Support post filter

Signed-off-by: chasingegg <[email protected]>
milvus-io · Nov 8, 2024 · 82aa236 · 82aa236
1 parent ebc3c82
commit 82aa236
Show file tree

Hide file tree

Showing 59 changed files with 6,177 additions and 1,560 deletions.
diff --git a/internal/core/src/common/Chunk.cpp b/internal/core/src/common/Chunk.cpp
@@ -27,6 +27,25 @@ StringChunk::StringViews() {
     return {ret, valid_};
 }
 
+std::pair<std::vector<std::string_view>, FixedVector<bool>>
+StringChunk::ViewsByOffsets(const FixedVector<int64_t>& offsets) {
+    std::vector<std::string_view> ret;
+    FixedVector<bool> valid_res;
+    size_t size = offsets.size();
+    ret.reserve(size);
+    valid_res.reserve(size);
+    for (auto i = 0; i < size; ++i) {
+        uint32_t string_size;
+        char* pos = data_;
+        pos += offsets_[offsets[i]];
+        string_size = *reinterpret_cast<uint32_t*>(pos);
+        pos += sizeof(uint32_t);
+        ret.emplace_back(std::string_view(pos, string_size));
+        valid_res.emplace_back(isValid(offsets[i]));
+    }
+    return {ret, valid_res};
+}
+
 void
 ArrayChunk::ConstructViews() {
     views_.reserve(row_nums_);

diff --git a/internal/core/src/common/Chunk.h b/internal/core/src/common/Chunk.h
@@ -170,6 +170,9 @@ class StringChunk : public Chunk {
         return result;
     }
 
+    std::pair<std::vector<std::string_view>, FixedVector<bool>>
+    ViewsByOffsets(const FixedVector<int64_t>& offsets);
+
     const char*
     ValueAt(int64_t idx) const override {
         return (*this)[idx].data();

diff --git a/internal/core/src/common/Consts.h b/internal/core/src/common/Consts.h
@@ -47,6 +47,7 @@ const char KMEANS_CLUSTER[] = "KMEANS";
 const char VEC_OPT_FIELDS[] = "opt_fields";
 const char PAGE_RETAIN_ORDER[] = "page_retain_order";
 const char TEXT_LOG_ROOT_PATH[] = "text_log";
+const char POST_FILTER[] = "post_filter";
 
 const char DEFAULT_PLANNODE_ID[] = "0";
 const char DEAFULT_QUERY_ID[] = "0";

diff --git a/internal/core/src/common/QueryInfo.h b/internal/core/src/common/QueryInfo.h
@@ -35,6 +35,7 @@ struct SearchInfo {
     std::optional<FieldId> group_by_field_id_;
     tracer::TraceContext trace_ctx_;
     bool materialized_view_involved = false;
+    bool post_filter_execution = false;
 };
 
 using SearchInfoPtr = std::shared_ptr<SearchInfo>;

diff --git a/internal/core/src/exec/Driver.cpp b/internal/core/src/exec/Driver.cpp
@@ -23,6 +23,7 @@
 #include "exec/operator/CallbackSink.h"
 #include "exec/operator/CountNode.h"
 #include "exec/operator/FilterBitsNode.h"
+#include "exec/operator/FilterNode.h"
 #include "exec/operator/MvccNode.h"
 #include "exec/operator/Operator.h"
 #include "exec/operator/VectorSearchNode.h"
@@ -52,11 +53,16 @@ DriverFactory::CreateDriver(std::unique_ptr<DriverContext> ctx,
     for (size_t i = 0; i < plannodes_.size(); ++i) {
         auto id = operators.size();
         auto plannode = plannodes_[i];
-        if (auto filternode =
+        if (auto filterbitsnode =
                 std::dynamic_pointer_cast<const plan::FilterBitsNode>(
                     plannode)) {
+            operators.push_back(std::make_unique<PhyFilterBitsNode>(
+                id, ctx.get(), filterbitsnode));
+        } else if (auto filternode =
+                       std::dynamic_pointer_cast<const plan::FilterNode>(
+                           plannode)) {
             operators.push_back(
-                std::make_unique<PhyFilterBitsNode>(id, ctx.get(), filternode));
+                std::make_unique<PhyFilterNode>(id, ctx.get(), filternode));
         } else if (auto mvccnode =
                        std::dynamic_pointer_cast<const plan::MvccNode>(
                            plannode)) {

diff --git a/internal/core/src/exec/QueryContext.h b/internal/core/src/exec/QueryContext.h
@@ -230,6 +230,11 @@ class QueryContext : public Context {
         return search_info_;
     }
 
+    knowhere::MetricType
+    get_metric_type() {
+        return search_info_.metric_type_;
+    }
+
     const query::PlaceholderGroup*
     get_placeholder_group() {
         return placeholder_group_;

diff --git a/internal/core/src/exec/expression/AlwaysTrueExpr.cpp b/internal/core/src/exec/expression/AlwaysTrueExpr.cpp
@@ -21,9 +21,13 @@ namespace exec {
 
 void
 PhyAlwaysTrueExpr::Eval(EvalCtx& context, VectorPtr& result) {
-    int64_t real_batch_size = current_pos_ + batch_size_ >= active_count_
-                                  ? active_count_ - current_pos_
-                                  : batch_size_;
+    auto input = context.get_offset_input();
+    has_offset_input_ = (input != nullptr);
+    int64_t real_batch_size = (has_offset_input_)
+                                  ? input->size()
+                                  : (current_pos_ + batch_size_ >= active_count_
+                                         ? active_count_ - current_pos_
+                                         : batch_size_);
 
     // always true no need to skip null
     if (real_batch_size == 0) {

diff --git a/internal/core/src/exec/expression/AlwaysTrueExpr.h b/internal/core/src/exec/expression/AlwaysTrueExpr.h
@@ -47,11 +47,14 @@ class PhyAlwaysTrueExpr : public Expr {
 
     void
     MoveCursor() override {
-        int64_t real_batch_size = current_pos_ + batch_size_ >= active_count_
-                                      ? active_count_ - current_pos_
-                                      : batch_size_;
+        if (!has_offset_input_) {
+            int64_t real_batch_size =
+                current_pos_ + batch_size_ >= active_count_
+                    ? active_count_ - current_pos_
+                    : batch_size_;
 
-        current_pos_ += real_batch_size;
+            current_pos_ += real_batch_size;
+        }
     }
 
  private: