#include "structure.h"
-static CanonImage *cim;
+static const 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}
#define MIN_COLUMNS 6
#define INTERESTING_COLUMNS 7
static int text_h=-1, columns=-1;
static OcrReader *rd;
+char *archipelago, *island;
+
+#define OTHERCOORD_x y
+#define OTHERCOORD_y x
+
const CanonColourInfo canoncolourinfos[]= {
{ 0x475A5E, '*' }, /* edge */
{ 0x2C5F7A, '*' }, /* edge just under box heading shadow */
+ { 0xC5C7AE, '*' }, /* blank area of partial commodities list */
+ { 0x6B828C, '*' }, /* background of ship status meter area */
+ { 0x934405, '*' }, /* border of ship meter area */
{ 0x7D9094, '+' }, /* interbox */
+ { 0x022158, 'O' }, /* ahoy /w output foreground */
+ { 0xB5B686, 'H' }, /* ahoy /w output heading background */
{ 0xBDC5BF, ' ' }, /* background - pale Sugar cane, etc. */
{ 0xADB5AF, ' ' }, /* background - dark */
static void mustfail1(const char *file, int line, const char *what) {
fprintf(stderr,
- "\n\n"
- "Unable to figure out contents YPP client display.\n"
- " Check that your client is logged in has the correct display.\n"
- " If that isn't the problem, please report this as a fault.\n\n"
+ "\n\n"
+ "Unable to figure out contents of YPP client display.\n"
+ "Please check the following:\n"
+ " * YPP client is showing commodity listing screen\n"
+ " * YPP client has antialiased text turned off (Ye / Options / General)\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:"
- " %s:%d: requirement failed: %s\n",
+ " %s:%d: requirement failed:\n"
+ " %s\n",
file, line, what);
}
static void mustfail2(void) NORET;
#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) {
+ 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}};
- MR(rm);MP(p);MS(ok);
+ MI(lineno),MR(rm);MP(p);MS(ok);
}));
}
}
-static void require_rectangle_r(Rect rr, const char *ok) {
- require_rectangle(rr.tl.x,rr.tl.y, rr.br.x,rr.br.y, 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 debug_rect(const char *what, int whati, Rect rr) {
debug_flush();
}
+static int commod_selector_matches(Rect search, const char *const *all,
+ int allh, int allw) {
+ int alloffy, alloffx;
+ for (alloffy=0; alloffy < search.br.y; alloffy++) {
+ if (alloffy+allh-1 < search.tl.y) continue;
+ for (alloffx=search.tl.x; alloffx+allw-1 <= search.br.x; alloffx++) {
+ int good=0, bad=0;
+ int x,y;
+ for (x=0; x<allw; x++)
+ for (y=0; y<allh; y++) {
+ int want= all[y][x];
+ if (want==' ') continue;
+ if (get(alloffx+x, alloffy+y) == want)
+ good++;
+ else
+ bad++;
+ }
+ debugf("CHECKCOMMOD alloff=%d,%d good=%d bad=%d\n",
+ alloffx,alloffy, good,bad);
+ if (good > 20*bad)
+ return 1;
+ }
+ }
+ return 0;
+}
+
#define WALK_UNTIL(point,coord,increm,last,edge) \
for (;;) { \
if ((point).coord == (last)+(increm)) break; \
); \
}while(0)
-void find_structure(CanonImage *im) {
+#define ADJUST_BOX(search,insidechrs,want, lim,LIMIT_MUST, TLBR,XY,increm) \
+ for (;;) { \
+ LIMIT_MUST( (search).tl.XY != (search).br.XY && \
+ (search).tl.XY != (lim), \
+ MR((search));MSB(#TLBR);MSB(#XY) ); \
+ int got=0; \
+ Point p=(search).tl; \
+ for (p.XY=(search).TLBR.XY; \
+ p.OTHERCOORD_##XY <= (search).br.OTHERCOORD_##XY; \
+ p.OTHERCOORD_##XY++) \
+ got += !!strchr(insidechrs, get_p(p)); \
+ if (got >= (want)) \
+ break; \
+ (search).TLBR.XY += increm; \
+ }
+
+void find_structure(const CanonImage *im, int *max_relevant_y_r) {
