X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?p=ypp-sc-tools.web-live.git;a=blobdiff_plain;f=pctb%2Fstructure.c;h=f10addc56eed9c65f77ede4809407333a4ca780a;hp=f0de5fa964d05f0c9e135b17fb99965200fff10b;hb=9d525e44bd9642a72eb7acd81bd9956b014e3872;hpb=64063c86071bbfc2de0fe032e372e1321c4b34e5 diff --git a/pctb/structure.c b/pctb/structure.c index f0de5fa..f10addc 100644 --- a/pctb/structure.c +++ b/pctb/structure.c @@ -27,11 +27,6 @@ #include "structure.h" -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); } - DEBUG_DEFINE_DEBUGF(struct) #define START_MAIN {200,200} @@ -40,19 +35,39 @@ DEBUG_DEFINE_DEBUGF(struct) #define TEXT_COLUMNS 2 #define MAX_COLUMNS 7 -static Rect mainr = { START_MAIN,START_MAIN }; -static int commbasey, comminty; -static int colrightx[INTERESTING_COLUMNS]; +struct PageStruct { + Rect mr; + int commbasey, comminty; + int colrightx[INTERESTING_COLUMNS]; +}; + +const CanonImage *page_images[MAX_PAGES]; +static PageStruct page_structs[MAX_PAGES]; +const RgbImage *page0_rgbimage; +int npages; + static int text_h=-1, columns=-1; + static OcrReader *rd; +static const CanonImage *cim; +static PageStruct s; + char *archipelago, *island; #define OTHERCOORD_x y #define OTHERCOORD_y x + +void select_page(int page) { + cim= page_images[page]; + s= page_structs[page]; + assert(cim); +} + + typedef struct { - Rgb rgb; /* on screen */ + Rgb rgbx; /* on screen, REVERSED BYTES ie r||g||b */ char c; /* canonical */ } CanonColourInfo; @@ -99,9 +114,9 @@ CanonColourInfoReds canoncolourinfo_tree; void canon_colour_prepare(void) { const CanonColourInfo *cci; for (cci=canoncolourinfo_table; cci->c; cci++) { - unsigned char r= cci->rgb >> 16; - unsigned char g= cci->rgb >> 8; - unsigned char b= cci->rgb; + unsigned char r= cci->rgbx >> 16; + unsigned char g= cci->rgbx >> 8; + unsigned char b= cci->rgbx; CanonColourInfoGreens *greens= canoncolourinfo_tree.red2[r]; if (!greens) { @@ -112,13 +127,16 @@ void canon_colour_prepare(void) { CanonColourInfoBlues *blues= greens->green2[g]; if (!blues) { blues= greens->green2[g]= mmalloc(sizeof(*blues)); - memset(blues, '?', sizeof(blues)); + memset(blues, '?', sizeof(*blues)); } blues->blue2[b]= cci->c; } } +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); } + static void mustfail1(const char *file, int line, const char *what) { fprintf(stderr, @@ -128,7 +146,6 @@ static void mustfail1(const char *file, int line, const char *what) { " * YPP client is showing commodity listing screen\n" " * YPP client window is on top (we try to raise it but your window\n" " manager might have prevented that from succeeding)\n" - " * Your X display is 24bpp (NB some VNC servers use 16bpp by default)\n" "\n" "If all of these are true, please report this as a fault.\n\n" "Technical details:" @@ -143,7 +160,7 @@ static void mustfail2(void) { } #define MUST(x, ifnot) do{ \ - if (!