imp2gbs.cpp File Reference

#include <ctype.h>
#include <stdio.h>
#include <errno.h>
#include <stdlib.h>
#include <entriesblk.h>
#include <iostream>
#include <treekeyidx.h>
#include <rawgenbook.h>
#include <utilstr.h>
#include <filemgr.h>
#include <utf8greekaccents.h>
#include <stringmgr.h>

Include dependency graph for imp2gbs.cpp:

Go to the source code of this file.

Functions
int	main (int argc, char **argv)
void	parseParams (int argc, char **argv)
void	usage (const char *app)
void	writeEntry (SWModule *book, SWBuf keyBuffer, SWBuf entBuffer)

Variables
bool	augEnt = true
bool	augMod = false
UTF8GreekAccents	greekAccentsFilter
bool	greekFilter = false
SWBuf	inFile
int	lexLevels = 0
SWBuf	outPath
bool	toUpper = false

Function Documentation

int main	(	int	argc,
		char **	argv
	)

Definition at line 222 of file imp2gbs.cpp.

                                {
    greekAccentsFilter.setOptionValue("Off");       // off = accents off
    parseParams(argc, argv);

    // Let's see if we can open our input file
    FileDesc *fd = FileMgr::getSystemFileMgr()->open(inFile, FileMgr::RDONLY);
    if (fd->getFd() < 0) {
        fprintf(stderr, "error: %s: couldn't open input file: %s \n", argv[0], inFile.c_str());
        exit(-2);
    }

    RawGenBook *book;

    // Do some initialization stuff
    if (!augMod) {
        RawGenBook::createModule(outPath);
    }
    book = new RawGenBook(outPath);

    SWBuf lineBuffer;
    SWBuf keyBuffer;
    SWBuf entBuffer;

    bool more = true;
    do {
        more = FileMgr::getLine(fd, lineBuffer)!=0;
        if (lineBuffer.startsWith("$$$")) {
            if ((keyBuffer.size()) && (entBuffer.size())) {
                writeEntry(book, keyBuffer, entBuffer);
            }
            keyBuffer = lineBuffer;
            keyBuffer << 3;
            keyBuffer.trim();
            entBuffer.size(0);
        }
        else {
            if (keyBuffer.size()) {
                entBuffer += lineBuffer;
                entBuffer += "\n";
            }
        }
    } while (more);
    if ((keyBuffer.size()) && (entBuffer.size())) {
        writeEntry(book, keyBuffer, entBuffer);
    }

    delete book;

    FileMgr::getSystemFileMgr()->close(fd);

    return 0;
}

void parseParams	(	int	argc,
		char **	argv
	)

Definition at line 80 of file imp2gbs.cpp.

                                        {

    if (argc < 2) {
        usage(*argv);
    }

    inFile = argv[1];

    for (int i = 2; i < argc; i++) {
        if (!strcmp(argv[i], "-o")) {
            if ((i+1 < argc) && (argv[i+1][0] != '-')) {
                outPath = argv[i+1];
                i++;
            }
            else usage(*argv);
        }
        else if (!strcmp(argv[i], "-U")) {
            if (StringMgr::hasUTF8Support()) {
                toUpper = true;
            }
            else {
                fprintf(stderr, "Error: %s.  Cannot reliably toUpper without UTF8 support\n\t(recompile with ICU enabled)\n\n", *argv);
                usage(*argv);
            }
        }
        else if (!strcmp(argv[i], "-g")) {
            greekFilter = true;
        }
        else if (!strcmp(argv[i], "-O")) {
            augEnt = false;
        }
        else if (!strcmp(argv[i], "-a")) {
            augMod = true;
        }
        else if (!strcmp(argv[i], "-l")) {
            if (i+1 < argc) {
                lexLevels = atoi(argv[i+1]);
                i++;
            }
            if (!lexLevels) usage(*argv);
        }
    }
    if (!outPath.size()) {
        outPath = inFile;
        unsigned int i;
        for (i = 0; (i < outPath.size() && outPath[i] != '.'); i++);
        outPath.size(i);
    }
}

void usage

(

const char *

app

)

Definition at line 65 of file imp2gbs.cpp.

                            {
    fprintf(stderr, "imp2gbs 1.0 General Book module creation tool for the SWORD Project\n\n");
    fprintf(stderr, "usage: %s <inFile> [OPTIONS]\n", app);
    fprintf(stderr, "\t-o <outPath>\n\t\tSpecify an output Path other than inFile location.\n");
    fprintf(stderr, "\t-a\n\t\tAugment Module [default: create new]\n");
    fprintf(stderr, "\t-O\n\t\tOverwrite entries of same key [default: append to]\n");
    fprintf(stderr, "\t-U\n\t\tKey filter: Convert toUpper\n");
    fprintf(stderr, "\t-g\n\t\tKey filter: Strip Greek diacritics\n");
    fprintf(stderr, "\t-l <levels>\n\t\tKey filter: Pseudo-Lexicon n-level generation using first character\n");
    fprintf(stderr, "\t\te.g. -l 2 \"Abbey\" -> \"A/AB/Abbey\"\n");
    fprintf(stderr, "\n");
    exit (-1);
}

void writeEntry	(	SWModule *	book,
		SWBuf	keyBuffer,
		SWBuf	entBuffer
	)

Definition at line 131 of file imp2gbs.cpp.

