Merge pull request #1835 from Expensify/main

Update expensify_prod branch

libstuff/libstuff.cpp

@@ -2805,7 +2805,7 @@ bool SIsValidSQLiteDateModifier(const string& modifier) {
     return true;
 }
 
-bool SREMatch(const string& regExp, const string& input, bool caseSensitive, bool partialMatch, vector<string>* matches) {
+bool SREMatch(const string& regExp, const string& input, bool caseSensitive, bool partialMatch, vector<string>* matches, size_t startOffset, size_t* matchOffset) {
     int errornumber = 0;
     PCRE2_SIZE erroroffset = 0;
     uint32_t matchFlags = 0;
@@ -2824,7 +2824,7 @@ bool SREMatch(const string& regExp, const string& input, bool caseSensitive, boo
     pcre2_set_depth_limit(matchContext, 1000); 
     pcre2_match_data* matchData = pcre2_match_data_create_from_pattern(re, 0);
 
-    int result = pcre2_match(re, (PCRE2_SPTR8)input.c_str(), input.size(), 0, matchFlags, matchData, matchContext); 
+    int result = pcre2_match(re, (PCRE2_SPTR8)input.c_str() + startOffset, input.size() - startOffset, 0, matchFlags, matchData, matchContext);
 
     // Clear out existing matches.
     if (matches) {
@@ -2841,7 +2841,10 @@ bool SREMatch(const string& regExp, const string& input, bool caseSensitive, boo
             if (start == PCRE2_UNSET || end == PCRE2_UNSET) {
                 continue;
             }
-            matches->push_back(input.substr(start, end - start));
+            matches->push_back(input.substr(startOffset + start, end - start));
+            if (i == 0 && matchOffset) {
+                *matchOffset = startOffset + start;
+            }
         }
     }
 
@@ -2852,6 +2855,19 @@ bool SREMatch(const string& regExp, const string& input, bool caseSensitive, boo
     return result > 0;
 }
 
+vector<vector <string>> SREMatchAll(const string& regExp, const string& input, bool caseSensitive) {
+    vector<vector<string>> returnValue;
+    vector<string> matches;
+    size_t startOffset = 0;
+    size_t matchOffset = 0;
+    while(SREMatch(regExp, input, caseSensitive, true, &matches, startOffset, &matchOffset)) {
+        returnValue.push_back(matches);
+        startOffset = matchOffset + matches[0].size();
+    }
+
+    return returnValue;
+}
+
 string SREReplace(const string& regExp, const string& input, const string& replacement, bool caseSensitive) {
     char* output = nullptr;
     size_t outSize = 0;

libstuff/libstuff.h

@@ -388,7 +388,21 @@ bool SConstantTimeEquals(const string& secret, const string& userInput);
 bool SConstantTimeIEquals(const string& secret, const string& userInput);
 
 // Unless `partialMatch` is specified, perform a full regex match (the '^' and '$' symbols are implicit).
-bool SREMatch(const string& regExp, const string& input, bool caseSensitive = true, bool partialMatch = false, vector<string>* matches = nullptr);
+//
+// If `matches` is supplied it will be cleared, and any matches to the expression will fill it. The first entry in matches will be the entire matched portion of the string,
+// and any following entries will be matched parenthesized subgroups.
+//
+// startOffset can be supplied to ignore the first part of the input string.
+//
+// If matchOffset is supplied, and a match is found, it will be set to the offset of the first character of the matched substring.
+// To find the end of the matched substring, you can do something like matchOffset + matches[0].size().
+bool SREMatch(const string& regExp, const string& input, bool caseSensitive = true, bool partialMatch = false, vector<string>* matches = nullptr, size_t startOffset = 0, size_t* matchOffset = nullptr);
+
+// Matches every instance of regExp in the input string. Returns a vector of vectors or strings.
+// The outer vector has one entry for each match found. The inner vectors contain first the entire matched substring, and following that, each match group
+vector<vector <string>> SREMatchAll(const string& regExp, const string& input, bool caseSensitive = true);
+
+// Replaces all instances of the matched `regExp` with `replacement` in `input`.
 string SREReplace(const string& regExp, const string& input, const string& replacement, bool caseSensitive = true);
 
 // Redact values that should not be logged.

test/tests/LibStuffTest.cpp

@@ -672,6 +672,14 @@ struct LibStuff : tpunit::TestFixture {
         ASSERT_EQUAL(matches[1], "this");
         ASSERT_EQUAL(matches[2], "contains");
         ASSERT_EQUAL(matches[3], "cat");
+
+        // Let's do multiple matches.
+        string catNames = "kitty Whiskers, kitty Mittens, kitty Snowball";
+        auto matchList = SREMatchAll(R"(kitty\s+(\w+))", catNames);
+        ASSERT_EQUAL(matchList.size(), 3);
+        ASSERT_EQUAL(matchList[0][1], "Whiskers");
+        ASSERT_EQUAL(matchList[1][1], "Mittens");
+        ASSERT_EQUAL(matchList[2][1], "Snowball");
     }
 
     void SREReplaceTest() {