sdext.pdfimport: Use string() in xpath for testTdf104597_textrun

...and remove the TODO comments.

Also, comment out the test for string ٱلسََّل which is related to tdf#152083 and will be fixed in a separate commit.

Change-Id: I075145be1ff58c16e976c710fbf8f55fc6752af1
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/142962
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>

2 files changed, 28 insertions, 26 deletions

diff --git a/sdext/source/pdfimport/test/testdocs/testTdf152083_rtl-ligatures.pdf b/sdext/source/pdfimport/test/testdocs/testTdf152083_rtl-ligatures.pdf
new file mode 100644
index 000000000000..635004c2e302
--- /dev/null
+++ b/sdext/source/pdfimport/test/testdocs/testTdf152083_rtl-ligatures.pdf
diff --git a/sdext/source/pdfimport/test/tests.cxx b/sdext/source/pdfimport/test/tests.cxx
index 786815941445..c199c2cc4547 100644
--- a/sdext/source/pdfimport/test/tests.cxx
+++ b/sdext/source/pdfimport/test/tests.cxx
@@ -796,28 +796,33 @@ namespace
             xmlDocUniquePtr pXmlDoc(xmlParseDoc(reinterpret_cast<xmlChar const *>(aOutput.getStr())));
 
             // Test for امُ عَلَيْكَ
-            // TODO: How to get the "عَلَيْكَ" in xpath, as shown after the <text:s> tag?
-            OString xpath = "//draw:frame[@draw:transform='matrix(917.222222222222 0 0 917.222222222222 14821.9583333333 2159.23861112778)']/draw:text-box/text:p/text:span";
-            OUString sContent = getXPathContent(pXmlDoc, xpath); // u"\nا\nُ\nم\n"
-            CPPUNIT_ASSERT_EQUAL_MESSAGE(aOutput.getStr(), OUString(u"اُم"), sContent.replaceAll("\n", ""));
-
-            // Test for ٱلَّسَل‬ . It appears in the 3rd frame, i.e. after the امُ عَلَيْكَ which is in the 2nd frame (from left to right)
-            // thus these two frames together appear as ٱلَّسَل امُ عَلَيْكَ in Draw‬.
-            xpath = "//draw:frame[@draw:transform='matrix(917.222222222222 0 0 917.222222222222 17420.1666666667 2159.23861112778)']/draw:text-box/text:p/text:span";
+            OString xpath = "string(//draw:frame[@draw:transform='matrix(917.222222222222 0 0 917.222222222222 14821.9583333333 2159.23861112778)']/draw:text-box/text:p/text:span)";
+            OUString sContent = getXPathContent(pXmlDoc, xpath);
+            CPPUNIT_ASSERT_EQUAL_MESSAGE(aOutput.getStr(), OUString(u"اُم َعَلْيَك"), sContent.replaceAll("\n\n", " ").replaceAll("\n", ""));
+
+            // Test for ٱلسََّل . It appears in the 3rd frame, i.e. after the امُ عَلَيْكَ which is in the 2nd frame (from left to right)
+            // thus these two frames together appear as ٱلسََّل امُ عَلَيْكَ in Draw‬.
+            // FIXME: Should be ٱلسَّلَامُ عَلَيْكَ (i.e. the two text frames should be merged into one so that the ل and the ا will show as لَا rather than ل ا)
+            // Note: this is commented due to ٱلسََّل is currently shown as ٱلَّسَل and will be fixed in a separate commit.
+            //xpath = "string(//draw:frame[@draw:transform='matrix(917.222222222222 0 0 917.222222222222 17420.1666666667 2159.23861112778)']/draw:text-box/text:p/text:span)";
+            //sContent = getXPathContent(pXmlDoc, xpath);
+            //CPPUNIT_ASSERT_EQUAL_MESSAGE(aOutput.getStr(), OUString(u"ٱلسََّل"), sContent.replaceAll("\n\n", " ").replaceAll("\n", ""));
+
+            // Test for "LibreOffice RTL"
+            xpath = "string(//draw:frame[@draw:transform='matrix(917.222222222222 0 0 917.222222222222 12779.375 5121.79583335)']/draw:text-box/text:p/text:span)";
             sContent = getXPathContent(pXmlDoc, xpath);
-            CPPUNIT_ASSERT_EQUAL_MESSAGE(aOutput.getStr(), OUString(u"ٱلَّسَل"), sContent.replaceAll("\n", ""));
+            CPPUNIT_ASSERT_EQUAL_MESSAGE(aOutput.getStr(), OUString(u"LibreOffice RTL"), sContent.replaceAll("\n\n", " ").replaceAll("\n", ""));
 
