michael@0: /*
michael@0: *******************************************************************************
michael@0: *   Copyright (C) 2011-2013, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: *******************************************************************************
michael@0: *   file name:  ppucd.cpp
michael@0: *   encoding:   US-ASCII
michael@0: *   tab size:   8 (not used)
michael@0: *   indentation:4
michael@0: *
michael@0: *   created on: 2011dec11
michael@0: *   created by: Markus W. Scherer
michael@0: */
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/uchar.h"
michael@0: #include "charstr.h"
michael@0: #include "cstring.h"
michael@0: #include "ppucd.h"
michael@0: #include "uassert.h"
michael@0: #include "uparse.h"
michael@0: 
michael@0: #include <stdio.h>
michael@0: #include <string.h>
michael@0: 
michael@0: #define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
michael@0: 
michael@0: U_NAMESPACE_BEGIN
michael@0: 
michael@0: PropertyNames::~PropertyNames() {}
michael@0: 
michael@0: int32_t
michael@0: PropertyNames::getPropertyEnum(const char *name) const {
michael@0:     return u_getPropertyEnum(name);
michael@0: }
michael@0: 
michael@0: int32_t
michael@0: PropertyNames::getPropertyValueEnum(int32_t property, const char *name) const {
michael@0:     return u_getPropertyValueEnum((UProperty)property, name);
michael@0: }
michael@0: 
michael@0: UniProps::UniProps()
michael@0:         : start(U_SENTINEL), end(U_SENTINEL),
michael@0:           bmg(U_SENTINEL), bpb(U_SENTINEL),
michael@0:           scf(U_SENTINEL), slc(U_SENTINEL), stc(U_SENTINEL), suc(U_SENTINEL),
michael@0:           digitValue(-1), numericValue(NULL),
michael@0:           name(NULL), nameAlias(NULL) {
michael@0:     memset(binProps, 0, sizeof(binProps));
michael@0:     memset(intProps, 0, sizeof(intProps));
michael@0:     memset(age, 0, 4);
michael@0: }
michael@0: 
michael@0: UniProps::~UniProps() {}
michael@0: 
michael@0: const int32_t PreparsedUCD::kNumLineBuffers;
michael@0: 
michael@0: PreparsedUCD::PreparsedUCD(const char *filename, UErrorCode &errorCode)
michael@0:         : icuPnames(new PropertyNames()), pnames(icuPnames),
michael@0:           file(NULL),
michael@0:           defaultLineIndex(-1), blockLineIndex(-1), lineIndex(0),
michael@0:           lineNumber(0),
michael@0:           lineType(NO_LINE),
michael@0:           fieldLimit(NULL), lineLimit(NULL) {
michael@0:     if(U_FAILURE(errorCode)) { return; }
michael@0: 
michael@0:     if(filename==NULL || *filename==0 || (*filename=='-' && filename[1]==0)) {
michael@0:         filename=NULL;
michael@0:         file=stdin;
michael@0:     } else {
michael@0:         file=fopen(filename, "r");
michael@0:     }
michael@0:     if(file==NULL) {
michael@0:         perror("error opening preparsed UCD");
michael@0:         fprintf(stderr, "error opening preparsed UCD file %s\n", filename ? filename : "\"no file name given\"");
michael@0:         errorCode=U_FILE_ACCESS_ERROR;
michael@0:         return;
michael@0:     }
michael@0: 
michael@0:     memset(ucdVersion, 0, 4);
michael@0:     lines[0][0]=0;
michael@0: }
michael@0: 
michael@0: PreparsedUCD::~PreparsedUCD() {
michael@0:     if(file!=stdin) {
michael@0:         fclose(file);
michael@0:     }
michael@0:     delete icuPnames;
michael@0: }
michael@0: 
michael@0: // Same order as the LineType values.