(x)) { \ + if (__builtin_expect(!(x), 0)) { \ mustfail1(__FILE__,__LINE__,#x); \ ifnot; \ mustfail2(); \ @@ -153,10 +170,10 @@ static void mustfail2(void) { #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 MRGB(v) fprintf(stderr," %s=%06"PRIx32, #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 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) @@ -243,7 +260,7 @@ static int commod_selector_matches(Rect search, const char *const *all, #define ADJUST_BOX(search,insidechrs,OP,want, lim,LIMIT_MUST, TLBR,XY,increm) \ for (;;) { \ LIMIT_MUST( (search).tl.XY != (search).br.XY && \ - (search).tl.XY != (lim), \ + (search).TLBR.XY != (lim), \ MR((search));MSB(#TLBR);MSB(#XY) ); \ int got=0; \ Point p=(search).tl; \ @@ -256,9 +273,15 @@ static int commod_selector_matches(Rect search, const char *const *all, (search).TLBR.XY += increm; \ } -void find_structure(CanonImage *im, int *max_relevant_y_r) { +void find_structure(const CanonImage *im, + PageStruct **pagestruct_r, + int *max_relevant_y_r, + Point *commod_focus_point_r, + Point *commod_page_point_r, + Point *commod_focuslast_point_r) { cim= im; - + + FILLZERO(s); Rect whole = { {0,0}, {cim->w-1,cim->h-1} }; if (DEBUGP(rect)) { @@ -273,82 +296,138 @@ void find_structure(CanonImage *im, int *max_relevant_y_r) { } } - WALK_UNTIL_MUST(mainr.tl, x,-1, whole.tl.x, '*'); - WALK_UNTIL_MUST(mainr.tl, y,-1, whole.tl.y, '*'); - WALK_UNTIL_MUST(mainr.br, x,+1, whole.br.x, '*'); - WALK_UNTIL_MUST(mainr.br, y,+1, whole.br.y, '*'); + Point mainr_tl= START_MAIN; + s.mr.tl= mainr_tl; + WALK_UNTIL_MUST(s.mr.tl, y,-1, whole.tl.y, ' '); + s.mr.br= s.mr.tl; + + WALK_UNTIL_MUST(s.mr.tl, x,-1, whole.tl.x, '*'); + WALK_UNTIL_MUST(s.mr.tl, y,-1, whole.tl.y, '*'); + WALK_UNTIL_MUST(s.mr.br, x,+1, whole.br.x, '*'); + WALK_UNTIL_MUST(s.mr.br, y,+1, whole.br.y, '*'); - REQUIRE_RECTANGLE(mainr.tl.x-1, mainr.tl.y, mainr.tl.x-1, mainr.br.y, "*"); - REQUIRE_RECTANGLE(mainr.br.x+1, mainr.tl.y, mainr.br.x+1, mainr.br.y, "*"); - REQUIRE_RECTANGLE(mainr.tl.x, mainr.tl.y-1, mainr.br.x, mainr.tl.y-1, "*"); - REQUIRE_RECTANGLE(mainr.tl.x, mainr.br.y+1, mainr.br.x, mainr.br.y+1, "*"); + REQUIRE_RECTANGLE(s.mr.tl.x-1, s.mr.tl.y, s.mr.tl.x-1, s.mr.br.y, "*"); + REQUIRE_RECTANGLE(s.mr.br.x+1, s.mr.tl.y, s.mr.br.x+1, s.mr.br.y, "*"); + REQUIRE_RECTANGLE(s.mr.tl.x, s.mr.tl.y-1, s.mr.br.x, s.mr.tl.y-1, "*"); + REQUIRE_RECTANGLE(s.mr.tl.x, s.mr.br.y+1, s.mr.br.x, s.mr.br.y+1, "*"); #define CHECK_STRIP_BORDER(tlbr,xy,increm) \ do { \ Point csb_p; \ Rect csb_r; \ - csb_p= mainr.tl; \ - csb_p.xy= mainr.tlbr.xy; \ + csb_p= s.mr.tl; \ + csb_p.x++; csb_p.y++; \ + csb_p.xy= s.mr.tlbr.xy; \ if (get_p(csb_p)=='+') { \ - csb_r= mainr; \ + csb_r= s.mr; \ csb_r.tl.xy= csb_p.xy; \ csb_r.br.xy= csb_p.xy; \ require_rectangle_r(csb_r, "+", __LINE__); \ - mainr.tlbr.xy += increm; \ + s.mr.tlbr.xy += increm; \ } \ } while(0) - debug_rect("mainr",0, mainr); + debug_rect("s.mr",0, s.mr); CHECK_STRIP_BORDER(tl,x,+1); CHECK_STRIP_BORDER(tl,y,+1); CHECK_STRIP_BORDER(br,x,-1); CHECK_STRIP_BORDER(br,y,-1); - debug_rect("mainr",1, mainr); + debug_rect("s.mr",1, s.mr); - Point up = START_MAIN; - WALK_UNTIL_MUST(up, y,-1, mainr.tl.y, '+'); + Rect updown= {START_MAIN,START_MAIN}; + const int chkw= 100; + updown.br.x += chkw-1; + updown.br.y++; + debug_rect("updown",__LINE__,updown); - Point down = START_MAIN; - down.y++; - WALK_UNTIL_MUST(down, y,+1, mainr.br.y, '+'); + ADJUST_BOX(updown, "+", >=,chkw, s.mr.tl.y, MUST, tl,y,-1); + debug_rect("updown",__LINE__,updown); + updown.br.y= updown.tl.y; + updown.tl.y= updown.tl.y-1; - commbasey= up.y; - comminty= down.y - up.y + 2; + ADJUST_BOX(updown, "+*",>=,chkw, s.mr.tl.y-1, MUST, tl,y,-1); + debug_rect("updown",__LINE__,updown); - Point across= { mainr.tl.x, commbasey }; + s.commbasey= updown.tl.y + 1; + s.comminty= updown.br.y - updown.tl.y; + + Rect across= {{ s.mr.tl.x - 1, s.commbasey }, + { s.mr.tl.x, s.commbasey + s.comminty-2 }}; int colno=0; for (;;) { - MUST( get_p(across) != '+', MI(colno);MP(across);MR(mainr);MI(commbasey) ); - WALK_UNTIL(across, x,+1, mainr.br.x, '+'); - MUST( colno < MAX_COLUMNS, MP(across);MR(mainr);MI(commbasey); ); - int colrx= across.x; - if (colrx > mainr.br.x) colrx= mainr.br.x; + +#define LIMIT_QUITEQ(cond,mp) { if (!(cond)) break; } + debug_rect("across",colno*1000000+__LINE__, across); + ADJUST_BOX(across, "+",>=,s.comminty-1,s.mr.br.x,LIMIT_QUITEQ,br,x,+1); + debug_rect("across",colno*1000000+__LINE__, across); + + MUST( colno < MAX_COLUMNS, + MI(colno);MR(across);MR(s.mr);MI(s.commbasey); ); + int colrx= across.br.x-1; + if (colrx >= s.mr.br.x) colrx= s.mr.br.x; if (colno < INTERESTING_COLUMNS) - colrightx[colno]= colrx; + s.colrightx[colno]= colrx; colno++; - if (across.x >= mainr.br.x-1) + if (across.br.x >= s.mr.br.x) break; - across.x++; - REQUIRE_RECTANGLE(across.x,mainr.tl.y, across.x,mainr.br.y, "+"); - across.x++; + REQUIRE_RECTANGLE(across.br.x,s.mr.tl.y, across.br.x,s.mr.br.y, "+"); + across.br.x++; } - MUST( colno >= MIN_COLUMNS, MI(colno);MR(mainr);MP(across); ); + MUST( colno >= MIN_COLUMNS, MI(colno);MR(s.mr);MR(across); ); + + const int pagerh= 6; + Rect pager= {{ s.mr.br.x, s.mr.br.y - (pagerh-1) }, + { s.mr.br.x + 1, s.mr.br.y }}; + + debug_rect("pager",__LINE__,pager); + ADJUST_BOX(pager, "o",>=,pagerh-2, whole.br.x,MUST, br,x,+1); + debug_rect("pager",__LINE__,pager); + + pager.tl.x= pager.br.x; + pager.br.x= pager.br.x + 1; + debug_rect("pager",__LINE__,pager); + ADJUST_BOX(pager, "o",>=,pagerh-2, whole.br.x,MUST, br,x,+1); + debug_rect("pager",__LINE__,pager); + + ADJUST_BOX(pager, "o",>=,RECT_W(pager)-2, s.mr.tl.y,LIMIT_QUITEQ, tl,y,-1); + debug_rect("pager",__LINE__,pager); #define