+ case Expression::Kind::kESQUERY: {
  LOG(FATAL) << "Should not decode text search expression";
  return exp;
  }
@@ -721,17 +718,8 @@ std::ostream& operator<<(std::ostream& os, Expression::Kind kind) {
  case Expression::Kind::kPathBuild:
  os << "PathBuild";
  break;
- case Expression::Kind::kTSPrefix:
- os << "Prefix";
- break;
- case Expression::Kind::kTSWildcard:
- os << "Wildcard";
- break;
- case Expression::Kind::kTSRegexp:
- os << "Regexp";
- break;
- case Expression::Kind::kTSFuzzy:
- os << "Fuzzy";
+ case Expression::Kind::kESQUERY:
+ os << "ESQuery";
  break;
  case Expression::Kind::kListComprehension:
  os << "ListComprehension";

diff --git a/src/common/expression/Expression.h b/src/common/expression/Expression.h
@@ -96,10 +96,7 @@ class Expression {
 
  kPathBuild,
  // text or key word search expression
- kTSPrefix,
- kTSWildcard,
- kTSRegexp,
- kTSFuzzy,
+ kESQUERY,
 
  kAggregate,
  kIsNull,

diff --git a/src/common/expression/TextSearchExpression.cpp b/src/common/expression/TextSearchExpression.cpp
@@ -10,29 +10,14 @@
 namespace nebula {
 
 bool TextSearchArgument::operator==(const TextSearchArgument& rhs) const {
- return val_ == rhs.val_ && op_ == rhs.op_ && fuzziness_ == rhs.fuzziness_ &&
- limit_ == rhs.limit_ && timeout_ == rhs.timeout_;
+ return index_ == rhs.index_ && query_ == rhs.query_;
 }
 
 std::string TextSearchArgument::toString() const {
  std::string buf;
  buf.reserve(64);
- buf = from_ + "." + prop_ + ", ";
- buf += "\"" + val_ + "\"";
- if (fuzziness_ == -1) {
- buf += ", AUTO, ";
- buf += ((op_ == "or") ? "OR" : "AND");
- } else if (fuzziness_ > -1) {
- buf += ", ";
- buf += folly::stringPrintf("%d, ", fuzziness_);
- buf += ((op_ == "or") ? "OR" : "AND");
- }
- if (limit_ != -1) {
- buf += folly::stringPrintf(", %d", limit_);
- }
- if (timeout_ != -1) {
- buf += folly::stringPrintf(", %d", timeout_);
- }
+ buf += index_ + ", ";
+ buf += "\"" + query_ + "\"";
  return buf;
 }
 
@@ -49,20 +34,8 @@ std::string TextSearchExpression::toString() const {
  std::string buf;
  buf.reserve(64);
  switch (kind_) {
- case Kind::kTSWildcard: {
- buf = "WILDCARD(";
- break;
- }
- case Kind::kTSPrefix: {
- buf = "PREFIX(";
- break;
- }
- case Kind::kTSFuzzy: {
- buf = "FUZZY(";
- break;
- }
- case Kind::kTSRegexp: {
- buf = "REGEXP(";
+ case Kind::kESQUERY: {
+ buf = "ES_QUERY(";
  break;
  }
  default: {

diff --git a/src/common/expression/TextSearchExpression.h b/src/common/expression/TextSearchExpression.h
@@ -14,60 +14,19 @@ namespace nebula {
 class TextSearchArgument final {
  public:
  static TextSearchArgument* make(ObjectPool* pool,
- const std::string& from,
- const std::string& prop,
- const std::string& val) {
- return pool->makeAndAdd<TextSearchArgument>(from, prop, val);
+ const std::string& index,
+ const std::string& query) {
+ return pool->makeAndAdd<TextSearchArgument>(index, query);
  }
 
  ~TextSearchArgument() = default;
 
- void setVal(const std::string& val) {
- val_ = val;
+ std::string& index() {
+ return index_;
  }
 
