/*
* convertFilesToPS()
*
* Input: dirin (input directory)
* substr (<optional> substring filter on filenames; can be NULL)
* res (typ. 300 or 600 ppi)
* fileout (output ps file)
* Return: 0 if OK, 1 on error
*
* Notes:
* (1) This generates a PS file for all image files in a specified
* directory that contain the substr pattern to be matched.
* (2) Each image is written to a separate page in the output PS file.
* (3) All images are written compressed:
* * if tiffg4 --> use ccittg4
* * if jpeg --> use dct
* * all others --> use flate
* If the image is jpeg or tiffg4, we use the existing compressed
* strings for the encoding; otherwise, we read the image into
* a pix and flate-encode the pieces.
* (4) The resolution is often confusing. It is interpreted
* as the resolution of the output display device: "If the
* input image were digitized at 300 ppi, what would it
* look like when displayed at res ppi." So, for example,
* if res = 100 ppi, then the display pixels are 3x larger
* than the 300 ppi pixels, and the image will be rendered
* 3x larger.
* (5) The size of the PostScript file is independent of the resolution,
* because the entire file is encoded. The res parameter just
* tells the PS decomposer how to render the page. Therefore,
* for minimum file size without loss of visual information,
* if the output res is less than 300, you should downscale
* the image to the output resolution before wrapping in PS.
* (6) The "canvas" on which the image is rendered, at the given
* output resolution, is a standard page size (8.5 x 11 in).
*/
l_int32
convertFilesToPS(const char *dirin,
const char *substr,
l_int32 res,
const char *fileout)
{
SARRAY *sa;
PROCNAME("convertFilesToPS");
if (!dirin)
return ERROR_INT("dirin not defined", procName, 1);
if (!fileout)
return ERROR_INT("fileout not defined", procName, 1);
if (res <= 0) {
L_INFO("setting res to 300 ppi", procName);
res = 300;
}
if (res < 10 || res > 4000)
L_WARNING("res is typically in the range 300-600 ppi", procName);
/* Get all filtered and sorted full pathnames. */
sa = getSortedPathnamesInDirectory(dirin, substr, 0, 0);
/* Generate the PS file. */
sarrayConvertFilesToPS(sa, res, fileout);
sarrayDestroy(&sa);
return 0;
}
/*
* convertFilesFittedToPS()
*
* Input: dirin (input directory)
* substr (<optional> substring filter on filenames; can be NULL)
* xpts, ypts (desired size in printer points; use 0 for default)
* fileout (output ps file)
* Return: 0 if OK, 1 on error
*
* Notes:
* (1) This generates a PS file for all files in a specified directory
* that contain the substr pattern to be matched.
* (2) Each image is written to a separate page in the output PS file.
* (3) All images are written compressed:
* * if tiffg4 --> use ccittg4
* * if jpeg --> use dct
* * all others --> use flate
* If the image is jpeg or tiffg4, we use the existing compressed
* strings for the encoding; otherwise, we read the image into
* a pix and flate-encode the pieces.
* (4) The resolution is internally determined such that the images
* are rendered, in at least one direction, at 100% of the given
* size in printer points. Use 0.0 for xpts or ypts to get
* the default value, which is 612.0 or 792.0, rsp.
* (5) The size of the PostScript file is independent of the resolution,
* because the entire file is encoded. The @xpts and @ypts
* parameter tells the PS decomposer how to render the page.
*/
l_int32
convertFilesFittedToPS(const char *dirin,
const char *substr,
l_float32 xpts,
l_float32 ypts,
const char *fileout)
{
SARRAY *sa;
PROCNAME("convertFilesFittedToPS");
if (!dirin)
return ERROR_INT("dirin not defined", procName, 1);
if (!fileout)
return ERROR_INT("fileout not defined", procName, 1);
if (xpts <= 0.0) {
L_INFO("setting xpts to 612.0 ppi", procName);
xpts = 612.0;
}
if (ypts <= 0.0) {
L_INFO("setting ypts to 792.0 ppi", procName);
ypts = 792.0;
}
if (xpts < 100.0 || xpts > 2000.0 || ypts < 100.0 || ypts > 2000.0)
L_WARNING("xpts,ypts are typically in the range 500-800", procName);
/* Get all filtered and sorted full pathnames. */
sa = getSortedPathnamesInDirectory(dirin, substr, 0, 0);
/* Generate the PS file. */
sarrayConvertFilesFittedToPS(sa, xpts, ypts, fileout);
sarrayDestroy(&sa);
return 0;
}
/*!
* jbCorrelation()
*
* Input: dirin (directory of input images)
* thresh (typically ~0.8)
* weight (typically ~0.6)
* components (JB_CONN_COMPS, JB_CHARACTERS, JB_WORDS)
* rootname (for output files)
* firstpage (0-based)
* npages (use 0 for all pages in dirin)
* renderflag (1 to render from templates; 0 to skip)
* Return: 0 if OK, 1 on error
*
* Notes:
* (1) The images must be 1 bpp. If they are not, you can convert
* them using convertFilesTo1bpp().
* (2) See prog/jbcorrelation for generating more output (e.g.,
* for debugging)
*/
l_int32
jbCorrelation(const char *dirin,
l_float32 thresh,
l_float32 weight,
l_int32 components,
const char *rootname,
l_int32 firstpage,
l_int32 npages,
l_int32 renderflag) {
char filename[L_BUF_SIZE];
l_int32 nfiles, i, numpages;
JBDATA *data;
JBCLASSER *classer;
PIX *pix;
PIXA *pixa;
SARRAY *safiles;
PROCNAME("jbCorrelation");
if (!dirin)
return ERROR_INT("dirin not defined", procName, 1);
if (!rootname)
return ERROR_INT("rootname not defined", procName, 1);
if (components != JB_CONN_COMPS && components != JB_CHARACTERS &&
components != JB_WORDS)
return ERROR_INT("components invalid", procName, 1);
safiles = getSortedPathnamesInDirectory(dirin, NULL, firstpage, npages);
nfiles = sarrayGetCount(safiles);
/* Classify components */
classer = jbCorrelationInit(components, 0, 0, thresh, weight);
jbAddPages(classer, safiles);
/* Save data */
data = jbDataSave(classer);
jbDataWrite(rootname, data);
/* Optionally, render pages using class templates */
if (renderflag) {
pixa = jbDataRender(data, FALSE);
numpages = pixaGetCount(pixa);
if (numpages != nfiles)
fprintf(stderr, "numpages = %d, nfiles = %d, not equal!\n",
numpages, nfiles);
for (i = 0; i < numpages; i++) {
pix = pixaGetPix(pixa, i, L_CLONE);
snprintf(filename, L_BUF_SIZE, "%s.%05d", rootname, i);
fprintf(stderr, "filename: %s\n", filename);
pixWrite(filename, pix, IFF_PNG);
pixDestroy(&pix);
}
pixaDestroy(&pixa);
}
sarrayDestroy(&safiles);
jbClasserDestroy(&classer);
jbDataDestroy(&data);
return 0;
}
int main(int argc,
char **argv)
{
char *dirin, *dirout, *infile, *outfile, *tail;
l_int32 i, nfiles, border, x, y, w, h, xb, yb, wb, hb;
BOX *box1, *box2;
BOXA *boxa1, *boxa2;
PIX *pixs, *pixt1, *pixd;
SARRAY *safiles;
static char mainName[] = "croptext";
if (argc != 4)
return ERROR_INT("Syntax: croptext dirin border dirout", mainName, 1);
dirin = argv[1];
border = atoi(argv[2]);
dirout = argv[3];
setLeptDebugOK(1);
safiles = getSortedPathnamesInDirectory(dirin, NULL, 0, 0);
nfiles = sarrayGetCount(safiles);
for (i = 0; i < nfiles; i++) {
infile = sarrayGetString(safiles, i, L_NOCOPY);
splitPathAtDirectory(infile, NULL, &tail);
outfile = genPathname(dirout, tail);
pixs = pixRead(infile);
pixt1 = pixMorphSequence(pixs, "r11 + c10.40 + o5.5 + x4", 0);
boxa1 = pixConnComp(pixt1, NULL, 8);
if (boxaGetCount(boxa1) == 0) {
fprintf(stderr, "Warning: no components on page %s\n", tail);
continue;
}
boxa2 = boxaSort(boxa1, L_SORT_BY_AREA, L_SORT_DECREASING, NULL);
box1 = boxaGetBox(boxa2, 0, L_CLONE);
boxGetGeometry(box1, &x, &y, &w, &h);
xb = L_MAX(0, x - border);
yb = L_MAX(0, y - border);
wb = w + 2 * border;
hb = h + 2 * border;
box2 = boxCreate(xb, yb, wb, hb);
pixd = pixClipRectangle(pixs, box2, NULL);
pixWrite(outfile, pixd, IFF_TIFF_G4);
pixDestroy(&pixs);
pixDestroy(&pixt1);
pixDestroy(&pixd);
boxaDestroy(&boxa1);
boxaDestroy(&boxa2);
}
return 0;
}
/*!
* pixaReadFiles()
*
* Input: dirname
* substr (<optional> substring filter on filenames; can be null)
* Return: pixa, or null on error
*
* Notes:
* (1) @dirname is the full path for the directory.
* (2) @substr is the part of the file name (excluding
* the directory) that is to be matched. All matching
* filenames are read into the Pixa. If substr is NULL,
* all filenames are read into the Pixa.
*/
PIXA *
pixaReadFiles(const char *dirname,
const char *substr)
{
PIXA *pixa;
SARRAY *sa;
PROCNAME("pixaReadFiles");
if (!dirname)
return (PIXA *)ERROR_PTR("dirname not defined", procName, NULL);
if ((sa = getSortedPathnamesInDirectory(dirname, substr, 0, 0)) == NULL)
return (PIXA *)ERROR_PTR("sa not made", procName, NULL);
pixa = pixaReadFilesSA(sa);
sarrayDestroy(&sa);
return pixa;
}
PIXA *MakeBootnum2(void)
{
char *fname;
l_int32 i, n, w, h;
BOX *box;
PIX *pix;
PIXA *pixa;
L_RECOG *recog;
SARRAY *sa;
/* Phase 1: generate recog from the digit data */
recog = recogCreate(20, 32, L_USE_ALL, 120, 1);
sa = getSortedPathnamesInDirectory("recog/bootnums", "png", 0, 0);
n = sarrayGetCount(sa);
for (i = 0; i < n; i++) {
/* Read each pix: grayscale, multi-character, labelled */
fname = sarrayGetString(sa, i, L_NOCOPY);
if ((pix = pixRead(fname)) == NULL) {
fprintf(stderr, "Can't read %s\n", fname);
continue;
}
/* Convert to a set of 1 bpp, single character, labelled */
pixGetDimensions(pix, &w, &h, NULL);
box = boxCreate(0, 0, w, h);
recogTrainLabelled(recog, pix, box, NULL, 1, 0);
pixDestroy(&pix);
boxDestroy(&box);
}
recogTrainingFinished(recog, 1);
sarrayDestroy(&sa);
/* Phase 2: generate pixa consisting of 1 bpp, single character pix */
recogWritePixa("/tmp/lept/recog/digits/bootnum2.pa", recog);
pixa = pixaRead("/tmp/lept/recog/digits/bootnum2.pa");
recogDestroy(&recog);
return pixa;
}
/*!
* regTestCompareFiles()
*
* Input: rp (regtest parameters)
* index1 (of one output file from reg test)
* index2 (of another output file from reg test)
* Return: 0 if OK, 1 on error (a failure in comparison is not an error)
*
* Notes:
* (1) This only does something in "compare" mode.
* (2) The canonical format of the golden filenames is:
* /tmp/golden/<root of main name>_golden.<index>.<ext of localname>
* e.g.,
* /tmp/golden/maze_golden.0.png
*/
l_int32
regTestCompareFiles(L_REGPARAMS *rp,
l_int32 index1,
l_int32 index2) {
char *name1, *name2;
char namebuf[256];
l_int32 same;
SARRAY *sa;
PROCNAME("regTestCompareFiles");
if (!rp)
return ERROR_INT("rp not defined", procName, 1);
if (index1 < 0 || index2 < 0) {
rp->success = FALSE;
return ERROR_INT("index1 and/or index2 is negative", procName, 1);
}
if (index1 == index2) {
rp->success = FALSE;
return ERROR_INT("index1 must differ from index2", procName, 1);
}
rp->index++;
if (rp->mode != L_REG_COMPARE) return 0;
/* Generate the golden file names */
snprintf(namebuf, sizeof(namebuf), "%s_golden.%02d.", rp->testname, index1);
sa = getSortedPathnamesInDirectory("/tmp/golden", namebuf, 0, 0);
if (sarrayGetCount(sa) != 1) {
sarrayDestroy(&sa);
rp->success = FALSE;
L_ERROR("golden file %s not found\n", procName, namebuf);
return 1;
}
name1 = sarrayGetString(sa, 0, L_COPY);
sarrayDestroy(&sa);
snprintf(namebuf, sizeof(namebuf), "%s_golden.%02d.", rp->testname, index2);
sa = getSortedPathnamesInDirectory("/tmp/golden", namebuf, 0, 0);
if (sarrayGetCount(sa) != 1) {
sarrayDestroy(&sa);
rp->success = FALSE;
FREE(name1);
L_ERROR("golden file %s not found\n", procName, namebuf);
return 1;
}
name2 = sarrayGetString(sa, 0, L_COPY);
sarrayDestroy(&sa);
/* Test and record on failure */
filesAreIdentical(name1, name2, &same);
if (!same) {
fprintf(rp->fp,
"Failure in %s_reg, index %d: comparing %s with %s\n",
rp->testname, rp->index, name1, name2);
fprintf(stderr,
"Failure in %s_reg, index %d: comparing %s with %s\n",
rp->testname, rp->index, name1, name2);
rp->success = FALSE;
}
FREE(name1);
FREE(name2);
return 0;
}
int main(int argc,
char **argv)
{
char filename[BUF_SIZE];
char *dirin, *rootname, *fname;
l_int32 i, j, w, h, firstpage, npages, nfiles, ncomp;
l_int32 index, ival, rval, gval, bval;
BOX *box;
BOXA *boxa;
BOXAA *baa;
JBDATA *data;
JBCLASSER *classer;
NUMA *nai;
NUMAA *naa;
SARRAY *safiles;
PIX *pixs, *pixt1, *pixt2, *pixd;
PIXCMAP *cmap;
static char mainName[] = "wordsinorder";
if (argc != 3 && argc != 5)
return ERROR_INT(
" Syntax: wordsinorder dirin rootname [firstpage, npages]",
mainName, 1);
dirin = argv[1];
rootname = argv[2];
if (argc == 3) {
firstpage = 0;
npages = 0;
}
else {
firstpage = atoi(argv[3]);
npages = atoi(argv[4]);
}
/* Compute the word bounding boxes at 2x reduction, along with
* the textlines that they are in. */
safiles = getSortedPathnamesInDirectory(dirin, NULL, firstpage, npages);
nfiles = sarrayGetCount(safiles);
baa = boxaaCreate(nfiles);
naa = numaaCreate(nfiles);
for (i = 0; i < nfiles; i++) {
fname = sarrayGetString(safiles, i, 0);
if ((pixs = pixRead(fname)) == NULL) {
L_WARNING("image file %d not read\n", mainName, i);
continue;
}
pixGetWordBoxesInTextlines(pixs, 2, MIN_WORD_WIDTH, MIN_WORD_HEIGHT,
MAX_WORD_WIDTH, MAX_WORD_HEIGHT,
&boxa, &nai);
boxaaAddBoxa(baa, boxa, L_INSERT);
numaaAddNuma(naa, nai, L_INSERT);
#if RENDER_PAGES
/* Show the results on a 2x reduced image, where each
* word is outlined and the color of the box depends on the
* computed textline. */
pixt1 = pixReduceRankBinary2(pixs, 2, NULL);
pixGetDimensions(pixt1, &w, &h, NULL);
pixd = pixCreate(w, h, 8);
cmap = pixcmapCreateRandom(8, 1, 1); /* first color is black */
pixSetColormap(pixd, cmap);
pixt2 = pixUnpackBinary(pixt1, 8, 1);
pixRasterop(pixd, 0, 0, w, h, PIX_SRC | PIX_DST, pixt2, 0, 0);
ncomp = boxaGetCount(boxa);
for (j = 0; j < ncomp; j++) {
box = boxaGetBox(boxa, j, L_CLONE);
numaGetIValue(nai, j, &ival);
index = 1 + (ival % 254); /* omit black and white */
pixcmapGetColor(cmap, index, &rval, &gval, &bval);
pixRenderBoxArb(pixd, box, 2, rval, gval, bval);
boxDestroy(&box);
}
snprintf(filename, BUF_SIZE, "%s.%05d", rootname, i);
fprintf(stderr, "filename: %s\n", filename);
pixWrite(filename, pixd, IFF_PNG);
pixDestroy(&pixt1);
pixDestroy(&pixt2);
pixDestroy(&pixs);
pixDestroy(&pixd);
#endif /* RENDER_PAGES */
}
boxaaDestroy(&baa);
numaaDestroy(&naa);
sarrayDestroy(&safiles);
return 0;
}
/*!
* convertFilesTo1bpp()
*
* Input: dirin
* substr (<optional> substring filter on filenames; can be NULL)
* upscaling (1, 2 or 4; only for input color or grayscale)
* thresh (global threshold for binarization; use 0 for default)
* firstpage
* npages (use 0 to do all from @firstpage to the end)
* dirout
* outformat (IFF_PNG, IFF_TIFF_G4)
* Return: 0 if OK, 1 on error
*
* Notes:
* (1) Images are sorted lexicographically, and the names in the
* output directory are retained except for the extension.
*/
l_int32
convertFilesTo1bpp(const char *dirin,
const char *substr,
l_int32 upscaling,
l_int32 thresh,
l_int32 firstpage,
l_int32 npages,
const char *dirout,
l_int32 outformat) {
l_int32 i, nfiles;
char buf[512];
char *fname, *tail, *basename;
PIX *pixs, *pixg1, *pixg2, *pixb;
SARRAY *safiles;
PROCNAME("convertFilesTo1bpp");
if (!dirin)
return ERROR_INT("dirin", procName, 1);
if (!dirout)
return ERROR_INT("dirout", procName, 1);
if (upscaling != 1 && upscaling != 2 && upscaling != 4)
return ERROR_INT("invalid upscaling factor", procName, 1);
if (thresh <= 0) thresh = 180;
if (firstpage < 0) firstpage = 0;
if (npages < 0) npages = 0;
if (outformat != IFF_TIFF_G4)
outformat = IFF_PNG;
safiles = getSortedPathnamesInDirectory(dirin, substr, firstpage, npages);
if (!safiles)
return ERROR_INT("safiles not made", procName, 1);
if ((nfiles = sarrayGetCount(safiles)) == 0) {
sarrayDestroy(&safiles);
return ERROR_INT("no matching files in the directory", procName, 1);
}
for (i = 0; i < nfiles; i++) {
fname = sarrayGetString(safiles, i, L_NOCOPY);
if ((pixs = pixRead(fname)) == NULL) {
L_WARNING("Couldn't read file %s\n", procName, fname);
continue;
}
if (pixGetDepth(pixs) == 32)
pixg1 = pixConvertRGBToLuminance(pixs);
else
pixg1 = pixClone(pixs);
pixg2 = pixRemoveColormap(pixg1, REMOVE_CMAP_TO_GRAYSCALE);
if (pixGetDepth(pixg2) == 1) {
pixb = pixClone(pixg2);
} else {
if (upscaling == 1)
pixb = pixThresholdToBinary(pixg2, thresh);
else if (upscaling == 2)
pixb = pixScaleGray2xLIThresh(pixg2, thresh);
else /* upscaling == 4 */
pixb = pixScaleGray4xLIThresh(pixg2, thresh);
}
pixDestroy(&pixs);
pixDestroy(&pixg1);
pixDestroy(&pixg2);
splitPathAtDirectory(fname, NULL, &tail);
splitPathAtExtension(tail, &basename, NULL);
if (outformat == IFF_TIFF_G4) {
snprintf(buf, sizeof(buf), "%s/%s.tif", dirout, basename);
pixWrite(buf, pixb, IFF_TIFF_G4);
} else {
snprintf(buf, sizeof(buf), "%s/%s.png", dirout, basename);
pixWrite(buf, pixb, IFF_PNG);
}
pixDestroy(&pixb);
FREE(tail);
FREE(basename);
}
sarrayDestroy(&safiles);
return 0;
}
int main(int argc,
char **argv)
{
char *fname, *filename;
const char *str;
char buffer[512];
l_int32 i, npages;
size_t length;
FILE *fp;
NUMA *naflags, *nasizes;
PIX *pix, *pix1, *pix2, *pixd;
PIXA *pixa;
PIXCMAP *cmap;
SARRAY *savals, *satypes, *sa;
static char mainName[] = "mtifftest";
if (argc != 1)
return ERROR_INT(" Syntax: mtifftest", mainName, 1);
lept_mkdir("tiff");
#if 1 /* ------------------ Test multipage I/O -------------------*/
/* This puts every image file in the directory with a string
* match to "weasel" into a multipage tiff file.
* Images with 1 bpp are coded as g4; the others as zip.
* It then reads back into a pix and displays. */
writeMultipageTiff(".", "weasel8.", "/tmp/tiff/weasel8.tif");
pixa = pixaReadMultipageTiff("/tmp/tiff/weasel8.tif");
pixd = pixaDisplayTiledInRows(pixa, 1, 1200, 0.5, 0, 15, 4);
pixDisplay(pixd, 100, 0);
pixDestroy(&pixd);
pixd = pixaDisplayTiledInRows(pixa, 8, 1200, 0.8, 0, 15, 4);
pixDisplay(pixd, 100, 200);
pixDestroy(&pixd);
pixd = pixaDisplayTiledInRows(pixa, 32, 1200, 1.2, 0, 15, 4);
pixDisplay(pixd, 100, 400);
pixDestroy(&pixd);
pixaDestroy(&pixa);
#endif
#if 1 /* ------------ Test single-to-multipage I/O -------------------*/
/* Read the files and generate a multipage tiff file of G4 images.
* Then convert that to a G4 compressed and ascii85 encoded PS file. */
sa = getSortedPathnamesInDirectory(".", "weasel4.", 0, 4);
sarrayWriteStream(stderr, sa);
sarraySort(sa, sa, L_SORT_INCREASING);
sarrayWriteStream(stderr, sa);
npages = sarrayGetCount(sa);
for (i = 0; i < npages; i++) {
fname = sarrayGetString(sa, i, 0);
filename = genPathname(".", fname);
pix1 = pixRead(filename);
if (!pix1) continue;
pix2 = pixConvertTo1(pix1, 128);
if (i == 0)
pixWriteTiff("/tmp/tiff/weasel4", pix2, IFF_TIFF_G4, "w+");
else
pixWriteTiff("/tmp/tiff/weasel4", pix2, IFF_TIFF_G4, "a");
pixDestroy(&pix1);
pixDestroy(&pix2);
lept_free(filename);
}
/* Write it out as a PS file */
convertTiffMultipageToPS("/tmp/tiff/weasel4", "/tmp/tiff/weasel4.ps",
NULL, 0.95);
sarrayDestroy(&sa);
#endif
#if 1 /* ------------------ Test multipage I/O -------------------*/
/* Read count of pages in tiff multipage file */
writeMultipageTiff(".", "weasel2", weasel_orig);
fp = lept_fopen(weasel_orig, "rb");
if (fileFormatIsTiff(fp)) {
tiffGetCount(fp, &npages);
fprintf(stderr, " Tiff: %d page\n", npages);
}
else
return ERROR_INT(" file not tiff", mainName, 1);
lept_fclose(fp);
/* Split into separate page files */
for (i = 0; i < npages + 1; i++) { /* read one beyond to catch error */
if (i == npages)
L_INFO("Errors in next 2 lines are intentional!\n", mainName);
pix = pixReadTiff(weasel_orig, i);
if (!pix) continue;
sprintf(buffer, "/tmp/tiff/%03d.tif", i);
pixWrite(buffer, pix, IFF_TIFF_ZIP);
pixDestroy(&pix);
}
/* Read separate page files and write reversed file */
for (i = npages - 1; i >= 0; i--) {
sprintf(buffer, "/tmp/tiff/%03d.tif", i);
pix = pixRead(buffer);
if (!pix) continue;
if (i == npages - 1)
pixWriteTiff(weasel_rev, pix, IFF_TIFF_ZIP, "w+");
else
pixWriteTiff(weasel_rev, pix, IFF_TIFF_ZIP, "a");
pixDestroy(&pix);
}
/* Read reversed file and reverse again */
pixa = pixaCreate(npages);
for (i = 0; i < npages; i++) {
pix = pixReadTiff(weasel_rev, i);
pixaAddPix(pixa, pix, L_INSERT);
}
for (i = npages - 1; i >= 0; i--) {
pix = pixaGetPix(pixa, i, L_CLONE);
if (i == npages - 1)
pixWriteTiff(weasel_rev_rev, pix, IFF_TIFF_ZIP, "w+");
else
pixWriteTiff(weasel_rev_rev, pix, IFF_TIFF_ZIP, "a");
pixDestroy(&pix);
}
pixaDestroy(&pixa);
#endif
#if 0 /* ----- test adding custom public tags to a tiff header ----- */
pix = pixRead("feyn.tif");
naflags = numaCreate(10);
savals = sarrayCreate(10);
satypes = sarrayCreate(10);
nasizes = numaCreate(10);
/* numaAddNumber(naflags, TIFFTAG_XMLPACKET); */ /* XMP: 700 */
numaAddNumber(naflags, 700);
str = "<xmp>This is a Fake XMP packet</xmp>\n<text>Guess what ...?</text>";
length = strlen(str);
sarrayAddString(savals, (char *)str, L_COPY);
sarrayAddString(satypes, (char *)"char*", L_COPY);
numaAddNumber(nasizes, length); /* get it all */
numaAddNumber(naflags, 269); /* DOCUMENTNAME */
sarrayAddString(savals, (char *)"One silly title", L_COPY);
sarrayAddString(satypes, (char *)"const char*", L_COPY);
numaAddNumber(naflags, 270); /* IMAGEDESCRIPTION */
sarrayAddString(savals, (char *)"One page of text", L_COPY);
sarrayAddString(satypes, (char *)"const char*", L_COPY);
/* the max sample is used by rendering programs
* to scale the dynamic range */
numaAddNumber(naflags, 281); /* MAXSAMPLEVALUE */
sarrayAddString(savals, (char *)"4", L_COPY);
sarrayAddString(satypes, (char *)"l_uint16", L_COPY);
/* note that date is required to be a 20 byte string */
numaAddNumber(naflags, 306); /* DATETIME */
sarrayAddString(savals, (char *)"2004:10:11 09:35:15", L_COPY);
sarrayAddString(satypes, (char *)"const char*", L_COPY);
/* note that page number requires 2 l_uint16 input */
numaAddNumber(naflags, 297); /* PAGENUMBER */
sarrayAddString(savals, (char *)"1-412", L_COPY);
sarrayAddString(satypes, (char *)"l_uint16-l_uint16", L_COPY);
pixWriteTiffCustom("/tmp/tiff/tags.tif", pix, IFF_TIFF_G4, "w", naflags,
savals, satypes, nasizes);
fprintTiffInfo(stderr, (char *)"/tmp/tiff/tags.tif");
fprintf(stderr, "num flags = %d\n", numaGetCount(naflags));
fprintf(stderr, "num sizes = %d\n", numaGetCount(nasizes));
fprintf(stderr, "num vals = %d\n", sarrayGetCount(savals));
fprintf(stderr, "num types = %d\n", sarrayGetCount(satypes));
numaDestroy(&naflags);
numaDestroy(&nasizes);
sarrayDestroy(&savals);
sarrayDestroy(&satypes);
pixDestroy(&pix);
#endif
return 0;
}
/*!
* \brief jbWordsInTextlines()
*
* \param[in] dirin directory of input pages
* \param[in] reduction 1 for full res; 2 for half-res
* \param[in] maxwidth of word mask components, to be kept
* \param[in] maxheight of word mask components, to be kept
* \param[in] thresh on correlation; 0.80 is reasonable
* \param[in] weight for handling thick text; 0.6 is reasonable
* \param[out] pnatl numa with textline index for each component
* \param[in] firstpage 0-based
* \param[in] npages use 0 for all pages in dirin
* \return classer for the set of pages
*
* <pre>
* Notes:
* (1) This is a high-level function. See prog/jbwords for example
* of usage.
* (2) Typically, words can be found reasonably well at a resolution
* of about 150 ppi. For highest accuracy, you should use 300 ppi.
* Assuming that the input images are 300 ppi, use reduction = 1
* for finding words at full res, and reduction = 2 for finding
* them at 150 ppi.
