From e44efbd7a53d32ee94cc2771860b1446c0b5cac8 Mon Sep 17 00:00:00 2001
From: Johannes Kalmbach <johannes.kalmbach@gmail.com>
Date: Thu, 1 Aug 2024 11:04:17 +0200
Subject: [PATCH] Fix a bug when using [blankNodePropertyLists] and
 (collections) together with the relaxed parser.

TODO: Probably the solution is to throw out the relaxed parser completely.
---
 src/parser/TurtleParser.cpp | 12 +++++++-----
 test/TurtleParserTest.cpp   | 33 ++++++++++++++++++++++++++++++---
 2 files changed, 37 insertions(+), 8 deletions(-)

diff --git a/src/parser/TurtleParser.cpp b/src/parser/TurtleParser.cpp
index e0725aa3c..edc8cdacf 100644
--- a/src/parser/TurtleParser.cpp
+++ b/src/parser/TurtleParser.cpp
@@ -189,8 +189,8 @@ bool TurtleParser<T>::object() {
   // check blank Node first because _: also could look like a prefix
   // TODO<joka921> Currently collections and blankNodePropertyLists do not work
   // on dblp when using the relaxed parser. Is this fixable?
-  if (blankNode() || literal() || iri() || collection() ||
-      blankNodePropertyList()) {
+  if (blankNode() || blankNodePropertyList() || collection() || literal() ||
+      iri()) {
     emitTriple();
     return true;
   } else {
@@ -632,14 +632,16 @@ bool TurtleParser<T>::pnameLnRelaxed() {
   // is ok
   tok_.skipWhitespaceAndComments();
   auto view = tok_.view();
-  auto pos = view.find(':');
-  if (pos == string::npos) {
+  auto pos = view.find_first_of(" \t\r\n,;)]:");
+  // If we find any whitespace or other characters that end a structure in
+  // Turtle before we find a ':', the parsing failed.
+  if (pos == string::npos || view[pos] != ':') {
     return false;
   }
   // these can also be part of a collection etc.
   // find any character that can end a pnameLn when assuming that no
   // escape sequences were used
-  auto posEnd = view.find_first_of(" \t\r\n,;", pos);
+  auto posEnd = view.find_first_of(" \t\r\n,;)]", pos);
   if (posEnd == string::npos) {
     // make tests work
     posEnd = view.size();
diff --git a/test/TurtleParserTest.cpp b/test/TurtleParserTest.cpp
index 44f1928bd..b8dc3d37f 100644
--- a/test/TurtleParserTest.cpp
+++ b/test/TurtleParserTest.cpp
@@ -29,15 +29,18 @@ auto iri = [](std::string_view s) {
 // `rule` (a member function of `Parser` that returns a bool). Return the
 // parser, if the call to `rule` returns true, else return `std::nullopt`.
 template <typename Parser, auto rule, size_t blankNodePrefix = 0>
-std::optional<Parser> parseRule(const std::string& input) {
+auto parseRule =
+    [](const std::string& input,
+       decltype(Parser::prefixMap_) prefixMap = {}) -> std::optional<Parser> {
   Parser parser;
+  parser.prefixMap_ = std::move(prefixMap);
   parser.setBlankNodePrefixOnlyForTesting(blankNodePrefix);
   parser.setInputStream(input);
   if (!std::invoke(rule, parser)) {
     return std::nullopt;
   }
   return parser;
-}
+};
 
 // Asserts that parsing the `rule` works and that the last parse result and the
 // emitted triples are as expected. Returns the `Parser` instance that is used
@@ -317,7 +320,7 @@ TEST(TurtleParserTest, blankNodePropertyList) {
 }
 
 TEST(TurtleParserTest, object) {
-  auto runCommonTests = [](auto p) {
+  auto runCommonTests = []<typename Parser>(Parser p) {
     auto sub = iri("<sub>");
     auto pred = iri("<pred>");
     p.activeSubject_ = sub;
@@ -344,6 +347,30 @@ TEST(TurtleParserTest, object) {
 
     exp = TurtleTriple{sub, pred, "_:u_someblank"};
     ASSERT_EQ(p.triples_.back(), exp);
+
+    {
+      auto map = decltype(Parser::prefixMap_){};
+      map["b"] = iri("<bla/>");
+      auto p = parseRule<Parser, &Parser::object>("[ b:blubb 42]", map);
+      EXPECT_TRUE(p.has_value());
+      EXPECT_EQ(p.value().lastParseResult_, "_:g_0_0");
+      EXPECT_EQ(p.value().triples_.size(), 2);
+      EXPECT_THAT(
+          p.value().triples_[0],
+          ::testing::Eq(TurtleTriple{"_:g_0_0", iri("<bla/blubb>"), 42}));
+    }
+    {
+      auto map = decltype(Parser::prefixMap_){};
+      map["b"] = iri("<bla/>");
+      auto p = parseRule<Parser, &Parser::object>("(b:blubb)]", map);
+      EXPECT_TRUE(p.has_value());
+      EXPECT_EQ(p.value().lastParseResult_, "_:g_0_0");
+      EXPECT_EQ(p.value().triples_.size(), 3);
+      auto first = iri("<http://www.w3.org/1999/02/22-rdf-syntax-ns#first>");
+      EXPECT_THAT(
+          p.value().triples_[0],
+          ::testing::Eq(TurtleTriple{"_:g_0_0", first, iri("<bla/blubb>")}));
+    }
   };
   runCommonTests(Re2Parser{});
   runCommonTests(CtreParser{});