cim= im;
Rect whole = { {0,0}, {cim->w-1,cim->h-1} };
+ if (DEBUGP(rect)) {
+ int xscaleunit, y,x;
+ for (y=0, xscaleunit=1; y<4; y++, xscaleunit*=10) {
+ fprintf(debug," ");
+ for (x=0; x<=cim->w; x++) {
+ if (x % xscaleunit) fputc(' ',debug);
+ else fprintf(debug,"%d",(x / xscaleunit)%10);
+ }
+ fputc('\n',debug);
+ }
+ }
+
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, '*');
- 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, "*");
-
-#define CHECK_STRIP_BORDER(tlbr,xy,increm) \
- do { \
- Point csb_p; \
- Rect csb_r; \
- csb_p= mainr.tl; \
- csb_p.xy= mainr.tlbr.xy; \
- if (get_p(csb_p)=='+') { \
- csb_r= mainr; \
- csb_r.tl.xy= csb_p.xy; \
- csb_r.br.xy= csb_p.xy; \
- require_rectangle_r(csb_r, "+"); \
- mainr.tlbr.xy += increm; \
- } \
+ 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, "*");
+
+#define CHECK_STRIP_BORDER(tlbr,xy,increm) \
+ do { \
+ Point csb_p; \
+ Rect csb_r; \
+ csb_p= mainr.tl; \
+ csb_p.xy= mainr.tlbr.xy; \
+ if (get_p(csb_p)=='+') { \
+ csb_r= mainr; \
+ csb_r.tl.xy= csb_p.xy; \
+ csb_r.br.xy= csb_p.xy; \
+ require_rectangle_r(csb_r, "+", __LINE__); \
+ mainr.tlbr.xy += increm; \
+ } \
} while(0)
debug_rect("mainr",0, mainr);
down.y++;
WALK_UNTIL_MUST(down, y,+1, mainr.br.y, '+');
- if (DEBUGP(rect)) {
- int xscaleunit, y,x;
- for (y=0, xscaleunit=1; y<4; y++, xscaleunit*=10) {
- fprintf(debug," ");
- for (x=0; x<=cim->w; x++) {
- if (x % xscaleunit) fputc(' ',debug);
- else fprintf(debug,"%d",(x / xscaleunit)%10);
- }
- fputc('\n',debug);
- }
- }
-
commbasey= up.y;
comminty= down.y - up.y + 2;
break;
across.x++;
- require_rectangle(across.x,mainr.tl.y, across.x,mainr.br.y, "+");
+ REQUIRE_RECTANGLE(across.x,mainr.tl.y, across.x,mainr.br.y, "+");
across.x++;
}
MUST( colno >= MIN_COLUMNS, MI(colno);MR(mainr);MP(across); );
SET_ONCE(columns, colno);
SET_ONCE(text_h, comminty - 1);
+ if (max_relevant_y_r)
+ SET_ONCE(*max_relevant_y_r, mainr.br.y + 10);
}
+void check_correct_commodities(void) {
+ Rect search= { { 50,39 }, { 130,59 } };
+
+ ADJUST_BOX(search,"_",10, cim->h, MUST, tl,y,+1);
+ ADJUST_BOX(search,"_",10, 0, MUST, br,y,-1);
+
+ debug_rect("commodselr",1, search);
+
+ static const char *all_small[]= {
+ " ___________________________________ ",
+ " ________X____X__X____________________ ",
+ " ________ X___ X_ X_____XXXXXXXXXXX_____ ",
+ "_________X_X__ X_ X______XXXXXXXXX_______",
+ "________ X X__ X_ X_______XXXXXXX________",
+ "________X_ _X_ X_ X________XXXXX_________",
+ "_______ X__ X_ X_ X_________XXX__________",
+ "_______XXXXXXX X_ X__________X___________",
+ " _____ X X X_ X______________________",
+ " ____X_____ _XX_ X______________________",
+ " __ _______ __ ______________________ ",
+ };
+ static const char *all_big[]= {
+ "???_______________________________________???",
+ "??_________________________________________??",
+ "?_________X______X___X______________________?",
+ "_________?X_____?X__?X______XXXXXXXXXXX______",
+ "_________X_X____?X__?X_______XXXXXXXXX_______",
+ "________?X?X____?X__?X________XXXXXXX________",
+ "________X_?_X___?X__?X_________XXXXX_________",
+ "_______?X__?X___?X__?X__________XXX__________",
+ "_______?XXXXX___?X__?X___________X___________",
+ "_______X????_X__?X__?X_______________________",
+ "?_____?X____?X__?X__?X_______________________",
+ "??____X_____?_X_?X__?X_______________________",
