[libxml] add parseOptions

Added the ability to explicitly specify parsing options.
commit_hash:1bd7947cfc298f0c3edc895a77c64f70504b78d5

3 files changed, 38 insertions, 11 deletions

diff --git a/library/cpp/xml/document/xml-document-decl.h b/library/cpp/xml/document/xml-document-decl.h
index 643ba664680..552304bb520 100644
--- a/library/cpp/xml/document/xml-document-decl.h
+++ b/library/cpp/xml/document/xml-document-decl.h
@@ -29,10 +29,11 @@ namespace NXml {
         /**
         * create TDocument
         * @param source: filename, XML string, or name for the root element (depends on @src)
-        * @param src: source type: File | String | RootName
+        * @param type: source type: File | String | RootName
+        * @param parseOptions: XML parser options
         * throws if file not found or cannot be parsed
         */
-        TDocument(const TString& source, Source type = File);
+        TDocument(const TString& source, Source type = File, int parseOptions = XML_PARSE_NOCDATA);
 
     public:
         TDocument(const TDocument& that) = delete;
@@ -71,8 +72,8 @@ namespace NXml {
         }
 
     private:
-        void ParseFile(const TString& file);
-        void ParseString(TZtStringBuf xml);
+        void ParseFile(const TString& file, int parseOptions);
+        void ParseString(TZtStringBuf xml, int parseOptions);
 
         TDocument(TDocHolder doc)
             : Doc(std::move(doc))
diff --git a/library/cpp/xml/document/xml-document.cpp b/library/cpp/xml/document/xml-document.cpp
index 25c0ed6a17e..8f9b303c916 100644
--- a/library/cpp/xml/document/xml-document.cpp
+++ b/library/cpp/xml/document/xml-document.cpp
@@ -17,13 +17,13 @@ namespace {
 }
 
 namespace NXml {
-    TDocument::TDocument(const TString& xml, Source type) {
+    TDocument::TDocument(const TString& xml, Source type, int parseOptions) {
         switch (type) {
             case File:
-                ParseFile(xml);
+                ParseFile(xml, parseOptions);
                 break;
             case String:
-                ParseString(xml);
+                ParseString(xml, parseOptions);
                 break;
             case RootName: {
                 TDocHolder doc(xmlNewDoc(XMLCHAR("1.0")));
@@ -55,7 +55,7 @@ namespace NXml {
         return *this;
     }
 
-    void TDocument::ParseFile(const TString& file) {
+    void TDocument::ParseFile(const TString& file, int parseOptions) {
         if (!NFs::Exists(file))
             THROW(XmlException, "File " << file << " doesn't exist");
 
@@ -63,7 +63,7 @@ namespace NXml {
         if (!pctx)
             THROW(XmlException, "Can't create parser context");
 
-        TDocHolder doc(xmlCtxtReadFile(pctx.Get(), file.c_str(), nullptr, XML_PARSE_NOCDATA));
+        TDocHolder doc(xmlCtxtReadFile(pctx.Get(), file.c_str(), nullptr, parseOptions));
         if (!doc)
             THROW(XmlException, "Can't parse file " << file);
 
@@ -75,12 +75,12 @@ namespace NXml {
         Doc = std::move(doc);
     }
 
-    void TDocument::ParseString(TZtStringBuf xml) {
+    void TDocument::ParseString(TZtStringBuf xml, int parseOptions) {
         TParserCtxtPtr pctx(xmlNewParserCtxt());
         if (pctx.Get() == nullptr)
             THROW(XmlException, "Can't create parser context");
 
-        TDocHolder doc(xmlCtxtReadMemory(pctx.Get(), xml.c_str(), (int)xml.size(), nullptr, nullptr, XML_PARSE_NOCDATA));
+        TDocHolder doc(xmlCtxtReadMemory(pctx.Get(), xml.c_str(), (int)xml.size(), nullptr, nullptr, parseOptions));
 
         if (!doc)
             THROW(XmlException, "Can't parse string");
diff --git a/library/cpp/xml/document/xml-document_ut.cpp b/library/cpp/xml/document/xml-document_ut.cpp
index 0ec1fc60838..b4af2cdccf3 100644
--- a/library/cpp/xml/document/xml-document_ut.cpp
+++ b/library/cpp/xml/document/xml-document_ut.cpp
@@ -338,4 +338,30 @@ Y_UNIT_TEST_SUITE(TestXmlDocument) {
         root.SetAttr("quux", "literal");
         root.SetAttr("frob", 500);
     }
+
+    Y_UNIT_TEST(Cdata) {
+        using namespace NXml;
+        const TString xml = "<?xml version=\"1.0\"?>\n"
+                            "<example><![CDATA[Some & data]]> test</example>\n";
+
+        TDocument docDefault(xml, TDocument::String);
+        UNIT_ASSERT(docDefault.Root().FirstChild().IsText());
+
+        TDocument docWithCdata(xml, TDocument::String, XML_PARSE_NOCDATA);
+        UNIT_ASSERT(docWithCdata.Root().FirstChild().IsText());
+
+        TDocument docWithoutCdata(xml, TDocument::String, 0);
+        UNIT_ASSERT(!docWithoutCdata.Root().FirstChild().IsText());
+    }
+
+    Y_UNIT_TEST(InvalidXml) {
+        using namespace NXml;
+        const TString xml = "<?xml version=\"1.0\"?>\n"
+                            "<root><item>value</root>\n";
+
+        UNIT_CHECK_GENERATED_EXCEPTION(TDocument(xml, TDocument::String), yexception);
+
+        TDocument doc(xml, TDocument::String, XML_PARSE_RECOVER);
+        UNIT_ASSERT_EQUAL(doc.Root().FirstChild().Value<TString>(), "value");
+    }
 }