- const std::string& from() {
- return from_;
- }
-
- const std::string& prop() {
- return prop_;
- }
-
- const std::string& val() const {
- return val_;
- }
-
- void setOP(const std::string& op) {
- op_ = op;
- }
-
- const std::string& op() const {
- return op_;
- }
-
- void setFuzziness(int32_t fuzz) {
- fuzziness_ = fuzz;
- }
-
- int32_t fuzziness() {
- return fuzziness_;
- }
-
- void setLimit(int32_t limit) {
- limit_ = limit;
- }
-
- int32_t limit() {
- return limit_;
- }
-
- void setTimeout(int32_t timeout) {
- timeout_ = timeout;
- }
-
- int32_t timeout() {
- return timeout_;
+ std::string& query() {
+ return query_;
  }
 
  bool operator==(const TextSearchArgument& rhs) const;
@@ -76,35 +35,18 @@ class TextSearchArgument final {
 
  private:
  friend ObjectPool;
- TextSearchArgument(const std::string& from, const std::string& prop, const std::string& val)
- : from_(from), prop_(prop), val_(val) {}
+ TextSearchArgument(const std::string& index, const std::string& query)
+ : index_(index), query_(query) {}
 
  private:
- std::string from_;
- std::string prop_;
- std::string val_;
- std::string op_;
- int32_t fuzziness_{-2};
- int32_t limit_{10000};
- int32_t timeout_{-1};
+ std::string index_;
+ std::string query_;
 };
 
 class TextSearchExpression : public Expression {
  public:
- static TextSearchExpression* makePrefix(ObjectPool* pool, TextSearchArgument* arg) {
- return pool->makeAndAdd<TextSearchExpression>(pool, Kind::kTSPrefix, arg);
- }
-
- static TextSearchExpression* makeWildcard(ObjectPool* pool, TextSearchArgument* arg) {
- return pool->makeAndAdd<TextSearchExpression>(pool, Kind::kTSWildcard, arg);
- }
-
- static TextSearchExpression* makeRegexp(ObjectPool* pool, TextSearchArgument* arg) {
- return pool->makeAndAdd<TextSearchExpression>(pool, Kind::kTSRegexp, arg);
- }
-
- static TextSearchExpression* makeFuzzy(ObjectPool* pool, TextSearchArgument* arg) {
- return pool->makeAndAdd<TextSearchExpression>(pool, Kind::kTSFuzzy, arg);
+ static TextSearchExpression* makeQuery(ObjectPool* pool, TextSearchArgument* arg) {
+ return pool->makeAndAdd<TextSearchExpression>(pool, Kind::kESQUERY, arg);
  }
 
  static TextSearchExpression* make(ObjectPool* pool, Kind kind, TextSearchArgument* arg) {
@@ -125,7 +67,7 @@ class TextSearchExpression : public Expression {
  std::string toString() const override;
 
  Expression* clone() const override {
- auto arg = TextSearchArgument::make(pool_, arg_->from(), arg_->prop(), arg_->val());
+ auto arg = TextSearchArgument::make(pool_, arg_->index(), arg_->query());
  return TextSearchExpression::make(pool_, kind_, arg);
  }
 

diff --git a/src/common/function/FunctionManager.cpp b/src/common/function/FunctionManager.cpp
@@ -436,6 +436,7 @@ std::unordered_map<std::string, std::vector<TypeSignature>> FunctionManager::typ
  {"json_extract",
  {TypeSignature({Value::Type::STRING}, Value::Type::MAP),
  TypeSignature({Value::Type::STRING}, Value::Type::NULLVALUE)}},
+ {"score", {TypeSignature({}, Value::Type::__EMPTY__)}},
 };
 
 // static
@@ -2937,6 +2938,17 @@ FunctionManager::FunctionManager() {
  return Value::kNullValue;
  };
  }
+ // Score function is used to identify the score of a full-text search
+ {
+ auto &attr = functions_["score"];
+ attr.minArity_ = 0;
+ attr.maxArity_ = 0;
+ attr.isAlwaysPure_ = true;
+ attr.body_ = [](const auto &) -> Value {
+ // Only placeholder, will be replaced by actual expression and need not to be evaluated
+ return Value::kNullValue;
+ };
+ }
 } // NOLINT
 
 // static