nidx-binding 6.3.7.post341__tar.gz → 6.3.7.post343__tar.gz

{nidx_binding-6.3.7.post341 → nidx_binding-6.3.7.post343}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nidx_binding
-Version: 6.3.7.post341
+Version: 6.3.7.post343
 Classifier: Programming Language :: Rust
 Classifier: Programming Language :: Python :: Implementation :: CPython
 Classifier: Programming Language :: Python :: Implementation :: PyPy

{nidx_binding-6.3.7.post341 → nidx_binding-6.3.7.post343}/nidx_protos/build.py

@@ -26,30 +26,30 @@ from grpc_tools import protoc
 
 def pdm_build_initialize(context):
     build_dir = context.ensure_build_dir()
-    python_dir = f"{build_dir}/nidx_protos"
-    try:
-        os.mkdir(python_dir)
-    except FileExistsError:
-        pass
 
     well_known_path = resources.files("grpc_tools") / "_proto"
 
     # Compile protos
-    for proto in [
-        "src/nidx.proto",
+    for proto, has_grpc in [
+        ("nidx_protos/nidx.proto", True),
+        ("nidx_protos/nodereader.proto", True),
+        ("nidx_protos/nodewriter.proto", True),
+        ("nidx_protos/noderesources.proto", False),
     ]:
         command = [
             "grpc_tools.protoc",
-            "--proto_path=src",
+            "--proto_path=..",
             "--proto_path=../../",
             f"--proto_path={well_known_path}",
-            f"--python_out={python_dir}",
-            f"--pyi_out={python_dir}",
-            f"--grpc_python_out={python_dir}",
+            f"--python_out={build_dir}",
+            f"--mypy_out={build_dir}",
             proto,
         ]
+        if has_grpc:
+            command.append(f"--grpc_python_out={build_dir}")
+
         if protoc.main(command) != 0:
             raise Exception("error: {} failed".format(command))
 
     # Create py.typed to enable type checking
-    open(f"{python_dir}/py.typed", "w")
+    open(f"{build_dir}/nidx_protos/py.typed", "w")

{nidx_binding-6.3.7.post341 → nidx_binding-6.3.7.post343}/nidx_protos/build.rs

@@ -24,14 +24,8 @@ fn main() -> Result<()> {
     println!("cargo:rerun-if-changed=../../nucliadb_protos");
     println!("cargo:rerun-if-changed=src");
     tonic_build::configure().emit_rerun_if_changed(false).compile_protos(
-        &[
-            "src/nidx.proto",
-            "../../nucliadb_protos/nodereader.proto",
-            "../../nucliadb_protos/noderesources.proto",
-            "../../nucliadb_protos/nodewriter.proto",
-            "../../nucliadb_protos/kb_usage.proto",
-        ],
-        &["../../", "src"],
+        &["nidx_protos/nidx.proto", "../../nucliadb_protos/kb_usage.proto"],
+        &["../../", ".."],
     )?;
     Ok(())
 }

{nidx_binding-6.3.7.post341/nidx_protos/src → nidx_binding-6.3.7.post343/nidx_protos}/nidx.proto

@@ -2,8 +2,8 @@ syntax = "proto3";
 
 package nidx;
 
-import "nucliadb_protos/nodereader.proto";
-import "nucliadb_protos/nodewriter.proto";
+import "nidx_protos/nodereader.proto";
+import "nidx_protos/nodewriter.proto";
 
 service NidxApi {
   rpc GetShard(nodereader.GetShardRequest) returns (noderesources.Shard) {}

