2 * Parsing of the structure of the YPP client's displayed image
5 * This is part of ypp-sc-tools, a set of third-party tools for assisting
6 * players of Yohoho Puzzle Pirates.
8 * Copyright (C) 2009 Ian Jackson <ijackson@chiark.greenend.org.uk>
10 * This program is free software: you can redistribute it and/or modify
11 * it under the terms of the GNU General Public License as published by
12 * the Free Software Foundation, either version 3 of the License, or
13 * (at your option) any later version.
15 * This program is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 * GNU General Public License for more details.
20 * You should have received a copy of the GNU General Public License
21 * along with this program. If not, see <http://www.gnu.org/licenses/>.
23 * Yohoho and Puzzle Pirates are probably trademarks of Three Rings and
24 * are used without permission. This program is not endorsed or
25 * sponsored by Three Rings.
28 #include "structure.h"
30 DEBUG_DEFINE_DEBUGF(struct)
32 #define START_MAIN {200,200}
34 #define INTERESTING_COLUMNS 7
35 #define TEXT_COLUMNS 2
40 int commbasey, comminty;
41 int colrightx[INTERESTING_COLUMNS];
44 const CanonImage *page_images[MAX_PAGES];
45 static PageStruct page_structs[MAX_PAGES];
46 const RgbImage *page0_rgbimage;
49 static int text_h=-1, columns=-1;
53 static const CanonImage *cim;
56 char *archipelago, *island;
58 #define OTHERCOORD_x y
59 #define OTHERCOORD_y x
62 void select_page(int page) {
63 cim= page_images[page];
64 s= page_structs[page];
70 Rgb rgb; /* on screen */
71 char c; /* canonical */
74 const CanonColourInfo canoncolourinfo_table[]= {
75 { 0x475A5E, '*' }, /* edge */
76 { 0x2C5F7A, '*' }, /* edge just under box heading shadow */
77 { 0xC5C7AE, '*' }, /* blank area of partial commodities list */
78 { 0x6B828C, '*' }, /* background of ship status meter area */
79 { 0x934405, '*' }, /* border of ship meter area */
80 { 0x7D9094, '+' }, /* interbox */
81 { 0x022158, 'O' }, /* ahoy /w output foreground */
82 { 0xB5B686, 'H' }, /* ahoy /w output heading background */
84 { 0xBDC5BF, ' ' }, /* background - pale Sugar cane, etc. */
85 { 0xADB5AF, ' ' }, /* background - dark */
86 { 0xC7E1C3, ' ' }, /* background - pale Swill, etc. */
87 { 0xB5CFB1, ' ' }, /* background - dark */
88 { 0xD6CEB0, ' ' }, /* background - pale Madder, etc. */
89 { 0xC8C0A2, ' ' }, /* background - dark */
90 { 0xE0E1D3, ' ' }, /* background - pale Lorandite, etc. */
91 { 0xD0D1C3, ' ' }, /* background - dark */
92 { 0xE5E6C1, ' ' }, /* background - pale Cloth */
93 { 0xD7D8B3, ' ' }, /* background - dark */
94 { 0xEDDED9, ' ' }, /* background - pale Dye */
95 { 0xDACBC6, ' ' }, /* background - dark */
96 { 0xD3DEDF, ' ' }, /* background - pale Paint */
97 { 0xC5D0D1, ' ' }, /* background - dark */
98 { 0xDCD1CF, ' ' }, /* background - pale Enamel */
99 { 0xCEC3C1, ' ' }, /* background - dark */
100 { 0xF3F6F5, ' ' }, /* background - pale fruit */
101 { 0xE2E7E5, ' ' }, /* background - dark */
103 { 0x000000, 'o' }, /* foreground */
104 { 0xD4B356, ' ' }, /* background (cursor) */
