From f31b04a303607cd82757e7c4f60bb536658c8a30 Mon Sep 17 00:00:00 2001
From: wolfbeast <mcwerewolf@wolfbeast.com>
Date: Mon, 18 Nov 2019 12:20:44 +0100
Subject: Issue #1284 - Implement /s (dotAll) for regular expressions.

Resolves #1284.
---
 js/src/irregexp/RegExpParser.cpp | 52 ++++++++++++++++++++++++++++++++++++----
 js/src/irregexp/RegExpParser.h   |  1 +
 2 files changed, 49 insertions(+), 4 deletions(-)

(limited to 'js/src/irregexp')

diff --git a/js/src/irregexp/RegExpParser.cpp b/js/src/irregexp/RegExpParser.cpp
index 9ef9fe3e2..28abdb0b4 100644
--- a/js/src/irregexp/RegExpParser.cpp
+++ b/js/src/irregexp/RegExpParser.cpp
@@ -1384,7 +1384,7 @@ UnicodeEverythingAtom(LifoAlloc* alloc)
 {
     RegExpBuilder* builder = alloc->newInfallible<RegExpBuilder>(alloc);
 
-    // everything except \x0a, \x0d, \u2028 and \u2029
+    // Everything except \x0a, \x0d, \u2028 and \u2029
 
     CharacterRangeVector* ranges = alloc->newInfallible<CharacterRangeVector>(*alloc);
     ranges->append(CharacterRange::Range(0x0, 0x09));
@@ -1414,6 +1414,38 @@ UnicodeEverythingAtom(LifoAlloc* alloc)
     return builder->ToRegExp();
 }
 
+static inline RegExpTree*
+UnicodeDotAllAtom(LifoAlloc* alloc)
+{
+    RegExpBuilder* builder = alloc->newInfallible<RegExpBuilder>(alloc);
+
+    // Full range excluding surrogates because /s was specified
+
+    CharacterRangeVector* ranges = alloc->newInfallible<CharacterRangeVector>(*alloc);
+    ranges->append(CharacterRange::Range(0x0, unicode::LeadSurrogateMin - 1));
+    ranges->append(CharacterRange::Range(unicode::TrailSurrogateMax + 1, unicode::UTF16Max));
+    builder->AddAtom(alloc->newInfallible<RegExpCharacterClass>(ranges, false));
+
+    builder->NewAlternative();
+
+    builder->AddAtom(RangeAtom(alloc, unicode::LeadSurrogateMin, unicode::LeadSurrogateMax));
+    builder->AddAtom(NegativeLookahead(alloc, unicode::TrailSurrogateMin,
+                                       unicode::TrailSurrogateMax));
+
+    builder->NewAlternative();
+
+    builder->AddAssertion(alloc->newInfallible<RegExpAssertion>(
+        RegExpAssertion::NOT_AFTER_LEAD_SURROGATE));
+    builder->AddAtom(RangeAtom(alloc, unicode::TrailSurrogateMin, unicode::TrailSurrogateMax));
+
+    builder->NewAlternative();
+
+    builder->AddAtom(RangeAtom(alloc, unicode::LeadSurrogateMin, unicode::LeadSurrogateMax));
+    builder->AddAtom(RangeAtom(alloc, unicode::TrailSurrogateMin, unicode::TrailSurrogateMax));
+
+    return builder->ToRegExp();
+}
+
 RegExpTree*
 UnicodeCharacterClassEscapeAtom(LifoAlloc* alloc, char16_t char_class, bool ignore_case)
 {
@@ -1541,13 +1573,25 @@ RegExpParser<CharT>::ParseDisjunction()
           }
           case '.': {
             Advance();
-            // everything except \x0a, \x0d, \u2028 and \u2029
+            
             if (unicode_) {
-                builder->AddAtom(UnicodeEverythingAtom(alloc));
+                if (dotall_) {
+                    // Everything
+                    builder->AddAtom(UnicodeDotAllAtom(alloc));
+                } else {
+                    // Everything except \x0a, \x0d, \u2028 and \u2029
+                    builder->AddAtom(UnicodeEverythingAtom(alloc));
+                }
                 break;
             }
             CharacterRangeVector* ranges = alloc->newInfallible<CharacterRangeVector>(*alloc);
-            CharacterRange::AddClassEscape(alloc, '.', ranges);
+            if (dotall_) {
+                // Everything
+                CharacterRange::AddClassEscape(alloc, '*', ranges);
+            } else {
+                // Everything except \x0a, \x0d, \u2028 and \u2029
+                CharacterRange::AddClassEscape(alloc, '.', ranges);
+            }
             RegExpTree* atom = alloc->newInfallible<RegExpCharacterClass>(ranges, false);
             builder->AddAtom(atom);
             break;
diff --git a/js/src/irregexp/RegExpParser.h b/js/src/irregexp/RegExpParser.h
index 2f02625b5..36c24cd7c 100644
--- a/js/src/irregexp/RegExpParser.h
+++ b/js/src/irregexp/RegExpParser.h
@@ -313,6 +313,7 @@ class RegExpParser
     bool multiline_;
     bool unicode_;
     bool ignore_case_;
+    bool dotall_;
     bool simple_;
     bool contains_anchor_;
     bool is_scanned_for_captures_;
-- 
cgit v1.2.3