Add parameter for maximum horizontal stretching for croppdf

* This allows better filling for a 8.5 x 11 inch printed page.

prog/croppdf.c

@@ -38,7 +38,8 @@
  *    is encoded with tiffg4.
  *
  *    Syntax:
- *       croppdf basedir lrclear tbclear edgeclean lradd tbadd title fileout
+ *       croppdf basedir lrclear tbclear edgeclean lradd tbadd maxwiden
+ *       title fileout
  *
  *    The %basedir is a directory where the input pdf files are located.
  *    The program will operate on every file in this directory with
@@ -50,6 +51,15 @@
  *    The %edgeclean parameter is used to remove edge noise, going from
  *    0 (default, no removal) to 15 (maximally aggressive removal).
  *
+ *    The suggested value for %lradd and %tbadd is 50.  Laser printers do not
+ *    print foreground pixels very close to the page edges, and using a
+ *    margin of 50 pixels (1/6" at 300 ppi) should allow all foregrounnd
+ *    pixels to be printed.
+ *
+ *    The %maxwiden parameter allows the foreground to better fill an
+ *    8.5 x 11 inch printed page.  It gives the maximum fractional horizontal
+ *    stretching allowed.  Suggested values are between 1.0 and 1.15.
+ *
  *    The %title is the title given to the pdf.  Use %title == "none"
  *    to omit the title.
  *
@@ -91,20 +101,22 @@ char       buf[256];
 char      *basedir, *fname, *tail, *basename, *imagedir, *title, *fileout;
 l_int32    lrclear, tbclear, edgeclean, lradd, tbadd;
 l_int32    render_res, i, n, ret;
+l_float32  maxwiden;
 SARRAY    *sa;
 
-    if (argc != 9)
+    if (argc != 10)
         return ERROR_INT(
             "Syntax: croppdf basedir lrclear tbclear edgeclean "
-            "lradd tbadd title fileout", __func__, 1);
+            "lradd tbadd maxwiden title fileout", __func__, 1);
     basedir = argv[1];
     lrclear = atoi(argv[2]);
     tbclear = atoi(argv[3]);
     edgeclean = atoi(argv[4]);
     lradd = atoi(argv[5]);
     tbadd = atoi(argv[6]);
-    title = argv[7];
-    fileout = argv[8];
+    maxwiden = atof(argv[7]);
+    title = argv[8];
+    fileout = argv[9];
     setLeptDebugOK(1);
 
         /* Set up a directory for temp images */
@@ -156,7 +168,7 @@ SARRAY    *sa;
     lept_free(imagedir);
     sarrayWriteStderr(sa);
     lept_stderr("cropping ...\n");
-    cropFilesToPdf(sa, lrclear, tbclear, edgeclean, lradd, tbadd,
+    cropFilesToPdf(sa, lrclear, tbclear, edgeclean, lradd, tbadd, maxwiden,
                    title, fileout);
 
     return 0;

prog/misctest1.c

@@ -416,7 +416,7 @@ PIXCMAP  *cmap, *cmapg;
 
         /* Page cropping */
     pix1 = pixRead("tel_3.tif");
-    pix2 = pixCropImage(pix1, 30, 30, 4, 25, 25,
+    pix2 = pixCropImage(pix1, 30, 30, 4, 25, 25, 1.15,
                         "/tmp/lept/misc/cropdebug.pdf", NULL);
     pixDestroy(&pix1);
     pixDestroy(&pix2);

src/allheaders.h

@@ -1375,7 +1375,7 @@ LEPT_DLL extern PIX * pixGenHalftoneMask ( PIX *pixs, PIX **ppixtext, l_int32 *p
 LEPT_DLL extern PIX * pixGenerateHalftoneMask ( PIX *pixs, PIX **ppixtext, l_int32 *phtfound, PIXA *pixadb );
 LEPT_DLL extern PIX * pixGenTextlineMask ( PIX *pixs, PIX **ppixvws, l_int32 *ptlfound, PIXA *pixadb );
 LEPT_DLL extern PIX * pixGenTextblockMask ( PIX *pixs, PIX *pixvws, PIXA *pixadb );
-LEPT_DLL extern PIX * pixCropImage ( PIX *pixs, l_int32 lr_clear, l_int32 tb_clear, l_int32 edgeclean, l_int32 lr_add, l_int32 tb_add, const char *debugfile, BOX **pcropbox );
+LEPT_DLL extern PIX * pixCropImage ( PIX *pixs, l_int32 lr_clear, l_int32 tb_clear, l_int32 edgeclean, l_int32 lr_add, l_int32 tb_add, l_float32 maxwiden, const char *debugfile, BOX **pcropbox );
 LEPT_DLL extern PIX * pixCleanImage ( PIX *pixs, l_int32 contrast, l_int32 rotation, l_int32 scale, l_int32 opensize );
 LEPT_DLL extern BOX * pixFindPageForeground ( PIX *pixs, l_int32 threshold, l_int32 mindist, l_int32 erasedist, l_int32 showmorph, PIXAC *pixac );
 LEPT_DLL extern l_ok pixSplitIntoCharacters ( PIX *pixs, l_int32 minw, l_int32 minh, BOXA **pboxa, PIXA **ppixa, PIX **ppixdebug );
@@ -1405,7 +1405,7 @@ LEPT_DLL extern l_ok partifyPixac ( PIXAC *pixac, l_int32 nparts, const char *ou
 LEPT_DLL extern BOXA * boxaGetWhiteblocks ( BOXA *boxas, BOX *box, l_int32 sortflag, l_int32 maxboxes, l_float32 maxoverlap, l_int32 maxperim, l_float32 fract, l_int32 maxpops );
 LEPT_DLL extern BOXA * boxaPruneSortedOnOverlap ( BOXA *boxas, l_float32 maxoverlap );
 LEPT_DLL extern l_ok compressFilesToPdf ( SARRAY *sa, l_int32 onebit, l_int32 savecolor, l_float32 scalefactor, l_int32 quality, const char *title, const char *fileout );
-LEPT_DLL extern l_ok cropFilesToPdf ( SARRAY *sa, l_int32 lr_clear, l_int32 tb_clear, l_int32 edgeclean, l_int32 lr_add, l_int32 tb_add, const char *title, const char *fileout );
+LEPT_DLL extern l_ok cropFilesToPdf ( SARRAY *sa, l_int32 lr_clear, l_int32 tb_clear, l_int32 edgeclean, l_int32 lr_add, l_int32 tb_add, l_float32 maxwiden, const char *title, const char *fileout );
 LEPT_DLL extern l_ok cleanTo1bppFilesToPdf ( SARRAY *sa, l_int32 res, l_int32 contrast, l_int32 rotation, l_int32 opensize, const char *title, const char *fileout );
 LEPT_DLL extern l_ok convertFilesToPdf ( const char *dirname, const char *substr, l_int32 res, l_float32 scalefactor, l_int32 type, l_int32 quality, const char *title, const char *fileout );
 LEPT_DLL extern l_ok saConvertFilesToPdf ( SARRAY *sa, l_int32 res, l_float32 scalefactor, l_int32 type, l_int32 quality, const char *title, const char *fileout );

src/pageseg.c

@@ -538,6 +538,7 @@ PIX     *pix1, *pix2, *pix3, *pixd;
  *                           15 is maximally aggressive
  * \param[in]    lr_add      full res expansion of crop box on left and right
  * \param[in]    tb_add      full res expansion of crop box on top and bottom
+ * \param[in]    maxwiden    max fractional horizontal stretch allowed
  * \param[in]   *debugfile   [optional] usually is NULL
  * \param[out]  *pcropbox    [optional] crop box at full resolution
  * \return  cropped pix, or NULL on error
@@ -555,8 +556,9 @@ PIX     *pix1, *pix2, *pix3, *pixd;
  *          (e) 2x expansion of the bounding box to full resolution. 
  *          (f) Crops the binarized image to the bounding box.
  *          (g) Slightly thickens long horizontal lines.
- *          (h) Does anamorphic horizontal upscaling to better fill
- *              an 8.5 x 11 inch printed page.
+ *          (h) Does anamorphic horizontal upscaling by %maxwiden
+ *              to better fill an 8.5 x 11 inch printed page.
+ *              Suggest not to exceed 1.15.
  *          Note that input parameters are given at full resolution and
  *          (a) - (c) are done at 2x reduction for efficiency.
  *      (2) The side clearing must not exceed 1/6 of the dimension on that side.
@@ -574,6 +576,7 @@ pixCropImage(PIX         *pixs,
              l_int32      edgeclean,
              l_int32      lr_add,
              l_int32      tb_add,
+             l_float32    maxwiden,
              const char  *debugfile,
              BOX        **pcropbox)
 {
@@ -607,6 +610,9 @@ BOX       *box1, *box2;
                 __func__, w / 6, h / 6);
         return NULL;
     }
+    if (maxwiden > 1.2)
+        L_WARNING("maxwiden = %f > 1.2; suggest between 1.0 and 1.15\n",
+                  __func__, maxwiden);
     pixa1 = (debugfile) ? pixaCreate(5) : NULL;
     if (pixa1) pixaAddPix(pixa1, pixs, L_COPY);
 
@@ -664,11 +670,11 @@ BOX       *box1, *box2;
     pixDestroy(&pix3);
 
         /* Widen the result to fit the standard page shape (8.5 x 11 inch).
-         * Do not stretch horizontally by more than 15%. */
+         * Do not stretch horizontally by more than %maxwiden. */
     pixGetDimensions(pix2, &w, &h, NULL);
     hscale = (l_float32)h / (1.2941f * (l_float32)w);
-    if (hscale > 1.0) {
-        hscale = L_MIN(hscale, 1.15);
+    if (hscale > 1.0 && maxwiden > 1.0) {
+        hscale = L_MIN(hscale, maxwiden);
         pix3 = pixScale(pix2, hscale, 1.0);
     } else {
         pix3 = pixClone(pix2);

src/pdfapp.c

@@ -226,6 +226,7 @@ PIXAC     *pixac1 = NULL;
  *                             default = 0 (no removal);
  * \param[in]    lr_add        full res expansion of crop box on left and right
  * \param[in]    tb_add        full res expansion of crop box on top and bottom
+ * \param[in]    maxwiden      max fractional horizontal stretch allowed
  * \param[in]    title         [optional] pdf title; can be null
  * \param[in]    fileout       pdf file of all images
  * \return  0 if OK, 1 on error
@@ -251,6 +252,7 @@ cropFilesToPdf(SARRAY      *sa,
                l_int32      edgeclean,
                l_int32      lr_add,
                l_int32      tb_add,
+               l_float32    maxwiden,
                const char  *title,
                const char  *fileout)
 {
@@ -280,7 +282,7 @@ PIXAC     *pixac1 = NULL;
         fname = sarrayGetString(sa, i, L_NOCOPY);
         pixs = pixRead(fname);
         pix1 = pixCropImage(pixs, lr_clear, tb_clear, edgeclean,
-                            lr_add, tb_add, NULL, NULL);
+                            lr_add, tb_add, maxwiden, NULL, NULL);
         if (n <= maxsmallset)
             pixaAddPix(pixa1, pix1, L_INSERT);
         else