105 { 0xFFFFFF, 'o' }, /* foreground (cursor) */
107 { 0x5B93BF, '_' }, /* selector dropdown background */
108 { 0xD7C94F, 'X' }, /* selector dropdown foreground */
112 CanonColourInfoReds canoncolourinfo_tree;
114 void canon_colour_prepare(void) {
115 const CanonColourInfo *cci;
116 for (cci=canoncolourinfo_table; cci->c; cci++) {
117 unsigned char r= cci->rgb >> 16;
118 unsigned char g= cci->rgb >> 8;
119 unsigned char b= cci->rgb;
121 CanonColourInfoGreens *greens= canoncolourinfo_tree.red2[r];
123 greens= canoncolourinfo_tree.red2[r]= mmalloc(sizeof(*greens));
127 CanonColourInfoBlues *blues= greens->green2[g];
129 blues= greens->green2[g]= mmalloc(sizeof(*blues));
130 memset(blues, '?', sizeof(*blues));
133 blues->blue2[b]= cci->c;
137 static inline char get(int x, int y) { return cim->d[y * cim->w + x]; }
138 static inline char get_p(Point p) { return get(p.x,p.y); }
141 static void mustfail1(const char *file, int line, const char *what) {
144 "Unable to figure out contents of YPP client display.\n"
145 "Please check the following:\n"
146 " * YPP client is showing commodity listing screen\n"
147 " * YPP client window is on top (we try to raise it but your window\n"
148 " manager might have prevented that from succeeding)\n"
150 "If all of these are true, please report this as a fault.\n\n"
152 " %s:%d: requirement failed:\n"
156 static void mustfail2(void) NORET;
157 static void mustfail2(void) {
158 fprintf(stderr, "\n\nGiving up.\n");
162 #define MUST(x, ifnot) do{ \
163 if (__builtin_expect(!(x), 0)) { \
164 mustfail1(__FILE__,__LINE__,#x); \
170 #define MP(v) fprintf(stderr," %s=%d,%d",#v,(v).x,(v).y)
171 #define MI(v) fprintf(stderr," %s=%d", #v,(v))
172 #define MIL(v) fprintf(stderr," %s=%ld", #v,(v))
173 #define MRGB(v) fprintf(stderr," %s=%06lx", #v,(v))
174 #define MC(v) fprintf(stderr," %s='%c'", #v,(v))
175 #define MS(v) fprintf(stderr," %s=\"%s\"", #v,(v))
176 #define MF(v) fprintf(stderr," %s=%f", #v,(v))
177 #define MSB(v) fprintf(stderr," %s", (v))
178 #define MR(v) fprintf(stderr," %s=%d,%d..%d,%d",\
179 #v,(v).tl.x,(v).tl.y,(v).br.x,(v).br.y)
182 #define REQUIRE_RECTANGLE(tlx,tly,brx,bry,ok) \
183 require_rectangle(tlx, tly, brx, bry, ok, __LINE__);
185 #define FOR_P_RECT(p,rr) \
186 for ((p).x=(rr).tl.x; (p).x<=(rr).br.x; (p).x++) \
187 for ((p).y=(rr).tl.y; (p).y<=(rr).br.y; (p).y++)
189 static void require_rectangle_r(Rect rr, const char *ok, int lineno) {
193 MUST( strchr(ok,c), ({
194 MI(lineno),MR(rr);MP(p);MS(ok);
198 static void require_rectangle(int tlx, int tly, int brx, int bry,
199 const char *ok, int lineno) {
200 Rect rr= {{tlx,tly},{brx,bry}};
201 require_rectangle_r(rr, ok, lineno);
204 static void debug_rect(const char *what, int whati, Rect rr) {
205 if (!DEBUGP(rect)) return;
207 fprintf(debug, "%s %d: %d,%d..%d,%d:\n", what, whati,
208 rr.tl.x,rr.tl.y, rr.br.x,rr.br.y);
209 w= rr.br.x - rr.tl.x + 1;
210 for (y=rr.tl.y; y<=rr.br.y; y++) {
211 fprintf(debug, "%4d%*s|", y, rr.tl.x,"");
212 fwrite(cim->d + y*cim->w + rr.tl.x, 1, w, debug);
219 static int commod_selector_matches(Rect search, const char *const *all,
220 int allh, int allw) {
221 int alloffy, alloffx;
222 for (alloffy=0; alloffy < search.br.y; alloffy++) {
223 if (alloffy+allh-1 < search.tl.y) continue;
224 for (alloffx=search.tl.x; alloffx+allw-1 <= search.br.x; alloffx++) {
227 for (x=0; x<allw; x++)
228 for (y=0; y<allh; y++) {
230 if (want==' ') continue;
231 if (get(alloffx+x, alloffy+y) == want)
236 debugf("CHECKCOMMOD alloff=%d,%d good=%d bad=%d\n",
237 alloffx,alloffy, good,bad);
245 #define WALK_UNTIL(point,coord,increm,last,edge) \
247 if ((point).coord == (last)+(increm)) break; \
248 if (get_p((point)) == (edge)) { (point).coord -= (increm); break; } \
249 (point).coord += (increm); \
252 #define WALK_UNTIL_MUST(point,coord,increm,last,edge) \
254 WALK_UNTIL(point,coord,increm,last,edge); \
255 MUST( (point).coord != (last)+(increm), \
256 MP(point); MI(increm); MI(last); MC(edge); \
260 #define ADJUST_BOX(search,insidechrs,OP,want, lim,LIMIT_MUST, TLBR,XY,increm) \
262 LIMIT_MUST( (search).tl.XY != (search).br.XY && \
263 (search).TLBR.XY != (lim), \
264 MR((search));MSB(#TLBR);MSB(#XY) ); \
266 Point p=(search).tl; \
267 for (p.XY=(search).TLBR.XY; \
268 p.OTHERCOORD_##XY <= (search).br.OTHERCOORD_##XY; \
269 p.OTHERCOORD_##XY++) \
270 got += !!strchr(insidechrs, get_p(p)); \
271 if ((got) OP (want)) \
273 (search).TLBR.XY += increm; \
276 void find_structure(const CanonImage *im,
277 PageStruct **pagestruct_r,
278 int *max_relevant_y_r,
279 Point *commod_focus_point_r,
280 Point *commod_page_point_r,
281 Point *commod_focuslast_point_r) {
285 Rect whole = { {0,0}, {cim->w-1,cim->h-1} };
289 for (y=0, xscaleunit=1; y<4; y++, xscaleunit*=10) {
291 for (x=0; x<=cim->w; x++) {
292 if (x % xscaleunit) fputc(' ',debug);
293 else fprintf(debug,"%d",(x / xscaleunit)%10);
299 Point mainr_tl= START_MAIN;
301 WALK_UNTIL_MUST(s.mr.tl, y,-1, whole.tl.y, ' ');
304 WALK_UNTIL_MUST(s.mr.tl, x,-1, whole.tl.x, '*');
305 WALK_UNTIL_MUST(s.mr.tl, y,-1, whole.tl.y, '*');
306 WALK_UNTIL_MUST(s.mr.br, x,+1, whole.br.x, '*');
307 WALK_UNTIL_MUST(s.mr.br, y,+1, whole.br.y, '*');
309 REQUIRE_RECTANGLE(s.mr.tl.x-1, s.mr.tl.y, s.mr.tl.x-1, s.mr.br.y, "*");
310 REQUIRE_RECTANGLE(s.mr.br.x+1, s.mr.tl.y, s.mr.br.x+1, s.mr.br.y, "*");
311 REQUIRE_RECTANGLE(s.mr.tl.x, s.mr.tl.y-1, s.mr.br.x, s.mr.tl.y-1, "*");
312 REQUIRE_RECTANGLE(s.mr.tl.x, s.mr.br.y+1, s.mr.br.x, s.mr.br.y+1, "*");
314 #define CHECK_STRIP_BORDER(tlbr,xy,increm) \
319 csb_p.x++; csb_p.y++; \
320 csb_p.xy= s.mr.tlbr.xy; \
321 if (get_p(csb_p)=='+') { \
323 csb_r.tl.xy= csb_p.xy; \
324 csb_r.br.xy= csb_p.xy; \
325 require_rectangle_r(csb_r, "+", __LINE__); \
326 s.mr.tlbr.xy += increm; \
330 debug_rect("s.mr",0, s.mr);
332 CHECK_STRIP_BORDER(tl,x,+1);
333 CHECK_STRIP_BORDER(tl,y,+1);
334 CHECK_STRIP_BORDER(br,x,-1);
335 CHECK_STRIP_BORDER(br,y,-1);
337 debug_rect("s.mr",1, s.mr);
339 Rect updown= {START_MAIN,START_MAIN};
341 updown.br.x += chkw-1;
343 debug_rect("updown",__LINE__,updown);
345 ADJUST_BOX(updown, "+", >=,chkw, s.mr.tl.y, MUST, tl,y,-1);
346 debug_rect("updown",__LINE__,updown);
