From ef7de92ccb6ba809d3d8e40c10fa9174548f582d Mon Sep 17 00:00:00 2001
From: Jolly Chen <jolly.chen@student.uva.nl>
Date: Mon, 15 Apr 2024 14:06:19 +0200
Subject: [PATCH] [Tree] Fix globbing of ? when also used as query When ? is
 used in TChain::Add in both the filename for globbing and as a starter for a
 query string, TChain::Add fails to find the intended file, because everything
 after the first ? is interpreted as the query string.

This commit changes the behaviour so that
any ? characters before the last '.root' are
interpreted as a wildcard characters.

Fixes #10239
---
 tree/tree/src/TChain.cxx         | 22 ++++++++++++++++++++++
 tree/tree/test/TChainParsing.cxx | 12 ++++++++++++
 2 files changed, 34 insertions(+)

diff --git a/tree/tree/src/TChain.cxx b/tree/tree/src/TChain.cxx
index fb0e6bd8e41a1..2533490418e3e 100644
--- a/tree/tree/src/TChain.cxx
+++ b/tree/tree/src/TChain.cxx
@@ -367,6 +367,28 @@ Int_t TChain::Add(const char *name, Long64_t nentries /* = TTree::kMaxEntries */
    TString basename, treename, query, suffix;
    ParseTreeFilename(name, basename, treename, query, suffix);
 
+   // Special case: ? used for query string AND as wildcard in the filename.
+   // In this case, everything after the first ? is parsed as query/suffix
+   // string in ParseTreeFilename. We assume that everything until the last
+   // occurence of .root should be part of the basename so we remove it
+   // from the suffix and add it back to the basename.
+   // See: https://github.com/root-project/root/issues/10239
+   static const char *dotr = ".root";
+   static Ssiz_t dotrl = strlen(dotr);
+   // Find the last one
+   Ssiz_t lastDotrIdx = kNPOS;
+   Ssiz_t dotrIdx = suffix.Index(dotr);
+   while (dotrIdx != kNPOS) {
+      lastDotrIdx = dotrIdx;
+      dotrIdx = suffix.Index(dotr, dotrIdx + 1);
+   }
+   if (lastDotrIdx != kNPOS) {
+      // Add the part up until '.root' to the basename for globbing
+      basename.Append(suffix, lastDotrIdx + dotrl);
+      // Remove the part up until '.root' from the suffix
+      suffix.Replace(0, lastDotrIdx + dotrl, "");
+   }
+
    // case with one single file
    if (!basename.MaybeWildcard()) {
       return AddFile(name, nentries);
diff --git a/tree/tree/test/TChainParsing.cxx b/tree/tree/test/TChainParsing.cxx
index 29b869359fa4b..92efeda7a6376 100644
--- a/tree/tree/test/TChainParsing.cxx
+++ b/tree/tree/test/TChainParsing.cxx
@@ -240,6 +240,7 @@ TEST(TChainParsing, RecursiveGlob)
    TChain none;
    TChain nested;
    TChain globDir;
+   TChain regex;
 
    nodir.Add("*");
    const auto *nodirFiles = nodir.GetListOfFiles();
@@ -274,6 +275,17 @@ TEST(TChainParsing, RecursiveGlob)
    for (std::size_t i = 0; i < expectedFileNamesGlobDir.size(); i++) {
       EXPECT_EQ(globDirChainFileNames[i], expectedFileNamesGlobDir[i]);
    }
+
+   regex.Add("test*/subdir?/[0-9]?.root?#events");
+   const auto *regexChainFiles = regex.GetListOfFiles();
+   ASSERT_TRUE(regexChainFiles);
+   EXPECT_EQ(regexChainFiles->GetEntries(), 3);
+   std::vector<std::string> expectedFileNamesRegex{
+      ConcatUnixFileName(cwd, "testglob/subdir1/1a.root"),
+      ConcatUnixFileName(cwd, "testglob/subdir1/1b.root"),
+      ConcatUnixFileName(cwd, "testglob/subdir3/3a.root")};
+   auto regexChainFileNames = GetFileNamesVec(regexChainFiles);
+   EXPECT_VEC_EQ(regexChainFileNames, expectedFileNamesRegex);
 }
 
 #if !defined(_MSC_VER) || defined(R__ENABLE_BROKEN_WIN_TESTS)