/*
*******************************************************************************
*
*   Copyright (C) 1998-2000, International Business Machines
*   Corporation and others.  All Rights Reserved.
*
*******************************************************************************
*
* File parse.c
*
* Modification History:
*
*   Date        Name        Description
*   05/26/99    stephen     Creation.
*   02/25/00    weiv        Overhaul to write udata
*******************************************************************************
*/

#include "ucolimp.h"
#include "parse.h"
#include "error.h"
#include "uhash.h"
#include "cmemory.h"
#include "read.h"
#include "unicode/ustdio.h"
#include "ustr.h"
#include "reslist.h"
#include "unicode/ustring.h"
#include "unicode/putil.h"

U_CAPI const UChar * U_EXPORT2 ucol_getDefaultRulesArray(uint32_t *size);

U_STRING_DECL(k_start_string, "string", 6);
U_STRING_DECL(k_start_binary, "binary", 6);
U_STRING_DECL(k_start_table, "table", 5);
U_STRING_DECL(k_start_int, "int", 3);
U_STRING_DECL(k_start_array, "array", 5);
U_STRING_DECL(k_start_intvector, "intvector", 9);
U_STRING_DECL(k_start_reserved, "reserved", 8);
U_STRING_DECL(rootName, "root", 4);

static UBool didInit=FALSE;
static UBool didInitRoot=FALSE;

/* Node IDs for the state transition table. */
enum ENode {
    eError,
    eInitial,   /* Next: Locale name */
    eGotLoc,    /* Next: { */
    eIdle,      /* Next: Tag name | } */
    eGotTag,    /* Next: { | : */
    eNode5,     /* Next: Data | Subtag */
    eNode6,     /* Next: } | { | , */
    eList,      /* Next: List data */
    eNode8,     /* Next: , */
    eTagList,   /* Next: Subtag data */
    eNode10,    /* Next: } */
    eNode11,    /* Next: Subtag */
    eNode12,    /* Next: { */
    e2dArray,   /* Next: Data | } */
    eNode14,    /* Next: , | } */
    eNode15,    /* Next: , | } */
    eNode16,     /* Next: { | } */
    eTypeStart, /* Next: Type name */
    eGotType    /* Next: { */
};

/* Action codes for the state transtiion table. */
enum EAction {
    /* Generic actions */
    eNOP       = 0x0100, /* Do nothing */
    eOpen      = 0x0200, /* Open a new locale data block with the data
                          string as the locale name */
    eClose     = 0x0300, /* Close a locale data block */
    eSetTag    = 0x0400, /* Record the last string as the tag name */

    /* Comma-delimited lists */
    eBegList   = 0x1100, /* Start a new string list with the last string
                          as the first element */
    eEndList   = 0x1200, /* Close a string list being built */
    eListStr   = 0x1300, /* Record the last string as a data string and
                          increment the index */
    eStr       = 0x1400, /* Record the last string as a singleton string */

    /* 2-d lists */
    eBeg2dList = 0x2100, /* Start a new 2d string list with no elements as yet */
    eEnd2dList = 0x2200, /* Close a 2d string list being built */
    e2dStr     = 0x2300, /* Record the last string as a 2d string */
    eNewRow    = 0x2400, /* Start a new row */

    /* Tagged lists */
    eBegTagged = 0x3100, /* Start a new tagged list with the last
                          string as the first subtag */
    eEndTagged = 0x3200, /* Close a tagged list being build */
    eSubtag    = 0x3300, /* Record the last string as the subtag */
    eTaggedStr = 0x3400,  /* Record the last string as a tagged string */

    /* Type support */
    eBegType = 0x4100, /* Start getting a type */
    eSetType = 0x4200 /* Record and init type */
};

/* A struct which encapsulates a node ID and an action. */
struct STransition {
    enum ENode fNext;
    enum EAction fAction;
};