nidx_binding-6.3.7.post343/nidx_protos/nodereader.proto

@@ -0,0 +1,473 @@
+syntax = "proto3";
+
+package nodereader;
+
+import public "nidx_protos/noderesources.proto";
+import "google/protobuf/timestamp.proto";
+import public "nucliadb_protos/utils.proto";
+
+message StreamFilter {
+    enum Conjunction {
+        AND = 0;
+        OR = 1;
+        NOT = 2;
+    }
+    Conjunction conjunction = 1;
+
+    repeated string labels = 2;
+}
+
+
+message Faceted {
+    repeated string labels = 1;
+}
+
+
+message OrderBy {
+    enum OrderType {
+      DESC = 0;
+      ASC = 1;
+    }
+    enum OrderField {
+        CREATED = 0;
+        MODIFIED = 1;
+    }
+    string field = 1 [deprecated = true];
+    OrderType type = 2;
+    OrderField sort_by = 3;
+}
+
+message FacetResult {
+    string tag = 1;
+    int32 total = 2;
+}
+message FacetResults {
+    repeated FacetResult facetresults = 1;
+}
+
+message ResultScore {
+    float bm25 = 1;
+    // In the case of two equal bm25 scores, booster
+    // decides
+    float booster = 2;
+}
+
+message DocumentResult {
+    string uuid = 1;
+    ResultScore score = 3;
+    string field = 4;
+    repeated string labels = 5;
+}
+
+message DocumentSearchResponse {
+    int32 total = 1;
+
+    repeated DocumentResult results = 2;
+
+    map<string, FacetResults> facets = 3;
+
+    int32 page_number = 4;
+    int32 result_per_page = 5;
+
+    // The text that lead to this results
+    string query = 6;
+
+    // Is there a next page
+    bool next_page = 7;
+    bool bm25 = 8;
+}
+
+message ParagraphResult {
+    string uuid = 1;
+    string field = 3;
+    uint64 start = 4;
+    uint64 end = 5;
+    string paragraph = 6;
+    string split = 7;
+    uint64 index = 8;
+    ResultScore score = 9;
+    repeated string matches = 10;
+
+    // Metadata that can't be searched with but is returned on search results
+    noderesources.ParagraphMetadata metadata = 11;
+
+    repeated string labels = 12;
+}
+
+message ParagraphSearchResponse {
+    int32 fuzzy_distance = 10;
+    int32 total = 1;
+    repeated ParagraphResult results = 2;
+
+    // For each field what facets are.
+    map<string, FacetResults> facets = 3;
+
+    // What page is the answer.
+    int32 page_number = 4;
+
+    // How many results are in this page.
+    int32 result_per_page = 5;
+
+    // The text that lead to this results
+    string query = 6;
+
+    // Is there a next page
+    bool next_page = 7;
+    bool bm25 = 8;
+    repeated string ematches = 9;
+}
+
+message DocumentVectorIdentifier {
+    string id = 1;
+}
+
+message DocumentScored {
+    DocumentVectorIdentifier doc_id = 1;
+    float score = 2;
+    noderesources.SentenceMetadata metadata = 3;
+    repeated string labels = 4;
+}
+
+message VectorSearchResponse {
+    // List of docs closer to the asked one.
+    repeated DocumentScored documents = 1;
+
+    // What page is the answer.
+    int32 page_number = 4;
+
+    // How many results are in this page.
+    int32 result_per_page = 5;
+}
+
+message RelationPrefixSearchResponse {
+    repeated utils.RelationNode nodes = 1;
+}
+
+message GraphQuery {
+    message Node {
+        optional string value = 1;
+        optional utils.RelationNode.NodeType node_type = 2;
+        optional string node_subtype = 3;
+
+        enum MatchKind {
+            DEPRECATED_EXACT = 0;
+            DEPRECATED_FUZZY = 1;
+        }
+        MatchKind match_kind = 4 [deprecated = true];
+
+        enum MatchLocation {
+            // Match full value. Search "Rosa Parks" to find "Rosa Parks"
+            FULL = 0;
+            // Match prefix. Search "Rosa P" or "Ros" to find "Rosa Parks"
+            PREFIX = 1;
+            // Match individual words. Search "Rosa" or "Parks" to find "Rosa Parks"
+            WORDS = 2;
