Apple-2-Tools
/
ciderpress
mirror of https://github.com/fadden/ciderpress.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

Add vestigal support for Gutenberg Word Processor formatting

This commit is contained in:
David Schmidt 2009-01-05 02:11:31 +00:00
parent abd3515424
commit 5637461d8a
9 changed files with 1401 additions and 1293 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
app/Main.cpp
View File

@ -2644,6 +2644,8 @@ MainWindow::ConfigureReformatFromPreferences(ReformatHolder* pReformat)
pPreferences->GetPrefBool(kPrConvGWP));
pReformat->SetReformatAllowed(ReformatHolder::kReformatMagicWindow,
pPreferences->GetPrefBool(kPrConvText8));
pReformat->SetReformatAllowed(ReformatHolder::kReformatGutenberg,
pPreferences->GetPrefBool(kPrConvGutenberg));
pReformat->SetReformatAllowed(ReformatHolder::kReformatAWP,
pPreferences->GetPrefBool(kPrConvAWP));
pReformat->SetReformatAllowed(ReformatHolder::kReformatAWP,
@ -2707,7 +2709,13 @@ MainWindow::ConfigureReformatFromPreferences(ReformatHolder* pReformat)
/*static*/ ReformatHolder::SourceFormat
MainWindow::ReformatterSourceFormat(DiskImg::FSFormat format)
{
if (DiskImg::UsesDOSFileStructure(format))
/*
* Gutenberg both UsesDOSFileStructure and is formatted with
* kFormatGutenberg, so check for the latter first.
*/
if (format == DiskImg::kFormatGutenberg)
return ReformatHolder::kSourceFormatGutenberg;
else if (DiskImg::UsesDOSFileStructure(format))
return ReformatHolder::kSourceFormatDOS;
else if (format == DiskImg::kFormatCPM)
return ReformatHolder::kSourceFormatCPM;

2
app/Preferences.cpp
View File

@ -118,6 +118,7 @@ const Preferences::PrefMap Preferences::fPrefMaps[kPrefNumLastEntry] = {
{ kPrConvBusiness, kBool, kPrefsSect, _T("conv-business") },
{ kPrConvGWP, kBool, kPrefsSect, _T("conv-gwp") },
{ kPrConvText8, kBool, kPrefsSect, _T("conv-text8") },
{ kPrConvGutenberg, kBool, kPrefsSect, _T("conv-gutenberg") },
{ kPrConvAWP, kBool, kPrefsSect, _T("conv-awp") },
{ kPrConvADB, kBool, kPrefsSect, _T("conv-adb") },
{ kPrConvASP, kBool, kPrefsSect, _T("conv-asp") },
@ -227,6 +228,7 @@ Preferences::Preferences(void)
SetPrefBool(kPrConvBusiness, true);
SetPrefBool(kPrConvGWP, true);
SetPrefBool(kPrConvText8, true);
SetPrefBool(kPrConvGutenberg, true);
SetPrefBool(kPrConvAWP, true);
SetPrefBool(kPrConvADB, true);
SetPrefBool(kPrConvASP, true);

1
app/Preferences.h
View File

@ -171,6 +171,7 @@ typedef enum {
kPrConvBusiness, // bool
kPrConvGWP, // bool
kPrConvText8, // bool
kPrConvGutenberg, // bool
kPrConvAWP, // bool
kPrConvADB, // bool
kPrConvASP, // bool

4
reformat/Reformat.cpp
View File

@ -86,6 +86,7 @@ ReformatHolder::GetReformatInstance(ReformatID id)
case kReformatSHR_DG3200: pReformat = new ReformatDG3200SHR; break;
case kReformatPrintShop: pReformat = new ReformatPrintShop; break;
case kReformatMacPaint: pReformat = new ReformatMacPaint; break;
case kReformatGutenberg: pReformat = new ReformatGutenberg; break;
case kReformatUnknown:
case kReformatMAX:
default: assert(false); break;
@ -191,6 +192,9 @@ ReformatHolder::GetReformatName(ReformatID id)
case kReformatMagicWindow:
descr = "Magic Window";
break;
case kReformatGutenberg:
descr = "Gutenberg Word Processor";
break;
case kReformatAWP:
descr = "AppleWorks Word Processor";
break;