SET_ONCE(var,val) do{ \ int v= (val); \ if ((var)==-1) (var)= v; \ - else MUST( (var) == v, MSB(#var);MI((var));MI(v);MR(mainr); ); \ + else MUST( (var) == v, MSB(#var);MI((var));MI(v);MR(s.mr); ); \ }while(0) SET_ONCE(columns, colno); - SET_ONCE(text_h, comminty - 1); + SET_ONCE(text_h, s.comminty - 1); + + if (pagestruct_r) { + *pagestruct_r= mmalloc(sizeof(s)); + **pagestruct_r= s; + } + if (max_relevant_y_r) - SET_ONCE(*max_relevant_y_r, mainr.br.y + 10); + SET_ONCE(*max_relevant_y_r, s.mr.br.y + 10); + + if (commod_focus_point_r) { + *commod_focus_point_r= s.mr.tl; + commod_focus_point_r->x += 10; + commod_focus_point_r->y += s.comminty/3; + } + if (commod_focuslast_point_r) { + *commod_focuslast_point_r= s.mr.br; + commod_focuslast_point_r->x -= 10; + commod_focuslast_point_r->y -= s.comminty/3; + } + if (commod_page_point_r) { + commod_page_point_r->x= (pager.tl.x + pager.br.x) / 2; + commod_page_point_r->y= pager.tl.y - 1; + } MUST( text_h <= OCR_MAX_H, MI(text_h) ); } @@ -412,6 +491,10 @@ static void file_read_image_ppm(FILE *f) { struct pam inpam; unsigned char rgb_buf[3]; CanonImage *im; + RgbImage *ri; + PageStruct *pstruct; + + progress("page %d reading ...",npages); pnm_readpaminit(f, &inpam, sizeof(inpam)); if (!(inpam.maxval == 255 && @@ -419,14 +502,12 @@ static void file_read_image_ppm(FILE *f) { inpam.format == RPPM_FORMAT)) fatal("PNM screenshot(s) file must be 8bpp 1 byte-per-sample RGB raw"); - CANONICALISE_IMAGE(im, inpam.width, inpam.height, { - int rr= fread(&rgb_buf,1,3,f); - sysassert(!ferror(f)); + CANONICALISE_IMAGE(im, inpam.width, inpam.height, ri, { + errno=0; int rr= fread_unlocked(&rgb_buf,1,3,f); + sysassert(rr==3); if (rr!=3) fatal("PNM screenshot(s) file ends unexpectedly"); - r= rgb_buf[0]; - g= rgb_buf[1]; - b= rgb_buf[2]; + rgb= rgb_buf[0] | (rgb_buf[1] << 8) | (rgb_buf[2] << 16); }); sysassert(!ferror(screenshot_file)); @@ -434,7 +515,21 @@ static void file_read_image_ppm(FILE *f) { if (!(npages < MAX_PAGES)) fatal("Too many images in screenshots file; max is %d.\n", MAX_PAGES); - page_images[npages++]= im; + find_structure(im,&pstruct, 0,0,0,0); + store_current_page(im,pstruct,ri); + npages++; +} + +void store_current_page(CanonImage *ci, PageStruct *pstruct, RgbImage *rgb) { + assert(ci==cim); + progress("page %d unantialiasing...",npages); + adjust_colours(ci, rgb); + progress("page %d storing ...",npages); + if (!npages) page0_rgbimage= rgb; + else free(rgb); + page_images[npages]= cim; + page_structs[npages]= *pstruct; + free(pstruct); } void read_one_screenshot(void) { @@ -457,139 +552,241 @@ void read_screenshots(void) { if (c==EOF) break; ungetc(c, screenshot_file); } - progress("reading screenshot %d...",npages); file_read_image_ppm(screenshot_file); } sysassert(!ferror(screenshot_file)); progress_log("read %d