-            // Test for "LibreOffice LTR"
-            // TODO: How to get the "LTR" as shown after the <text:s> tag?
-            xpath = "//draw:frame[@draw:transform='matrix(917.222222222222 0 0 917.222222222222 12779.375 5121.79583335)']/draw:text-box/text:p/text:span";
+            // Test for "LibreOffice LTR (test)"
+            xpath = "string(//draw:frame[last()-1]/draw:text-box/text:p/text:span[last()])";
             sContent = getXPathContent(pXmlDoc, xpath);
-            CPPUNIT_ASSERT_EQUAL_MESSAGE(aOutput.getStr(), OUString(u"LibreOffice"), sContent.replaceAll("\n", ""));
+            CPPUNIT_ASSERT_EQUAL_MESSAGE(aOutput.getStr(), OUString(u"LibreOffice LTR (test)"), sContent.replaceAll("\n\n", " ").replaceAll("\n", ""));
 
             /* Test for Chinese characters */
             // Use last() instead of matrix below, because the matrix may be different on different OS due to fallback of Chinese fonts.
-            xpath = "//draw:frame[last()]/draw:text-box/text:p/text:span";
+            xpath = "string(//draw:frame[last()]/draw:text-box/text:p/text:span)";
             sContent = getXPathContent(pXmlDoc, xpath);
-            CPPUNIT_ASSERT_EQUAL_MESSAGE(aOutput.getStr(), OUString(u"中文测试，中文"), sContent.replaceAll("\n", ""));
+            CPPUNIT_ASSERT_EQUAL_MESSAGE(aOutput.getStr(), OUString(u"中文测试，中文"), sContent.replaceAll("\n\n", " ").replaceAll("\n", ""));
 
             // Test pdf text run in the Writer PDF import filter
             xAdaptor->setTreeVisitorFactory(createWriterTreeVisitorFactory());
@@ -825,19 +830,16 @@ namespace
             xAdaptor->odfConvert(m_directories.getURLFromSrc(u"/sdext/source/pdfimport/test/testdocs/tdf104597_textrun.pdf"),
                     new OutputWrapString(aOutput2),
                     nullptr);
-            // FIXME: the same draw:frame is duplicated in the xml output,
-            // e.g. there are two draw:frame with draw:z-index="3" with the same content.
             xmlDocUniquePtr pXmlDoc2(xmlParseDoc(reinterpret_cast<xmlChar const *>(aOutput2.getStr())));
-            xpath = "//draw:frame[@draw:z-index='3'][1]/draw:text-box/text:p/text:span";
-            sContent = getXPathContent(pXmlDoc2, xpath).replaceAll("\n", "");
+            xpath = "string(//draw:frame[@draw:z-index='3'][1]/draw:text-box/text:p/text:span)";
+            sContent = getXPathContent(pXmlDoc2, xpath).replaceAll("\n\n", " ").replaceAll("\n", "");
             CPPUNIT_ASSERT_EQUAL_MESSAGE(aOutput2.getStr(), OUString(u"ٱلَّسَل"), sContent);
-            xpath = "//draw:frame[@draw:z-index='2'][1]/draw:text-box/text:p/text:span";
-            sContent = getXPathContent(pXmlDoc2, xpath).replaceAll("\n", "");
-            // need to use اُم rather than اُم َعَلْيَك here, because this node may be different on different systems
-            CPPUNIT_ASSERT_EQUAL(true, sContent.match(u"اُم"));
-            xpath = "//draw:frame[last()]/draw:text-box/text:p/text:span";
-            sContent = getXPathContent(pXmlDoc2, xpath);
-            CPPUNIT_ASSERT_EQUAL_MESSAGE(aOutput2.getStr(), OUString(u"中文测试，中文"), sContent.replaceAll("\n", ""));
+            xpath = "string(//draw:frame[@draw:z-index='2'][1]/draw:text-box/text:p/text:span)";
+            sContent = getXPathContent(pXmlDoc2, xpath).replaceAll("\n\n", " ").replaceAll("\n", "");
+            CPPUNIT_ASSERT_EQUAL(true, sContent.match(u"اُم َعَلْيَك"));
+            xpath = "string(//draw:frame[last()]/draw:text-box/text:p/text:span)";
+            sContent = getXPathContent(pXmlDoc2, xpath).replaceAll("\n\n", " ").replaceAll("\n", "");
+            CPPUNIT_ASSERT_EQUAL_MESSAGE(aOutput2.getStr(), OUString(u"中文测试，中文"), sContent);
 #endif
         }