Add multidoc support to fyamlcpp

add multidoc support to libfyaml

3 files changed, 137 insertions, 8 deletions

diff --git a/library/cpp/yaml/fyamlcpp/fyamlcpp.cpp b/library/cpp/yaml/fyamlcpp/fyamlcpp.cpp
index 42ee6b830d..fc804128e4 100644
--- a/library/cpp/yaml/fyamlcpp/fyamlcpp.cpp
+++ b/library/cpp/yaml/fyamlcpp/fyamlcpp.cpp
@@ -280,6 +280,11 @@ TNode TNodeRef::Copy() const {
     return TNode(fy_node_copy(fy_node_document(Node_), Node_));
 }
 
+TNode TNodeRef::Copy(const TDocument& to) const {
+    ENSURE_NODE_NOT_EMPTY(Node_);
+    return TNode(fy_node_copy(to.Document_.get(), Node_));
+}
+
 TString TNodeRef::Path() const {
     ENSURE_NODE_NOT_EMPTY(Node_);
     char* path = fy_node_get_path(Node_);
@@ -801,14 +806,77 @@ void TDocument::UnregisterUserDataCleanup() {
     fy_document_unregister_meta(Document_.get());
 }
 
+TMark TDocument::BeginMark() const {
+    ENSURE_DOCUMENT_NOT_EMPTY(Document_);
+    auto* fyds = fy_document_get_document_state(Document_.get());
+    auto* mark = fy_document_state_start_mark(fyds);
+    return TMark{
+        mark->input_pos,
+        mark->line,
+        mark->column,
+    };
+}
+
+TMark TDocument::EndMark() const {
+    ENSURE_DOCUMENT_NOT_EMPTY(Document_);
+    auto* fyds = fy_document_get_document_state(Document_.get());
+    auto* mark = fy_document_state_end_mark(fyds);
+    return TMark{
+        mark->input_pos,
+        mark->line,
+        mark->column,
+    };
+}
+
 std::unique_ptr<char, void(*)(char*)> TJsonEmitter::EmitToCharArray() const {
     std::unique_ptr<char, void(*)(char*)> res(
-        fy_emit_document_to_string(
-            Document_.Document_.get(),
+        fy_emit_node_to_string(
+            Node_.Node_,
             (fy_emitter_cfg_flags)(FYECF_DEFAULT | FYECF_SORT_KEYS | FYECF_MODE_JSON_TP)), &NDetail::FreeChar);
     return res;
 }
 
+TParser::TParser(fy_parser* parser, fy_diag* diag)
+    : Parser_(parser, fy_parser_destroy)
+    , Diag_(diag, fy_diag_destroy)
+{}
+
+TParser TParser::Create(const char* stream)
+{
+    fy_diag_cfg dcfg;
+    fy_diag_cfg_default(&dcfg);
+    std::unique_ptr<fy_diag, void(*)(fy_diag*)> diag(fy_diag_create(&dcfg), fy_diag_destroy);
+    fy_diag_set_collect_errors(diag.get(), true);
+    fy_parse_cfg cfg{
+        "",
+        // FYPCF_PARSE_COMMENTS,
+        FYPCF_QUIET,
+        nullptr,
+        diag.get()
+    };
+    auto* parser = fy_parser_create(&cfg);
+    if (!parser) {
+        fy_diag_error* err;
+        void *iter = nullptr;
+        while ((err = fy_diag_errors_iterate(diag.get(), &iter)) != nullptr) {
+            ythrow yexception() << err->file << ":" << err->line << ":" << err->column << " " << err->msg;
+        }
+    }
+
+    fy_parser_set_string(parser, stream, -1);
+
+    return TParser(parser, diag.release());
+}
+
+std::optional<TDocument> TParser::NextDocument() {
+    auto* doc = fy_parse_load_document(Parser_.get());
+    if (!doc) {
+        return std::nullopt;
+    }
+
+    return TDocument(doc, fy_document_get_diag(doc));
+}
+
 namespace NDetail {
 
 void RethrowError(fy_diag* diag) {
diff --git a/library/cpp/yaml/fyamlcpp/fyamlcpp.h b/library/cpp/yaml/fyamlcpp/fyamlcpp.h
index 7437b74ef7..5b645a557c 100644
--- a/library/cpp/yaml/fyamlcpp/fyamlcpp.h
+++ b/library/cpp/yaml/fyamlcpp/fyamlcpp.h
@@ -8,6 +8,7 @@
 #include <memory>
 #include <optional>
 
+struct fy_parser;
 struct fy_node;
 struct fy_document;
 struct fy_diag;
@@ -67,6 +68,8 @@ class TSequenceIterator;
 class TReverseSequenceIterator;
 class TSequence;
 class TJsonEmitter;
+class TParser;
+struct TMark;
 
 class TDocumentIterator {
     friend class TDocument;
@@ -88,6 +91,7 @@ class TNodeRef {
     friend class TSequence;
     friend class TSequenceIterator;
     friend class TReverseSequenceIterator;
+    friend class TJsonEmitter;
 
