bool TessPDFRenderer::imageToPDFObj(Pix *pix,
char *filename,
long int objnum,
char **pdf_object,
long int *pdf_object_size) {
size_t n;
char b0[kBasicBufSize];
char b1[kBasicBufSize];
char b2[kBasicBufSize];
if (!pdf_object_size || !pdf_object)
return false;
*pdf_object = nullptr;
*pdf_object_size = 0;
if (!filename)
return false;
L_Compressed_Data *cid = nullptr;
const int kJpegQuality = 85;
int format, sad;
findFileFormat(filename, &format);
if (pixGetSpp(pix) == 4 && format == IFF_PNG) {
Pix *p1 = pixAlphaBlendUniform(pix, 0xffffff00);
sad = pixGenerateCIData(p1, L_FLATE_ENCODE, 0, 0, &cid);
pixDestroy(&p1);
} else {
sad = l_generateCIDataForPdf(filename, pix, kJpegQuality, &cid);
}
if (sad || !cid) {
l_CIDataDestroy(&cid);
return false;
}
const char *group4 = "";
const char *filter;
switch(cid->type) {
case L_FLATE_ENCODE:
filter = "/FlateDecode";
break;
case L_JPEG_ENCODE:
filter = "/DCTDecode";
break;
case L_G4_ENCODE:
filter = "/CCITTFaxDecode";
group4 = " /K -1\n";
break;
case L_JP2K_ENCODE:
filter = "/JPXDecode";
break;
default:
l_CIDataDestroy(&cid);
return false;
}
// Maybe someday we will accept RGBA but today is not that day.
// It requires creating an /SMask for the alpha channel.
// http://stackoverflow.com/questions/14220221
const char *colorspace;
if (cid->ncolors > 0) {
n = snprintf(b0, sizeof(b0),
" /ColorSpace [ /Indexed /DeviceRGB %d %s ]\n",
cid->ncolors - 1, cid->cmapdatahex);
if (n >= sizeof(b0)) {
l_CIDataDestroy(&cid);
return false;
}
colorspace = b0;
} else {
switch (cid->spp) {
case 1:
colorspace = " /ColorSpace /DeviceGray\n";
break;
case 3:
colorspace = " /ColorSpace /DeviceRGB\n";
break;
default:
l_CIDataDestroy(&cid);
return false;
}
}
int predictor = (cid->predictor) ? 14 : 1;
// IMAGE
n = snprintf(b1, sizeof(b1),
"%ld 0 obj\n"
"<<\n"
" /Length %ld\n"
" /Subtype /Image\n",
objnum, (unsigned long) cid->nbytescomp);
if (n >= sizeof(b1)) {
l_CIDataDestroy(&cid);
return false;
}
n = snprintf(b2, sizeof(b2),
" /Width %d\n"
" /Height %d\n"
" /BitsPerComponent %d\n"
" /Filter %s\n"
" /DecodeParms\n"
" <<\n"
" /Predictor %d\n"
" /Colors %d\n"
"%s"
" /Columns %d\n"
" /BitsPerComponent %d\n"
" >>\n"
">>\n"
"stream\n",
cid->w, cid->h, cid->bps, filter, predictor, cid->spp,
group4, cid->w, cid->bps);
if (n >= sizeof(b2)) {
l_CIDataDestroy(&cid);
return false;
}
const char *b3 =
"endstream\n"
"endobj\n";
size_t b1_len = strlen(b1);
size_t b2_len = strlen(b2);
size_t b3_len = strlen(b3);
size_t colorspace_len = strlen(colorspace);
*pdf_object_size =
b1_len + colorspace_len + b2_len + cid->nbytescomp + b3_len;
*pdf_object = new char[*pdf_object_size];
char *p = *pdf_object;
memcpy(p, b1, b1_len);
p += b1_len;
memcpy(p, colorspace, colorspace_len);
p += colorspace_len;
memcpy(p, b2, b2_len);
p += b2_len;
memcpy(p, cid->datacomp, cid->nbytescomp);
p += cid->nbytescomp;
memcpy(p, b3, b3_len);
l_CIDataDestroy(&cid);
return true;
}
bool TessPDFRenderer::imageToPDFObj(Pix *pix,
char *filename,
long int objnum,
char **pdf_object,
long int *pdf_object_size) {
size_t n;
char b0[kBasicBufSize];
char b1[kBasicBufSize];
char b2[kBasicBufSize];
if (!pdf_object_size || !pdf_object)
return false;
*pdf_object = NULL;
*pdf_object_size = 0;
if (!filename)
return false;
L_COMP_DATA *cid = NULL;
const int kJpegQuality = 85;
// TODO(jbreiden) Leptonica 1.71 doesn't correctly handle certain
// types of PNG files, especially if there are 2 samples per pixel.
// We can get rid of this logic after Leptonica 1.72 is released and
// has propagated everywhere. Bug discussion as follows.
// https://code.google.com/p/tesseract-ocr/issues/detail?id=1300
int format, sad;
findFileFormat(filename, &format);
if (pixGetSpp(pix) == 4 && format == IFF_PNG) {
pixSetSpp(pix, 3);
sad = pixGenerateCIData(pix, L_FLATE_ENCODE, 0, 0, &cid);
} else {
sad = l_generateCIDataForPdf(filename, pix, kJpegQuality, &cid);
}
if (sad || !cid) {
l_CIDataDestroy(&cid);
return false;
}
const char *group4 = "";
const char *filter;
switch (cid->type) {
case L_FLATE_ENCODE:
filter = "/FlateDecode";
break;
case L_JPEG_ENCODE:
filter = "/DCTDecode";
break;
case L_G4_ENCODE:
filter = "/CCITTFaxDecode";
group4 = " /K -1\n";
break;
case L_JP2K_ENCODE:
filter = "/JPXDecode";
break;
default:
l_CIDataDestroy(&cid);
return false;
}
// Maybe someday we will accept RGBA but today is not that day.