/* This table describes an ATM (state machine) which parses resource
   bundle text files rather strictly. Each row represents a node. The
   columns of that row represent transitions into other nodes. Most
   transitions are "eError" because most transitions are
   disallowed. For example, if the parser has just seen a tag name, it
   enters node 4 ("eGotTag"). The state table then marks only one
   valid transition, which is into node 5, upon seeing an eOpenBrace
   token. We allow an extra comma after the last element in a
   comma-delimited list (transition from eList to eIdle on
   kCloseBrace). */
static struct STransition gTransitionTable [] = {
  /*                kString           kOpenBrace            kCloseBrace         kComma       */
  /*eError*/    {eError,eNOP},       {eError,eNOP},        {eError,eNOP},      {eError,eNOP},

  /*eInitial*/  {eGotLoc,eOpen},     {eError,eNOP},        {eError,eNOP},      {eError,eNOP},
  /*eGotLoc*/   {eError,eNOP},       {eIdle,eNOP},         {eError,eNOP},      {eError,eNOP},

  /*eIdle*/     {eGotTag,eSetTag},   {eError,eNOP},        {eInitial,eClose},  {eError,eNOP},
  /*eGotTag*/   {eError,eNOP},       {eNode5,eNOP},        {eError,eNOP},      {eError,eNOP},
  /*eNode5*/    {eNode6,eNOP},       {e2dArray,eBeg2dList},{eError,eNOP},      {eError,eNOP},
  /*eNode6*/    {eError,eNOP},       {eTagList,eBegTagged},{eIdle,eStr},       {eList,eBegList},

  /*eList*/     {eNode8,eListStr},   {eError,eNOP},         {eIdle,eEndList},  {eError,eNOP},
  /*eNode8*/    {eError,eNOP},       {eError,eNOP},         {eIdle,eEndList},  {eList,eNOP},

  /*eTagList*/  {eNode10,eTaggedStr},{eError,eNOP},         {eError,eNOP},     {eError,eNOP},
  /*eNode10*/   {eError,eNOP},       {eError,eNOP},         {eNode11,eNOP},    {eError,eNOP},
  /*eNode11*/   {eNode12,eNOP},      {eError,eNOP},         {eIdle,eEndTagged},{eError,eNOP},
  /*eNode12*/   {eError,eNOP},       {eTagList,eSubtag},    {eError,eNOP},     {eError,eNOP},

  /*e2dArray*/  {eNode14,e2dStr},    {eError,eNOP},         {eNode15,eNOP},    {eError,eNOP},
  /*eNode14*/   {eError,eNOP},       {eError,eNOP},         {eNode15,eNOP},    {e2dArray,eNOP},
  /*eNode15*/   {eError,eNOP},       {e2dArray,eNewRow},    {eIdle,eEnd2dList},{eNode16,eNOP},
  /*eNode16*/   {eError,eNOP},       {e2dArray,eNewRow},    {eIdle,eEnd2dList},{eError,eNOP},
  /*eTypeStart*/{eGotType,eSetType}, {eError,eNOP},         {eError,eNOP},     {eError,eNOP},
  /*eGotType*/  {eError,eNOP},       {eError,eNOP},         {eError,eNOP},     {eError,eNOP}
};

/* Row length is 4 */
#define GETTRANSITION(row,col) (gTransitionTable[col + (row<<2)])
/* Not anymore, it is 5 now */
/*#define GETTRANSITION(row,col) (gTransitionTable[col + (row*5)])*/