* </pre>
*/
JBCLASSER *
jbWordsInTextlines(const char *dirin,
l_int32 reduction,
l_int32 maxwidth,
l_int32 maxheight,
l_float32 thresh,
l_float32 weight,
NUMA **pnatl,
l_int32 firstpage,
l_int32 npages)
{
char *fname;
l_int32 nfiles, i, w, h;
BOXA *boxa;
JBCLASSER *classer;
NUMA *nai, *natl;
PIX *pix;
PIXA *pixa;
SARRAY *safiles;
PROCNAME("jbWordsInTextlines");
if (!pnatl)
return (JBCLASSER *)ERROR_PTR("&natl not defined", procName, NULL);
*pnatl = NULL;
if (!dirin)
return (JBCLASSER *)ERROR_PTR("dirin not defined", procName, NULL);
if (reduction != 1 && reduction != 2)
return (JBCLASSER *)ERROR_PTR("reduction not in {1,2}", procName, NULL);
safiles = getSortedPathnamesInDirectory(dirin, NULL, firstpage, npages);
nfiles = sarrayGetCount(safiles);
/* Classify components */
classer = jbCorrelationInit(JB_WORDS, maxwidth, maxheight, thresh, weight);
classer->safiles = sarrayCopy(safiles);
natl = numaCreate(0);
*pnatl = natl;
for (i = 0; i < nfiles; i++) {
fname = sarrayGetString(safiles, i, L_NOCOPY);
if ((pix = pixRead(fname)) == NULL) {
L_WARNING("image file %d not read\n", procName, i);
continue;
}
pixGetDimensions(pix, &w, &h, NULL);
if (reduction == 1) {
classer->w = w;
classer->h = h;
} else { /* reduction == 2 */
classer->w = w / 2;
classer->h = h / 2;
}
pixGetWordsInTextlines(pix, reduction, JB_WORDS_MIN_WIDTH,
JB_WORDS_MIN_HEIGHT, maxwidth, maxheight,
&boxa, &pixa, &nai);
jbAddPageComponents(classer, pix, boxa, pixa);
numaJoin(natl, nai, 0, -1);
pixDestroy(&pix);
numaDestroy(&nai);
boxaDestroy(&boxa);
pixaDestroy(&pixa);
}
sarrayDestroy(&safiles);
return classer;
}
l_int32 main(int argc,
char **argv)
{
char buf[256], dirname[256];
char *dirin, *pattern, *subdirout, *fname, *tail, *basename;
l_int32 thresh, i, n;
l_float32 scalefactor;
PIX *pix1, *pix2, *pix3, *pix4;
SARRAY *sa;
static char mainName[] = "binarizefiles.c";
if (argc != 6) {
fprintf(stderr,
"Syntax: binarizefiles dirin pattern thresh scalefact dirout\n"
" dirin: input directory for image files\n"
" pattern: use 'allfiles' to convert all files\n"
" in the directory\n"
" thresh: 0 for adaptive; > 0 for global thresh (e.g., 128)\n"
" scalefactor: in (0.0 ... 4.0]; use 1.0 to prevent scaling\n"
" subdirout: subdirectory of /tmp for output files\n");
return 1;
}
dirin = argv[1];
pattern = argv[2];
thresh = atoi(argv[3]);
scalefactor = atof(argv[4]);
subdirout = argv[5];
if (!strcmp(pattern, "allfiles"))
pattern = NULL;
if (scalefactor <= 0.0 || scalefactor > 4.0) {
L_WARNING("invalid scalefactor: setting to 1.0\n", mainName);
scalefactor = 1.0;
}
/* Get the input filenames */
sa = getSortedPathnamesInDirectory(dirin, pattern, 0, 0);
sarrayWriteStream(stderr, sa);
n = sarrayGetCount(sa);
/* Write the output files */
makeTempDirname(dirname, 256, subdirout);
fprintf(stderr, "dirname: %s\n", dirname);
lept_mkdir(subdirout);
for (i = 0; i < n; i++) {
fname = sarrayGetString(sa, i, L_NOCOPY);
if ((pix1 = pixRead(fname)) == NULL) {
L_ERROR("file %s not read as image", mainName, fname);
continue;
}
splitPathAtDirectory(fname, NULL, &tail);
splitPathAtExtension(tail, &basename, NULL);
snprintf(buf, sizeof(buf), "%s/%s.tif", dirname, basename);
FREE(tail);
FREE(basename);
fprintf(stderr, "fileout: %s\n", buf);
if (scalefactor != 1.0)
pix2 = pixScale(pix1, scalefactor, scalefactor);
else
pix2 = pixClone(pix1);
if (thresh == 0) {
pix4 = pixConvertTo8(pix2, 0);
pix3 = pixAdaptThresholdToBinary(pix4, NULL, 1.0);
pixDestroy(&pix4);
} else {
pix3 = pixConvertTo1(pix2, thresh);
}
pixWrite(buf, pix3, IFF_TIFF_G4);
pixDestroy(&pix1);
pixDestroy(&pix2);
pixDestroy(&pix3);
}
sarrayDestroy(&sa);
return 0;
}
l_int32 main(int argc,
char **argv)
{
char buf[256], rootname[256];
char *dir, *pattern, *psdir, *imagedir;
char *fname, *tail, *filename;
l_int32 i, n, ret;
SARRAY *sa, *saps;
static char mainName[] = "concatpdf";
if (argc != 2 && argc != 3)
return ERROR_INT("Syntax: concatpdf dir [pattern]", mainName, 1);
dir = argv[1];
pattern = (argc == 3) ? argv[2] : NULL;
/* Get the names of the pdf files */
sa = getSortedPathnamesInDirectory(dir, pattern, 0, 0);
sarrayWriteStream(stderr, sa);
n = sarrayGetCount(sa);
#if 1
/* Convert to ps */
psdir = genPathname("/tmp/ps", NULL);
lept_rmdir("ps");
lept_mkdir("ps");
saps = sarrayCreate(n);
for (i = 0; i < n; i++) {
fname = sarrayGetString(sa, i, L_NOCOPY);
splitPathAtDirectory(fname, NULL, &tail);
splitPathAtExtension(tail, &filename, NULL);
snprintf(buf, sizeof(buf), "acroread -toPostScript -annotsOff %s %s",
fname, psdir);
fprintf(stderr, "%s\n", buf);
ret = system(buf); /* acroread -toPostScript -annotsOff */
snprintf(buf, sizeof(buf), "%s/%s.ps", psdir, filename);