+            // Match individual word by prefix. Search "Ros" or "Par" to find "Rosa Parks"
+            PREFIX_WORDS = 3;
+        }
+
+        message ExactMatch {
+            MatchLocation kind = 1;
+        }
+
+        message FuzzyMatch {
+            MatchLocation kind = 1;
+            uint32 distance = 2;
+        }
+
+        oneof new_match_kind {
+            ExactMatch exact = 5;
+            FuzzyMatch fuzzy = 6;
+        }
+    }
+
+    message Relation {
+        optional string value = 1;
+        optional utils.Relation.RelationType relation_type = 2;
+    }
+
+    message Path {
+        optional Node source = 1;
+        optional Relation relation = 2;
+        optional Node destination = 3;
+        bool undirected = 4;
+    }
+
+    message BoolQuery {
+        repeated PathQuery operands = 1;
+    }
+
+    message FacetFilter {
+        string facet = 1;
+    }
+
+    message PathQuery {
+        oneof query {
+            GraphQuery.Path path = 1;
+            PathQuery bool_not = 2;
+            BoolQuery bool_and = 3;
+            BoolQuery bool_or = 4;
+            FacetFilter facet = 5;
+        }
+    }
+
+    // Search directed/undirected paths (triplets of node-relation-node). It
+    // can also be used to express node and relation queries
+    PathQuery path = 1;
+}
+
+message GraphSearchRequest {
+    string shard = 1;
+    GraphQuery query = 2;
+
+    // Decides what do we want in return: paths (triplets), nodes or relations.
+    // Returning paths implies returning nodes and relations.
+    enum QueryKind {
+        PATH = 0;
+        NODES = 1;
+        RELATIONS = 2;
+    }
+    QueryKind kind = 3;
+
+    uint32 top_k = 4;
+
+    optional utils.Security security = 5;
+    optional FilterExpression field_filter = 6;
+}
+
+message GraphSearchResponse {
+    message Relation {
+        utils.Relation.RelationType relation_type = 1;
+        string label = 2;
+    }
+
+    // To optimize proto size, paths are built using pointers to nodes and
+    // relations. This avoid repetition of nodes/relations and allow smaller
+    // protos for big graphs.
+    message Path {
+        uint32 source = 1;
+        uint32 relation = 2;
+        uint32 destination = 3;
+        optional utils.RelationMetadata metadata = 4;
+        optional string resource_field_id = 5;
+        repeated string facets = 6;
+    }
+
+    repeated utils.RelationNode nodes = 1;
+    repeated Relation relations = 2;
+    repeated Path graph = 3;
+}
+
+message FilterExpression {
+    message FilterExpressionList {
+        repeated FilterExpression operands = 1;
+    }
+    message ResourceFilter {
+        string resource_id = 1;
+    }
+    message FieldFilter {
+        string field_type = 1;
+        optional string field_id = 2;
+    }
+    message DateRangeFilter {
+        enum DateField {
+            CREATED = 0;
+            MODIFIED = 1;
+        }
+        DateField field = 1;
+        optional google.protobuf.Timestamp since = 2;
+        optional google.protobuf.Timestamp until = 3;
+    }
+    message KeywordFilter {
+        string keyword = 1;
+    }
+    message FacetFilter {
+        string facet = 1;
+    }
+
+    oneof expr {
+        FilterExpressionList bool_and = 1;
+        FilterExpressionList bool_or = 2;
+        FilterExpression bool_not = 3;
+
+        ResourceFilter resource = 4;
+        FieldFilter field = 5;
+        KeywordFilter keyword = 6;
+        DateRangeFilter date = 7;
+        FacetFilter facet = 8;
+    }
+}
+
+enum FilterOperator {
+    AND = 0;
+    OR = 1;
+}
+
+message SearchRequest {
+    reserved 20, 21;
+
+    string shard = 1;
+
+    // query this text in all the paragraphs
+    string body = 3;
+
+    OrderBy order = 5;
+
+    Faceted faceted = 6;
+
