TEIPlain Class Reference

#include <teiplain.h>

Inheritance diagram for TEIPlain:

Collaboration diagram for TEIPlain:

Classes
class	MyUserData

Public Member Functions
virtual const char *	getHeader () const
virtual char	processText (SWBuf &text, const SWKey *key=0, const SWModule *module=0)
	TEIPlain ()

Protected Member Functions
void	addAllowedEscapeString (const char *findString)
void	addEscapeStringSubstitute (const char *findString, const char *replaceString)
void	addTokenSubstitute (const char *findString, const char *replaceString)
void	appendEscapeString (SWBuf &buf, const char *escString)
virtual BasicFilterUserData *	createUserData (const SWModule *module, const SWKey *key)
virtual bool	handleEscapeString (SWBuf &buf, const char *escString, BasicFilterUserData *userData)
virtual bool	handleNumericEscapeString (SWBuf &buf, const char *escString)
virtual bool	handleToken (SWBuf &buf, const char *token, BasicFilterUserData *userData)
bool	passAllowedEscapeString (SWBuf &buf, const char *escString)
virtual bool	processStage (char, SWBuf &, char *&, BasicFilterUserData *)
void	removeAllowedEscapeString (const char *findString)
void	removeEscapeStringSubstitute (const char *findString)
void	removeTokenSubstitute (const char *findString)
void	setEscapeEnd (const char *escEnd)
void	setEscapeStart (const char *escStart)
void	setEscapeStringCaseSensitive (bool val)
void	setPassThruNumericEscapeString (bool val)
void	setPassThruUnknownEscapeString (bool val)
void	setPassThruUnknownToken (bool val)
virtual void	setStageProcessing (char stages)
void	setTokenCaseSensitive (bool val)
void	setTokenEnd (const char *tokenEnd)
void	setTokenStart (const char *tokenStart)
bool	substituteEscapeString (SWBuf &buf, const char *escString)
bool	substituteToken (SWBuf &buf, const char *token)

Static Protected Attributes
static const char	FINALIZE = 8
static const char	INITIALIZE = 1
static const char	POSTCHAR = 4
static const char	PRECHAR = 2

Detailed Description

this filter converts TEI text to plain text

Definition at line 33 of file teiplain.h.

Constructor & Destructor Documentation

SWORD_NAMESPACE_START TEIPlain::TEIPlain

(

)

Definition at line 29 of file teiplain.cpp.

                   {
    setTokenStart("<");
    setTokenEnd(">");

    setEscapeStart("&");
    setEscapeEnd(";");

    setEscapeStringCaseSensitive(true);

    addEscapeStringSubstitute("amp", "&");
    addEscapeStringSubstitute("apos", "'");
    addEscapeStringSubstitute("lt", "<");
    addEscapeStringSubstitute("gt", ">");
    addEscapeStringSubstitute("quot", "\"");

    setTokenCaseSensitive(true);
}

Member Function Documentation

void SWBasicFilter::addAllowedEscapeString ( const char *  findString )

protectedinherited

Registers an esc control sequence that can pass unchanged

Definition at line 156 of file swbasicfilter.cpp.

                                                                 {
    char *buf = 0;

    if (!escStringCaseSensitive) {
        stdstr(&buf, findString);
        toupperstr(buf);
        p->escPassSet.insert(StringSet::value_type(buf));
        delete [] buf;
    }
    else p->escPassSet.insert(StringSet::value_type(findString));
}

void SWBasicFilter::addEscapeStringSubstitute ( const char *  findString, const char *  replaceString  )

protectedinherited

Registers an esc control sequence

Definition at line 176 of file swbasicfilter.cpp.

                                                                                               {
    char *buf = 0;

    if (!escStringCaseSensitive) {
        stdstr(&buf, findString);
        toupperstr(buf);
        p->escSubMap.insert(DualStringMap::value_type(buf, replaceString));
        delete [] buf;
    }
    else p->escSubMap.insert(DualStringMap::value_type(findString, replaceString));
}

void SWBasicFilter::addTokenSubstitute ( const char *  findString, const char *  replaceString  )

protectedinherited

Registers a simple token substitutions. Usually called from the c-tor of a subclass

Definition at line 136 of file swbasicfilter.cpp.

                                                                                        {
    char *buf = 0;

    if (!tokenCaseSensitive) {
        stdstr(&buf, findString);
        toupperstr(buf);
        p->tokenSubMap[buf] = replaceString;
        delete [] buf;
    }
    else p->tokenSubMap[findString] = replaceString;
}

void SWBasicFilter::appendEscapeString ( SWBuf &  buf, const char *  escString  )

protectedinherited

This appends escString to buf as an entity

Definition at line 216 of file swbasicfilter.cpp.

                                                                        {
    buf += escStart;
    buf += escString;
    buf += escEnd;
}

virtual BasicFilterUserData* TEIPlain::createUserData ( const SWModule *  module, const SWKey *  key  )

inlineprotectedvirtual

Reimplemented from SWBasicFilter.

Definition at line 42 of file teiplain.h.

                                                                                          {
        return new MyUserData(module, key);
    }

virtual const char* SWFilter::getHeader ( ) const

inlinevirtualinherited

This method can supply a header associated with the processing done with this filter. A typical example is a suggested CSS style block for classed containers.

Reimplemented in OSISLaTeX, OSISXHTML, ThMLLaTeX, ThMLXHTML, TEIXHTML, GBFLaTeX, and GBFXHTML.

Definition at line 62 of file swfilter.h.

{ return ""; }

bool SWBasicFilter::handleEscapeString ( SWBuf &  buf, const char *  escString, BasicFilterUserData *  userData  )

protectedvirtualinherited

This function is called for every escape sequence encountered in the input text.

Parameters

buf	the output buffer
escString	the escape sequence (e.g. "amp" for &)
userData	user storage space for data transient to 1 full buffer parse

Returns

false if was not handled and should be handled in

subclasses should return true if they handled the esc seq, or false if they did not.

Definition at line 286 of file swbasicfilter.cpp.

                                                                                                       {
    return substituteEscapeString(buf, escString);
}

bool SWBasicFilter::handleNumericEscapeString ( SWBuf &  buf, const char *  escString  )

protectedvirtualinherited

This function is called for all numeric escape sequences. If passThrough

Parameters

buf	the output buffer
escString	the escape sequence (e.g. "#235" for &235;)

Returns

subclasses should return true if they handled the esc seq, or false if they did not.

Definition at line 244 of file swbasicfilter.cpp.

                                                                               {
    if (passThruNumericEsc) {
        appendEscapeString(buf, escString);
        return true;
    }
    return false;
}

bool TEIPlain::handleToken ( SWBuf &  buf, const char *  token, BasicFilterUserData *  userData  )

protectedvirtual

This function is called for every token encountered in the input text.

Parameters

buf	the output buffer
token	the token (e.g. "p align='left'"
userData	user storage space for data transient to 1 full buffer parse

Returns

subclasses should return true if they handled the token, or false if they did not.

Reimplemented from SWBasicFilter.

Definition at line 48 of file teiplain.cpp.

                                                                                       {
  // manually process if it wasn't a simple substitution
    if (!substituteToken(buf, token)) {
        //MyUserData *u = (MyUserData *)userData;
        XMLTag tag(token);

        // <p> paragraph tag
        if (!strcmp(tag.getName(), "p")) {
            if ((!tag.isEndTag()) && (!tag.isEmpty())) {    // non-empty start tag
                buf += "\n";
            }
            else if (tag.isEndTag()) {  // end tag
                buf += "\n";
                userData->supressAdjacentWhitespace = true;
            }
            else {                  // empty paragraph break marker
                buf += "\n\n";
                userData->supressAdjacentWhitespace = true;
            }
        }

        // <entryFree>
        else if (!strcmp(tag.getName(), "entryFree")) {
            SWBuf n = tag.getAttribute("n");
            if ((!tag.isEndTag()) && (!tag.isEmpty())) {
                    if (n != "") {
                    buf += n;
                    buf += ". ";
                }
            }
        }