347 updown.br.y= updown.tl.y;
348 updown.tl.y= updown.tl.y-1;
350 ADJUST_BOX(updown, "+*",>=,chkw, s.mr.tl.y-1, MUST, tl,y,-1);
351 debug_rect("updown",__LINE__,updown);
353 s.commbasey= updown.tl.y + 1;
354 s.comminty= updown.br.y - updown.tl.y;
356 Rect across= {{ s.mr.tl.x - 1, s.commbasey },
357 { s.mr.tl.x, s.commbasey + s.comminty-2 }};
361 #define LIMIT_QUITEQ(cond,mp) { if (!(cond)) break; }
362 debug_rect("across",colno*1000000+__LINE__, across);
363 ADJUST_BOX(across, "+",>=,s.comminty-1,s.mr.br.x,LIMIT_QUITEQ,br,x,+1);
364 debug_rect("across",colno*1000000+__LINE__, across);
366 MUST( colno < MAX_COLUMNS,
367 MI(colno);MR(across);MR(s.mr);MI(s.commbasey); );
368 int colrx= across.br.x-1;
369 if (colrx >= s.mr.br.x) colrx= s.mr.br.x;
370 if (colno < INTERESTING_COLUMNS)
371 s.colrightx[colno]= colrx;
375 if (across.br.x >= s.mr.br.x)
378 REQUIRE_RECTANGLE(across.br.x,s.mr.tl.y, across.br.x,s.mr.br.y, "+");
381 MUST( colno >= MIN_COLUMNS, MI(colno);MR(s.mr);MR(across); );
384 Rect pager= {{ s.mr.br.x, s.mr.br.y - (pagerh-1) },
385 { s.mr.br.x + 1, s.mr.br.y }};
387 debug_rect("pager",__LINE__,pager);
388 ADJUST_BOX(pager, "o",>=,pagerh-2, whole.br.x,MUST, br,x,+1);
389 debug_rect("pager",__LINE__,pager);
391 pager.tl.x= pager.br.x;
392 pager.br.x= pager.br.x + 1;
393 debug_rect("pager",__LINE__,pager);
394 ADJUST_BOX(pager, "o",>=,pagerh-2, whole.br.x,MUST, br,x,+1);
395 debug_rect("pager",__LINE__,pager);
397 ADJUST_BOX(pager, "o",>=,RECT_W(pager)-2, s.mr.tl.y,LIMIT_QUITEQ, tl,y,-1);
398 debug_rect("pager",__LINE__,pager);
400 #define SET_ONCE(var,val) do{ \
402 if ((var)==-1) (var)= v; \
403 else MUST( (var) == v, MSB(#var);MI((var));MI(v);MR(s.mr); ); \
406 SET_ONCE(columns, colno);
407 SET_ONCE(text_h, s.comminty - 1);
410 *pagestruct_r= mmalloc(sizeof(s));
414 if (max_relevant_y_r)
415 SET_ONCE(*max_relevant_y_r, s.mr.br.y + 10);
417 if (commod_focus_point_r) {
418 *commod_focus_point_r= s.mr.tl;
419 commod_focus_point_r->x += 10;
420 commod_focus_point_r->y += s.comminty/3;
422 if (commod_focuslast_point_r) {
423 *commod_focuslast_point_r= s.mr.br;
424 commod_focuslast_point_r->x -= 10;
425 commod_focuslast_point_r->y -= s.comminty/3;
427 if (commod_page_point_r) {
428 commod_page_point_r->x= (pager.tl.x + pager.br.x) / 2;
429 commod_page_point_r->y= pager.tl.y - 1;
432 MUST( text_h <= OCR_MAX_H, MI(text_h) );
435 void check_correct_commodities(void) {
436 Rect search= { { 50,39 }, { 130,59 } };
438 ADJUST_BOX(search,"_",>=,10, cim->h, MUST, tl,y,+1);
439 ADJUST_BOX(search,"_",>=,10, 0, MUST, br,y,-1);
441 debug_rect("commodselr",1, search);
443 static const char *all_small[]= {
444 " ___________________________________ ",
445 " ________X____X__X____________________ ",
446 " ________ X___ X_ X_____XXXXXXXXXXX_____ ",
447 "_________X_X__ X_ X______XXXXXXXXX_______",
448 "________ X X__ X_ X_______XXXXXXX________",
449 "________X_ _X_ X_ X________XXXXX_________",
450 "_______ X__ X_ X_ X_________XXX__________",
451 "_______XXXXXXX X_ X__________X___________",
452 " _____ X X X_ X______________________",
