From 6f79979ab8847a264459213a61cc501d22fe2964 Mon Sep 17 00:00:00 2001
From: Bas Leijdekkers <basleijdekkers@gmail.com>
Date: Fri, 24 Feb 2017 12:31:55 +0100
Subject: [PATCH] RegExp: fix python regex conditional reference parsing

---
 .../intellij/lang/regexp/RegExpParser.java    | 13 ++-
 .../lang/regexp/RegExpParsingTest.java        |  1 +
 RegExpSupport/testData/psi/Groups39.txt       | 91 +++++++++----------
 RegExpSupport/testData/psi/Groups40.txt       | 53 ++++++-----
 RegExpSupport/testData/psi/Groups43.txt       | 85 +++++++++++++++++
 5 files changed, 169 insertions(+), 74 deletions(-)
 create mode 100644 RegExpSupport/testData/psi/Groups43.txt

diff --git a/RegExpSupport/src/org/intellij/lang/regexp/RegExpParser.java b/RegExpSupport/src/org/intellij/lang/regexp/RegExpParser.java
index 4650f62df3ee..3412ead6d63d 100644
--- a/RegExpSupport/src/org/intellij/lang/regexp/RegExpParser.java
+++ b/RegExpSupport/src/org/intellij/lang/regexp/RegExpParser.java
@@ -391,7 +391,18 @@ public class RegExpParser implements PsiParser {
         builder.error("Group name or number expected");
       }
       checkMatches(builder, RegExpTT.GROUP_END, "Unclosed group reference");
-      parseGroupEnd(builder);
+      if (!parseBranch(builder)) {
+        patternExpected(builder);
+      }
+      else {
+        if (builder.getTokenType() == RegExpTT.UNION) {
+          builder.advanceLexer();
+          if (!parseBranch(builder)) {
+            patternExpected(builder);
+          }
+        }
+        checkMatches(builder, RegExpTT.GROUP_END, "Unclosed group");
+      }
       marker.done(RegExpElementTypes.PY_COND_REF);
     }
     else if (type == RegExpTT.PROPERTY) {
diff --git a/RegExpSupport/test/org/intellij/lang/regexp/RegExpParsingTest.java b/RegExpSupport/test/org/intellij/lang/regexp/RegExpParsingTest.java
index a81dd79e2622..4b809d887999 100644
--- a/RegExpSupport/test/org/intellij/lang/regexp/RegExpParsingTest.java
+++ b/RegExpSupport/test/org/intellij/lang/regexp/RegExpParsingTest.java
@@ -193,6 +193,7 @@ public class RegExpParsingTest extends ParsingTestCase {
   public void testGroups40() throws IOException { doCodeTest("(?(name)yes-pattern|{"); }
   public void testGroups41() throws IOException { doCodeTest("(?>atomic)"); }
   public void testGroups42() throws IOException { doCodeTest("(?:non-capturing)"); }
+  public void testGroups43() throws IOException { doCodeTest("(?(name)yes-pattern|no_pattern|maybe-pattern)"); }
 
   public void testEscapes1() throws IOException { doCodeTest("\\q"); }
   public void testEscapes2() throws IOException { doCodeTest("\\#"); }
diff --git a/RegExpSupport/testData/psi/Groups39.txt b/RegExpSupport/testData/psi/Groups39.txt
index 876b3d744dd5..f473d213aa49 100644
--- a/RegExpSupport/testData/psi/Groups39.txt
+++ b/RegExpSupport/testData/psi/Groups39.txt
@@ -5,50 +5,49 @@ REGEXP_FILE
         PsiElement(PYTHON_COND_REF)('(?(')
         PsiElement(NAME)('name')
         PsiElement(GROUP_END)(')')
-        RegExpPatternImpl: <yes-pattern|no-pattern>
-          RegExpBranchImpl: <yes-pattern>
-            RegExpCharImpl: <y>
-              PsiElement(CHARACTER)('y')
-            RegExpCharImpl: <e>
-              PsiElement(CHARACTER)('e')
-            RegExpCharImpl: <s>
-              PsiElement(CHARACTER)('s')
-            RegExpCharImpl: <->
-              PsiElement(CHARACTER)('-')
-            RegExpCharImpl: <p>
-              PsiElement(CHARACTER)('p')
-            RegExpCharImpl: <a>
-              PsiElement(CHARACTER)('a')
-            RegExpCharImpl: <t>
-              PsiElement(CHARACTER)('t')
-            RegExpCharImpl: <t>
-              PsiElement(CHARACTER)('t')
-            RegExpCharImpl: <e>
-              PsiElement(CHARACTER)('e')
-            RegExpCharImpl: <r>
-              PsiElement(CHARACTER)('r')
-            RegExpCharImpl: <n>
-              PsiElement(CHARACTER)('n')
-          PsiElement(UNION)('|')
-          RegExpBranchImpl: <no-pattern>
-            RegExpCharImpl: <n>
-              PsiElement(CHARACTER)('n')