        // <sense>
        else if (!strcmp(tag.getName(), "sense")) {
            SWBuf n = tag.getAttribute("n");
            if ((!tag.isEndTag()) && (!tag.isEmpty())) {
                    if (n != "") {
                    buf += n;
                    buf += ". ";
                }
            }
            else if (tag.isEndTag()) {
                            buf += "\n";
            }
        }

        // <div>
        else if (!strcmp(tag.getName(), "div")) {

            if ((!tag.isEndTag()) && (!tag.isEmpty())) {
                buf.append("\n\n\n");
            }
            else if (tag.isEndTag()) {
            }
        }

        // <etym>
        else if (!strcmp(tag.getName(), "etym")) {
            if ((!tag.isEndTag()) && (!tag.isEmpty())) {
                buf += "[";
            }
            else if (tag.isEndTag()) {
                    buf += "]";
            }
        }

        // <list> <item>  This implementation does not distinguish between forms of lists
        // it would be nice if a numbered list could be added
        
        else if (!strcmp(tag.getName(), "list")) {
            if ((!tag.isEndTag()) && (!tag.isEmpty())) {

                buf += "\n";
            }
            else if (tag.isEndTag()) {
                buf += "\n";
            }
        }
        else if (!strcmp(tag.getName(), "item")) {
            if ((!tag.isEndTag()) && (!tag.isEmpty())) {
                buf += "\t* ";
            }
            else if (tag.isEndTag()) {
                buf += "\n";
            }
        }
        else {
            return false;  // we still didn't handle token
        }
    }
    return true;
}

bool SWBasicFilter::passAllowedEscapeString ( SWBuf &  buf, const char *  escString  )

protectedinherited

This passes allowed escapeStrings

Definition at line 223 of file swbasicfilter.cpp.

                                                                             {
    StringSet::iterator it;

    if (!escStringCaseSensitive) {
            char *tmp = 0;
        stdstr(&tmp, escString);
        toupperstr(tmp);
        it = p->escPassSet.find(tmp);
        delete [] tmp;
    } else 
        it = p->escPassSet.find(escString);

    if (it != p->escPassSet.end()) {
        appendEscapeString(buf, escString);
        return true;
    }

    return false;
}

virtual bool SWBasicFilter::processStage ( char  , SWBuf &  , char *&  , BasicFilterUserData *    )

inlineprotectedvirtualinherited

Reimplemented in OSISPlain.

Definition at line 185 of file swbasicfilter.h.

{ return false; }

char SWBasicFilter::processText ( SWBuf &  text, const SWKey *  key = 0, const SWModule *  module = 0  )

virtualinherited

This method processes and appropriately modifies the text given it for a particular filter task

Parameters

text	The text to be filtered/converted
key	Current key That was used.
module	Current module.

Returns

0

Implements SWFilter.

Reimplemented in ThMLRTF, OSISOSIS, and OSISRTF.

Definition at line 315 of file swbasicfilter.cpp.

                                                                                     {
    char *from;
    char token[4096];
    int tokpos = 0;
    bool intoken = false;
    bool inEsc = false;
    int escStartPos = 0, escEndPos = 0;
    int tokenStartPos = 0, tokenEndPos = 0;
    SWBuf lastTextNode;
    BasicFilterUserData *userData = createUserData(module, key);

    SWBuf orig = text;
    from = orig.getRawData();
    text = "";

    if (processStages & INITIALIZE) {
        if (processStage(INITIALIZE, text, from, userData)) {   // processStage handled it all
            delete userData;
            return 0;
        }
    }

    for (;*from; from++) {

        if (processStages & PRECHAR) {
            if (processStage(PRECHAR, text, from, userData))    // processStage handled this char
                continue;
        }

        if (*from == tokenStart[tokenStartPos]) {
            if (tokenStartPos == (tokenStartLen - 1)) {
                intoken = true;
                tokpos = 0;
                token[0] = 0;
                token[1] = 0;
                token[2] = 0;
                inEsc = false;
            }
            else tokenStartPos++;
            continue;
        }

        if (*from == escStart[escStartPos]) {
            if (escStartPos == (escStartLen - 1)) {
                intoken = true;
                tokpos = 0;
                token[0] = 0;
                token[1] = 0;
                token[2] = 0;
                inEsc = true;
            }
            else escStartPos++;
            continue;
        }