453 " ____X_____ _XX_ X______________________",
454 " __ _______ __ ______________________ ",
456 static const char *all_big[]= {
457 "???_______________________________________???",
458 "??_________________________________________??",
459 "?_________X______X___X______________________?",
460 "_________?X_____?X__?X______XXXXXXXXXXX______",
461 "_________X_X____?X__?X_______XXXXXXXXX_______",
462 "________?X?X____?X__?X________XXXXXXX________",
463 "________X_?_X___?X__?X_________XXXXX_________",
464 "_______?X__?X___?X__?X__________XXX__________",
465 "_______?XXXXX___?X__?X___________X___________",
466 "_______X????_X__?X__?X_______________________",
467 "?_____?X____?X__?X__?X_______________________",
468 "??____X_____?_X_?X__?X_______________________",
469 "???__?_______?__?___?_______________________?",
472 #define COMMOD_SELECTOR_MATCHES(all) \
473 commod_selector_matches(search, all, \
474 sizeof((all))/sizeof((all)[0]), \
477 if (!(COMMOD_SELECTOR_MATCHES(all_small) ||
478 COMMOD_SELECTOR_MATCHES(all_big)))
479 fatal("Commodities selector not set to `All'.");
482 CanonImage *alloc_canon_image(int w, int h) {
483 CanonImage *im= mmalloc(sizeof(CanonImage) + w*h);
486 memset(im->d,'?',w*h);
490 static void file_read_image_ppm(FILE *f) {
492 unsigned char rgb_buf[3];
497 progress("page %d reading...",npages);
499 pnm_readpaminit(f, &inpam, sizeof(inpam));
500 if (!(inpam.maxval == 255 &&
501 inpam.bytes_per_sample == 1 &&
502 inpam.format == RPPM_FORMAT))
503 fatal("PNM screenshot(s) file must be 8bpp 1 byte-per-sample RGB raw");
505 CANONICALISE_IMAGE(im, inpam.width, inpam.height, rgb, {
506 int rr= fread(&rgb_buf,1,3,f);
507 sysassert(!ferror(f));
508 if (rr!=3) fatal("PNM screenshot(s) file ends unexpectedly");
515 sysassert(!ferror(screenshot_file));
517 if (!(npages < MAX_PAGES))
518 fatal("Too many images in screenshots file; max is %d.\n", MAX_PAGES);
520 find_structure(im,&pstruct, 0,0,0,0);
521 store_current_page(im,pstruct,rgb);
525 void store_current_page(CanonImage *ci, PageStruct *pstruct, RgbImage *rgb) {
527 progress("page %d condensing...",npages);
528 adjust_colours(ci, rgb);
529 progress("page %d storing...",npages);
530 if (!npages) page0_rgbimage= rgb;
532 page_images[npages]= cim;
533 page_structs[npages]= *pstruct;
537 void read_one_screenshot(void) {
538 progress("reading screenshot...");
539 progress_log("read screenshot.");
542 void read_screenshots(void) {
545 sysassert(! fstat(fileno(screenshot_file), &stab) );
548 if (S_ISREG(stab.st_mode)) {
549 long pos= ftell(screenshot_file);
550 if (pos == stab.st_size) break;
552 int c= fgetc(screenshot_file);
554 ungetc(c, screenshot_file);
556 file_read_image_ppm(screenshot_file);
558 sysassert(!ferror(screenshot_file));
559 progress_log("read %d screenshots.",npages);
562 static inline double find_aa_density(const RgbImage *ri,
563 Point p, long background,
564 long foreground, int fg_extra) {
565 Rgb here= ri_rgb(ri, p.x, p.y);
567 double alpha[3], alpha_mean=0;
569 for (i=0; i<3; i++) {
570 unsigned char here_chan= here >> (i*8);