-            RegExpCharImpl: <o>
-              PsiElement(CHARACTER)('o')
-            RegExpCharImpl: <->
-              PsiElement(CHARACTER)('-')
-            RegExpCharImpl: <p>
-              PsiElement(CHARACTER)('p')
-            RegExpCharImpl: <a>
-              PsiElement(CHARACTER)('a')
-            RegExpCharImpl: <t>
-              PsiElement(CHARACTER)('t')
-            RegExpCharImpl: <t>
-              PsiElement(CHARACTER)('t')
-            RegExpCharImpl: <e>
-              PsiElement(CHARACTER)('e')
-            RegExpCharImpl: <r>
-              PsiElement(CHARACTER)('r')
-            RegExpCharImpl: <n>
-              PsiElement(CHARACTER)('n')
+        RegExpBranchImpl: <yes-pattern>
+          RegExpCharImpl: <y>
+            PsiElement(CHARACTER)('y')
+          RegExpCharImpl: <e>
+            PsiElement(CHARACTER)('e')
+          RegExpCharImpl: <s>
+            PsiElement(CHARACTER)('s')
+          RegExpCharImpl: <->
+            PsiElement(CHARACTER)('-')
+          RegExpCharImpl: <p>
+            PsiElement(CHARACTER)('p')
+          RegExpCharImpl: <a>
+            PsiElement(CHARACTER)('a')
+          RegExpCharImpl: <t>
+            PsiElement(CHARACTER)('t')
+          RegExpCharImpl: <t>
+            PsiElement(CHARACTER)('t')
+          RegExpCharImpl: <e>
+            PsiElement(CHARACTER)('e')
+          RegExpCharImpl: <r>
+            PsiElement(CHARACTER)('r')
+          RegExpCharImpl: <n>
+            PsiElement(CHARACTER)('n')
+        PsiElement(UNION)('|')
+        RegExpBranchImpl: <no-pattern>
+          RegExpCharImpl: <n>
+            PsiElement(CHARACTER)('n')
+          RegExpCharImpl: <o>
+            PsiElement(CHARACTER)('o')
+          RegExpCharImpl: <->
+            PsiElement(CHARACTER)('-')
+          RegExpCharImpl: <p>
+            PsiElement(CHARACTER)('p')
+          RegExpCharImpl: <a>
+            PsiElement(CHARACTER)('a')
+          RegExpCharImpl: <t>
+            PsiElement(CHARACTER)('t')
+          RegExpCharImpl: <t>
+            PsiElement(CHARACTER)('t')
+          RegExpCharImpl: <e>
+            PsiElement(CHARACTER)('e')
+          RegExpCharImpl: <r>
+            PsiElement(CHARACTER)('r')
+          RegExpCharImpl: <n>
+            PsiElement(CHARACTER)('n')
         PsiElement(GROUP_END)(')')
\ No newline at end of file
diff --git a/RegExpSupport/testData/psi/Groups40.txt b/RegExpSupport/testData/psi/Groups40.txt
index 2b06a7323922..12e58af412e9 100644
--- a/RegExpSupport/testData/psi/Groups40.txt
+++ b/RegExpSupport/testData/psi/Groups40.txt
@@ -6,33 +6,32 @@ REGEXP_FILE
           PsiElement(PYTHON_COND_REF)('(?(')
           PsiElement(NAME)('name')
           PsiElement(GROUP_END)(')')
-          RegExpPatternImpl: <yes-pattern|>
-            RegExpBranchImpl: <yes-pattern>
-              RegExpCharImpl: <y>
-                PsiElement(CHARACTER)('y')
-              RegExpCharImpl: <e>
-                PsiElement(CHARACTER)('e')
-              RegExpCharImpl: <s>
-                PsiElement(CHARACTER)('s')
-              RegExpCharImpl: <->
-                PsiElement(CHARACTER)('-')
-              RegExpCharImpl: <p>
-                PsiElement(CHARACTER)('p')
-              RegExpCharImpl: <a>
-                PsiElement(CHARACTER)('a')
-              RegExpCharImpl: <t>
-                PsiElement(CHARACTER)('t')
-              RegExpCharImpl: <t>
-                PsiElement(CHARACTER)('t')
-              RegExpCharImpl: <e>
-                PsiElement(CHARACTER)('e')
-              RegExpCharImpl: <r>
-                PsiElement(CHARACTER)('r')
-              RegExpCharImpl: <n>
-                PsiElement(CHARACTER)('n')
-            PsiElement(UNION)('|')
-            PsiErrorElement:Dangling metacharacter
-              <empty list>
+          RegExpBranchImpl: <yes-pattern>
+            RegExpCharImpl: <y>
+              PsiElement(CHARACTER)('y')
+            RegExpCharImpl: <e>
+              PsiElement(CHARACTER)('e')
+            RegExpCharImpl: <s>
+              PsiElement(CHARACTER)('s')