sarrayAddString(saps, buf, L_COPY);
lept_free(tail);
lept_free(filename);
}
sarrayDestroy(&sa);
#endif
#if 1
/* Rasterize */
imagedir = genPathname("/tmp/image", NULL);
lept_rmdir("image");
lept_mkdir("image");
sarrayWriteStream(stderr, saps);
n = sarrayGetCount(saps);
for (i = 0; i < n; i++) {
fname = sarrayGetString(saps, i, L_NOCOPY);
snprintf(rootname, sizeof(rootname), "%s/r%d", imagedir, i);
snprintf(buf, sizeof(buf), "ps2png-gray %s %s", fname, rootname);
fprintf(stderr, "%s\n", buf);
ret = system(buf); /* ps2png-gray */
}
#endif
#if 1
/* Generate the pdf */
convertFilesToPdf(imagedir, "png", RESOLUTION, 1.0, L_FLATE_ENCODE,
0, "", "/tmp/output.pdf");
#endif
return 0;
}
int main(int argc,
char **argv)
{
char filename[BUF_SIZE];
char *dirin, *rootname, *fname;
l_int32 i, firstpage, npages, nfiles;
l_float32 thresh, weight;
JBDATA *data;
JBCLASSER *classer;
SARRAY *safiles;
PIX *pix, *pixt;
PIXA *pixa, *pixadb;
static char mainName[] = "jbcorrelation";
if (argc != 5 && argc != 7)
return ERROR_INT(" Syntax: jbcorrelation dirin thresh weight "
"rootname [firstpage, npages]", mainName, 1);
dirin = argv[1];
thresh = atof(argv[2]);
weight = atof(argv[3]);
rootname = argv[4];
if (argc == 5) {
firstpage = 0;
npages = 0;
}
else {
firstpage = atoi(argv[5]);
npages = atoi(argv[6]);
}
#if 0
/*--------------------------------------------------------------*/
jbCorrelation(dirin, thresh, weight, COMPONENTS, rootname,
firstpage, npages, 1);
/*--------------------------------------------------------------*/
#else
/*--------------------------------------------------------------*/
safiles = getSortedPathnamesInDirectory(dirin, NULL, firstpage, npages);
nfiles = sarrayGetCount(safiles);
sarrayWriteStream(stderr, safiles);
/* Classify components on requested pages */
startTimer();
classer = jbCorrelationInit(COMPONENTS, 0, 0, thresh, weight);
jbAddPages(classer, safiles);
fprintf(stderr, "Time to generate classes: %6.3f sec\n", stopTimer());
/* Save and write out the result */
data = jbDataSave(classer);
jbDataWrite(rootname, data);
fprintf(stderr, "Number of classes: %d\n", classer->nclass);
/* Render the pages from the classifier data.
* Use debugflag == FALSE to omit outlines of each component. */
pixa = jbDataRender(data, FALSE);
/* Write the pages out */
npages = pixaGetCount(pixa);
if (npages != nfiles)
fprintf(stderr, "npages = %d, nfiles = %d, not equal!\n",
npages, nfiles);
for (i = 0; i < npages; i++) {
pix = pixaGetPix(pixa, i, L_CLONE);
snprintf(filename, BUF_SIZE, "%s.%05d", rootname, i);
fprintf(stderr, "filename: %s\n", filename);
pixWrite(filename, pix, IFF_PNG);
pixDestroy(&pix);
}
#if DISPLAY_DIFFERENCE
fname = sarrayGetString(safiles, 0, 0);
pixt = pixRead(fname);
pix = pixaGetPix(pixa, 0, L_CLONE);
pixXor(pixt, pixt, pix);
pixWrite("junk_output_diff", pixt, IFF_PNG);
pixDestroy(&pix);
pixDestroy(&pixt);
#endif /* DISPLAY_DIFFERENCE */
#if DEBUG_TEST_DATA_IO
{ JBDATA *newdata;
PIX *newpix;
PIXA *newpixa;
l_int32 same, iofail;
/* Read the data back in and render the pages */
newdata = jbDataRead(rootname);
newpixa = jbDataRender(newdata, FALSE);
iofail = FALSE;
for (i = 0; i < npages; i++) {
pix = pixaGetPix(pixa, i, L_CLONE);
newpix = pixaGetPix(newpixa, i, L_CLONE);
pixEqual(pix, newpix, &same);
if (!same) {
iofail = TRUE;
fprintf(stderr, "pix on page %d are unequal!\n", i);
}
pixDestroy(&pix);
pixDestroy(&newpix);
}
if (iofail)
fprintf(stderr, "read/write for jbdata fails\n");
else
fprintf(stderr, "read/write for jbdata succeeds\n");
jbDataDestroy(&newdata);
pixaDestroy(&newpixa);
}
#endif /* DEBUG_TEST_DATA_IO */
#if RENDER_DEBUG
/* Use debugflag == TRUE to see outlines of each component. */
pixadb = jbDataRender(data, TRUE);
/* Write the debug pages out */
npages = pixaGetCount(pixadb);
for (i = 0; i < npages; i++) {
pix = pixaGetPix(pixadb, i, L_CLONE);
snprintf(filename, BUF_SIZE, "%s.db.%05d", rootname, i);
fprintf(stderr, "filename: %s\n", filename);
pixWrite(filename, pix, IFF_PNG);
pixDestroy(&pix);
}
pixaDestroy(&pixadb);
#endif /* RENDER_DEBUG */
#if DISPLAY_ALL_INSTANCES
/* display all instances, organized by template */
pix = pixaaDisplayByPixa(classer->pixaa,
X_SPACING, Y_SPACING, MAX_OUTPUT_WIDTH);
pixWrite("output_instances", pix, IFF_PNG);
pixDestroy(&pix);
#endif /* DISPLAY_ALL_INSTANCES */
pixaDestroy(&pixa);
sarrayDestroy(&safiles);
jbClasserDestroy(&classer);
jbDataDestroy(&data);
/*--------------------------------------------------------------*/
#endif
return 0;
}
int main(int argc,
char **argv) {
const char *name;
l_int32 i, n;
BOX *box;
PIX *pix0, *pix1, *pixd;
PIXA *pixa;
SARRAY *sa1, *sa2, *sa3, *sa4;
L_REGPARAMS *rp;
if (regTestSetup(argc, argv, &rp))