571 unsigned char bg_chan= background >> (i*8);
572 unsigned char fg_chan= foreground >> (i*8);
574 ((double)here_chan - (double)bg_chan) /
575 ((fg_chan + fg_extra) - (double)bg_chan);
576 alpha[i]= alpha_chan;
577 alpha_mean += alpha_chan * (1/3.0);
580 double thresh= 1.5/AAMAXVAL;
581 double alpha_min= alpha_mean - thresh;
582 double alpha_max= alpha_mean + thresh;
584 MUST( alpha_min <= alpha[i] && alpha[i] <= alpha_max,
586 MRGB(here);MRGB(background);MRGB(foreground);MI(fg_extra);
587 MF(alpha_min); MI(i);MF(alpha[i]);MF(alpha_max) );
589 if ( -1e-5 < alpha_mean && alpha_mean <= 0.0 ) alpha_mean= 0.0;
590 if (1.0 <= alpha_mean && alpha_mean <= 1.0+1e-5) alpha_mean= 1.0;
592 MUST( 0 <= alpha_mean &&
593 (fg_extra ? (alpha_mean < 0.999) : (alpha_mean <= 1.0)),
595 MRGB(here);MRGB(background);MRGB(foreground);MI(fg_extra);
596 MF(alpha_mean); MF(alpha[0]);MF(alpha[1]);MF(alpha[2]); );
601 static void find_commodity(int offset, Rect *rr) {
602 /* rr->tl.x==-1 if offset out of range */
603 rr->tl.y= s.commbasey - offset*s.comminty;
604 rr->br.y= rr->tl.y + s.comminty-2;
605 if (rr->tl.y < s.mr.tl.y || rr->br.y > s.mr.br.y) { rr->tl.x=-1; return; }
610 if (rr->tl.y > s.mr.tl.y)
611 REQUIRE_RECTANGLE(rr->tl.x,rr->tl.y-1, rr->br.x,rr->tl.y-1, "+");
612 if (rr->br.y < s.mr.tl.y)
613 REQUIRE_RECTANGLE(rr->tl.x,rr->br.y+1, rr->br.x,rr->br.y+1, "+");
616 static void compute_table_location(Rect commod, int colno, Rect *cell) {
617 cell->tl.y= commod.tl.y;
618 cell->br.y= commod.br.y;
619 cell->tl.x= !colno ? commod.tl.x : s.colrightx[colno-1]+2;
620 cell->br.x= s.colrightx[colno];
621 debug_rect("cell", colno, *cell);
624 static void ocr_rectangle(Rect r, const OcrCellType ct, FILE *tsv_output) {
625 OcrResultGlyph *results, *res;
627 int w= r.br.x - r.tl.x + 1;
630 for (x=0; x<w; x++) {
632 for (y=0; y<text_h; y++) {
633 Point here= { x+r.tl.x, y+r.tl.y };
634 int pixel= get_p(here);
635 if (pixel==' ') pixel= '0';
636 MUST( pixel >= '0' && pixel <= '0'+AAMAXVAL,
637 MC(pixel);MP(here);MSB(ocr_celltype_name(ct));MR(r); );
638 pixcol_p_add(&cols[x], y, pixel-'0');
643 results= ocr(rd,ct,w,cols);
644 for (res=results; res->s; res++)
645 fputs(res->s,tsv_output);
648 #define FOR_COMMODITY_CELL(ROW_START, CELL, ROW_END) do{ \
650 int tryrect, colno; \
652 for (tryrect= +cim->h; tryrect >= -cim->h; tryrect--) { \
653 find_commodity(tryrect, &rowr); \
656 debug_rect("commod",tryrect, rowr); \
660 for (colno=0; colno<columns; colno++) { \
661 compute_table_location(rowr,colno,&cell); \
670 static void adjust_colours_cell(CanonImage *ci, const RgbImage *ri,
671 int colno, Rect cell) {
673 unsigned char chanbg[3];
674 long bg_count=0, light_count=0, dark_count=0;
677 background= ri_rgb(ri, cell.br.x, cell.br.y);
678 memcpy(chanbg, RI_PIXEL(ri, cell.br.x, cell.br.y), 3);
681 const unsigned char *here_pixel= RI_PIXEL(ri, p.x, p.y);
683 for (i=0; i<3; i++) {
684 unsigned here= here_pixel[i];
685 if (here == chanbg[i]) bg_count++;
686 else if (here < chanbg[i]) dark_count += (chanbg[i] - here)/4 + 1;
687 else if (here > chanbg[i]) light_count += (here - chanbg[i])/4 + 1;
