ICU-23267 Fix static analyzer errors

icu4c/source/common/putil.cpp

@@ -915,8 +915,15 @@ static UBool compareBinaryFiles(const char* defaultTZFileName, const char* TZFil
             fseek(tzInfo->defaultTZFilePtr, 0, SEEK_END);
             tzInfo->defaultTZFileSize = ftell(tzInfo->defaultTZFilePtr);
         }
-        fseek(file, 0, SEEK_END);
+        if (fseek(file, 0, SEEK_END) != 0) {
+            fclose(file);
+            return false;
+        }
         sizeFile = ftell(file);
+        if (sizeFile == -1) {
+            fclose(file);
+            return false;
+        }
         sizeFileLeft = sizeFile;
 
         if (sizeFile != tzInfo->defaultTZFileSize) {
@@ -926,16 +933,26 @@ static UBool compareBinaryFiles(const char* defaultTZFileName, const char* TZFil
              * compare each byte to determine equality.
              */
             if (tzInfo->defaultTZBuffer == nullptr) {
-                rewind(tzInfo->defaultTZFilePtr);
+                if (fseek(tzInfo->defaultTZFilePtr, 0, SEEK_SET) != 0) {
+                    fclose(file);
+                    return false;
+                }
                 tzInfo->defaultTZBuffer = static_cast<char*>(uprv_malloc(sizeof(char) * tzInfo->defaultTZFileSize));
                 sizeFileRead = fread(tzInfo->defaultTZBuffer, 1, tzInfo->defaultTZFileSize, tzInfo->defaultTZFilePtr);
             }
-            rewind(file);
+            if (fseek(file, 0, SEEK_SET) != 0) {
+                fclose(file);
+                return false;
+            }
             while(sizeFileLeft > 0) {
                 uprv_memset(bufferFile, 0, MAX_READ_SIZE);
                 sizeFileToRead = sizeFileLeft < MAX_READ_SIZE ? sizeFileLeft : MAX_READ_SIZE;
 
                 sizeFileRead = fread(bufferFile, 1, sizeFileToRead, file);
+                if (sizeFileRead == -1) {
+                    fclose(file);
+                    return false;
+                }
                 if (memcmp(tzInfo->defaultTZBuffer + tzInfo->defaultTZPosition, bufferFile, sizeFileRead) != 0) {
                     result = false;
                     break;

icu4c/source/common/ucnv.cpp

@@ -877,6 +877,10 @@ _fromUnicodeWithCallback(UConverterFromUnicodeArgs *pArgs, UErrorCode *err) {
          * and failed to consume all of them.
          * We need to "replay" them from a temporary buffer and convert them first.
          */
+        if (!pArgs->source) {
+            *err=U_ILLEGAL_ARGUMENT_ERROR;
+            return;
+        }
         realSource=pArgs->source;
         realSourceLimit=pArgs->sourceLimit;
         realFlush=pArgs->flush;
@@ -1322,6 +1326,10 @@ _toUnicodeWithCallback(UConverterToUnicodeArgs *pArgs, UErrorCode *err) {
          * and failed to consume all of them.
          * We need to "replay" them from a temporary buffer and convert them first.
          */
+        if (!pArgs->source) {
+            *err=U_ILLEGAL_ARGUMENT_ERROR;
+            return;
+        }
         realSource=pArgs->source;
         realSourceLimit=pArgs->sourceLimit;
         realFlush=pArgs->flush;

icu4c/source/common/udata.cpp

@@ -1291,7 +1291,7 @@ doOpenChoice(const char *path, const char *type, const char *name,
     dataPath = u_getDataDirectory();
 
     /****    Time zone individual files override  */
-    if (isICUData && isTimeZoneFile(name, type)) {
+    if (isICUData && type != nullptr && isTimeZoneFile(name, type)) {
         const char *tzFilesDir = u_getTimeZoneFilesDirectory(pErrorCode);
         if (tzFilesDir[0] != 0) {
 #ifdef UDATA_DEBUG

icu4c/source/common/uenum.cpp

@@ -99,7 +99,8 @@ uenum_unextDefault(UEnumeration* en,
             if (ustr == nullptr) {
                 *status = U_MEMORY_ALLOCATION_ERROR;
             } else {
-                u_charsToUChars(cstr, ustr, len+1);
+                u_charsToUChars(cstr, ustr, len);
+                ustr[len] = 0;
             }
         }
     } else {
@@ -129,7 +130,8 @@ uenum_nextDefault(UEnumeration* en,
             *status = U_MEMORY_ALLOCATION_ERROR;
             return nullptr;
         }
-        u_UCharsToChars(tempUCharVal, tempCharVal, *resultLength + 1);
+        u_UCharsToChars(tempUCharVal, tempCharVal, *resultLength);
+        tempCharVal[*resultLength] = 0;
         return tempCharVal;
     } else {
         *status = U_UNSUPPORTED_ERROR;

icu4c/source/common/uidna.cpp

@@ -915,6 +915,7 @@ uidna_compare(  const char16_t *s1, int32_t length1,
 
     if (U_FAILURE(bufferStatus)) {
         *status = bufferStatus;
+        goto CLEANUP;
     }
 
     // when toASCII is applied all label separators are replaced with FULL_STOP

icu4c/source/common/uniset.cpp

@@ -241,7 +241,7 @@ UnicodeSet& UnicodeSet::copyFrom(const UnicodeSet& o, UBool asThawed) {
     } else if (hasStrings()) {
         strings_->removeAllElements();
     }
-    if (o.stringSpan != nullptr && !asThawed) {
+    if (o.stringSpan != nullptr && strings_ != nullptr && !asThawed) {
         stringSpan = new UnicodeSetStringSpan(*o.stringSpan, *strings_);
         if (stringSpan == nullptr) { // Check for memory allocation error.
             setToBogus();

icu4c/source/common/ushape.cpp

@@ -982,7 +982,7 @@ expandCompositCharAtEnd(char16_t *dest, int32_t sourceLength, int32_t destSize,U
 
     uprv_memset(tempbuffer, 0, (sourceLength+1)*U_SIZEOF_UCHAR);
 
-    while(dest[inpsize-1] == SPACE_CHAR) {
+    while(inpsize>0 && dest[inpsize-1] == SPACE_CHAR) {
         countr++;
         inpsize--;
     }
@@ -1525,6 +1525,11 @@ u_shapeArabic(const char16_t *source, int32_t sourceLength,
                 prevLink = currLink;
                 currLink = getLink(source[i]);
                 if (aggregate_tashkeel && ((prevLink|currLink)&COMBINE) == COMBINE && aggregation_possible) {
+                    if (j<0 || j>=2*sourceLength) {
+                        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
+                        if (tempsource != nullptr) uprv_free(tempsource);
+                        return 0;
+                    }
                     aggregation_possible = 0;
                     tempsource[j] = (prev<source[i]?prev:source[i])-0x064C+0xFC5E;
                     currLink = getLink(tempsource[j]);

icu4c/source/i18n/anytrans.cpp

@@ -359,7 +359,9 @@ void AnyTransliterator::registerIDs() {
 
     UErrorCode ec = U_ZERO_ERROR;
     Hashtable seen(true, ec);
-
+    if (U_FAILURE(ec)) {
+        return;
+    }
     int32_t sourceCount = Transliterator::_countAvailableSources();
     for (int32_t s=0; s<sourceCount; ++s) {
         UnicodeString source;

icu4c/source/i18n/collationsettings.cpp

@@ -195,6 +195,10 @@ CollationSettings::setReorderArrays(const int32_t *codes, int32_t codesLength,
     int32_t *ownedCodes;
     int32_t totalLength = codesLength + rangesLength;
     U_ASSERT(totalLength > 0);
+    if (totalLength <= 0) {
+        errorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
     if(totalLength <= reorderCodesCapacity) {
         ownedCodes = const_cast<int32_t *>(reorderCodes);
     } else {

icu4c/source/i18n/nfrs.cpp

@@ -602,6 +602,9 @@ NFRuleSet::findFractionRuleSetRule(double number) const
     // all the precision we need, and we can do all of the rest
     // of the math using integer arithmetic
     int64_t leastCommonMultiple = rules[0]->getBaseValue();
+    if (leastCommonMultiple == 0) {
+        return nullptr;
+    }
     int64_t numerator;
     {
         for (uint32_t i = 1; i < rules.size(); ++i) {

icu4c/source/i18n/nfrule.cpp

@@ -903,7 +903,8 @@ NFRule::shouldRollBack(int64_t number) const
     // a modulus substitution, its base value isn't an even multiple
     // of 100, and the value we're trying to format _is_ an even
     // multiple of 100.  This is called the "rollback rule."
-    if ((sub1 != nullptr && sub1->isModulusSubstitution()) || (sub2 != nullptr && sub2->isModulusSubstitution())) {
+    if (radix != 0 && ((sub1 != nullptr && sub1->isModulusSubstitution()) ||
+                       (sub2 != nullptr && sub2->isModulusSubstitution()))) {
         int64_t re = util64_pow(radix, exponent);
         return (number % re) == 0 && (baseValue % re) != 0;
     }

icu4c/source/i18n/number_capi.cpp

@@ -380,6 +380,9 @@ usnumf_formatInt64(
         return;
     }
     auto localResult = formatter->fFormatter.formatInt64(value, *ec);
+    if (U_FAILURE(*ec)) {
+        return;
+    }
     result->setTo(std::move(localResult)); 
 }
 

icu4c/source/i18n/scriptset.cpp

@@ -225,9 +225,8 @@ ScriptSet &ScriptSet::parseScripts(const UnicodeString &scriptString, UErrorCode
             }
         }
         if (oneScriptName.length() > 0) {
-            char buf[40];
+            char buf[40] = {};
             oneScriptName.extract(0, oneScriptName.length(), buf, sizeof(buf)-1, US_INV);
-            buf[sizeof(buf)-1] = 0;
             int32_t sc = u_getPropertyValueEnum(UCHAR_SCRIPT, buf);
             if (sc == UCHAR_INVALID_CODE) {
                 status = U_ILLEGAL_ARGUMENT_ERROR;

icu4c/source/i18n/timezone.cpp

@@ -739,7 +739,9 @@ void TimeZone::getOffset(UDate date, UBool local, int32_t& rawOffset,
                               static_cast<uint8_t>(dow), millis,
                               Grego::monthLength(year, month),
                               ec) - rawOffset;
-
+        if (U_FAILURE(ec)) {
+            return;
+        }
         // Recompute if local==true, dstOffset!=0.
         if (pass!=0 || !local || dstOffset == 0) {
             break;

icu4c/source/i18n/tzfmt.cpp

@@ -697,8 +697,8 @@ TimeZoneFormat::format(UTimeZoneFormatStyle style, const TimeZone& tz, UDate dat
         UErrorCode status = U_ZERO_ERROR;
         int32_t rawOffset, dstOffset;
         tz.getOffset(date, false, rawOffset, dstOffset, status);
-        int32_t offset = rawOffset + dstOffset;
         if (U_SUCCESS(status)) {
+            int32_t offset = rawOffset + dstOffset;
             switch (style) {
             case UTZFMT_STYLE_GENERIC_LOCATION:
             case UTZFMT_STYLE_GENERIC_LONG:

icu4c/source/i18n/ucol_sit.cpp

@@ -578,7 +578,10 @@ ucol_normalizeShortDefinitionString(const char *definition,
     if(U_FAILURE(*status)) {
         return 0;
     }
-
+    if(!destination && capacity != 0) {
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
     if(destination) {
         uprv_memset(destination, 0, capacity*sizeof(char));
     }

icu4c/source/i18n/uregex.cpp

@@ -1877,7 +1877,7 @@ int32_t RegexCImpl::split(RegularExpression     *regexp,
                 i++;
 
                 // Set up to extract the capture group contents into the dest buffer.
-                destFields[i] = &destBuf[destIdx];
+                destFields[i] = (destBuf == nullptr) ? nullptr :  &destBuf[destIdx];
                 tStatus = U_ZERO_ERROR;
                 int32_t t = uregex_group(reinterpret_cast<URegularExpression*>(regexp),
                                          groupNum,