screenshots.",npages); } +#define FIXPT_SHIFT 15 + +typedef long Fixpt; +static inline Fixpt int2fixpt(int x) { return x<> (i*8)) & 0xff ); + fg_chan[i]= aa_foreground >> (i*8); + + aa_scale_chan[i]= 1.0 / (int2fixpt(fg_chan[i]) + fg_extra - aa_bg_chan[i]) + * dbl2fixpt(1) * dbl2fixpt(1); + } +} + +static inline Fixpt find_aa_density(const RgbImage *ri, Point p) { + Rgb here= ri_rgb(ri, p.x, p.y); + + if (here==aa_background) return 0; + + Fixpt alpha[3], alpha_total=0; + int i; + for (i=0; i<3; i++) { + unsigned char here_chan= here >> (i*8); + + Fixpt alpha_chan= fixpt_mul(int2fixpt(here_chan) - aa_bg_chan[i], + aa_scale_chan[i]); + alpha[i]= alpha_chan; + alpha_total += alpha_chan; + } + + Fixpt one_third= dbl2fixpt(1/3.0); + Fixpt alpha_mean= fixpt_mul(alpha_total, one_third); + + Fixpt thresh= dbl2fixpt(1.5/AAMAXVAL); + Fixpt alpha_min= alpha_mean - thresh*2; + Fixpt alpha_max= alpha_mean + thresh*2; + + for (i=0; i<3; i++) + MUST( alpha_min <= alpha[i] && alpha[i] <= alpha_max, + MP(p); + MRGB(here);MRGB(aa_background);MRGB(aa_foreground); + MFP(aa_alpha_mean_max);MFP(thresh);MFP(alpha_mean); + MFP(alpha_min);MI(i);MFP(alpha[i]);MFP(alpha_max) ); + + MUST( -thresh <= alpha_mean && alpha_mean <= aa_alpha_mean_max + thresh, + MP(p); + MRGB(here);MRGB(aa_background);MRGB(aa_foreground); + MFP(aa_alpha_mean_max);MFP(thresh); + MFP(alpha_mean); MFP(alpha[0]);MFP(alpha[1]);MFP(alpha[2]); ); + + if (alpha_mean < 0) alpha_mean= 0; + if (alpha_mean > aa_alpha_mean_max) alpha_mean= aa_alpha_mean_max; + + return alpha_mean; +} + static void find_commodity(int offset, Rect *rr) { /* rr->tl.x==-1 if offset out of range */ - rr->tl.y= commbasey - offset*comminty; - rr->br.y= rr->tl.y + comminty-2; - if (rr->tl.y < mainr.tl.y || rr->br.y > mainr.br.y) { rr->tl.x=-1; return; } + rr->tl.y= s.commbasey - offset*s.comminty; + rr->br.y= rr->tl.y + s.comminty-2; + if (rr->tl.y < s.mr.tl.y || rr->br.y > s.mr.br.y) { rr->tl.x=-1; return; } - rr->tl.x= mainr.tl.x; - rr->br.x= mainr.br.x; + rr->tl.x= s.mr.tl.x; + rr->br.x= s.mr.br.x; - if (rr->tl.y > mainr.tl.y) + if (rr->tl.y > s.mr.tl.y) REQUIRE_RECTANGLE(rr->tl.x,rr->tl.y-1, rr->br.x,rr->tl.y-1, "+"); - if (rr->br.y < mainr.tl.y) + if (rr->br.y < s.mr.tl.y) 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 *cell) { +static void compute_table_location(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]; + cell->tl.x= !colno ? commod.tl.x : s.colrightx[colno-1]+2; + cell->br.x= s.colrightx[colno]; debug_rect("cell", colno, *cell); +} - const RgbImage *ri= cim->rgb; - - Rgb background= ri_rgb(ri, cell->br.x, cell->br.y); +static void ocr_rectangle(Rect r, const OcrCellType ct, FILE *tsv_output) { + OcrResultGlyph *results, *res; + + int w= r.br.x - r.tl.x + 1; + Pixcol cols[w+1]; + int x,y; + for (x=0; x= '0' && pixel <= '0'+AAMAXVAL, + MC(pixel);MP(here);MSB(ocr_celltype_name(ct));MR(r); ); + pixcol_p_add(&cols[x], y, pixel-'0'); + } + } + FILLZERO(cols[w]); + + results= ocr(rd,ct,w,cols); + for (res=results; res->s; res++) + fputs(res->s,tsv_output); +} + +#define FOR_COMMODITY_CELL(ROW_START, CELL, ROW_END) do{ \ + Rect rowr, cell; \ + int tryrect, colno; \ + \ + for (tryrect= +cim->h; tryrect >= -cim->h; tryrect--) { \ + find_commodity(tryrect, &rowr); \ + if (rowr.tl.x < 0) \ + continue; \ + debug_rect("commod",tryrect, rowr); \ + \ + ROW_START; \ + \ + for (colno=0; colno background) light_count++; + + background= ri_rgb(ri, cell.br.x, cell.br.y); + for (i=0; i<3; i++) + chanbg[i]= background >> (i*8); + + FOR_P_RECT(p,cell) { + Rgb herergb= ri_rgb(ri, p.x, p.y); + if (herergb==background) { + bg_count+=3; + } else { + for (i=0; i<3; i++) { + unsigned char here= herergb >> (i*8); + if (here == chanbg[i]) bg_count++; + else if (here < chanbg[i]) dark_count += (chanbg[i] - here)/4 + 1; + else if (here > chanbg[i]) light_count += (here - chanbg[i])/4 + 1; + } + } } - long total_count= RECT_W(*cell) * RECT_H(*cell); + long total_count= RECT_W(cell) * RECT_H(cell) * 3; + MUST( bg_count > total_count / 2, - MR(*cell);MIL(total_count);MIL(bg_count); + 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; double fg_extra; - if (light_count) { + + if (light_count/16 > dark_count) { foreground= 0xffffffU; fg_extra= +1; - } else { + } else if (dark_count/16 > light_count) { foreground= 0; fg_extra= -1; + } else { + MUST( !"tell light from dark", + MR(cell);MIL(total_count);MIL(bg_count); + MIL(light_count);MIL(dark_count);MRGB(background); ); } + + 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); + 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) / - ((fg_chan + fg_extra) - (double)bg_chan); - alpha[i]= alpha_chan; - alpha_mean += alpha_chan * (1/3.0); - } + find_aa_density_prep(background, foreground, fg_extra); + + FOR_P_RECT(p,cell) { + Fixpt alpha= find_aa_density(ri,p); - double thresh= 1.5/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 < 0.999, - MRGB(here);MRGB(background);MRGB(foreground); - MF(alpha_mean); MF(alpha[0]);MF(alpha[1]);MF(alpha[2]); ); - int here_int= floor((AAMAXVAL+1)*alpha_mean); + int here_int= alpha >> (FIXPT_SHIFT - AADEPTH); assert(here_int <= AAMAXVAL); if (!(here_int==0 || here_int==AAMAXVAL)) monochrome=0; - cim->d[p.y * cim->w + p.x]= '0' + here_int; + ci->d[p.y * ci->w + p.x]= '0' + here_int; } - debug_rect("cell0M", colno, *cell); + debug_rect("cell0M", colno, cell); - require_rectangle_r(*cell, "0123456789", __LINE__); + require_rectangle_r(cell, "0123456789", __LINE__); } -static void ocr_rectangle(Rect r, const OcrCellType ct, FILE *tsv_output) { - OcrResultGlyph *results, *res; +void adjust_colours(CanonImage *ci, const RgbImage *ri) { + if (!