690 long total_count= RECT_W(cell) * RECT_H(cell) * 3;
692 MUST( bg_count > total_count / 2,
693 MR(cell);MIL(total_count);MIL(bg_count);
694 MIL(light_count);MIL(dark_count) );
696 if (bg_count == total_count)
702 if (light_count/16 > dark_count) {
703 foreground= 0xffffffU;
705 } else if (dark_count/16 > light_count) {
709 MUST( !"tell light from dark",
710 MR(cell);MIL(total_count);MIL(bg_count);
711 MIL(light_count);MIL(dark_count);MRGB(background); );
714 debugf("TABLEENTRY col=%d %d,%d..%d,%d bg=%ld light=%ld dark=%ld\n",
715 colno, cell.tl.x,cell.tl.y, cell.br.x,cell.br.y,
716 bg_count, light_count, dark_count);
721 double alpha= find_aa_density(ri,p,background,foreground,fg_extra);
723 int here_int= floor((AAMAXVAL+1)*alpha);
724 assert(here_int <= AAMAXVAL);
725 if (!(here_int==0 || here_int==AAMAXVAL)) monochrome=0;
726 ci->d[p.y * ci->w + p.x]= '0' + here_int;
729 debug_rect("cell0M", colno, cell);
731 require_rectangle_r(cell, "0123456789", __LINE__);
734 void adjust_colours(CanonImage *ci, const RgbImage *ri) {
735 if (!(o_mode & mf_analyse))
740 FOR_COMMODITY_CELL({},({
741 adjust_colours_cell(ci,ri,colno,cell);
745 void analyse(FILE *tsv_output) {
748 for (page=0; page<npages; page++) {
752 check_correct_commodities();
755 rd= ocr_init(text_h);
757 progress("Scanning page %d...",page);
764 fputs(tab, tsv_output);
768 : &ocr_celltype_number,
772 fputs("\n", tsv_output);
773 sysassert(!ferror(tsv_output));
774 sysassert(!fflush(tsv_output));
777 progress("Commodity table scan complete.");
780 //static Rect islandnamer;
782 DEBUG_DEFINE_SOME_DEBUGF(structcolon,colondebugf)
784 Rect find_sunshine_widget(void) {
787 sunshiner.tl.x= cim->w - 1034 + 885;
788 sunshiner.br.x= cim->w - 1034 + 1020;
792 ADJUST_BOX(sunshiner,"o*",>=,30, 100,MUST, tl,y,-1);
793 ADJUST_BOX(sunshiner,"o*",>=,30, 100,MUST, br,y,+1);
794 debug_rect("sunshiner",0, sunshiner);
796 MUST(sunshiner.br.y - sunshiner.tl.y > 20, MR(sunshiner));
799 ADJUST_BOX(sunshiner,"o",>=,20, (cim->w - 1034 + 700), MUST, tl,x,-1);
800 ADJUST_BOX(sunshiner,"o",>=,20, cim->w, MUST, br,x,+1);
801 debug_rect("sunshiner",1, sunshiner);
805 void find_islandname(void) {
806 const RgbImage *rgbsrc= page0_rgbimage;
809 RgbImage *ri= alloc_rgb_image(rgbsrc->w, rgbsrc->h);
810 memcpy(ri->data, rgbsrc->data, ri->w * ri->h * 3);
812 Rect sunshiner= find_sunshine_widget();
813 char sunshine[MAXIMGIDENT], archisland[MAXIMGIDENT];
815 const unsigned char *srcp;
816 unsigned char *destp, *endp;
817 for (srcp= rgbsrc->data, destp=ri->data,
818 endp= ri->data + 3 * ri->w * ri->h;
821 unsigned char c= *srcp & 0xf0;
825 identify_rgbimage(ri, sunshiner, sunshine, "sunshine widget");
827 if (!memcmp(sunshine,"Vessel ",5)) {
830 islandnamer.tl.x= cim->w - 1034 + 885;
831 islandnamer.br.x= cim->w - 1034 + 1020;
832 islandnamer.tl.y= 128;
833 islandnamer.br.y= 156;
835 ADJUST_BOX(islandnamer,"o",>=,5, 0, MUST, tl,y,+1);
836 ADJUST_BOX(islandnamer,"o",>=,5, cim->h, MUST, br,y,-1);
838 ADJUST_BOX(islandnamer,"o",>=,1, 0, MUST, tl,x,+1);
839 ADJUST_BOX(islandnamer,"o",>=,1, cim->w, MUST, br,x,-1);
841 debug_rect("islandnamer",0, islandnamer);
842 // int larger_islandnamebry= islandnamer.tl.y + 25;
843 // MUST(islandnamer.br.y < larger_islandnamebry,
844 // MR(islandnamer);MI(larger_islandnamebry));
845 // islandnamer.br.y = larger_islandnamebry;
846 debug_rect("islandnamer",1, islandnamer);
849 for (x=islandnamer.tl.x; x<=islandnamer.br.x; x++)
850 for (y=islandnamer.tl.y; y<=islandnamer.br.y; y++) {
851 if (RI_PIXEL(ri,x,y)[0] < 0x40) {
852 for (i=0; i<3; i++) {
853 RI_PIXEL(ri,x,y)[i]= 0;
858 identify_rgbimage(ri, islandnamer, archisland, "island");
859 } else if (!strcmp(sunshine,"Land - Ahoy!")) {
862 islandnamer.tl.x= (sunshiner.tl.x + sunshiner.br.x) / 2;
863 islandnamer.tl.y= sunshiner.tl.y + 100;
864 islandnamer.br= islandnamer.tl;
865 debug_rect("islandnamer",__LINE__, islandnamer);
867 WALK_UNTIL_MUST(islandnamer.tl,y, -1, sunshiner.br.y, 'H');
868 WALK_UNTIL_MUST(islandnamer.tl,x, -1, 0, 'o');
869 WALK_UNTIL_MUST(islandnamer.br,x, +1, cim->w, 'o');
870 debug_rect("islandnamer",__LINE__, islandnamer);
872 #define RW (RECT_W(islandnamer))
873 #define RH (RECT_H(islandnamer))
875 ADJUST_BOX(islandnamer,"O",>=,RW-4, cim->h, MUST,br,y,+1);
876 debug_rect("islandnamer",__LINE__, islandnamer);
878 islandnamer.br.y += 2;
880 ADJUST_BOX(islandnamer,"*",<,RW, cim->h, MUST,br,y,+1);
881 debug_rect("islandnamer",__LINE__, islandnamer);
883 islandnamer.tl.y= islandnamer.br.y-1;
884 islandnamer.br.y= islandnamer.br.y+1;
885 debug_rect("islandnamer",__LINE__, islandnamer);
887 ADJUST_BOX(islandnamer,"*",>=,RW, cim->h, MUST,br,y,+1);
888 debug_rect("islandnamer",__LINE__, islandnamer);
890 ADJUST_BOX(islandnamer,"*",<, RH, cim->w, MUST,tl,x,+1);
891 debug_rect("islandnamer",__LINE__, islandnamer);
893 MUST( RECT_H(islandnamer) <= 30, MR(islandnamer));
896 int nspaces=1, might_be_colon=0;
897 uint32_t colon_pattern= 0;
900 for (p.x=islandnamer.br.x; p.x>islandnamer.tl.x; p.x--) {
901 colondebugf("structcolon: x=%4d nsp=%2d mbc=%d cp=%08"PRIx32" ",
902 p.x, nspaces, might_be_colon, colon_pattern);
905 int runs[32], nruns=0;
906 runs[0]=0; runs[1]=0;
908 for (p.y=islandnamer.tl.y; p.y<=islandnamer.br.y; p.y++) {
910 double alpha= find_aa_density(ri,p, 0xCCCCAA,0x002255,0);
922 colondebugf(" pat=%08"PRIx32" nruns=%d runs[]={%d,%d..} ",
923 pattern, nruns, runs[0],runs[1]);
927 /* omg it _is_ a colon */
932 if (nruns==2 && runs[1]==runs[0]) {
934 if (pattern==colon_pattern)
935 goto ok_might_be_colon;
936 } else if (nspaces>=2) {
937 colon_pattern= pattern;
939 goto ok_might_be_colon;
941 } else if (nruns==1 && runs[0]==1 && might_be_colon) {
948 colondebugf(" nsp=%2d mbc=%d\n", nspaces, might_be_colon);
950 MUST(!"colon found", MP(p);MR(islandnamer) );
953 colondebugf(" found\n");
954 islandnamer.br.x= p.x;
956 identify_rgbimage(ri, islandnamer, archisland, "island");
959 MUST(!"sunshine shows ship or ahoy", MS(sunshine) );
963 char *delim= strstr(archisland," - ");
965 archipelago= masprintf("%.*s", (int)(delim-archisland), archisland);
966 island= masprintf("%s", delim+3);