                                                                  {


    if (greekFilter) {
        greekAccentsFilter.processText(keyBuffer);
    }

    if (toUpper) {
        unsigned size = (keyBuffer.size()+5)*3;
        keyBuffer.setFillByte(0);
        keyBuffer.resize(size);
        StringMgr::getSystemStringMgr()->upperUTF8(keyBuffer.getRawData(), size-2);
    }

// Added for Hesychius, but this stuff should be pushed back into new StringMgr
// functionality
#ifdef _ICU_
//  if (lexLevels) {
    if (lexLevels && !keyBuffer.startsWith("/Intro")) {
        unsigned size = (keyBuffer.size()+(lexLevels*2));
        keyBuffer.setFillByte(0);
        keyBuffer.resize(size);
            
        UErrorCode err = U_ZERO_ERROR;
        
        int max = (size+5)*3;
        UChar *ubuffer = new UChar[max+10];
        int32_t len;
        
        u_strFromUTF8(ubuffer, max+9, &len, keyBuffer.c_str(), -1, &err);
        if (err == U_ZERO_ERROR) {
            UChar *upper = new UChar[(lexLevels+1)*3];
            memcpy(upper, ubuffer, lexLevels*sizeof(UChar));
            upper[lexLevels] = 0;
            len = u_strToUpper(upper, (lexLevels+1)*3, upper, -1, 0, &err);
            memmove(ubuffer+len+1, ubuffer, (max-len)*sizeof(UChar));
            memcpy(ubuffer, upper, len*sizeof(UChar));
            ubuffer[len] = '/';
            delete [] upper;

            int totalShift = 0;
            for (int i = lexLevels-1; i; i--) {
                int shift = (i < len)? i : len;
                memmove(ubuffer+(shift+1), ubuffer, (max-shift)*sizeof(UChar));
                ubuffer[shift] = '/';
                totalShift += (shift+1);
            }
            u_strToUTF8(keyBuffer.getRawData(), max, 0, ubuffer, -1, &err);
        }
        
/*
        u_strFromUTF8(ubuffer, max+9, &len, keyBuffer.c_str(), -1, &err);
        if (err == U_ZERO_ERROR) {
            int totalShift = 0;
            for (int i = lexLevels; i; i--) {
                int shift = (i < len)? i : len;
                memmove(ubuffer+(shift+1), ubuffer, (max-shift)*sizeof(UChar));
                ubuffer[shift] = '/';
                totalShift += (shift+1);
            }
            UChar *upper = new UChar[(totalShift+1)*3];
            memcpy(upper, ubuffer, totalShift*sizeof(UChar));
            upper[totalShift] = 0;
            len = u_strToUpper(upper, (totalShift+1)*3, upper, -1, 0, &err);
            memmove(ubuffer+len, ubuffer+totalShift, (max-totalShift)*sizeof(UChar));
            memcpy(ubuffer, upper, len*sizeof(UChar));
            delete [] upper;
            u_strToUTF8(keyBuffer.getRawData(), max, 0, ubuffer, -1, &err);
        }
*/
        
        delete [] ubuffer;
    }
#endif

    std::cout << keyBuffer << std::endl;

    book->setKey(keyBuffer.c_str());

    // check to see if we already have an entry
    for (int i = 2; book->getKey()->popError() != KEYERR_OUTOFBOUNDS; i++) {
        SWBuf key;
        key.setFormatted("%s {%d}", keyBuffer.c_str(), i);
        std::cout << "dup key, trying: " << key << std::endl;
        book->setKey(key.c_str());
    }

    book->setEntry(entBuffer);
}

Variable Documentation

bool augEnt = true

Definition at line 60 of file imp2gbs.cpp.

bool augMod = false

Definition at line 59 of file imp2gbs.cpp.

UTF8GreekAccents greekAccentsFilter

Definition at line 62 of file imp2gbs.cpp.

bool greekFilter = false

Definition at line 58 of file imp2gbs.cpp.

SWBuf inFile

Definition at line 56 of file imp2gbs.cpp.

int lexLevels = 0

Definition at line 61 of file imp2gbs.cpp.

SWBuf outPath

Definition at line 55 of file imp2gbs.cpp.

bool toUpper = false

Definition at line 57 of file imp2gbs.cpp.