7
reformat/Reformat.h
View File

@ -1,5 +1,6 @@
/*
* CiderPress
* Copyright (C) 2009 by CiderPress authors. All Rights Reserved.
* Copyright (C) 2007, 2008 by faddenSoft, LLC. All Rights Reserved.
* See the file LICENSE for distribution terms.
*/
@ -96,6 +97,7 @@ public:
kReformatGWP,
kReformatMagicWindow,
kReformatGutenberg,
kReformatAWP,
kReformatADB,
@ -182,12 +184,14 @@ public:
*
* We want to know if it's DOS so we can relax some file-type checking,
* and we want to know if it's CP/M so we can adjust the way we think
* about text files.
* about text files. We want to know if it's Gutenberg because they only
* have one type of file, and it's indistingusihable from any other text file!
*/
typedef enum SourceFormat {
kSourceFormatGeneric = 0,
kSourceFormatDOS,
kSourceFormatCPM,
kSourceFormatGutenberg,
} SourceFormat;
@ -268,6 +272,7 @@ public:
friend class ReformatTeach;
friend class ReformatGWP;
friend class ReformatMagicWindow;
friend class ReformatGutenberg;
friend class ReformatAWP;
friend class ReformatADB;
friend class ReformatASP;

2187
reformat/ReformatBase.cpp
View File

File diff suppressed because it is too large Load Diff

2
reformat/ReformatBase.h
View File

@ -314,6 +314,8 @@ protected:
void ConvertEOL(const unsigned char* srcBuf, long srcLen,
bool stripHiBits);
void ConvertEOL(const unsigned char* srcBuf, long srcLen,
bool stripHiBits, bool stripNulls);
void BufHexDump(const unsigned char* srcBuf, long srcLen);
void SetResultBuffer(ReformatOutput* pOutput, bool multiFont = false);

