ICU-10336 add leniency controls for lenient like behavior

X-SVN-Rev: 35319

diff --git a/icu4c/source/i18n/smpdtfmt.cpp b/icu4c/source/i18n/smpdtfmt.cpp
index 086a7462288150bb9d1858c4eb2e1162e574ad52..62c7d22131ee964eccfe80690c50ad31cfa089fd 100644 (file)
--- a/icu4c/source/i18n/smpdtfmt.cpp
+++ b/icu4c/source/i18n/smpdtfmt.cpp
@@ -819,6 +819,7 @@ void SimpleDateFormat::initializeBooleanAttributes()
     setBooleanAttribute(UDAT_PARSE_ALLOW_WHITESPACE, true, status);
     setBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, true, status);
     setBooleanAttribute(UDAT_PARSE_PARTIAL_MATCH, true, status);
+    setBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, true, status);
 }
 
 /* Define one-century window into which to disambiguate dates using
@@ -2722,17 +2723,25 @@ int32_t SimpleDateFormat::subParse(const UnicodeString& text, int32_t& start, UC
             }
             int32_t newStart = 0;
             if (patternCharIndex==UDAT_MONTH_FIELD) {
-                newStart = matchString(text, start, UCAL_MONTH, fSymbols->fMonths, fSymbols->fMonthsCount, wideMonthPat, cal); // try MMMM
-                if (newStart > 0) {
-                    return newStart;
+                if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
+                    newStart = matchString(text, start, UCAL_MONTH, fSymbols->fMonths, fSymbols->fMonthsCount, wideMonthPat, cal); // try MMMM
+                    if (newStart > 0) {
+                        return newStart;
+                    }
+                }
+                if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
+                    newStart = matchString(text, start, UCAL_MONTH, fSymbols->fShortMonths, fSymbols->fShortMonthsCount, shortMonthPat, cal); // try MMM
                 }
-                newStart = matchString(text, start, UCAL_MONTH, fSymbols->fShortMonths, fSymbols->fShortMonthsCount, shortMonthPat, cal); // try MMM
             } else {
-                newStart = matchString(text, start, UCAL_MONTH, fSymbols->fStandaloneMonths, fSymbols->fStandaloneMonthsCount, wideMonthPat, cal); // try LLLL
-                if (newStart > 0) {
-                    return newStart;
+                if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
+                    newStart = matchString(text, start, UCAL_MONTH, fSymbols->fStandaloneMonths, fSymbols->fStandaloneMonthsCount, wideMonthPat, cal); // try LLLL
+                    if (newStart > 0) {
+                        return newStart;
+                    }
+                }
+                if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
+                    newStart = matchString(text, start, UCAL_MONTH, fSymbols->fStandaloneShortMonths, fSymbols->fStandaloneShortMonthsCount, shortMonthPat, cal); // try LLL
                 }
-                newStart = matchString(text, start, UCAL_MONTH, fSymbols->fStandaloneShortMonths, fSymbols->fStandaloneShortMonthsCount, shortMonthPat, cal); // try LLL
             }
             if (newStart > 0 || !getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))  // currently we do not try to parse MMMMM/LLLLL: #8860
                 return newStart;
@@ -2784,22 +2793,30 @@ int32_t SimpleDateFormat::subParse(const UnicodeString& text, int32_t& start, UC
             // Want to be able to parse both short and long forms.
             // Try count == 4 (EEEE) wide first:
             int32_t newStart = 0;
-            if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
-                                      fSymbols->fWeekdays, fSymbols->fWeekdaysCount, NULL, cal)) > 0)
-                return newStart;
+            if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
+                if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
+                                          fSymbols->fWeekdays, fSymbols->fWeekdaysCount, NULL, cal)) > 0)
+                    return newStart;
+            }
             // EEEE wide failed, now try EEE abbreviated