michael@0: static const char *lineTypeStrings[]={
michael@0:     NULL,
michael@0:     NULL,
michael@0:     "ucd",
michael@0:     "property",
michael@0:     "binary",
michael@0:     "value",
michael@0:     "defaults",
michael@0:     "block",
michael@0:     "cp",
michael@0:     "algnamesrange"
michael@0: };
michael@0: 
michael@0: PreparsedUCD::LineType
michael@0: PreparsedUCD::readLine(UErrorCode &errorCode) {
michael@0:     if(U_FAILURE(errorCode)) { return NO_LINE; }
michael@0:     // Select the next available line buffer.
michael@0:     while(!isLineBufferAvailable(lineIndex)) {
michael@0:         ++lineIndex;
michael@0:         if (lineIndex == kNumLineBuffers) {
michael@0:             lineIndex = 0;
michael@0:         }
michael@0:     }
michael@0:     char *line=lines[lineIndex];
michael@0:     *line=0;
michael@0:     lineLimit=fieldLimit=line;
michael@0:     lineType=NO_LINE;
michael@0:     char *result=fgets(line, sizeof(lines[0]), file);
michael@0:     if(result==NULL) {
michael@0:         if(ferror(file)) {
michael@0:             perror("error reading preparsed UCD");
michael@0:             fprintf(stderr, "error reading preparsed UCD before line %ld\n", (long)lineNumber);
michael@0:             errorCode=U_FILE_ACCESS_ERROR;
michael@0:         }
michael@0:         return NO_LINE;
michael@0:     }
michael@0:     ++lineNumber;
michael@0:     if(*line=='#') {
michael@0:         fieldLimit=strchr(line, 0);
michael@0:         return lineType=EMPTY_LINE;
michael@0:     }
michael@0:     // Remove trailing /r/n.
michael@0:     char c;
michael@0:     char *limit=strchr(line, 0);
michael@0:     while(line<limit && ((c=*(limit-1))=='\n' || c=='\r')) { --limit; }
michael@0:     // Remove trailing white space.
michael@0:     while(line<limit && ((c=*(limit-1))==' ' || c=='\t')) { --limit; }
michael@0:     *limit=0;
michael@0:     lineLimit=limit;
michael@0:     if(line==limit) {
michael@0:         fieldLimit=limit;
michael@0:         return lineType=EMPTY_LINE;
michael@0:     }
michael@0:     // Split by ';'.
michael@0:     char *semi=line;
michael@0:     while((semi=strchr(semi, ';'))!=NULL) { *semi++=0; }
michael@0:     fieldLimit=strchr(line, 0);
michael@0:     // Determine the line type.
michael@0:     int32_t type;
michael@0:     for(type=EMPTY_LINE+1;; ++type) {
michael@0:         if(type==LINE_TYPE_COUNT) {
michael@0:             fprintf(stderr,
michael@0:                     "error in preparsed UCD: unknown line type (first field) '%s' on line %ld\n",
michael@0:                     line, (long)lineNumber);
michael@0:             errorCode=U_PARSE_ERROR;
michael@0:             return NO_LINE;
michael@0:         }
michael@0:         if(0==strcmp(line, lineTypeStrings[type])) {
michael@0:             break;
michael@0:         }
michael@0:     }
michael@0:     lineType=(LineType)type;
michael@0:     if(lineType==UNICODE_VERSION_LINE && fieldLimit<lineLimit) {
michael@0:         u_versionFromString(ucdVersion, fieldLimit+1);
michael@0:     }
michael@0:     return lineType;
michael@0: }
michael@0: 
michael@0: const char *
michael@0: PreparsedUCD::firstField() {
michael@0:     char *field=lines[lineIndex];
michael@0:     fieldLimit=strchr(field, 0);
michael@0:     return field;
michael@0: }
michael@0: 
michael@0: const char *
michael@0: PreparsedUCD::nextField() {
michael@0:     if(fieldLimit==lineLimit) { return NULL; }
michael@0:     char *field=fieldLimit+1;
michael@0:     fieldLimit=strchr(field, 0);
michael@0:     return field;
michael@0: }
michael@0: 
michael@0: const UniProps *
michael@0: PreparsedUCD::getProps(UnicodeSet &newValues, UErrorCode &errorCode) {
michael@0:     if(U_FAILURE(errorCode)) { return NULL; }
michael@0:     newValues.clear();
michael@0:     if(!lineHasPropertyValues()) {
michael@0:         errorCode=U_ILLEGAL_ARGUMENT_ERROR;
michael@0:         return NULL;
michael@0:     }
michael@0:     firstField();
michael@0:     const char *field=nextField();
michael@0:     if(field==NULL) {
michael@0:         // No range field after the type.