        if (inEsc) {
            if (*from == escEnd[escEndPos]) {
                if (escEndPos == (escEndLen - 1)) {
                    intoken = inEsc = false;
                    userData->lastTextNode = lastTextNode;
                    
                    if (!userData->suspendTextPassThru)  { //if text through is disabled no tokens should pass, too
                        if ((!handleEscapeString(text, token, userData)) && (passThruUnknownEsc)) {
                            appendEscapeString(text, token);
                        }
                    }
                    escEndPos = escStartPos = tokenEndPos = tokenStartPos = 0;
                    lastTextNode = "";
                    continue;
                }
            }
        }

        if (!inEsc) {
            if (*from == tokenEnd[tokenEndPos]) {
                if (tokenEndPos == (tokenEndLen - 1)) {
                    intoken = false;
                    userData->lastTextNode = lastTextNode;
                    if ((!handleToken(text, token, userData)) && (passThruUnknownToken)) {
                        text += tokenStart;
                        text += token;
                        text += tokenEnd;
                    }
                    escEndPos = escStartPos = tokenEndPos = tokenStartPos = 0;
                    lastTextNode = "";
                    if (!userData->suspendTextPassThru) {
                        userData->lastSuspendSegment.size(0);
                    }
                    continue;
                }
            }
        }

        if (intoken) {
            if (tokpos < 4090) {
                token[tokpos++] = *from;
                token[tokpos+2] = 0;
            }
        }
        else {
            if ((!userData->supressAdjacentWhitespace) || (*from != ' ')) {
                if (!userData->suspendTextPassThru) {
                    text.append(*from);
                }
                else    userData->lastSuspendSegment.append(*from);
                lastTextNode.append(*from);
            }
            userData->supressAdjacentWhitespace = false;
        }

        if (processStages & POSTCHAR)
            processStage(POSTCHAR, text, from, userData);

    }

    if (processStages & FINALIZE)
        processStage(FINALIZE, text, from, userData);

    delete userData;
    return 0;
}

void SWBasicFilter::removeAllowedEscapeString ( const char *  findString )

protectedinherited

Unregisters an esc control sequence that can pass unchanged

Definition at line 169 of file swbasicfilter.cpp.

                                                                    {
    if (p->escPassSet.find(findString) != p->escPassSet.end()) {
        p->escPassSet.erase( p->escPassSet.find(findString) );
    }
}

void SWBasicFilter::removeEscapeStringSubstitute ( const char *  findString )

protectedinherited

Unregisters an esc control sequence

Definition at line 189 of file swbasicfilter.cpp.

                                                                       {
    if (p->escSubMap.find(findString) != p->escSubMap.end()) {
        p->escSubMap.erase( p->escSubMap.find(findString) );
    }
}

void SWBasicFilter::removeTokenSubstitute ( const char *  findString )

protectedinherited

Unregisters a simple token substitute

Definition at line 149 of file swbasicfilter.cpp.

                                                                {
    if (p->tokenSubMap.find(findString) != p->tokenSubMap.end()) {
        p->tokenSubMap.erase( p->tokenSubMap.find(findString) );
    }
}

void SWBasicFilter::setEscapeEnd ( const char *  escEnd )

protectedinherited

Sets the end of escape sequence (by default ";").

Definition at line 297 of file swbasicfilter.cpp.

                                                   {
    stdstr(&(this->escEnd), escEnd);
    escEndLen   = strlen(escEnd);
}

void SWBasicFilter::setEscapeStart ( const char *  escStart )

protectedinherited

Sets the beginning of escape sequence (by default "&").

Definition at line 291 of file swbasicfilter.cpp.

                                                       {
    stdstr(&(this->escStart), escStart);
    escStartLen = strlen(escStart);
}

void SWBasicFilter::setEscapeStringCaseSensitive ( bool  val )

protectedinherited

Are escapeStrings case sensitive or not? Call this function before addEscapeStingSubstitute()

Definition at line 131 of file swbasicfilter.cpp.