-            else if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
-                                   fSymbols->fShortWeekdays, fSymbols->fShortWeekdaysCount, NULL, cal)) > 0)
-                return newStart;
+            if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
+                if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
+                                       fSymbols->fShortWeekdays, fSymbols->fShortWeekdaysCount, NULL, cal)) > 0)
+                    return newStart;
+            }
             // EEE abbreviated failed, now try EEEEEE short
-            else if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
-                                   fSymbols->fShorterWeekdays, fSymbols->fShorterWeekdaysCount, NULL, cal)) > 0)
-                return newStart;
+            if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 6) {
+                if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
+                                       fSymbols->fShorterWeekdays, fSymbols->fShorterWeekdaysCount, NULL, cal)) > 0)
+                    return newStart;
+            }
             // EEEEEE short failed, now try EEEEE narrow
-            else if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
-                                   fSymbols->fNarrowWeekdays, fSymbols->fNarrowWeekdaysCount, NULL, cal)) > 0)
-                return newStart;
-            else if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status) || patternCharIndex == UDAT_DAY_OF_WEEK_FIELD)
+            if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 5) {
+                if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
+                                       fSymbols->fNarrowWeekdays, fSymbols->fNarrowWeekdaysCount, NULL, cal)) > 0)
+                    return newStart;
+            }
+            if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status) || patternCharIndex == UDAT_DAY_OF_WEEK_FIELD)
                 return newStart;
             // else we allowing parsing as number, below
         }
@@ -2816,16 +2833,22 @@ int32_t SimpleDateFormat::subParse(const UnicodeString& text, int32_t& start, UC
             // Want to be able to parse both short and long forms.
             // Try count == 4 (cccc) first:
             int32_t newStart = 0;
-            if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
+            if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
+                if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
                                       fSymbols->fStandaloneWeekdays, fSymbols->fStandaloneWeekdaysCount, NULL, cal)) > 0)
-                return newStart;
-            else if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
+                    return newStart;
+            }
+            if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
+                if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
                                           fSymbols->fStandaloneShortWeekdays, fSymbols->fStandaloneShortWeekdaysCount, NULL, cal)) > 0)
-                return newStart;
-            else if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
+                    return newStart;
+            }
+            if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 6) {
+                if ((newStart = matchString(text, start, UCAL_DAY_OF_WEEK,
                                           fSymbols->fStandaloneShorterWeekdays, fSymbols->fStandaloneShorterWeekdaysCount, NULL, cal)) > 0)
-                return newStart;
-            else if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))
+                    return newStart;
+            }
+            if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))
                 return newStart;
             // else we allowing parsing as number, below
         }
@@ -2859,15 +2882,21 @@ int32_t SimpleDateFormat::subParse(const UnicodeString& text, int32_t& start, UC
             // Try count == 4 first:
             int32_t newStart = 0;
 