michael@0:         fprintf(stderr,
michael@0:                 "error in preparsed UCD: missing default/block/cp range field "
michael@0:                 "(no second field) on line %ld\n",
michael@0:                 (long)lineNumber);
michael@0:         errorCode=U_PARSE_ERROR;
michael@0:         return NULL;
michael@0:     }
michael@0:     UChar32 start, end;
michael@0:     if(!parseCodePointRange(field, start, end, errorCode)) { return NULL; }
michael@0:     UniProps *props;
michael@0:     switch(lineType) {
michael@0:     case DEFAULTS_LINE:
michael@0:         if(defaultLineIndex>=0) {
michael@0:             fprintf(stderr,
michael@0:                     "error in preparsed UCD: second line with default properties on line %ld\n",
michael@0:                     (long)lineNumber);
michael@0:             errorCode=U_PARSE_ERROR;
michael@0:             return NULL;
michael@0:         }
michael@0:         if(start!=0 || end!=0x10ffff) {
michael@0:             fprintf(stderr,
michael@0:                     "error in preparsed UCD: default range must be 0..10FFFF, not '%s' on line %ld\n",
michael@0:                     field, (long)lineNumber);
michael@0:             errorCode=U_PARSE_ERROR;
michael@0:             return NULL;
michael@0:         }
michael@0:         props=&defaultProps;
michael@0:         defaultLineIndex=lineIndex;
michael@0:         break;
michael@0:     case BLOCK_LINE:
michael@0:         blockProps=defaultProps;  // Block inherits default properties.
michael@0:         props=&blockProps;
michael@0:         blockLineIndex=lineIndex;
michael@0:         break;
michael@0:     case CP_LINE:
michael@0:         if(blockProps.start<=start && end<=blockProps.end) {
michael@0:             // Code point range fully inside the last block inherits the block properties.
michael@0:             cpProps=blockProps;
michael@0:         } else if(start>blockProps.end || end<blockProps.start) {
michael@0:             // Code point range fully outside the last block inherits the default properties.
michael@0:             cpProps=defaultProps;
michael@0:         } else {
michael@0:             // Code point range partially overlapping with the last block is illegal.
michael@0:             fprintf(stderr,
michael@0:                     "error in preparsed UCD: cp range %s on line %ld only "
michael@0:                     "partially overlaps with block range %04lX..%04lX\n",
michael@0:                     field, (long)lineNumber, (long)blockProps.start, (long)blockProps.end);
michael@0:             errorCode=U_PARSE_ERROR;
michael@0:             return NULL;
michael@0:         }
michael@0:         props=&cpProps;
michael@0:         break;
michael@0:     default:
michael@0:         // Will not occur because of the range check above.
michael@0:         errorCode=U_ILLEGAL_ARGUMENT_ERROR;
michael@0:         return NULL;
michael@0:     }
michael@0:     props->start=start;
michael@0:     props->end=end;
michael@0:     while((field=nextField())!=NULL) {
michael@0:         if(!parseProperty(*props, field, newValues, errorCode)) { return NULL; }
michael@0:     }
michael@0:     return props;
michael@0: }
michael@0: 
michael@0: static const struct {
michael@0:     const char *name;
michael@0:     int32_t prop;
michael@0: } ppucdProperties[]={
michael@0:     { "Name_Alias", PPUCD_NAME_ALIAS },
michael@0:     { "Conditional_Case_Mappings", PPUCD_CONDITIONAL_CASE_MAPPINGS },
michael@0:     { "Turkic_Case_Folding", PPUCD_TURKIC_CASE_FOLDING }
michael@0: };
michael@0: 
michael@0: // Returns TRUE for "ok to continue parsing fields".