return 1;
/* ---------------- Find all the jpg and tif images --------------- */
sa1 = getSortedPathnamesInDirectory(".", ".jpg", 0, 0);
sa2 = getSortedPathnamesInDirectory(".", ".tif", 0, 0);
sa3 = sarraySelectByRange(sa1, 0, 9);
sa4 = sarraySelectByRange(sa2, 0, 9);
sarrayConcatenate(sa3, sa4);
n = sarrayGetCount(sa3);
sarrayDestroy(&sa1);
sarrayDestroy(&sa2);
sarrayDestroy(&sa4);
/* ---------------- Use replace to fill up a pixa -------------------*/
pixa = pixaCreate(1);
pixaExtendArrayToSize(pixa, n);
if ((pix0 = pixRead("marge.jpg")) == NULL)
rp->success = FALSE;
pix1 = pixScaleToSize(pix0, 144, 108); /* scale 0.25 */
pixDestroy(&pix0);
pixaInitFull(pixa, pix1, NULL); /* fill it up */
pixd = pixaDisplayTiledInRows(pixa, 32, 1000, 1.0, 0, 25, 2);
pixDisplayWithTitle(pixd, 100, 100, NULL, rp->display);
pixWrite("/tmp/regout/pix1.jpg", pixd, IFF_JFIF_JPEG);
pixDestroy(&pix1);
pixDestroy(&pixd);
/* ---------------- And again with jpgs and tifs -------------------*/
for (i = 0; i < n; i++) {
name = sarrayGetString(sa3, i, L_NOCOPY);
if ((pix0 = pixRead(name)) == NULL)
rp->success = FALSE;
pix1 = pixScaleToSize(pix0, 144, 108);
pixaReplacePix(pixa, i, pix1, NULL);
pixDestroy(&pix0);
}
pixd = pixaDisplayTiledInRows(pixa, 32, 1000, 1.0, 0, 25, 2);
pixDisplayWithTitle(pixd, 400, 100, NULL, rp->display);
pixWrite("/tmp/regout/pix2.jpg", pixd, IFF_JFIF_JPEG);
pixDestroy(&pixd);
/* ---------------- And again, reversing the order ------------------*/
box = boxCreate(0, 0, 0, 0);
pixaInitFull(pixa, NULL, box);
boxDestroy(&box);
for (i = 0; i < n; i++) {
name = sarrayGetString(sa3, i, L_NOCOPY);
if ((pix0 = pixRead(name)) == NULL)
rp->success = FALSE;
pix1 = pixScaleToSize(pix0, 144, 108);
pixaReplacePix(pixa, n - 1 - i, pix1, NULL);
pixDestroy(&pix0);
}
pixd = pixaDisplayTiledInRows(pixa, 32, 1000, 1.0, 0, 25, 2);
pixDisplayWithTitle(pixd, 700, 100, NULL, rp->display);
pixWrite("/tmp/regout/pix3.jpg", pixd, IFF_JFIF_JPEG);
pixDestroy(&pixd);
sarrayDestroy(&sa3);
pixaDestroy(&pixa);
return regTestCleanup(rp);
}
/*!
* regTestCompareFiles()
*
* Input: stream (for output; use NULL to generate golden files)
* argv ([0] == name of reg test)
* index1 (of one output file from reg test)
* index2 (of another output file from reg test)
* &success (<return> 0 on if different; input value on success)
* Return: 0 if OK, 1 on error (a failure in comparison is not an error)
*
* Notes:
* (1) If @fp != NULL, this function compares two golden files to
* determine if they are the same. If @fp == NULL, this is a
* "generate" operation; don't do the comparison.
* (2) This function can be called repeatedly in a single reg test.
* (3) The value for @success is initialized to TRUE in the reg test
* setup before this function is called for the first time.
* A failure in any file comparison is registered as a failure
* of the regression test.
* (4) The canonical format of the golden filenames is:
* /tmp/<root of main name>_golden.<index>.<ext of localname>
* e.g.,
* /tmp/maze_golden.0.png
*/
l_int32
regTestCompareFiles(FILE *fp,
char **argv,
l_int32 index1,
l_int32 index2,
l_int32 *psuccess)
{
char *root, *name1, *name2;
char namebuf[64];
l_int32 error,same;
SARRAY *sa;
PROCNAME("regTestCompareFiles");
if (!psuccess)
return ERROR_INT("&success not defined", procName, 1);
if (index1 < 0 || index2 < 0)
return ERROR_INT("index1 and/or index2 is negative", procName, 1);
if (index1 == index2)
return ERROR_INT("index1 must differ from index2", procName, 1);
if (!fp) /* no-op */
return 0;
/* Generate partial golden file names and find the actual
* paths to them. */
error = FALSE;
name1 = name2 = NULL;
if ((root = getRootNameFromArgv0(argv[0])) == NULL)
return ERROR_INT("invalid root", procName, 1);
snprintf(namebuf, sizeof(namebuf), "%s_golden.%d.", root, index1);
sa = getSortedPathnamesInDirectory("/tmp", namebuf, 0, 0);
if (sarrayGetCount(sa) != 1)
error = TRUE;
else
name1 = sarrayGetString(sa, 0, L_COPY);
sarrayDestroy(&sa);
snprintf(namebuf, sizeof(namebuf), "%s_golden.%d.", root, index2);
sa = getSortedPathnamesInDirectory("/tmp", namebuf, 0, 0);
if (sarrayGetCount(sa) != 1)
error = TRUE;
else
name2 = sarrayGetString(sa, 0, L_COPY);
sarrayDestroy(&sa);
FREE(root);
if (error == TRUE) {
if (name1) FREE(name1);
if (name2) FREE(name2);
L_ERROR("golden files not found", procName);
return 1;
}
/* Test and record on failure */
filesAreIdentical(name1, name2, &same);
if (!same) {
fprintf(fp, "Failure in %s: comparing %s with %s\n", argv[0],
name1, name2);
fprintf(stderr, "Failure in %s: comparing %s with %s\n", argv[0],
name1, name2);
*psuccess = 0;
}
FREE(name1);
FREE(name2);
return 0;
}