-            if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
+            if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
+                if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
                                       fSymbols->fQuarters, fSymbols->fQuartersCount, cal)) > 0)
-                return newStart;
-            else if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
+                    return newStart;
+            }
+            if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
+                if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
                                           fSymbols->fShortQuarters, fSymbols->fShortQuartersCount, cal)) > 0)
-                return newStart;
-            else if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))
+                    return newStart;
+            }
+            if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))
                 return newStart;
             // else we allowing parsing as number, below
+            if(!getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status))
+                return -start;
         }
         break;
 
@@ -2885,15 +2914,21 @@ int32_t SimpleDateFormat::subParse(const UnicodeString& text, int32_t& start, UC
             // Try count == 4 first:
             int32_t newStart = 0;
 
-            if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
+            if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 4) {
+                if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
                                       fSymbols->fStandaloneQuarters, fSymbols->fStandaloneQuartersCount, cal)) > 0)
-                return newStart;
-            else if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
+                    return newStart;
+            }
+            if(getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status) || count == 3) {
+                if ((newStart = matchQuarterString(text, start, UCAL_MONTH,
                                           fSymbols->fStandaloneShortQuarters, fSymbols->fStandaloneShortQuartersCount, cal)) > 0)
-                return newStart;
-            else if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))
+                    return newStart;
+            }
+            if (!getBooleanAttribute(UDAT_PARSE_ALLOW_NUMERIC, status))
                 return newStart;
             // else we allowing parsing as number, below
+            if(!getBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, status))
+                return -start;
         }
         break;
 

diff --git a/icu4c/source/i18n/unicode/udat.h b/icu4c/source/i18n/unicode/udat.h
index 765941542c1e70b4632b988af8600662134a1eca..477c0fe3abfc5f4d391b6ed9ef3acaeb4fe25659 100644 (file)
--- a/icu4c/source/i18n/unicode/udat.h
+++ b/icu4c/source/i18n/unicode/udat.h
@@ -862,12 +862,18 @@ typedef enum UDateFormatBooleanAttribute {
      * @draft ICU 53
      */
     UDAT_PARSE_PARTIAL_MATCH = 2,
+    /**  
+     * indicates tolerance of pattern mismatch between input data and specified format pattern. 
+     * e.g. accepting "September" for a month pattern of MMM ("Sep")   
+     * @draft ICU 53
+     */ 
+    UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH = 3,
 #endif /* U_HIDE_DRAFT_API */
     /**
      * count boolean date format constants
      * @draft ICU 53
      */
-    UDAT_BOOLEAN_ATTRIBUTE_COUNT = 3
+    UDAT_BOOLEAN_ATTRIBUTE_COUNT = 4
 } UDateFormatBooleanAttribute;
 
 #ifndef U_HIDE_DRAFT_API

diff --git a/icu4c/source/test/intltest/dtfmttst.cpp b/icu4c/source/test/intltest/dtfmttst.cpp
index 807c545dffd902e8a9d55e193d15b176002103ed..44e44583ed527c52af035fc17437bdd50737c00f 100644 (file)
--- a/icu4c/source/test/intltest/dtfmttst.cpp
+++ b/icu4c/source/test/intltest/dtfmttst.cpp
@@ -101,6 +101,8 @@ void DateFormatTest::runIndexedTest( int32_t index, UBool exec, const char* &nam
     */
     TESTCASE_AUTO(TestDotAndAtLeniency);
     TESTCASE_AUTO(TestDateFormatLeniency);
+    TESTCASE_AUTO(TestParseMultiPatternMatch);
+
     TESTCASE_AUTO_END;
 }
 
@@ -4199,7 +4201,7 @@ typedef struct {
     UBool leniency;
     UnicodeString parseString;
     UnicodeString pattern;
-    UnicodeString expectedResult;       // null indicates expected error
+    UnicodeString expectedResult;       // empty string indicates expected error
 } TestDateFormatLeniencyItem;
 
 void DateFormatTest::TestDateFormatLeniency() {
@@ -4252,6 +4254,98 @@ void DateFormatTest::TestDateFormatLeniency() {
     }
 }
 