michael@0: UBool
michael@0: PreparsedUCD::parseProperty(UniProps &props, const char *field, UnicodeSet &newValues,
michael@0:                             UErrorCode &errorCode) {
michael@0:     CharString pBuffer;
michael@0:     const char *p=field;
michael@0:     const char *v=strchr(p, '=');
michael@0:     int binaryValue;
michael@0:     if(*p=='-') {
michael@0:         if(v!=NULL) {
michael@0:             fprintf(stderr,
michael@0:                     "error in preparsed UCD: mix of binary-property-no and "
michael@0:                     "enum-property syntax '%s' on line %ld\n",
michael@0:                     field, (long)lineNumber);
michael@0:             errorCode=U_PARSE_ERROR;
michael@0:             return FALSE;
michael@0:         }
michael@0:         binaryValue=0;
michael@0:         ++p;
michael@0:     } else if(v==NULL) {
michael@0:         binaryValue=1;
michael@0:     } else {
michael@0:         binaryValue=-1;
michael@0:         // Copy out the property name rather than modifying the field (writing a NUL).
michael@0:         pBuffer.append(p, (int32_t)(v-p), errorCode);
michael@0:         p=pBuffer.data();
michael@0:         ++v;
michael@0:     }
michael@0:     int32_t prop=pnames->getPropertyEnum(p);
michael@0:     if(prop<0) {
michael@0:         for(int32_t i=0;; ++i) {
michael@0:             if(i==LENGTHOF(ppucdProperties)) {
michael@0:                 // Ignore unknown property names.
michael@0:                 return TRUE;
michael@0:             }
michael@0:             if(0==uprv_stricmp(p, ppucdProperties[i].name)) {
michael@0:                 prop=ppucdProperties[i].prop;
michael@0:                 U_ASSERT(prop>=0);
michael@0:                 break;
michael@0:             }
michael@0:         }
michael@0:     }
michael@0:     if(prop<UCHAR_BINARY_LIMIT) {
michael@0:         if(binaryValue>=0) {
michael@0:             props.binProps[prop]=(UBool)binaryValue;
michael@0:         } else {
michael@0:             // No binary value for a binary property.
michael@0:             fprintf(stderr,
michael@0:                     "error in preparsed UCD: enum-property syntax '%s' "
michael@0:                     "for binary property on line %ld\n",
michael@0:                     field, (long)lineNumber);
michael@0:             errorCode=U_PARSE_ERROR;
michael@0:         }
michael@0:     } else if(binaryValue>=0) {
michael@0:         // Binary value for a non-binary property.
michael@0:         fprintf(stderr,
michael@0:                 "error in preparsed UCD: binary-property syntax '%s' "
michael@0:                 "for non-binary property on line %ld\n",
michael@0:                 field, (long)lineNumber);
michael@0:         errorCode=U_PARSE_ERROR;
michael@0:     } else if (prop < UCHAR_INT_START) {
michael@0:         fprintf(stderr,
michael@0:                 "error in preparsed UCD: prop value is invalid: '%d' for line %ld\n",
michael@0:                 prop, (long)lineNumber);
michael@0:         errorCode=U_PARSE_ERROR;
michael@0:     } else if(prop<UCHAR_INT_LIMIT) {
michael@0:         int32_t value=pnames->getPropertyValueEnum(prop, v);
michael@0:         if(value==UCHAR_INVALID_CODE && prop==UCHAR_CANONICAL_COMBINING_CLASS) {
michael@0:             // TODO: Make getPropertyValueEnum(UCHAR_CANONICAL_COMBINING_CLASS, v) work.