/*********************************************************************
 * Hashtable glue
 ********************************************************************/

static UBool get(UHashtable *hash, const struct UString* tag) {
    return (UBool)(uhash_get(hash, tag) != NULL);
}

static void put(UHashtable *hash, const struct UString *tag,
                UErrorCode* status) {
    struct UString* key = (struct UString*)uprv_malloc(sizeof(struct UString));
    ustr_init(key);
    ustr_cpy(key, tag, status);
    uhash_put(hash, key, (void*)1, status);
}

static void freeUString(void* ustr) {
    ustr_deinit((struct UString*)ustr);
    uprv_free(ustr);
}

static int32_t hashUString(const void* ustr) {
    return uhash_hashUChars(((struct UString*)ustr)->fChars);
}

static UBool compareUString(const void* ustr1, const void* ustr2) {
    return uhash_compareUChars(((struct UString*)ustr1)->fChars,
                               ((struct UString*)ustr2)->fChars);
}

char *getModificationData(struct UFILE *file, UErrorCode *status) {
    enum ETokenType modType;
    struct UString modToken;
    char *retValue = NULL;

    ustr_init(&modToken);
    modType = getNextToken(file, &modToken, status);
    if(U_SUCCESS(*status) && modType == tok_open_brace) {
        modType = getNextToken(file, &modToken, status);
        if(U_SUCCESS(*status) && modType == tok_string) {
            retValue = uprv_malloc(u_strlen(modToken.fChars)+1);
            u_UCharsToChars(modToken.fChars, retValue, u_strlen(modToken.fChars)+1);
            modType = getNextToken(file, &modToken, status);
            if(U_SUCCESS(*status) && modType == tok_close_brace) {
                return retValue;
            } else {
                uprv_free(retValue);
            }
        }
    } 
    setErrorText("Invalid modificator directive");
    *status = U_INVALID_FORMAT_ERROR;

    return NULL;
}

/*********************************************************************
 * parse
 ********************************************************************/
int32_t lineCount = 0;
char lastTag[200] = "";

struct SRBRoot*
parse(FileStream *f, const char *cp, const char *inputDir,
      UErrorCode *status)
{
    struct UFILE *file;
    enum ETokenType type;
    enum ENode node;
    struct STransition t;

    struct UString token;
    struct UString tag;

    char cTag[1024];
    char cSubTag[1024];
    struct SRBRoot *bundle = NULL;
    struct SResource *rootTable = NULL;
    struct SResource *temp = NULL;
    struct SResource *temp1 = NULL;
    struct SResource *temp2 = NULL;
    UBool colEl = FALSE, colOverride = FALSE;
    UChar trueValue[] = {0x0054, 0x0052, 0x0055, 0x0045, 0x0000}; /* Just to store "TRUE" and "FALSE" */
    UChar falseValue[] = {0x0046, 0x0041, 0x004C, 0x0053, 0x0045, 0x0000}; 

    /* Hashtable for keeping track of seen tag names */
    struct UHashtable *data;

    strcpy(lastTag, "<none>");

    if(U_FAILURE(*status)) return NULL;

    /* setup */

    ustr_init(&token);
    ustr_init(&tag);
/*
    cTag = uprv_malloc(1024);
    if(cTag == NULL) {
        *status = U_MEMORY_ALLOCATION_ERROR;
        return NULL;
    }
    cSubTag = uprv_malloc(1024);
    if(cSubTag == NULL) {
        *status = U_MEMORY_ALLOCATION_ERROR;
        return NULL;
    }
*/

    node = eInitial;
    data = 0;

    file = u_finit((FILE *)f, 0, cp);
    lineCount = 1;
/*  file = u_finit(f, cp, status); */
    if(file == NULL) {
        setErrorText("Could not initialize input file - most probably because of wrong converter\n");
        *status = U_INVALID_FORMAT_ERROR;
        goto finish;
    }

    bundle = bundle_open(status);
    rootTable = bundle -> fRoot;

    if(U_FAILURE(*status) || file == NULL) {
        goto finish;
    }

    /* iterate through the stream */
    for(;;) {

        /* get next token from stream */
        type = getNextToken(file, &token, status);
        if(U_FAILURE(*status)) {
            goto finish;
        }

        switch(type) {
        case tok_EOF:
            *status = (node == eInitial) ? U_ZERO_ERROR : U_INVALID_FORMAT_ERROR;
            if(U_FAILURE(*status)) {
                setErrorText("Unexpected EOF encountered");
            }
            goto finish;
            /*break;*/

        case tok_error:
            *status = U_INVALID_FORMAT_ERROR;
            goto finish;
            /*break;*/

        default:
            break;
        }

        t = GETTRANSITION(node, type);
        node = t.fNext;

        if(node == eError) {
            *status = U_INVALID_FORMAT_ERROR;
            goto finish;
        }

        switch(t.fAction) {
        case eNOP:
            break;