+
+typedef struct {
+    UBool leniency;
+    UnicodeString parseString;
+    UnicodeString pattern;
+    UnicodeString expectedResult;       // empty string indicates expected error
+} TestMultiPatternMatchItem;
+
+void DateFormatTest::TestParseMultiPatternMatch() {
+        // For details see http://bugs.icu-project.org/trac/ticket/10336 
+        
+    const TestMultiPatternMatchItem items[] = {
+             // leniency    parse String                                 pattern                               expected result 
+            {true,       UnicodeString("2013-Sep 13"),                UnicodeString("yyyy-MMM dd"),         UnicodeString("2013-Sep 13")}, 
+            {true,       UnicodeString("2013-September 14"),          UnicodeString("yyyy-MMM dd"),         UnicodeString("2013-Sep 14")}, 
+            {false,      UnicodeString("2013-September 15"),          UnicodeString("yyyy-MMM dd"),         UnicodeString("")}, 
+            {false,      UnicodeString("2013-September 16"),          UnicodeString("yyyy-MMMM dd"),        UnicodeString("2013-September 16")}, 
+            {true,       UnicodeString("2013-Sep 17"),                UnicodeString("yyyy-LLL dd"),         UnicodeString("2013-Sep 17")}, 
+            {true,       UnicodeString("2013-September 18"),          UnicodeString("yyyy-LLL dd"),         UnicodeString("2013-Sep 18")}, 
+            {false,      UnicodeString("2013-September 19"),          UnicodeString("yyyy-LLL dd"),         UnicodeString("")}, 
+            {false,      UnicodeString("2013-September 20"),          UnicodeString("yyyy-LLLL dd"),        UnicodeString("2013-September 20")}, 
+            {true,       UnicodeString("2013 Sat Sep 21"),            UnicodeString("yyyy EEE MMM dd"),     UnicodeString("2013 Sat Sep 21")}, 
+            {true,       UnicodeString("2013 Sunday Sep 22"),         UnicodeString("yyyy EEE MMM dd"),     UnicodeString("2013 Sun Sep 22")}, 
+            {false,      UnicodeString("2013 Monday Sep 23"),         UnicodeString("yyyy EEE MMM dd"),     UnicodeString("")}, 
+            {false,      UnicodeString("2013 Tuesday Sep 24"),        UnicodeString("yyyy EEEE MMM dd"),    UnicodeString("2013 Tuesday Sep 24")}, 
+            {true,       UnicodeString("2013 Wed Sep 25"),            UnicodeString("yyyy eee MMM dd"),     UnicodeString("2013 Wed Sep 25")}, 
+            {true,       UnicodeString("2013 Thu Sep 26"),            UnicodeString("yyyy eee MMM dd"),     UnicodeString("2013 Thu Sep 26")}, 
+            {false,      UnicodeString("2013 Friday Sep 27"),         UnicodeString("yyyy eee MMM dd"),     UnicodeString("")}, 
+            {false,      UnicodeString("2013 Saturday Sep 28"),       UnicodeString("yyyy eeee MMM dd"),    UnicodeString("2013 Saturday Sep 28")}, 
+            {true,       UnicodeString("2013 Sun Sep 29"),            UnicodeString("yyyy ccc MMM dd"),     UnicodeString("2013 Sun Sep 29")}, 
+            {true,       UnicodeString("2013 Monday Sep 30"),         UnicodeString("yyyy ccc MMM dd"),     UnicodeString("2013 Mon Sep 30")}, 
+            {false,      UnicodeString("2013 Sunday Oct 13"),         UnicodeString("yyyy ccc MMM dd"),     UnicodeString("")}, 
+            {false,      UnicodeString("2013 Monday Oct 14"),         UnicodeString("yyyy cccc MMM dd"),    UnicodeString("2013 Monday Oct 14")}, 
+            {true,       UnicodeString("2013 Oct 15 Q4"),             UnicodeString("yyyy MMM dd QQQ"),     UnicodeString("2013 Oct 15 Q4")}, 
+            {true,       UnicodeString("2013 Oct 16 4th quarter"),    UnicodeString("yyyy MMM dd QQQ"),     UnicodeString("2013 Oct 16 Q4")}, 
+            {false,      UnicodeString("2013 Oct 17 4th quarter"),    UnicodeString("yyyy MMM dd QQQ"),     UnicodeString("")}, 