                                                         {
    escStringCaseSensitive = val;
}

void SWBasicFilter::setPassThruNumericEscapeString ( bool  val )

protectedinherited

Sets whether to pass thru a numeric escape sequence unchanged or allow it to be handled otherwise. Default is false.

Definition at line 121 of file swbasicfilter.cpp.

                                                           {
    passThruUnknownEsc = val;
}

void SWBasicFilter::setPassThruUnknownEscapeString ( bool  val )

protectedinherited

Sets whether to pass thru an unknown escape sequence unchanged or just remove it. Default is false.

Definition at line 116 of file swbasicfilter.cpp.

                                                           {
    passThruUnknownEsc = val;
}

void SWBasicFilter::setPassThruUnknownToken ( bool  val )

protectedinherited

Sets whether to pass thru an unknown token unchanged or just remove it. Default is false.

Definition at line 111 of file swbasicfilter.cpp.

                                                    {
    passThruUnknownToken = val;
}

virtual void SWBasicFilter::setStageProcessing ( char  stages )

inlineprotectedvirtualinherited

Definition at line 186 of file swbasicfilter.h.

{ processStages = stages; } // see STATICs up above

void SWBasicFilter::setTokenCaseSensitive ( bool  val )

protectedinherited

Are tokens case sensitive (like in GBF) or not? Call this function before addTokenSubstitute()

Definition at line 126 of file swbasicfilter.cpp.

                                                  {
    tokenCaseSensitive = val;
}

void SWBasicFilter::setTokenEnd ( const char *  tokenEnd )

protectedinherited

Sets the end of token start sequence (by default ">").

Definition at line 309 of file swbasicfilter.cpp.

                                                    {
    stdstr(&(this->tokenEnd), tokenEnd);
    tokenEndLen   = strlen(tokenEnd);
}

void SWBasicFilter::setTokenStart ( const char *  tokenStart )

protectedinherited

Sets the beginning of token start sequence (by default "<").

Definition at line 303 of file swbasicfilter.cpp.

                                                        {
    stdstr(&(this->tokenStart), tokenStart);
    tokenStartLen = strlen(tokenStart);
}

bool SWBasicFilter::substituteEscapeString ( SWBuf &  buf, const char *  escString  )

protectedinherited

This function performs the substitution of escapeStrings

Definition at line 253 of file swbasicfilter.cpp.

                                                                            {
    DualStringMap::iterator it;

    if (*escString == '#') {
        return handleNumericEscapeString(buf, escString);
    }

    if (passAllowedEscapeString(buf, escString)) {
        return true;
    }

    if (!escStringCaseSensitive) {
            char *tmp = 0;
        stdstr(&tmp, escString);
        toupperstr(tmp);
        it = p->escSubMap.find(tmp);
        delete [] tmp;
    } else 
    it = p->escSubMap.find(escString);

    if (it != p->escSubMap.end()) {
        buf += it->second.c_str();
        return true;
    }
    return false;
}

bool SWBasicFilter::substituteToken ( SWBuf &  buf, const char *  token  )

protectedinherited

This function performs the substitution of tokens

Definition at line 196 of file swbasicfilter.cpp.

                                                                 {
    DualStringMap::iterator it;

    if (!tokenCaseSensitive) {
            char *tmp = 0;
        stdstr(&tmp, token);
        toupperstr(tmp);
        it = p->tokenSubMap.find(tmp);
        delete [] tmp;
    } else
    it = p->tokenSubMap.find(token);

    if (it != p->tokenSubMap.end()) {
        buf += it->second.c_str();
        return true;
    }
    return false;
}

Member Data Documentation

const char SWBasicFilter::FINALIZE = 8

staticprotectedinherited

Definition at line 99 of file swbasicfilter.h.

const char SWBasicFilter::INITIALIZE = 1

staticprotectedinherited

Definition at line 96 of file swbasicfilter.h.

const char SWBasicFilter::POSTCHAR = 4

staticprotectedinherited

Definition at line 98 of file swbasicfilter.h.

const char SWBasicFilter::PRECHAR = 2

staticprotectedinherited

Definition at line 97 of file swbasicfilter.h.

---

The documentation for this class was generated from the following files:

• include/teiplain.h
• src/modules/filters/teiplain.cpp