          /* Record the last string as the tag name */
        case eSetTag:
            ustr_cpy(&tag, &token, status);
            u_UCharsToChars(tag.fChars, cTag, u_strlen(tag.fChars)+1);
            if(U_FAILURE(*status)) {
                goto finish;
            }
             strcpy(lastTag, cTag);
           /*  fprintf(stderr, "%d: %s\n", lineCount,  lastTag); //[prints all tags]
            */

            if(get(data, &tag)) {
                char *s;
                *status = U_INVALID_FORMAT_ERROR;
                s = uprv_malloc(1024);
                strcpy(s, "Duplicate tag name detected: ");
                u_austrcpy(s+strlen(s), tag.fChars);
                setErrorText(s);
                goto finish;
            }
            {
                char *modificator = uprv_strchr(cTag, ':');
                if(modificator != NULL) {
                    /* type modificator - do the type modification*/
                    *modificator = '\0';
                    ustr_deinit(&tag);
                    ustr_setlen(&tag, uprv_strlen(cTag), status);
                    u_charsToUChars(cTag, tag.fChars, uprv_strlen(cTag));
                    /* we need to test whether we have the same name, different type here */
                    if(get(data, &tag)) {
                      char *s;
                      *status = U_INVALID_FORMAT_ERROR;
                      s = uprv_malloc(1024);
                      strcpy(s, "Duplicate tag name detected: ");
                      u_austrcpy(s+strlen(s), tag.fChars);
                      setErrorText(s);
                      goto finish;
                    }
                    modificator++;
                    /* including streams of binary data */
                    if(uprv_strcmp(modificator, "bin") == 0) {
                      char *binaryValue;
                      char toConv[3];
                      uint32_t i = 0, bytesConverted = 0;
                      uint8_t val = 0;
                      uint8_t *newValue;
                      fprintf(stderr, "bin\n");
                      binaryValue = getModificationData(file, status);
                      if(U_SUCCESS(*status) && binaryValue != NULL) {
                        /* do the parsing & outputing of the data */
                        fprintf(stderr, "Will parse binary value  %s and store it in tag: %s\n", binaryValue, cTag);
                        newValue = uprv_malloc(sizeof(uint8_t)*uprv_strlen(binaryValue));
                        for(i = 0; i<uprv_strlen(binaryValue); i+=2) {
                          toConv[0] = *(binaryValue+i);
                          toConv[1] = *(binaryValue+i+1);
                          toConv[2] = '\0';
                          val = (uint8_t)uprv_strtoul(toConv, NULL, 16);
                          newValue[bytesConverted] = val;
                          bytesConverted++;
                        }
                        temp1 = bin_open(bundle, cTag, bytesConverted, newValue, status);
                        table_add(rootTable, temp1, status);

                        uprv_free(newValue);
                        uprv_free(binaryValue);
                        node = eIdle;
                      } else {
                        if(binaryValue != NULL) {
                          uprv_free(binaryValue);
                        }
                        node = eError;
                      }
                    }
                    /* including integers */
                    else if(uprv_strcmp(modificator, "int") == 0) {
                      char *intValue;
                      int32_t val;
                      fprintf(stderr, "int\n");
                      intValue = getModificationData(file, status);
                      if(U_SUCCESS(*status) && intValue != NULL) {
                        /* do the parsing & outputing of the data */
                        fprintf(stderr, "Will parse integer value  %s and store it in tag: %s\n", intValue, cTag);
                        val = uprv_strtol(intValue, NULL, 10);
                        uprv_free(intValue);
                        temp1 = int_open(bundle, cTag, val, status);
                        fprintf(stderr, "Added integer %s, value %d -> %s\n", cTag, val,
                          u_errorName(*status) );
                        table_add(rootTable, temp1, status);

                        put(data, &tag, status);
                        node = eIdle;
                      } else {
                        if(intValue != NULL) {
                          uprv_free(intValue);
                        }
                        node = eError;
                      }
                    }
                    /* importing a file and storing it in a binary object */
                    else if(uprv_strcmp(modificator, "import") == 0) {
                      FileStream *importFile;
                      int32_t len; 
                      uint8_t *binData;
                      char *fileName;
                      fprintf(stderr, "import\n");
                      fileName = getModificationData(file, status);
                      if(U_SUCCESS(*status) && fileName != NULL) {
                        /* do the reading & outputing of the file */
                        fprintf(stderr, "Will read %s and store it in tag:  %s\n", fileName, cTag);
                        /* Open the input file for reading */
                        if(inputDir == NULL) {
                          importFile = T_FileStream_open(fileName, "rb");
                        } else {
                          char *openFileName = NULL;
                          int32_t dirlen = uprv_strlen(inputDir);
                          int32_t filelen = uprv_strlen(fileName);
                          if(inputDir[dirlen-1] != U_FILE_SEP_CHAR) {
                            openFileName = (char *) uprv_malloc(dirlen+filelen+2);
                            uprv_strcpy(openFileName, inputDir);
                            openFileName[dirlen] = U_FILE_SEP_CHAR;
                            openFileName[dirlen+1] = '\0';
                            uprv_strcat(openFileName, fileName);
                          } else {
                            openFileName = (char *) uprv_malloc(dirlen+filelen+1);
                            uprv_strcpy(openFileName, inputDir);
                            uprv_strcat(openFileName, fileName);
                          }
                          importFile = T_FileStream_open(openFileName, "rb");
                          uprv_free(openFileName);
                        }
                        if(importFile == NULL) {
                          fprintf(stderr, "Error! Couldn't open input file %s for tag %s\n", fileName, cTag);

                          node = eError;
                          continue;
                        }

                        len = T_FileStream_size(importFile);
                        binData = uprv_malloc(len);
                        T_FileStream_read(importFile,binData,len);
                        T_FileStream_close(importFile);

                        temp1 = bin_open(bundle, cTag, len, binData, status);
                        fprintf(stderr, "Added %s, len %d -> %s\n", cTag, len,
                                  u_errorName(*status) );
                        table_add(rootTable, temp1, status);
                        uprv_free(binData);
                        uprv_free(fileName);
                        put(data, &tag, status);
                        node = eIdle;
                      } else {
                        if(fileName != NULL) {
                          uprv_free(fileName);
                        }
                        node = eError;
                      }
                    }
                    /* array of integers, still unimplemented */
                    else if(uprv_strcmp(modificator, "intarray") == 0) {
                      fprintf(stderr, "intarray\n");
                    }
                    /* unknown tupe - an error */
                    else {
                      fprintf(stderr, "Unknown\n");
                    }

                } else if(uprv_strcmp(cTag, "CollationElements") == 0) {
                    colEl = TRUE;
                }
            }
            break;