michael@0:             char *end;
michael@0:             unsigned long ccc=uprv_strtoul(v, &end, 10);
michael@0:             if(v<end && *end==0 && ccc<=254) {
michael@0:                 value=(int32_t)ccc;
michael@0:             }
michael@0:         }
michael@0:         if(value==UCHAR_INVALID_CODE) {
michael@0:             fprintf(stderr,
michael@0:                     "error in preparsed UCD: '%s' is not a valid value on line %ld\n",
michael@0:                     field, (long)lineNumber);
michael@0:             errorCode=U_PARSE_ERROR;
michael@0:         } else {
michael@0:             props.intProps[prop-UCHAR_INT_START]=value;
michael@0:         }
michael@0:     } else if(*v=='<') {
michael@0:         // Do not parse default values like <code point>, just set null values.
michael@0:         switch(prop) {
michael@0:         case UCHAR_BIDI_MIRRORING_GLYPH:
michael@0:             props.bmg=U_SENTINEL;
michael@0:             break;
michael@0:         case UCHAR_BIDI_PAIRED_BRACKET:
michael@0:             props.bpb=U_SENTINEL;
michael@0:             break;
michael@0:         case UCHAR_SIMPLE_CASE_FOLDING:
michael@0:             props.scf=U_SENTINEL;
michael@0:             break;
michael@0:         case UCHAR_SIMPLE_LOWERCASE_MAPPING:
michael@0:             props.slc=U_SENTINEL;
michael@0:             break;
michael@0:         case UCHAR_SIMPLE_TITLECASE_MAPPING:
michael@0:             props.stc=U_SENTINEL;
michael@0:             break;
michael@0:         case UCHAR_SIMPLE_UPPERCASE_MAPPING:
michael@0:             props.suc=U_SENTINEL;
michael@0:             break;
michael@0:         case UCHAR_CASE_FOLDING:
michael@0:             props.cf.remove();
michael@0:             break;
michael@0:         case UCHAR_LOWERCASE_MAPPING:
michael@0:             props.lc.remove();
michael@0:             break;
michael@0:         case UCHAR_TITLECASE_MAPPING:
michael@0:             props.tc.remove();
michael@0:             break;
michael@0:         case UCHAR_UPPERCASE_MAPPING:
michael@0:             props.uc.remove();
michael@0:             break;
michael@0:         case UCHAR_SCRIPT_EXTENSIONS:
michael@0:             props.scx.clear();
michael@0:             break;
michael@0:         default:
michael@0:             fprintf(stderr,
michael@0:                     "error in preparsed UCD: '%s' is not a valid default value on line %ld\n",
michael@0:                     field, (long)lineNumber);
michael@0:             errorCode=U_PARSE_ERROR;
michael@0:         }
michael@0:     } else {
michael@0:         char c;
michael@0:         switch(prop) {
michael@0:         case UCHAR_NUMERIC_VALUE:
michael@0:             props.numericValue=v;
michael@0:             c=*v;
michael@0:             if('0'<=c && c<='9' && v[1]==0) {
michael@0:                 props.digitValue=c-'0';
michael@0:             } else {
michael@0:                 props.digitValue=-1;
michael@0:             }
michael@0:             break;
michael@0:         case UCHAR_NAME:
michael@0:             props.name=v;
michael@0:             break;
michael@0:         case UCHAR_AGE:
michael@0:             u_versionFromString(props.age, v);  // Writes 0.0.0.0 if v is not numeric.