+            RegExpCharImpl: <->
+              PsiElement(CHARACTER)('-')
+            RegExpCharImpl: <p>
+              PsiElement(CHARACTER)('p')
+            RegExpCharImpl: <a>
+              PsiElement(CHARACTER)('a')
+            RegExpCharImpl: <t>
+              PsiElement(CHARACTER)('t')
+            RegExpCharImpl: <t>
+              PsiElement(CHARACTER)('t')
+            RegExpCharImpl: <e>
+              PsiElement(CHARACTER)('e')
+            RegExpCharImpl: <r>
+              PsiElement(CHARACTER)('r')
+            RegExpCharImpl: <n>
+              PsiElement(CHARACTER)('n')
+          PsiElement(UNION)('|')
+          PsiErrorElement:Dangling metacharacter
+            <empty list>
         RegExpQuantifierImpl: <{>
           PsiElement(LBRACE)('{')
           PsiErrorElement:Number expected
diff --git a/RegExpSupport/testData/psi/Groups43.txt b/RegExpSupport/testData/psi/Groups43.txt
new file mode 100644
index 000000000000..5c375073825e
--- /dev/null
+++ b/RegExpSupport/testData/psi/Groups43.txt
@@ -0,0 +1,85 @@
+REGEXP_FILE
+  RegExpPatternImpl: <(?(name)yes-pattern|no_pattern|maybe-pattern>
+    RegExpBranchImpl: <(?(name)yes-pattern|no_pattern>
+      RegExpPyCondRefImpl: <(?(name)yes-pattern|no_pattern>
+        PsiElement(PYTHON_COND_REF)('(?(')
+        PsiElement(NAME)('name')
+        PsiElement(GROUP_END)(')')
+        RegExpBranchImpl: <yes-pattern>
+          RegExpCharImpl: <y>
+            PsiElement(CHARACTER)('y')
+          RegExpCharImpl: <e>
+            PsiElement(CHARACTER)('e')
+          RegExpCharImpl: <s>
+            PsiElement(CHARACTER)('s')
+          RegExpCharImpl: <->
+            PsiElement(CHARACTER)('-')
+          RegExpCharImpl: <p>
+            PsiElement(CHARACTER)('p')
+          RegExpCharImpl: <a>
+            PsiElement(CHARACTER)('a')
+          RegExpCharImpl: <t>
+            PsiElement(CHARACTER)('t')
+          RegExpCharImpl: <t>
+            PsiElement(CHARACTER)('t')
+          RegExpCharImpl: <e>
+            PsiElement(CHARACTER)('e')
+          RegExpCharImpl: <r>
+            PsiElement(CHARACTER)('r')
+          RegExpCharImpl: <n>
+            PsiElement(CHARACTER)('n')
+        PsiElement(UNION)('|')
+        RegExpBranchImpl: <no_pattern>
+          RegExpCharImpl: <n>
+            PsiElement(CHARACTER)('n')
+          RegExpCharImpl: <o>
+            PsiElement(CHARACTER)('o')
+          RegExpCharImpl: <_>
+            PsiElement(CHARACTER)('_')
+          RegExpCharImpl: <p>
+            PsiElement(CHARACTER)('p')
+          RegExpCharImpl: <a>
+            PsiElement(CHARACTER)('a')
+          RegExpCharImpl: <t>
+            PsiElement(CHARACTER)('t')
+          RegExpCharImpl: <t>
+            PsiElement(CHARACTER)('t')
+          RegExpCharImpl: <e>
+            PsiElement(CHARACTER)('e')
+          RegExpCharImpl: <r>
+            PsiElement(CHARACTER)('r')
+          RegExpCharImpl: <n>
+            PsiElement(CHARACTER)('n')
+        PsiErrorElement:Unclosed group
+          <empty list>
+    PsiElement(UNION)('|')
+    RegExpBranchImpl: <maybe-pattern>
+      RegExpCharImpl: <m>
+        PsiElement(CHARACTER)('m')
+      RegExpCharImpl: <a>
+        PsiElement(CHARACTER)('a')
+      RegExpCharImpl: <y>
+        PsiElement(CHARACTER)('y')
+      RegExpCharImpl: <b>
+        PsiElement(CHARACTER)('b')
+      RegExpCharImpl: <e>
+        PsiElement(CHARACTER)('e')
+      RegExpCharImpl: <->
+        PsiElement(CHARACTER)('-')
+      RegExpCharImpl: <p>
+        PsiElement(CHARACTER)('p')
+      RegExpCharImpl: <a>
+        PsiElement(CHARACTER)('a')
+      RegExpCharImpl: <t>
+        PsiElement(CHARACTER)('t')
+      RegExpCharImpl: <t>
+        PsiElement(CHARACTER)('t')
+      RegExpCharImpl: <e>
+        PsiElement(CHARACTER)('e')
+      RegExpCharImpl: <r>
+        PsiElement(CHARACTER)('r')
+      RegExpCharImpl: <n>
+        PsiElement(CHARACTER)('n')
+  PsiErrorElement:Unmatched closing ')'
+    <empty list>
+  PsiElement(GROUP_END)(')')
\ No newline at end of file