          /* Record a singleton string */
        case eStr:
            if(temp != NULL) {
                *status = U_INTERNAL_PROGRAM_ERROR;
                goto finish;
            }
            temp = string_open(bundle, cTag, token.fChars, token.fLength, status);
            table_add(rootTable, temp, status);
#if 0
            if(colEl == TRUE) {
                const UChar * defaultRulesArray;
                UErrorCode intStatus = U_ZERO_ERROR;
                uint32_t defaultRulesArrayLength = 0;
                /* do the collation elements */
                int32_t len = 0;
                uint8_t *binColData = NULL;
                UCollator *coll = NULL; 
                UChar *rules = NULL;
                defaultRulesArray = ucol_getDefaultRulesArray(&defaultRulesArrayLength);
                rules = uprv_malloc(sizeof(defaultRulesArray[0])*(defaultRulesArrayLength + token.fLength));
                uprv_memcpy(rules, defaultRulesArray, defaultRulesArrayLength*sizeof(defaultRulesArray[0]));
                uprv_memcpy(rules + defaultRulesArrayLength, token.fChars, token.fLength*sizeof(token.fChars[0]));

                coll = ucol_openRules(rules, defaultRulesArrayLength + token.fLength, UCOL_DECOMP_CAN, 0, &intStatus);

                if(U_SUCCESS(intStatus) && coll !=NULL) {
                    ucol_setNormalization(coll, UCOL_NO_NORMALIZATION);
                    binColData = ucol_cloneRuleData(coll, &len, &intStatus);
                    if(U_SUCCESS(*status) && data != NULL) {
                        temp1 = bin_open(bundle, "%%Collation", len, binColData, status);
                        table_add(rootTable, temp1, status);
                        uprv_free(binColData);
                    }
                    ucol_close(coll);
                } else {
                    setErrorText("Warning: %%Collation could not be constructed from CollationElements - check context!");
                }
                uprv_free(rules);
                colEl = FALSE;
            }
#endif

            /*uhash_put(data, tag.fChars, status);*/
            put(data, &tag, status);
            if(U_FAILURE(*status)) {
                goto finish;
            }
            temp = NULL;
            break;
          /* Begin a string list */
        case eBegList:
            if(temp != NULL) {
                *status = U_INTERNAL_PROGRAM_ERROR;
                goto finish;
            }
            temp = array_open(bundle, cTag, status);
            temp1 = string_open(bundle, NULL, token.fChars, token.fLength, status);
            array_add(temp, temp1, status);
            temp1 = NULL;
            if(U_FAILURE(*status)) {
                goto finish;
            }
            break;
      
          /* Record a comma-delimited list string */      
        case eListStr:
            temp1 = string_open(bundle, NULL, token.fChars, token.fLength, status);
            array_add(temp, temp1, status);
            temp1 = NULL;
            if(U_FAILURE(*status)) {
                goto finish;
            }
            break;
      
          /* End a string list */
        case eEndList:
            /*uhash_put(data, tag.fChars, status);*/
            put(data, &tag, status);
            table_add(rootTable, temp, status);
            temp = NULL;
            if(U_FAILURE(*status)) {
                goto finish;
            }
            break;
      
        case eBeg2dList:
            if(temp != NULL) {
                *status = U_INTERNAL_PROGRAM_ERROR;
                goto finish;
            }
            temp = array_open(bundle, cTag, status);
            temp1 = array_open(bundle, NULL, status);
            if(U_FAILURE(*status)) {
                goto finish;
            }
            break;
      
        case eEnd2dList:
            /*uhash_put(data, tag.fChars, status);*/
            put(data, &tag, status);
            array_add(temp, temp1, status);
            table_add(rootTable, temp, status);
            temp1 = NULL;
            temp = NULL;
            if(U_FAILURE(*status)) {
                goto finish;
            }
            break;
      
        case e2dStr:
            temp2 = string_open(bundle, NULL, token.fChars, token.fLength, status);
            array_add(temp1, temp2, status);
            temp2 = NULL;
            if(U_FAILURE(*status)) {
                goto finish;
            }
            break;
      
        case eNewRow:
            array_add(temp, temp1, status);
            temp1 = array_open(bundle, NULL, status);
            if(U_FAILURE(*status)) {
                goto finish;
            }
            break;
      