michael@0:             break;
michael@0:         case UCHAR_BIDI_MIRRORING_GLYPH:
michael@0:             props.bmg=parseCodePoint(v, errorCode);
michael@0:             break;
michael@0:         case UCHAR_BIDI_PAIRED_BRACKET:
michael@0:             props.bpb=parseCodePoint(v, errorCode);
michael@0:             break;
michael@0:         case UCHAR_SIMPLE_CASE_FOLDING:
michael@0:             props.scf=parseCodePoint(v, errorCode);
michael@0:             break;
michael@0:         case UCHAR_SIMPLE_LOWERCASE_MAPPING:
michael@0:             props.slc=parseCodePoint(v, errorCode);
michael@0:             break;
michael@0:         case UCHAR_SIMPLE_TITLECASE_MAPPING:
michael@0:             props.stc=parseCodePoint(v, errorCode);
michael@0:             break;
michael@0:         case UCHAR_SIMPLE_UPPERCASE_MAPPING:
michael@0:             props.suc=parseCodePoint(v, errorCode);
michael@0:             break;
michael@0:         case UCHAR_CASE_FOLDING:
michael@0:             parseString(v, props.cf, errorCode);
michael@0:             break;
michael@0:         case UCHAR_LOWERCASE_MAPPING:
michael@0:             parseString(v, props.lc, errorCode);
michael@0:             break;
michael@0:         case UCHAR_TITLECASE_MAPPING:
michael@0:             parseString(v, props.tc, errorCode);
michael@0:             break;
michael@0:         case UCHAR_UPPERCASE_MAPPING:
michael@0:             parseString(v, props.uc, errorCode);
michael@0:             break;
michael@0:         case PPUCD_NAME_ALIAS:
michael@0:             props.nameAlias=v;
michael@0:             break;
michael@0:         case PPUCD_CONDITIONAL_CASE_MAPPINGS:
michael@0:         case PPUCD_TURKIC_CASE_FOLDING:
michael@0:             // No need to parse their values: They are hardcoded in the runtime library.
michael@0:             break;
michael@0:         case UCHAR_SCRIPT_EXTENSIONS:
michael@0:             parseScriptExtensions(v, props.scx, errorCode);
michael@0:             break;
michael@0:         default:
michael@0:             // Ignore unhandled properties.
michael@0:             return TRUE;
michael@0:         }
michael@0:     }
michael@0:     if(U_SUCCESS(errorCode)) {
michael@0:         newValues.add((UChar32)prop);
michael@0:         return TRUE;
michael@0:     } else {
michael@0:         return FALSE;
michael@0:     }
michael@0: }
michael@0: 
michael@0: UBool
michael@0: PreparsedUCD::getRangeForAlgNames(UChar32 &start, UChar32 &end, UErrorCode &errorCode) {
michael@0:     if(U_FAILURE(errorCode)) { return FALSE; }
michael@0:     if(lineType!=ALG_NAMES_RANGE_LINE) {
michael@0:         errorCode=U_ILLEGAL_ARGUMENT_ERROR;
michael@0:         return FALSE;
michael@0:     }
michael@0:     firstField();
michael@0:     const char *field=nextField();
michael@0:     if(field==NULL) {
michael@0:         // No range field after the type.