399
reformat/Text8.cpp
View File

@ -1,174 +1,225 @@
/*
* CiderPress
* Copyright (C) 2007 by faddenSoft, LLC. All Rights Reserved.
* See the file LICENSE for distribution terms.
*/
/*
* Convert 8-bit word processor files.
*
* Most formats convert reasonably well with "Converted Text", but this
* allows the files to be handled more transparently (e.g. Magic Window
* "formatted files", which can be mistaken for code.
*/
#include "StdAfx.h"
#include "Text8.h"
/*
* ===========================================================================
* Magic Window / Magic Window II
* ===========================================================================
*/
/*
* Magic Window and Magic Window II appear to use the same format for their
* "formatted files". The files are of type 'B', with a valid address field,
* and what looks like junk in the length field. The files have a 256-byte
* header that seems to hold some sort of title string as well as some
* binary goodies that I'm not sure what they are.
*
* The data from offset 256 on is entirely mixed-case high-ASCII text. It
* may contain printer-specific escape codes for bold, italic, etc.
*
* A ".MW" filename suffix is enforced by the program.
*/
/*
* Decide whether or not we want to handle this file.
*/
void
ReformatMagicWindow::Examine(ReformatHolder* pHolder)
{
if (pHolder->GetFileType() == kTypeBIN) {
bool isMW = ReformatMagicWindow::IsFormatted(pHolder);
bool isDotMW = strcasecmp(pHolder->GetNameExt(), ".MW") == 0;
if (isMW && isDotMW) {
/* gotta be */
pHolder->SetApplic(ReformatHolder::kReformatMagicWindow,
ReformatHolder::kApplicYes,
ReformatHolder::kApplicNot, ReformatHolder::kApplicNot);
} else if (isDotMW) {
/* right type and name; maybe our test is broken? */
pHolder->SetApplic(ReformatHolder::kReformatMagicWindow,
ReformatHolder::kApplicProbably,
ReformatHolder::kApplicNot, ReformatHolder::kApplicNot);
} else if (isMW) {
/* not likely, but offer it as non-default option */
pHolder->SetApplic(ReformatHolder::kReformatMagicWindow,
ReformatHolder::kApplicProbablyNot,
ReformatHolder::kApplicNot, ReformatHolder::kApplicNot);
} else {
/* not one of ours */
pHolder->SetApplic(ReformatHolder::kReformatMagicWindow,
ReformatHolder::kApplicNot,
ReformatHolder::kApplicNot, ReformatHolder::kApplicNot);
}
} else {
/* "unformatted" text even if ".MW"; nothing special required */
pHolder->SetApplic(ReformatHolder::kReformatMagicWindow,
ReformatHolder::kApplicNot,
ReformatHolder::kApplicNot, ReformatHolder::kApplicNot);
}
}
/*
* Figure out if this is a Magic Window "formatted" file.
*
* I don't know much about the format, so this is based on the similarities
* observed between half a dozen documents from different sources.
*/
/*static*/ bool
ReformatMagicWindow::IsFormatted(const ReformatHolder* pHolder)
{
const unsigned char* ptr = pHolder->GetSourceBuf(ReformatHolder::kPartData);
long srcLen = pHolder->GetSourceLen(ReformatHolder::kPartData);
int i, count00, count20;
/* want 256-byte header, plus a few bytes to check text */
if (srcLen < kHeaderLen+8)
return false;
/*
* First byte always seems to be 0x8d.
*/
if (ptr[0x00] != 0x8d)
return false;
/*
* 0x58 - 0xa0 is mostly filled with 0x00 (for Magic Window) or 0x20
* (for Magic Window II). Both seem to have space for the title in the
* preceeding part, but it's high-ASCII for MW and low-ASCII for MW2.
*
* Expect 50 out of 72 to match. If this is actually just uninitialized
* data then this test will be bogus.
*/
count00 = count20 = 0;
for (i = 0x58; i < 0xa0; i++) {
if (ptr[i] == 0x00)
count00++;
if (ptr[i] == 0x20)
count20++;
}
if (count00 < 50 && count20 < 50)
return false;
/*
* 0xa2 has some recognizeable bytes; sample values:
* MW 42 06 36 50 08 40
* MW2 42 06 36 55 08 40
* MW2 42 04 3a 50 00 50
* Not really sure what to make of these. If we can bracket these
* values we might have something.
*/
if (ptr[0xa2] != 0x42 ||
(ptr[0xa3] < 2 && ptr[0xa3] > 10) ||
(ptr[0xa4] < 0x30 && ptr[0xa4] > 0x40))
return false;
/*
* Make sure the rest of the file is 100% high ASCII.
*/
ptr += kHeaderLen;
srcLen -= kHeaderLen;
while (srcLen--) {
if ((*ptr & 0x80) == 0)
return false;
}
return true;
}
/*
* Skip the header and text-convert the reset.
*/
int
ReformatMagicWindow::Process(const ReformatHolder* pHolder,
ReformatHolder::ReformatID id, ReformatHolder::ReformatPart part,
ReformatOutput* pOutput)
{
const unsigned char* srcPtr = pHolder->GetSourceBuf(part);
long srcLen = pHolder->GetSourceLen(part);
long length = srcLen;
int retval = -1;
fUseRTF = false;
RTFBegin();
if (srcLen <= kHeaderLen)
goto bail;
ConvertEOL(srcPtr + kHeaderLen, srcLen - kHeaderLen, true);
//done:
RTFEnd();
SetResultBuffer(pOutput);
retval = 0;
bail:
return retval;
}
/*
* CiderPress
* Copyright (C) 2009 by CiderPress authors. All Rights Reserved.
* Copyright (C) 2007 by faddenSoft, LLC. All Rights Reserved.
* See the file LICENSE for distribution terms.
*/
/*
* Convert 8-bit word processor files.
*
* Most formats convert reasonably well with "Converted Text", but this
* allows the files to be handled more transparently (e.g. Magic Window
* "formatted files", which can be mistaken for code.
*/
#include "StdAfx.h"
#include "Text8.h"
/*
* ===========================================================================
