From e2a22a699458dae9f46140ac8c4753d951e0ba4c Mon Sep 17 00:00:00 2001
From: Antonin Delpeuch <antonin@delpeuch.eu>
Date: Fri, 20 Oct 2017 15:32:58 +0100
Subject: [PATCH] Forbid pipe characters in URL references to ease parsing.

This is a temporary fix before we do full Wikitext parsing inside references
(this needs a change upstream). See https://github.com/sweble/sweble-wikitext/issues/67 .
---
 .../refine/importers/WikitextImporter.java    |  2 +-
 .../importers/WikitextImporterTests.java      | 29 +++++++++++++++++++
 2 files changed, 30 insertions(+), 1 deletion(-)

diff --git a/main/src/com/google/refine/importers/WikitextImporter.java b/main/src/com/google/refine/importers/WikitextImporter.java
index 1106256b3..9927df528 100644
--- a/main/src/com/google/refine/importers/WikitextImporter.java
+++ b/main/src/com/google/refine/importers/WikitextImporter.java
@@ -156,7 +156,7 @@ public class WikitextImporter extends TabularImportingParserBase {
         private int spanningCellIdx;
         private List<String> internalLinksInCell;
         
-        private final Pattern urlPattern = Pattern.compile("\\b(https?|ftp)://[-a-zA-Z0-9+&@#/%?=~_|!:,.;]*[-a-zA-Z0-9+&@#/%=~_|]",
+        private final Pattern urlPattern = Pattern.compile("\\b(https?|ftp)://[-a-zA-Z0-9+&@#/%?=~_!:,.;]*[-a-zA-Z0-9+&@#/%=~_]",
                 Pattern.CASE_INSENSITIVE);
         
         public WikitextTableVisitor(boolean blankSpanningCells, boolean includeRawTemplates) {
diff --git a/main/tests/server/src/com/google/refine/tests/importers/WikitextImporterTests.java b/main/tests/server/src/com/google/refine/tests/importers/WikitextImporterTests.java
index ec0173189..1686e89dc 100644
--- a/main/tests/server/src/com/google/refine/tests/importers/WikitextImporterTests.java
+++ b/main/tests/server/src/com/google/refine/tests/importers/WikitextImporterTests.java
@@ -224,6 +224,35 @@ public class WikitextImporterTests extends ImporterTest {
         Assert.assertEquals(project.rows.get(1).cells.get(4).value, "http://gnu.org");
         Assert.assertEquals(project.rows.get(1).cells.get(2).value, "http://microsoft.com/");
     }
+
+    @Test
+    public void readTableWithReferencesTemplates() {
+        // inspired from https://www.mediawiki.org/wiki/Help:Tables
+        String input = "{|\n"
+        +"! price\n"
+        +"! fruit\n"
+        +"! merchant\n"
+        +"|-\n"
+        +"| a || b <ref name=\"myref\">{{cite web|url=http://gnu.org|accessdate=2017-08-30}}</ref>  || c <ref name=\"ms\"> or {{cite journal|url=http://microsoft.com/|title=BLah}} </ref>\n"
+        +"|-\n"
+        +"| d || e <ref name=\"ms\"/>|| f <ref name=\"myref\" />\n"
+        +"|-\n"
+        +"|}\n";
+        
+        try {
+            prepareOptions(-1, true, true, null);
+            parse(input);
+        } catch (Exception e) {
+                Assert.fail("Parsing failed", e);
+        }
+        Assert.assertEquals(project.columnModel.columns.size(), 5);
+        Assert.assertEquals(project.rows.get(0).cells.get(1).value, "b");
+        Assert.assertEquals(project.rows.get(0).cells.get(2).value, "http://gnu.org");
+        Assert.assertEquals(project.rows.get(0).cells.get(4).value, "http://microsoft.com/");
+        Assert.assertEquals(project.rows.get(1).cells.get(4).value, "http://gnu.org");
+        Assert.assertEquals(project.rows.get(1).cells.get(2).value, "http://microsoft.com/");
+    }
+
     //--helpers--
     
     private void parse(String wikitext) {