// It requires creating an /SMask for the alpha channel.
// http://stackoverflow.com/questions/14220221
const char *colorspace;
if (cid->ncolors > 0) {
n = snprintf(b0, sizeof(b0),
" /ColorSpace [ /Indexed /DeviceRGB %d %s ]\n",
cid->ncolors - 1, cid->cmapdatahex);
if (n >= sizeof(b0)) {
l_CIDataDestroy(&cid);
return false;
}
colorspace = b0;
} else {
switch (cid->spp) {
case 1:
colorspace = " /ColorSpace /DeviceGray\n";
break;
case 3:
colorspace = " /ColorSpace /DeviceRGB\n";
break;
default:
l_CIDataDestroy(&cid);
return false;
}
}
int predictor = (cid->predictor) ? 14 : 1;
// IMAGE
n = snprintf(b1, sizeof(b1),
"%ld 0 obj\n"
"<<\n"
" /Length %ld\n"
" /Subtype /Image\n",
objnum, (unsigned long) cid->nbytescomp);
if (n >= sizeof(b1)) {
l_CIDataDestroy(&cid);
return false;
}
n = snprintf(b2, sizeof(b2),
" /Width %d\n"
" /Height %d\n"
" /BitsPerComponent %d\n"
" /Filter %s\n"
" /DecodeParms\n"
" <<\n"
" /Predictor %d\n"
" /Colors %d\n"
"%s"
" /Columns %d\n"
" /BitsPerComponent %d\n"
" >>\n"
">>\n"
"stream\n",
cid->w, cid->h, cid->bps, filter, predictor, cid->spp,
group4, cid->w, cid->bps);
if (n >= sizeof(b2)) {
l_CIDataDestroy(&cid);
return false;
}
const char *b3 =
"endstream\n"
"endobj\n";
size_t b1_len = strlen(b1);
size_t b2_len = strlen(b2);
size_t b3_len = strlen(b3);
size_t colorspace_len = strlen(colorspace);
*pdf_object_size =
b1_len + colorspace_len + b2_len + cid->nbytescomp + b3_len;
*pdf_object = new char[*pdf_object_size];
if (!pdf_object) {
l_CIDataDestroy(&cid);
return false;
}
char *p = *pdf_object;
memcpy(p, b1, b1_len);
p += b1_len;
memcpy(p, colorspace, colorspace_len);
p += colorspace_len;
memcpy(p, b2, b2_len);
p += b2_len;
memcpy(p, cid->datacomp, cid->nbytescomp);
p += cid->nbytescomp;
memcpy(p, b3, b3_len);
l_CIDataDestroy(&cid);
return true;
}
bool TessPDFRenderer::pixToPDFObj(Pix *pix, long int objnum,
char **pdf_object,
long int *pdf_object_size) {
if (!pdf_object_size || !pdf_object)
return false;
*pdf_object = NULL;
*pdf_object_size = 0;
char b0[kBasicBufSize];
char b1[kBasicBufSize * 2];
char b2[kBasicBufSize];
L_COMP_DATA *cid;
int encoding_type;
const int kJpegQuality = 85;
if (selectDefaultPdfEncoding(pix, &encoding_type) != 0)
return false;
if (pixGenerateCIData(pix, encoding_type, kJpegQuality, 0, &cid) != 0)
return false;
const char *filter;
switch(encoding_type) {
case L_FLATE_ENCODE:
filter = "/FlateDecode";
break;
case L_JPEG_ENCODE:
filter = "/DCTDecode";
break;
case L_G4_ENCODE:
filter = "/CCITTFaxDecode";
break;
default:
return false;
}
const char *colorspace;
if (cid->ncolors > 0) {
snprintf(b0, sizeof(b0), "[ /Indexed /DeviceRGB %d %s ]",
cid->ncolors - 1, cid->cmapdatahex);
colorspace = b0;
} else {
switch (cid->spp) {
case 1:
colorspace = "/DeviceGray";
break;
case 3:
colorspace = "/DeviceRGB";
break;
default:
return false;
}
}
snprintf(b1, sizeof(b1),
"%ld 0 obj\n"
"<<\n"
" /Length %lu\n"
" /Subtype /Image\n"
" /ColorSpace %s\n"
" /Width %d\n"
" /Height %d\n"
" /BitsPerComponent %d\n"
" /Filter %s\n"
" /DecodeParms\n"
" <<\n"
" /K -1\n"
" /Columns %d\n"
" >>\n"
">>\n"
"stream\n",
objnum, (unsigned long) cid->nbytescomp, colorspace,
cid->w, cid->h, cid->bps, filter, cid->w);
size_t b1_len = strlen(b1);
snprintf(b2, sizeof(b2),
"\n"
"endstream\n"
"endobj\n");
size_t b2_len = strlen(b2);
*pdf_object_size = b1_len + cid->nbytescomp + b2_len;
*pdf_object = new char[*pdf_object_size];
if (!pdf_object)
return false;
memcpy(*pdf_object, b1, b1_len);
memcpy(*pdf_object + b1_len, cid->datacomp, cid->nbytescomp);
memcpy(*pdf_object + b1_len + cid->nbytescomp, b2, b2_len);
return true;
}