     TNodeRef(fy_node* node)
         : Node_(node)
@@ -112,6 +116,8 @@ public:
 
     TNode Copy() const;
 
+    TNode Copy(const TDocument& to) const;
+
     bool IsAlias() const;
 
     TNodeRef ResolveAlias() const;
@@ -500,7 +506,9 @@ private:
 
 class TDocument {
     friend class TNode;
+    friend class TNodeRef;
     friend class TJsonEmitter;
+    friend class TParser;
 
     TDocument(fy_document* doc = nullptr, fy_diag* diag = nullptr);
 
@@ -550,7 +558,6 @@ public:
         return it;
     }
 
-
     TDocumentNodeIterator end() {
         return TDocumentNodeIterator(TNodeRef(nullptr));
     }
@@ -559,6 +566,10 @@ public:
 
     std::unique_ptr<char, void(*)(char*)> EmitToCharArray() const;
 
+    TMark BeginMark() const;
+
+    TMark EndMark() const;
+
 private:
     std::unique_ptr<fy_document, void(*)(fy_document*)> Document_;
     std::unique_ptr<fy_diag, void(*)(fy_diag*)> Diag_;
@@ -578,16 +589,30 @@ private:
 
 class TJsonEmitter {
 public:
-    TJsonEmitter(const TDocument& doc) : Document_(doc) {}
+    TJsonEmitter(TDocument& doc) : Node_(doc.Root()) {}
+    TJsonEmitter(const TNodeRef& node) : Node_(node) {}
 
     std::unique_ptr<char, void(*)(char*)> EmitToCharArray() const;
 
 private:
-    const TDocument& Document_;
+    const TNodeRef Node_;
+};
+
+class TParser {
+    TParser(fy_parser* doc, fy_diag* diag);
+public:
+    static TParser Create(const char* cstr);
+
+    std::optional<TDocument> NextDocument();
+private:
+    std::unique_ptr<fy_parser, void(*)(fy_parser*)> Parser_;
+    std::unique_ptr<fy_diag, void(*)(fy_diag*)> Diag_;
+};
 
-    fy_document* Document() const {
-        return Document_.Document_.get();
-    }
+struct TMark {
+    size_t InputPos;
+    int Line;
+    int Column;
 };
 
 } // namesapce NFyaml
diff --git a/library/cpp/yaml/fyamlcpp/fyamlcpp_ut.cpp b/library/cpp/yaml/fyamlcpp/fyamlcpp_ut.cpp
index f9aab7b011..94093c4157 100644
--- a/library/cpp/yaml/fyamlcpp/fyamlcpp_ut.cpp
+++ b/library/cpp/yaml/fyamlcpp/fyamlcpp_ut.cpp
@@ -50,4 +50,40 @@ test: output
 
         UNIT_ASSERT_VALUES_EQUAL(ss.Str(), "test: output\n");
     }
+
+    Y_UNIT_TEST(Parser) {
+        const char *yaml = R"(
+test: a
+---
+test: b
+)";
+        auto parser = NFyaml::TParser::Create(yaml);
+
+        TStringStream ss;
+
+        auto docOpt = parser.NextDocument();
+        UNIT_ASSERT(docOpt);
+        ss << *docOpt;
+        UNIT_ASSERT_VALUES_EQUAL(ss.Str(), "test: a\n");
+        auto beginMark = docOpt->BeginMark();
+        UNIT_ASSERT_VALUES_EQUAL(beginMark.InputPos, 1);
+        auto endMark = docOpt->EndMark();
+        UNIT_ASSERT_VALUES_EQUAL(endMark.InputPos, 12);
+        UNIT_ASSERT_VALUES_EQUAL(TStringBuf(yaml).SubStr(beginMark.InputPos, endMark.InputPos - 4), ss.Str());
+
+        ss.clear();
+
+        auto docOpt2 = parser.NextDocument();
+        UNIT_ASSERT(docOpt2);
+        ss << *docOpt2;
+        UNIT_ASSERT_VALUES_EQUAL(ss.Str(), "---\ntest: b\n");
+        beginMark = docOpt2->BeginMark();
+        UNIT_ASSERT_VALUES_EQUAL(beginMark.InputPos, 9);
+        endMark = docOpt2->EndMark();
+        UNIT_ASSERT_VALUES_EQUAL(endMark.InputPos, 21);
+        UNIT_ASSERT_VALUES_EQUAL(TStringBuf(yaml).SubStr(beginMark.InputPos, endMark.InputPos), ss.Str());
+
+        auto docOpt3 = parser.NextDocument();
+        UNIT_ASSERT(!docOpt3);
+    }
 }