ICU-20921 Adding find and compare to StringPiece

icu4c/source/common/stringpiece.cpp

@@ -51,6 +51,47 @@ void StringPiece::set(const char* str) {
     length_ = 0;
 }
 
+int32_t StringPiece::find(StringPiece needle, int32_t offset) {
+  if (length() == 0 && needle.length() == 0) {
+    return 0;
+  }
+  // TODO: Improve to be better than O(N^2)?
+  for (int32_t i = offset; i < length(); i++) {
+    int32_t j = 0;
+    for (; j < needle.length(); i++, j++) {
+      if (data()[i] != needle.data()[j]) {
+        i -= j;
+        goto outer_end;
+      }
+    }
+    return i - j;
+    outer_end: void();
+  }
+  return -1;
+}
+
+int32_t StringPiece::compare(StringPiece other) {
+  int32_t i = 0;
+  for (; i < length(); i++) {
+    if (i == other.length()) {
+      // this is longer
+      return 1;
+    }
+    char a = data()[i];
+    char b = other.data()[i];
+    if (a < b) {
+      return -1;
+    } else if (a > b) {
+      return 1;
+    }
+  }
+  if (i < other.length()) {
+    // other is longer
+    return -1;
+  }
+  return 0;
+}
+
 U_EXPORT UBool U_EXPORT2
 operator==(const StringPiece& x, const StringPiece& y) {
   int32_t len = x.size();

icu4c/source/common/unicode/stringpiece.h

@@ -212,6 +212,26 @@ class U_COMMON_API StringPiece : public UMemory {
     }
   }
 
+#ifndef U_HIDE_DRAFT_API
+  /**
+   * Searches the StringPiece for the given search string (needle);
+   * @param needle The string for which to search.
+   * @param offset Where to start searching within this string (haystack).
+   * @return The offset of needle in haystack, or -1 if not found.
+   * @draft ICU 67
+   */
+  int32_t find(StringPiece needle, int32_t offset);
+
+  /**
+   * Compares this StringPiece with the other StringPiece, with semantics
+   * similar to std::string::compare().
+   * @param other The string to compare to.
+   * @return below zero if this < other; above zero if this > other; 0 if this == other.
+   * @draft ICU 67
+   */
+  int32_t compare(StringPiece other);
+#endif  // U_HIDE_DRAFT_API
+
   /**
    * Maximum integer, used as a default value for substring methods.
    * @stable ICU 4.2

icu4c/source/test/intltest/strtest.cpp

@@ -243,6 +243,7 @@ void StringTest::runIndexedTest(int32_t index, UBool exec, const char *&name, ch
     TESTCASE_AUTO(TestSTLCompatibility);
     TESTCASE_AUTO(TestStringPiece);
     TESTCASE_AUTO(TestStringPieceComparisons);
+    TESTCASE_AUTO(TestStringPieceFind);
     TESTCASE_AUTO(TestStringPieceOther);
 #ifdef U_HAVE_STRING_VIEW
     TESTCASE_AUTO(TestStringPieceStringView);
@@ -407,6 +408,35 @@ StringTest::TestStringPieceComparisons() {
     if(abc==abcd) {
         errln("abc==abcd");
     }
+
+    assertTrue("null<abc", null.compare(abc) < 0);
+    assertTrue("abc>null", abc.compare(null) > 0);
+    assertTrue("abc<abcd", abc.compare(abcd) < 0);
+    assertTrue("abcd>abc", abcd.compare(abc) > 0);
+    assertTrue("abc<abx", abc.compare(abx) < 0);
+    assertTrue("abx>abc", abx.compare(abc) > 0);
+    assertTrue("abx>abcd", abx.compare(abcd) > 0);
+    assertTrue("abcd<abx", abcd.compare(abx) < 0);
+    assertTrue("abx==abx", abx.compare(abx) == 0);
+
+    // Behavior should be the same as std::string::compare
+    {
+        std::string null("");
+        std::string abc("abc");
+        std::string abcd("abcdefg", 4);
+        std::string abx("abx");
+
+        assertTrue("std: null<abc", null.compare(abc) < 0);
+        assertTrue("std: abc>null", abc.compare(null) > 0);
+        assertTrue("std: abc<abcd", abc.compare(abcd) < 0);
+        assertTrue("std: abcd>abc", abcd.compare(abc) > 0);
+        assertTrue("std: abc<abx", abc.compare(abx) < 0);
+        assertTrue("std: abx>abc", abx.compare(abc) > 0);
+        assertTrue("std: abx>abcd", abx.compare(abcd) > 0);
+        assertTrue("std: abcd<abx", abcd.compare(abx) < 0);
+        assertTrue("std: abx==abx", abx.compare(abx) == 0);
+    }
+
     abcd.remove_suffix(1);
     if(abc!=abcd) {
         errln("abc!=abcd.remove_suffix(1)");
@@ -416,6 +446,51 @@ StringTest::TestStringPieceComparisons() {
     }
 }
 
+void
+StringTest::TestStringPieceFind() {
+    struct TestCase {
+        const char* haystack;
+        const char* needle;
+        int32_t expected;
+    } cases[] = {
+        { "", "", 0 },
+        { "", "x", -1 },
+        { "x", "", 0 },
+        { "x", "x", 0 },
+        { "xy", "x", 0 },
+        { "xy", "y", 1 },
+        { "xy", "xy", 0 },
+        { "xy", "xyz", -1 },
+        { "qwerty", "qqw", -1 },
+        { "qwerty", "qw", 0 },
+        { "qwerty", "er", 2 },
+        { "qwerty", "err", -1 },
+        { "qwerty", "ert", 2 },
+        { "qwerty", "ty", 4 },
+        { "qwerty", "tyy", -1 },
+        { "qwerty", "a", -1 },
+        { "qwerty", "abc", -1 }
+    };
+    int32_t caseNumber = 0;
+    for (auto& cas : cases) {
+        StringPiece haystack(cas.haystack);
+        StringPiece needle(cas.needle);
+        assertEquals(Int64ToUnicodeString(caseNumber),
+            cas.expected, haystack.find(needle, 0));
+        // Should be same as std::string::find
+        std::string stdhaystack(cas.haystack);
+        std::string stdneedle(cas.needle);
+        assertEquals(Int64ToUnicodeString(caseNumber) + u" (std)",
+            cas.expected, stdhaystack.find(stdneedle, 0));
+        // Test offsets against std::string::find
+        for (int32_t offset = 0; offset < haystack.length(); offset++) {
+            assertEquals(Int64ToUnicodeString(caseNumber) + "u @ " + Int64ToUnicodeString(offset),
+                stdhaystack.find(stdneedle, offset), haystack.find(needle, offset));
+        }
+        caseNumber++;
+    }
+}
+
 void
 StringTest::TestStringPieceOther() {
     static constexpr char msg[] = "Kapow!";

icu4c/source/test/intltest/strtest.h

@@ -43,6 +43,7 @@ private:
     void TestLowerOrdinal();
     void Test_UTF8_COUNT_TRAIL_BYTES();
     void TestStringPiece();
+    void TestStringPieceFind();
     void TestStringPieceComparisons();
     void TestStringPieceOther();
 #ifdef U_HAVE_STRING_VIEW