* Magic Window / Magic Window II
* ===========================================================================
*/
/*
* Magic Window and Magic Window II appear to use the same format for their
* "formatted files". The files are of type 'B', with a valid address field,
* and what looks like junk in the length field. The files have a 256-byte
* header that seems to hold some sort of title string as well as some
* binary goodies that I'm not sure what they are.
*
* The data from offset 256 on is entirely mixed-case high-ASCII text. It
* may contain printer-specific escape codes for bold, italic, etc.
*
* A ".MW" filename suffix is enforced by the program.
*/
/*
* Decide whether or not we want to handle this file.
*/
void
ReformatMagicWindow::Examine(ReformatHolder* pHolder)
{
if (pHolder->GetFileType() == kTypeBIN) {
bool isMW = ReformatMagicWindow::IsFormatted(pHolder);
bool isDotMW = strcasecmp(pHolder->GetNameExt(), ".MW") == 0;
if (isMW && isDotMW) {
/* gotta be */
pHolder->SetApplic(ReformatHolder::kReformatMagicWindow,
ReformatHolder::kApplicYes,
ReformatHolder::kApplicNot, ReformatHolder::kApplicNot);
} else if (isDotMW) {
/* right type and name; maybe our test is broken? */
pHolder->SetApplic(ReformatHolder::kReformatMagicWindow,
ReformatHolder::kApplicProbably,
ReformatHolder::kApplicNot, ReformatHolder::kApplicNot);
} else if (isMW) {
/* not likely, but offer it as non-default option */
pHolder->SetApplic(ReformatHolder::kReformatMagicWindow,
ReformatHolder::kApplicProbablyNot,
ReformatHolder::kApplicNot, ReformatHolder::kApplicNot);
} else {
/* not one of ours */
pHolder->SetApplic(ReformatHolder::kReformatMagicWindow,
ReformatHolder::kApplicNot,
ReformatHolder::kApplicNot, ReformatHolder::kApplicNot);
}
} else {
/* "unformatted" text even if ".MW"; nothing special required */
pHolder->SetApplic(ReformatHolder::kReformatMagicWindow,
ReformatHolder::kApplicNot,
ReformatHolder::kApplicNot, ReformatHolder::kApplicNot);
}
}
/*
* Figure out if this is a Magic Window "formatted" file.
*
* I don't know much about the format, so this is based on the similarities
* observed between half a dozen documents from different sources.
*/
/*static*/ bool
ReformatMagicWindow::IsFormatted(const ReformatHolder* pHolder)
{
const unsigned char* ptr = pHolder->GetSourceBuf(ReformatHolder::kPartData);
long srcLen = pHolder->GetSourceLen(ReformatHolder::kPartData);
int i, count00, count20;
/* want 256-byte header, plus a few bytes to check text */
if (srcLen < kHeaderLen+8)
return false;
/*
* First byte always seems to be 0x8d.
*/
if (ptr[0x00] != 0x8d)
return false;
/*
* 0x58 - 0xa0 is mostly filled with 0x00 (for Magic Window) or 0x20
* (for Magic Window II). Both seem to have space for the title in the
* preceeding part, but it's high-ASCII for MW and low-ASCII for MW2.
*
* Expect 50 out of 72 to match. If this is actually just uninitialized
* data then this test will be bogus.
*/
count00 = count20 = 0;
for (i = 0x58; i < 0xa0; i++) {
if (ptr[i] == 0x00)
count00++;
if (ptr[i] == 0x20)
count20++;
}
if (count00 < 50 && count20 < 50)
return false;
/*
* 0xa2 has some recognizeable bytes; sample values:
* MW 42 06 36 50 08 40
* MW2 42 06 36 55 08 40
* MW2 42 04 3a 50 00 50
* Not really sure what to make of these. If we can bracket these
* values we might have something.
*/
if (ptr[0xa2] != 0x42 ||
(ptr[0xa3] < 2 && ptr[0xa3] > 10) ||
(ptr[0xa4] < 0x30 && ptr[0xa4] > 0x40))
return false;
/*
* Make sure the rest of the file is 100% high ASCII.
*/
ptr += kHeaderLen;
srcLen -= kHeaderLen;
while (srcLen--) {
if ((*ptr & 0x80) == 0)
return false;
}
return true;
}
/*
* Skip the header and text-convert the rest.
*/
int
ReformatMagicWindow::Process(const ReformatHolder* pHolder,
ReformatHolder::ReformatID id, ReformatHolder::ReformatPart part,
ReformatOutput* pOutput)
{
const unsigned char* srcPtr = pHolder->GetSourceBuf(part);
long srcLen = pHolder->GetSourceLen(part);
long length = srcLen;
int retval = -1;
fUseRTF = false;
RTFBegin();
if (srcLen <= kHeaderLen)
goto bail;
ConvertEOL(srcPtr + kHeaderLen, srcLen - kHeaderLen, true);
//done:
RTFEnd();
SetResultBuffer(pOutput);
retval = 0;
bail:
return retval;
}
/*
* ===========================================================================
* Gutenberg Word Processor
* ===========================================================================
*/
/*
* Decide whether or not we want to handle this file.
*/
void
ReformatGutenberg::Examine(ReformatHolder* pHolder)
{
if ((pHolder->GetFileType() == kTypeTXT) &&
(pHolder->GetSourceFormat() == ReformatHolder::kSourceFormatGutenberg)) {
pHolder->SetApplic(ReformatHolder::kReformatGutenberg,
ReformatHolder::kApplicYes,
ReformatHolder::kApplicNot, ReformatHolder::kApplicNot);
}
}
/*
* Convert the text.
*/
int
ReformatGutenberg::Process(const ReformatHolder* pHolder,
ReformatHolder::ReformatID id, ReformatHolder::ReformatPart part,
ReformatOutput* pOutput)
{
const unsigned char* srcPtr = pHolder->GetSourceBuf(part);
long srcLen = pHolder->GetSourceLen(part);
long length = srcLen;
int retval = -1;
fUseRTF = false;
RTFBegin();
ConvertEOL(srcPtr, srcLen, true, true);
RTFEnd();
SetResultBuffer(pOutput);
retval = 0;
return retval;
}

82
reformat/Text8.h
View File

@ -1,33 +1,49 @@
/*
* CiderPress
* Copyright (C) 2007 by faddenSoft, LLC. All Rights Reserved.
* See the file LICENSE for distribution terms.
*/
/*
* Reformat 8-bit word processor files.
*/
#ifndef __LR_TEXT8__
#define __LR_TEXT8__
#include "ReformatBase.h"
/*
* Magic Window / Magic Window II
*/
class ReformatMagicWindow : public ReformatText {
public:
ReformatMagicWindow(void) {}
virtual ~ReformatMagicWindow(void) {}
virtual void Examine(ReformatHolder* pHolder);
virtual int Process(const ReformatHolder* pHolder,
ReformatHolder::ReformatID id, ReformatHolder::ReformatPart part,
ReformatOutput* pOutput);
private:
static bool IsFormatted(const ReformatHolder* pHolder);
enum { kHeaderLen = 256 };
};
#endif /*__LR_TEXT8__*/
/*
* CiderPress
* Copyright (C) 2009 by Ciderpress authors. All Rights Reserved.
* Copyright (C) 2007 by faddenSoft, LLC. All Rights Reserved.
* See the file LICENSE for distribution terms.
*/
/*
* Reformat 8-bit word processor files.
*/
#ifndef __LR_TEXT8__
#define __LR_TEXT8__
#include "ReformatBase.h"
/*
* Magic Window / Magic Window II
*/
class ReformatMagicWindow : public ReformatText {
public:
ReformatMagicWindow(void) {}
virtual ~ReformatMagicWindow(void) {}
virtual void Examine(ReformatHolder* pHolder);
virtual int Process(const ReformatHolder* pHolder,
ReformatHolder::ReformatID id, ReformatHolder::ReformatPart part,
ReformatOutput* pOutput);
private:
static bool IsFormatted(const ReformatHolder* pHolder);
enum { kHeaderLen = 256 };
};
/*
* Guterberg Word Processor
*/
class ReformatGutenberg : public ReformatText {
public:
ReformatGutenberg(void) {}
virtual ~ReformatGutenberg(void) {}
virtual void Examine(ReformatHolder* pHolder);
virtual int Process(const ReformatHolder* pHolder,
ReformatHolder::ReformatID id, ReformatHolder::ReformatPart part,
ReformatOutput* pOutput);
};
#endif /*__LR_TEXT8__*/