+    int32 page_number = 7;
+    int32 result_per_page = 8;
+
+    repeated float vector = 10; // Embedded vector search.
+
+    string vectorset = 15;
+
+    bool paragraph = 12;
+    bool document = 13;
+
+    bool with_duplicates = 14;
+    bool only_faceted = 16;
+
+    optional string advanced_query = 18;
+
+    float min_score_semantic = 23;
+    float min_score_bm25 = 25;
+
+    optional utils.Security security = 24;
+
+    optional FilterExpression field_filter = 26;
+    optional FilterExpression paragraph_filter = 27;
+    FilterOperator filter_operator = 28;
+
+    // Reduced graph search for specific use case of text block search. It only
+    // supports path search (as it's the only one that can return paragraph_ids)
+    // and filters are inherited from the main request
+    message GraphSearch {
+        GraphQuery query = 1;
+    }
+    GraphSearch graph_search = 29;
+}
+
+
+enum SuggestFeatures {
+    ENTITIES = 0;
+    PARAGRAPHS = 1;
+}
+
+message SuggestRequest {
+    string shard = 1;
+
+    string body = 2;
+    repeated SuggestFeatures features = 6;
+
+    optional FilterExpression field_filter = 7;
+    optional FilterExpression paragraph_filter = 8;
+    FilterOperator filter_operator = 9;
+}
+
+
+message RelatedEntities {
+    repeated string entities = 1;
+    uint32 total = 2;
+}
+
+message SuggestResponse {
+    int32 total = 1;
+    repeated ParagraphResult results = 2;
+
+    // The text that lead to this results
+    string query = 3;
+    repeated string ematches = 4;
+
+    // Entities related with the query
+    RelationPrefixSearchResponse entity_results = 6;
+}
+
+message SearchResponse {
+    reserved 4;
+    DocumentSearchResponse document = 1;
+    ParagraphSearchResponse paragraph = 2;
+    VectorSearchResponse vector = 3;
+    GraphSearchResponse graph = 5;
+}
+
+message IdCollection {
+    repeated string ids = 1;
+}
+
+message RelationEdge {
+    utils.Relation.RelationType edge_type = 1;
+    string property = 2;
+}
+
+message EdgeList {
+    repeated RelationEdge list = 1;
+}
+
+message GetShardRequest {
+    noderesources.ShardId shard_id = 1;
+    string vectorset = 2;
+}
+
+message ParagraphItem {
+    string id = 1;
+    repeated string labels = 2;
+}
+
+message DocumentItem {
+    string uuid = 1;
+    string field = 2;
+    repeated string labels = 3;
+}
+
+message StreamRequest {
+    noderesources.ShardId shard_id = 3;
+    StreamFilter filter = 4;
+}
+
+message GetShardFilesRequest {
+    string shard_id = 1;
+}
+
+message ShardFileList {
+  repeated ShardFile files = 2;
+}
+
+message ShardFile {
+    string relative_path = 1;
+    uint64 size = 2;
+}
+
+message DownloadShardFileRequest {
+    string shard_id = 1;
+    string relative_path = 2;
+}
+
+message ShardFileChunk {
+    bytes data = 1;
+    int32 index = 2;
+}
+
+// Implemented at nucliadb_object_storage
+
+service NodeReader {
+    rpc GetShard(GetShardRequest) returns (noderesources.Shard) {}
+
+    rpc DocumentIds(noderesources.ShardId) returns (IdCollection) {}
+    rpc ParagraphIds(noderesources.ShardId) returns (IdCollection) {}
+    rpc VectorIds(noderesources.VectorSetID) returns (IdCollection) {}
+    rpc RelationIds(noderesources.ShardId) returns (IdCollection) {}
+
+    rpc RelationEdges(noderesources.ShardId) returns (EdgeList) {}
+
+    rpc Search(SearchRequest) returns (SearchResponse) {}
+    rpc Suggest(SuggestRequest) returns (SuggestResponse) {}
+
+    // Streams
+    rpc Paragraphs(StreamRequest) returns (stream ParagraphItem) {}
+    rpc Documents(StreamRequest) returns (stream DocumentItem) {}
+
+    // Shard Download
+    rpc GetShardFiles(GetShardFilesRequest) returns (ShardFileList) {};
+    rpc DownloadShardFile(DownloadShardFileRequest) returns (stream ShardFileChunk) {};
+}