        case eBegTagged:
            if(temp != NULL) {
                *status = U_INTERNAL_PROGRAM_ERROR;
                goto finish;
            }
            temp = table_open(bundle, cTag, status);
            u_UCharsToChars(token.fChars, cSubTag, u_strlen(token.fChars)+1);
            if(U_FAILURE(*status)) {
                goto finish;
            }
            break;
      
        case eEndTagged:
            /*uhash_put(data, tag.fChars, status);*/
            put(data, &tag, status);
            table_add(rootTable, temp, status);
            temp = NULL;
            if(U_FAILURE(*status)) {
                goto finish;
            }
            break;
      
        case eTaggedStr:
            temp1 = string_open(bundle, cSubTag, token.fChars, token.fLength, status);
            table_add(temp, temp1, status);
            temp1 = NULL;
            if(U_FAILURE(*status)) {
                goto finish;
            }
            /* We have seen the Override tag aleady, now checks if the value is "TRUE" or "FALSE". */
            if (uprv_strcmp(cSubTag, "Override") == 0) 
            {
                if (u_strncmp(token.fChars, trueValue, u_strlen(trueValue)) == 0)
                {
                    colOverride = TRUE;
                } else {
                    colOverride = FALSE;
                }
            }
            if (colEl && (uprv_strcmp(cSubTag, "Sequence") == 0))
            {
                const UChar * defaultRulesArray;
                UErrorCode intStatus = U_ZERO_ERROR;
                uint32_t defaultRulesArrayLength = 0;
                /* do the collation elements */
                int32_t len = 0;
                uint8_t *binColData = NULL;
                UCollator *coll = NULL; 
                UChar *rules = NULL;
                struct UString newTag;

                if (colOverride == FALSE)
                {
                    defaultRulesArray = ucol_getDefaultRulesArray(&defaultRulesArrayLength);
                    rules = uprv_malloc(sizeof(defaultRulesArray[0])*(defaultRulesArrayLength + token.fLength));
                    uprv_memcpy(rules, defaultRulesArray, defaultRulesArrayLength*sizeof(defaultRulesArray[0]));
                    uprv_memcpy(rules + defaultRulesArrayLength, token.fChars, token.fLength*sizeof(token.fChars[0]));

                    coll = ucol_openRules(rules, defaultRulesArrayLength + token.fLength, UCOL_DECOMP_CAN, 0, &intStatus);
                } else {
                    coll = ucol_openRules(token.fChars, token.fLength, UCOL_DECOMP_CAN, 0, &intStatus);
                }


                if(U_SUCCESS(intStatus) && coll !=NULL) {
                    ucol_setNormalization(coll, UCOL_NO_NORMALIZATION);
                    binColData = ucol_cloneRuleData(coll, &len, &intStatus);
                    if(U_SUCCESS(*status) && data != NULL) {
                        temp1 = bin_open(bundle, "%%Collation", len, binColData, status);
                        table_add(rootTable, temp1, status);
                        uprv_free(binColData);
                    }
                    ucol_close(coll);
                } else {
                    setErrorText("Warning: %%Collation could not be constructed from CollationElements - check context!");
                }
                uprv_free(rules);
                colEl = FALSE;
                colOverride = FALSE;
                intStatus = U_ZERO_ERROR;
                ustr_initChars(&newTag, "CollationElements", -1, &intStatus);
                if(U_FAILURE(intStatus)) {
                    goto finish;
                } 
                put(data, &newTag, &intStatus);
                ustr_deinit(&newTag);
                if(U_FAILURE(intStatus)) {
                    goto finish;
                }
            }
            break;
      
          /* Record the last string as the subtag */
        case eSubtag:
            u_UCharsToChars(token.fChars, cSubTag, u_strlen(token.fChars)+1);
            if(U_FAILURE(*status)) {
                goto finish;
            }
            if(table_get(temp, cSubTag, status) != 0) {
                *status = U_INVALID_FORMAT_ERROR;
                setErrorText("Duplicate subtag found in tagged list");
                goto finish;
            }
            break;
      
        case eOpen:
            if(data != 0) {
                *status = U_INTERNAL_PROGRAM_ERROR;
                goto finish;
            }
            bundle_setlocale(bundle, token.fChars, status);

/* trying to get rid of collation elements in other 'root' bundles */
#if 0
            if(didInitRoot == FALSE) {
                U_STRING_INIT(rootName, "root", 4);
                didInitRoot = TRUE;
            }

            if(u_strcmp(token.fChars, rootName) == 0) {
                const UChar * defaultRulesArray;
                uint32_t defaultRulesArrayLength = 0;
                /* do the collation elements */
                int32_t len = 0;
/*                uint8_t *data = NULL;*/
                uint8_t *data2 = NULL;
                UCollator *coll = NULL; 

                UChar *rules = NULL;
                defaultRulesArray = ucol_getDefaultRulesArray(&defaultRulesArrayLength);
                rules = uprv_malloc(sizeof(defaultRulesArray[0])*(defaultRulesArrayLength));
                uprv_memcpy(rules, defaultRulesArray, defaultRulesArrayLength*sizeof(defaultRulesArray[0]));

                coll = ucol_openRules(rules, defaultRulesArrayLength, UCOL_DECOMP_CAN, 0, status);
                ucol_setNormalization(coll, UCOL_DEFAULT_NORMALIZATION);

                if(U_SUCCESS(*status) && coll !=NULL) {
                    data2 = ucol_cloneRuleData(coll, &len, status);
                    if(U_SUCCESS(*status) && data2 != NULL) {
                        temp1 = bin_open(bundle, "%%Collation", len, data2, status);
                        table_add(rootTable, temp1, status);
                        uprv_free(data2);
                    }
                    ucol_close(coll);
                }
                uprv_free(rules);
            }
#endif
            if(U_FAILURE(*status)) {
                goto finish;
            }
            data = uhash_open(hashUString, compareUString, status);
            uhash_setKeyDeleter(data, freeUString);
            break;
      
        case eClose:
            if(data == 0) {
                *status = U_INTERNAL_PROGRAM_ERROR;
                goto finish;
            }
            break;
        case eSetType:
            /* type recognition */
            if(!didInit) {
                U_STRING_INIT(k_start_string, "string", 6);
                U_STRING_INIT(k_start_binary, "binary", 6);
                U_STRING_INIT(k_start_table, "table", 5);
                U_STRING_INIT(k_start_int, "int", 3);
                U_STRING_INIT(k_start_array, "array", 5);
                U_STRING_INIT(k_start_intvector, "intvector", 9);
                U_STRING_INIT(k_start_reserved, "reserved", 8);
                didInit=TRUE;
            }
            if(u_strcmp(token.fChars, k_start_string) == 0) {
                node = eGotTag;
            } else if(u_strcmp(token.fChars, k_start_array) == 0) {
                node = eGotTag;
            } else if(u_strcmp(token.fChars, k_start_table) == 0) {
                node = eGotTag;
            } else if(u_strcmp(token.fChars, k_start_binary) == 0) {
                /* start of binary */
            } else if(u_strcmp(token.fChars, k_start_int) == 0) {
                /* start of integer */
            } else if(u_strcmp(token.fChars, k_start_intvector) == 0) {
                /* start of intvector */
            } else if(u_strcmp(token.fChars, k_start_reserved) == 0) {
                /* start of reserved */
            } else {
                *status = U_INTERNAL_PROGRAM_ERROR;
                goto finish;
            }
            break;
        }
    }

finish:

    /* clean  up */

    if(data != 0)
        uhash_close(data);

    ustr_deinit(&token);
    ustr_deinit(&tag);

    /*uprv_free(cTag);*/
    /*uprv_free(cSubTag);*/

    if(file != 0)
        u_fclose(file);

    return bundle;
}