+            {false,      UnicodeString("2013 Oct 18 Q4"),             UnicodeString("yyyy MMM dd QQQ"),     UnicodeString("2013 Oct 18 Q4")}, 
+            {true,       UnicodeString("2013 Oct 19 Q4"),             UnicodeString("yyyy MMM dd qqqq"),    UnicodeString("2013 Oct 19 4th quarter")}, 
+            {true,       UnicodeString("2013 Oct 20 4th quarter"),    UnicodeString("yyyy MMM dd qqqq"),    UnicodeString("2013 Oct 20 4th quarter")}, 
+            {false,      UnicodeString("2013 Oct 21 Q4"),             UnicodeString("yyyy MMM dd qqqq"),    UnicodeString("")}, 
+            {false,      UnicodeString("2013 Oct 22 4th quarter"),    UnicodeString("yyyy MMM dd qqqq"),    UnicodeString("2013 Oct 22 4th quarter")},
+            {false,      NULL,                                        NULL,                                 NULL},
+       };      
+
+    UErrorCode status = U_ZERO_ERROR;
+    LocalPointer<Calendar> cal(Calendar::createInstance(status));
+    if (U_FAILURE(status)) {
+        dataerrln(UnicodeString("FAIL: Unable to create Calendar for default timezone and locale."));
+        return;
+    }
+    const TestMultiPatternMatchItem * itemPtr;
+    DateFormat* sdmft = DateFormat::createDateInstance();
+    int32_t cnt = 0;
+    for (itemPtr = items; itemPtr->parseString != NULL; itemPtr++ ) {                                        
+       status = U_ZERO_ERROR;
+       ParsePosition pos(0);
+       ((SimpleDateFormat*) sdmft)->applyPattern(itemPtr->pattern);
+       if (U_FAILURE(status)) {
+           dataerrln("Unable to create SimpleDateFormat - %s", u_errorName(status));
+           continue;
+       }
+       ++cnt;
+       sdmft->setLenient(itemPtr->leniency);
+       sdmft->setBooleanAttribute(UDAT_PARSE_MULTIPLE_PATTERNS_FOR_MATCH, itemPtr->leniency, status);
+       UDate d = sdmft->parse(itemPtr->parseString, pos); 
+       
+       if(itemPtr->expectedResult.length() == 0) {
+           if(pos.getErrorIndex() != -1) {
+               continue;
+           } else {
+                errln("error: unexpected parse success - " + itemPtr->parseString + 
+                    " - error index " + pos.getErrorIndex() + 
+                    " - leniency " + itemPtr->leniency);
+                continue;
+           }
+       }
+       if(pos.getErrorIndex() != -1) { 
+               errln("error: parse error for string - " +itemPtr->parseString + " -- idx["+pos.getIndex()+"] errIdx["+pos.getErrorIndex()+"]"); 
+               continue; 
+           }
+
+       UnicodeString formatResult(""); 
+       sdmft->format(d, formatResult);
+       if(formatResult.compare(itemPtr->expectedResult) != 0) { 
+               errln("error: unexpected format result. expected[" + itemPtr->expectedResult + "]  but result was[" + formatResult + "]"); 
+           } else { 
+            logln("formatted results match! - " + formatResult);  
+           } 
+    }
+    delete sdmft;
+ }
+
 #endif /* #if !UCONFIG_NO_FORMATTING */
 
 //eof

diff --git a/icu4c/source/test/intltest/dtfmttst.h b/icu4c/source/test/intltest/dtfmttst.h
index 1cdfe8f77726b83c9297d83831925bc278c1b934..3d04ba7346e73d573607d9d4f9989831a768dbe2 100644 (file)
--- a/icu4c/source/test/intltest/dtfmttst.h
+++ b/icu4c/source/test/intltest/dtfmttst.h
@@ -1,6 +1,6 @@
 /********************************************************************
  * COPYRIGHT: 
- * Copyright (c) 1997-2013, International Business Machines Corporation and
+ * Copyright (c) 1997-2014, International Business Machines Corporation and
  * others. All Rights Reserved.
  ********************************************************************/
 
@@ -231,6 +231,8 @@ public:
 
     void TestDateFormatLeniency();
 
+    void TestParseMultiPatternMatch();
+
 private:
     UBool showParse(DateFormat &format, const UnicodeString &formattedString);