michael@0:         fprintf(stderr,
michael@0:                 "error in preparsed UCD: missing algnamesrange range field "
michael@0:                 "(no second field) on line %ld\n",
michael@0:                 (long)lineNumber);
michael@0:         errorCode=U_PARSE_ERROR;
michael@0:         return FALSE;
michael@0:     }
michael@0:     return parseCodePointRange(field, start, end, errorCode);
michael@0: }
michael@0: 
michael@0: UChar32
michael@0: PreparsedUCD::parseCodePoint(const char *s, UErrorCode &errorCode) {
michael@0:     char *end;
michael@0:     uint32_t value=(uint32_t)uprv_strtoul(s, &end, 16);
michael@0:     if(end<=s || *end!=0 || value>=0x110000) {
michael@0:         fprintf(stderr,
michael@0:                 "error in preparsed UCD: '%s' is not a valid code point on line %ld\n",
michael@0:                 s, (long)lineNumber);
michael@0:         errorCode=U_PARSE_ERROR;
michael@0:         return U_SENTINEL;
michael@0:     }
michael@0:     return (UChar32)value;
michael@0: }
michael@0: 
michael@0: UBool
michael@0: PreparsedUCD::parseCodePointRange(const char *s, UChar32 &start, UChar32 &end, UErrorCode &errorCode) {
michael@0:     uint32_t st, e;
michael@0:     u_parseCodePointRange(s, &st, &e, &errorCode);
michael@0:     if(U_FAILURE(errorCode)) {
michael@0:         fprintf(stderr,
michael@0:                 "error in preparsed UCD: '%s' is not a valid code point range on line %ld\n",
michael@0:                 s, (long)lineNumber);
michael@0:         return FALSE;
michael@0:     }
michael@0:     start=(UChar32)st;
michael@0:     end=(UChar32)e;
michael@0:     return TRUE;
michael@0: }
michael@0: 
michael@0: void
michael@0: PreparsedUCD::parseString(const char *s, UnicodeString &uni, UErrorCode &errorCode) {
michael@0:     UChar *buffer=uni.getBuffer(-1);
michael@0:     int32_t length=u_parseString(s, buffer, uni.getCapacity(), NULL, &errorCode);
michael@0:     if(errorCode==U_BUFFER_OVERFLOW_ERROR) {
michael@0:         errorCode=U_ZERO_ERROR;
michael@0:         uni.releaseBuffer(0);
michael@0:         buffer=uni.getBuffer(length);
michael@0:         length=u_parseString(s, buffer, uni.getCapacity(), NULL, &errorCode);
michael@0:     }
michael@0:     uni.releaseBuffer(length);
michael@0:     if(U_FAILURE(errorCode)) {
michael@0:         fprintf(stderr,
michael@0:                 "error in preparsed UCD: '%s' is not a valid Unicode string on line %ld\n",
michael@0:                 s, (long)lineNumber);
michael@0:     }
michael@0: }
michael@0: 
michael@0: void
michael@0: PreparsedUCD::parseScriptExtensions(const char *s, UnicodeSet &scx, UErrorCode &errorCode) {
michael@0:     if(U_FAILURE(errorCode)) { return; }
michael@0:     scx.clear();
michael@0:     CharString scString;
michael@0:     for(;;) {
michael@0:         const char *scs;
michael@0:         const char *scLimit=strchr(s, ' ');
michael@0:         if(scLimit!=NULL) {
michael@0:             scs=scString.clear().append(s, (int32_t)(scLimit-s), errorCode).data();
michael@0:             if(U_FAILURE(errorCode)) { return; }
michael@0:         } else {
michael@0:             scs=s;
michael@0:         }
michael@0:         int32_t script=pnames->getPropertyValueEnum(UCHAR_SCRIPT, scs);
michael@0:         if(script==UCHAR_INVALID_CODE) {
michael@0:             fprintf(stderr,
michael@0:                     "error in preparsed UCD: '%s' is not a valid script code on line %ld\n",
michael@0:                     scs, (long)lineNumber);
michael@0:             errorCode=U_PARSE_ERROR;
michael@0:             return;
michael@0:         } else if(scx.contains(script)) {
michael@0:             fprintf(stderr,
michael@0:                     "error in preparsed UCD: scx has duplicate '%s' codes on line %ld\n",
michael@0:                     scs, (long)lineNumber);
michael@0:             errorCode=U_PARSE_ERROR;
michael@0:             return;
michael@0:         } else {
michael@0:             scx.add(script);
michael@0:         }
michael@0:         if(scLimit!=NULL) {
michael@0:             s=scLimit+1;
michael@0:         } else {
michael@0:             break;
michael@0:         }
michael@0:     }
michael@0:     if(scx.isEmpty()) {
michael@0:         fprintf(stderr, "error in preparsed UCD: empty scx= on line %ld\n", (long)lineNumber);
michael@0:         errorCode=U_PARSE_ERROR;
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_NAMESPACE_END