(o_mode & mf_analyse)) + return; - int w= r.br.x - r.tl.x + 1; - Pixcol cols[w+1]; - int x,y; - for (x=0; x= '0' && pixel <= '0'+AAMAXVAL, - MC(pixel);MP(here);MSB(ocr_celltype_name(ct));MR(r); ); - pixcol_p_add(&cols[x], y, pixel-'0'); - } - } - FILLZERO(cols[w]); + cim= ci; - results= ocr(rd,ct,w,cols); - for (res=results; res->s; res++) - fputs(res->s,tsv_output); + FOR_COMMODITY_CELL({},({ + adjust_colours_cell(ci,ri,colno,cell); + }),{}); } void analyse(FILE *tsv_output) { - Rect thisr, entryr; - int page, tryrect, colno; + int page; for (page=0; pageh; tryrect >= -cim->h; tryrect--) { - find_commodity(tryrect, &thisr); - if (thisr.tl.x < 0) - continue; - debug_rect("commod",tryrect, thisr); - - const char *tab= ""; - for (colno=0; colnow, rgbsrc->h); + memcpy(ri->data, rgbsrc->data, ri->w * ri->h * 3); + Rect sunshiner= find_sunshine_widget(); char sunshine[MAXIMGIDENT], archisland[MAXIMGIDENT]; - const unsigned char *srcp; - unsigned char *destp, *endp; - for (srcp=page_images[0]->rgb->data, destp=ri->data, - endp= ri->data + 3 * ri->w * ri->h; + const Rgb *srcp; + Rgb *destp, *endp; + for (srcp= rgbsrc->data, destp=ri->data, + endp= ri->data + ri->w * ri->h; destp < endp; srcp++, destp++) { - unsigned char c= *srcp & 0xf0; - *destp= c | (c>>4); + Rgb new= *srcp & 0xf0f0f0; + *destp= new | (new>>4); } identify_rgbimage(ri, sunshiner, sunshine, "sunshine widget"); @@ -688,13 +885,11 @@ void find_islandname(RgbImage *ri) { // islandnamer.br.y = larger_islandnamebry; debug_rect("islandnamer",1, islandnamer); - int x,y,i; + int x,y; for (x=islandnamer.tl.x; x<=islandnamer.br.x; x++) for (y=islandnamer.tl.y; y<=islandnamer.br.y; y++) { - if (RI_PIXEL(ri,x,y)[0] < 0x40) { - for (i=0; i<3; i++) { - RI_PIXEL(ri,x,y)[i]= 0; - } + if ((ri_rgb(ri,x,y) & 0xff) < 0x40) { + *RI_PIXEL32(ri,x,y)= 0; } } @@ -747,13 +942,20 @@ void find_islandname(RgbImage *ri) { uint32_t pattern=0; int runs[32], nruns=0; runs[0]=0; runs[1]=0; + + find_aa_density_prep(0xCCCCAA,0x002255,0); for (p.y=islandnamer.tl.y; p.y<=islandnamer.br.y; p.y++) { pattern <<= 1; - switch (get_p(p)) { - case '*': if (runs[nruns]) { nruns++; runs[nruns]=0; } break; - case 'O': runs[nruns]++; pattern |= 1u; break; - default: pattern |= 1UL<<31; break; + Fixpt alpha= find_aa_density(ri,p); + if (alpha >= dbl2fixpt(0.49)) { + runs[nruns]++; + pattern |= 1u; + } else { + if (runs[nruns]) { + nruns++; + runs[nruns]=0; + } } } @@ -767,7 +969,7 @@ void find_islandname(RgbImage *ri) { nspaces++; might_be_colon=0; } else { - if (!(pattern & 1UL<<31) && nruns==2 && runs[1]==runs[0]) { + if (nruns==2 && runs[1]==runs[0]) { if (!nspaces) { if (pattern==colon_pattern) goto ok_might_be_colon; @@ -776,6 +978,8 @@ void find_islandname(RgbImage *ri) { might_be_colon=1; goto ok_might_be_colon; } + } else if (nruns==1 && runs[0]==1 && might_be_colon) { + goto colon_found; } might_be_colon=0; ok_might_be_colon: