#include "structure.h"
-static const CanonImage *cim;
+static CanonImage *cim;
static inline char get(int x, int y) { return cim->d[y * cim->w + x]; }
static inline char get_p(Point p) { return get(p.x,p.y); }
#define MP(v) fprintf(stderr," %s=%d,%d",#v,(v).x,(v).y)
#define MI(v) fprintf(stderr," %s=%d", #v,(v))
+#define MIL(v) fprintf(stderr," %s=%ld", #v,(v))
+#define MRGB(v) fprintf(stderr," %s=%06lx", #v,(v))
#define MC(v) fprintf(stderr," %s='%c'", #v,(v))
#define MS(v) fprintf(stderr," %s=\"%s\"", #v,(v))
+#define MF(v) fprintf(stderr," %s=\"%f\"", #v,(v))
#define MSB(v) fprintf(stderr," %s", (v))
#define MR(v) fprintf(stderr," %s=%d,%d..%d,%d",\
#v,(v).tl.x,(v).tl.y,(v).br.x,(v).br.y)
#define REQUIRE_RECTANGLE(tlx,tly,brx,bry,ok) \
require_rectangle(tlx, tly, brx, bry, ok, __LINE__);
-static void require_rectangle(int tlx, int tly, int brx, int bry,
- const char *ok, int lineno) {
+#define FOR_P_RECT(p,rr) \
+ for ((p).x=(rr).tl.x; (p).x<=(rr).br.x; (p).x++) \
+ for ((p).y=(rr).tl.y; (p).y<=(rr).br.y; (p).y++)
+
+static void require_rectangle_r(Rect rr, const char *ok, int lineno) {
Point p;
- for (p.x=tlx; p.x<=brx; p.x++)
- for (p.y=tly; p.y<=bry; p.y++) {
- int c= get_p(p);
- MUST( strchr(ok,c), ({
- Rect rm={{tlx,tly},{brx,bry}};
- MI(lineno),MR(rm);MP(p);MS(ok);
- }));
- }
+ FOR_P_RECT(p,rr) {
+ int c= get_p(p);
+ MUST( strchr(ok,c), ({
+ MI(lineno),MR(rr);MP(p);MS(ok);
+ }));
+ }
}
-static void require_rectangle_r(Rect rr, const char *ok, int lineno) {
- require_rectangle(rr.tl.x,rr.tl.y, rr.br.x,rr.br.y, ok, lineno);
+static void require_rectangle(int tlx, int tly, int brx, int bry,
+ const char *ok, int lineno) {
+ Rect rr= {{tlx,tly},{brx,bry}};
+ require_rectangle_r(rr, ok, lineno);
}
static void debug_rect(const char *what, int whati, Rect rr) {
(search).TLBR.XY += increm; \
}
-void find_structure(const CanonImage *im, int *max_relevant_y_r) {
+void find_structure(CanonImage *im, int *max_relevant_y_r) {
cim= im;
Rect whole = { {0,0}, {cim->w-1,cim->h-1} };
SET_ONCE(text_h, comminty - 1);
if (max_relevant_y_r)
SET_ONCE(*max_relevant_y_r, mainr.br.y + 10);
+
+ MUST( text_h <= OCR_MAX_H, MI(text_h) );
}
void check_correct_commodities(void) {
struct pam inpam;
unsigned char rgb_buf[3];
CanonImage *im;
- RgbImage *ri=0;
pnm_readpaminit(f, &inpam, sizeof(inpam));
if (!(inpam.maxval == 255 &&
inpam.format == RPPM_FORMAT))
fatal("PNM screenshot(s) file must be 8bpp 1 byte per sample RGB");
- if (!npages)
- page0_rgbimage= ri= alloc_rgb_image(inpam.width, inpam.height);
-
CANONICALISE_IMAGE(im, inpam.width, inpam.height, {
int r= fread(&rgb_buf,1,3,f);
sysassert(!ferror(f));
((unsigned long)rgb_buf[1]<<8) |
(rgb_buf[2]);
- if (ri)
- CANONIMG_ALSO_STORERGB(ri);
});
sysassert(!ferror(screenshot_file));
REQUIRE_RECTANGLE(rr->tl.x,rr->br.y+1, rr->br.x,rr->br.y+1, "+");
}
-static void find_table_entry(Rect commod, int colno, Rect *cellr) {
- cellr->tl.y= commod.tl.y;
- cellr->br.y= commod.br.y;
- cellr->tl.x= !colno ? commod.tl.x : colrightx[colno-1]+2;
- cellr->br.x= colrightx[colno];
- debug_rect("cell", colno, *cellr);
- require_rectangle_r(*cellr, " o", __LINE__);
+static void find_table_entry(Rect commod, int colno, Rect *cell) {
+ cell->tl.y= commod.tl.y;
+ cell->br.y= commod.br.y;
+ cell->tl.x= !colno ? commod.tl.x : colrightx[colno-1]+2;
+ cell->br.x= colrightx[colno];
+ debug_rect("cell", colno, *cell);
+
+ const RgbImage *ri= cim->rgb;
+
+ Rgb background= ri_rgb(ri, cell->br.x, cell->br.y);
+ long bg_count=0, light_count=0, dark_count=0;
+ Point p;
+ FOR_P_RECT(p,*cell) {
+ Rgb here= ri_rgb(ri, p.x, p.y);
+ if (here == background) bg_count++;
+ else if (here < background) dark_count++;
+ else if (here > background) light_count++;
+ }
+ long total_count= RECT_W(*cell) * RECT_H(*cell);
+ MUST( bg_count > total_count / 2,
+ MR(*cell);MIL(total_count);MIL(bg_count);
+ MIL(light_count);MIL(dark_count) );
+ if (bg_count == total_count)
+ return;
+
+ MUST( !!dark_count != !!light_count,
+ MR(*cell);MIL(total_count);MIL(bg_count);
+ MIL(light_count);MIL(dark_count) );
+
+ debugf("TABLEENTRY col=%d %d,%d..%d,%d bg=%ld light=%ld dark=%ld\n",
+ colno, cell->tl.x,cell->tl.y, cell->br.x,cell->br.y,
+ bg_count, light_count, dark_count);
+
+ Rgb foreground= light_count ? 0xffffffU : 0;
+ int monochrome= 1;
+
+ FOR_P_RECT(p,*cell) {
+ Rgb here= ri_rgb(ri, p.x, p.y);
+ double alpha[3], alpha_mean=0;
+ int i;
+ for (i=0; i<3; i++) {
+ unsigned char here_chan= here >> (i*8);
+ unsigned char bg_chan= background >> (i*8);
+ unsigned char fg_chan= foreground >> (i*8);
+ double alpha_chan=
+ ((double)here_chan - (double)bg_chan) /
+ ((double)fg_chan - (double)bg_chan);
+ alpha[i]= alpha_chan;
+ alpha_mean += alpha_chan * (1/3.0);
+ }
+
+ double thresh= 1.0/AAMAXVAL;
+ double alpha_min= alpha_mean - thresh;
+ double alpha_max= alpha_mean + thresh;
+ for (i=0; i<3; i++)
+ MUST( alpha_min <= alpha[i] && alpha[i] <= alpha_max,
+ MI(i);MRGB(here);MRGB(background);MRGB(foreground);
+ MF(alpha_min);MF(alpha[i]);MF(alpha_max) );
+
+ MUST( 0 <= alpha_mean && alpha_mean <= 1,
+ MRGB(here);MRGB(background);MRGB(foreground);
+ MF(alpha_mean) );
+ int here_int= AAMAXVAL*alpha_mean;
+ if (!(here_int==0 || here_int==AAMAXVAL)) monochrome=0;
+ cim->d[p.y * cim->w + p.x]= '0' + here_int;
+ }
+
+ debug_rect("cell0M", colno, *cell);
+
+ require_rectangle_r(*cell, "0123456789", __LINE__);
}
static void ocr_rectangle(Rect r, const OcrCellType ct, FILE *tsv_output) {
Pixcol cols[w+1];
int x,y;
for (x=0; x<w; x++) {
- Pixcol cx, rv;
- for (y=0, cx=0, rv=1; y<text_h; y++, rv<<=1) {
+ FILLZERO(cols[x]);
+ for (y=0; y<text_h; y++) {
Point here= { x+r.tl.x, y+r.tl.y };
int pixel= get_p(here);
- switch (pixel) {
- case ' ': break;
- case 'o': cx |= rv; break;
- default:
- MUST(!"wrong pixel",
- MC(pixel);MP(here);MSB(ocr_celltype_name(ct));MR(r); );
- }
+ if (pixel==' ') pixel= '0';
+ MUST( pixel >= '0' && pixel <= '0'+AAMAXVAL,
+ MC(pixel);MP(here);MSB(ocr_celltype_name(ct));MR(r); );
+ pixcol_p_add(&cols[x], y, pixel-'0');
}
- cols[x]= cx;
}
- cols[w]= 0;
+ FILLZERO(cols[w]);
results= ocr(rd,ct,w,cols);
for (res=results; res->s; res++)
const unsigned char *srcp;
unsigned char *destp, *endp;
- for (srcp=page0_rgbimage->data, destp=ri->data,
+ for (srcp=page_images[0]->rgb->data, destp=ri->data,
endp= ri->data + 3 * ri->w * ri->h;
destp < endp;
srcp++, destp++) {
char *delim= strstr(archisland," - ");
assert(delim);
- archipelago= masprintf("%.*s", delim-archisland, archisland);
+ archipelago= masprintf("%.*s", (int)(delim-archisland), archisland);
island= masprintf("%s", delim+3);
- free(ri);
}
~yarrgweb / ypp-sc-tools.db-test.git / blobdiff