commit e8d7bf954fe74ce85bdd084d6e12d27027a4c379
author Baole Fang <baole.fang@gmail.com> Thu Apr 27 14:16:18 2023 -0400
committer Stephan Bergmann <sbergman@redhat.com> Wed May 03 07:55:02 2023 +0200
tree eab59d3e6aeb6fa4aa5650b91abe731077f00142
parent a21e151508918c39189a18d46fcba490100e5890

tdf#145925: Support AutoCapitalize in DOI recognition

Sometimes, the first character of the doi string is auto capitalized, which isn't recognized as DOI.
Now, the doi detection is able to recognize doi string with the first character capitalized,
like what is done in url recognition.

Change-Id: I95334941dc4cda3095f1750fea927640dea55e23
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/151142
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>

svl/qa/unit/test_URIHelper.cxx

diff --git a/svl/qa/unit/test_URIHelper.cxx b/svl/qa/unit/test_URIHelper.cxx
index df9e5d5..33a0899 100644
--- a/svl/qa/unit/test_URIHelper.cxx
+++ b/svl/qa/unit/test_URIHelper.cxx
@@ -410,6 +410,11 @@ void Test::testFindFirstDOIInText() {
    };
    static Data const tests[] = {
        { "doi:10.1000/182", "https://doi.org/10.1000/182", 0, 15 }, // valid doi suffix with only digits
        { "Doi:10.1000/182", "https://doi.org/10.1000/182", 0, 15 }, // valid doi suffix with some of the the first three character being capitalized
        { "DoI:10.1000/182", "https://doi.org/10.1000/182", 0, 15 }, // valid doi suffix with some of the the first three character being capitalized
        { "DOI:10.1000/182", "https://doi.org/10.1000/182", 0, 15 }, // valid doi suffix with some of the the first three character being capitalized
        { "dOI:10.1000/182", "https://doi.org/10.1000/182", 0, 15 }, // valid doi suffix with some of the the first three character being capitalized
        { "dOi:10.1000/182", "https://doi.org/10.1000/182", 0, 15 }, // valid doi suffix with some of the the first three character being capitalized
        { "doi:10.1038/nature03001", "https://doi.org/10.1038/nature03001", 0, 23 }, // valid doi suffix with alphanumeric characters
        { "doi:10.1093/ajae/aaq063", "https://doi.org/10.1093/ajae/aaq063", 0, 23 }, // valid doi suffix with multiple slash
        { "doi:10.1016/S0735-1097(98)00347-7", "https://doi.org/10.1016/S0735-1097(98)00347-7", 0, 33 }, // valid doi suffix with characters apart from alphanumeric

svl/source/misc/urihelper.cxx

diff --git a/svl/source/misc/urihelper.cxx b/svl/source/misc/urihelper.cxx
index 0043b78..9aa78a58 100644
--- a/svl/source/misc/urihelper.cxx
+++ b/svl/source/misc/urihelper.cxx
@@ -757,7 +757,7 @@ OUString URIHelper::FindFirstDOIInText(OUString const & rText,
    sal_Int32 count = rEnd-rBegin;
    OUString candidate(rText.subView(rBegin, count));
    // Match with regex "doi:10\.\d{4,9}\/[-._;()\/:a-zA-Z0-9]+"
    if (candidate.startsWith("doi:10."))
    if (candidate.startsWithIgnoreAsciiCase("doi:10."))
    {
        bool flag = true;
        sal_Int32 digit = 0;
@@ -797,7 +797,7 @@ OUString URIHelper::FindFirstDOIInText(OUString const & rText,
        }
        if (flag && digit==-1)
        {
            return candidate.replaceFirst("doi:","https://doi.org/");
            return OUString::Concat("https://doi.org/")+candidate.subView(4);
        }
    }
    rBegin = rEnd;