+ "???__?_______?__?___?_______________________?",
+ };
+
+#define COMMOD_SELECTOR_MATCHES(all) \
+ commod_selector_matches(search, all, \
+ sizeof((all))/sizeof((all)[0]), \
+ strlen((all)[0]))
+
+ if (!(COMMOD_SELECTOR_MATCHES(all_small) ||
+ COMMOD_SELECTOR_MATCHES(all_big)))
+ fatal("Commodities selector not set to `All'.");
+}
+
CanonImage *alloc_canon_image(int w, int h) {
CanonImage *im= mmalloc(sizeof(CanonImage) + w*h);
im->w= w;
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[0]<<16) |
((unsigned long)rgb_buf[1]<<8) |
(rgb_buf[2]);
+
+ if (ri)
+ CANONIMG_ALSO_STORERGB(ri);
});
- sysassert(!ferror(screenshots_file));
+ sysassert(!ferror(screenshot_file));
if (!(npages < MAX_PAGES))
fatal("Too many images in screenshots file; max is %d.\n", MAX_PAGES);
void read_one_screenshot(void) {
progress("reading screenshot...");
- file_read_image_ppm(screenshots_file);
+ file_read_image_ppm(screenshot_file);
progress_log("read screenshot.");
}
void read_screenshots(void) {
struct stat stab;
- sysassert(! fstat(fileno(screenshots_file), &stab) );
+ sysassert(! fstat(fileno(screenshot_file), &stab) );
for (;;) {
if (S_ISREG(stab.st_mode)) {
- long pos= ftell(screenshots_file);
+ long pos= ftell(screenshot_file);
if (pos == stab.st_size) break;
} else {
- int c= fgetc(screenshots_file);
+ int c= fgetc(screenshot_file);
if (c==EOF) break;
- ungetc(c, screenshots_file);
+ ungetc(c, screenshot_file);
}
progress("reading screenshot %d...",npages);
- file_read_image_ppm(screenshots_file);
+ file_read_image_ppm(screenshot_file);
}
- sysassert(!ferror(screenshots_file));
+ sysassert(!ferror(screenshot_file));
progress_log("read %d screenshots.",npages);
}
rr->br.x= mainr.br.x;
if (rr->tl.y > mainr.tl.y)
- require_rectangle(rr->tl.x,rr->tl.y-1, rr->br.x,rr->tl.y-1, "+");
+ REQUIRE_RECTANGLE(rr->tl.x,rr->tl.y-1, rr->br.x,rr->tl.y-1, "+");
if (rr->br.y < mainr.tl.y)
- require_rectangle(rr->tl.x,rr->br.y+1, rr->br.x,rr->br.y+1, "+");
+ 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.x= !colno ? commod.tl.x : colrightx[colno-1]+2;
cellr->br.x= colrightx[colno];
debug_rect("cell", colno, *cellr);
- require_rectangle_r(*cellr, " o");
+ require_rectangle_r(*cellr, " o", __LINE__);
}
-static void ocr_rectangle(Rect r, const OcrCellType ct) {
+static void ocr_rectangle(Rect r, const OcrCellType ct, FILE *tsv_output) {
OcrResultGlyph *results, *res;
int w= r.br.x - r.tl.x + 1;
results= ocr(rd,ct,w,cols);
for (res=results; res->s; res++)
- printf("%s",res->s);
+ fputs(res->s,tsv_output);
}
-void analyse(void) {
+void analyse(FILE *tsv_output) {
Rect thisr, entryr;
int page, tryrect, colno;
for (page=0; page<npages; page++) {
- find_structure(page_images[page]);
+ find_structure(page_images[page], 0);
+
+ if (!page)
+ check_correct_commodities();
if (!rd)
rd= ocr_init(text_h);
+ progress("Scanning page %d...",page);
+
for (tryrect= +cim->h; tryrect >= -cim->h; tryrect--) {
find_commodity(tryrect, &thisr);
if (thisr.tl.x < 0)
const char *tab= "";
for (colno=0; colno<columns; colno++) {
find_table_entry(thisr,colno,&entryr);
- fputs(tab, stdout);
+ fputs(tab, tsv_output);
ocr_rectangle(entryr,
colno<TEXT_COLUMNS
? &ocr_celltype_text
- : &ocr_celltype_number);
+ : &ocr_celltype_number,
+ tsv_output);
tab= "\t";
}
- fputs("\n", stdout);
- sysassert(!ferror(stdout));
- sysassert(!fflush(stdout));
+ fputs("\n", tsv_output);
+ sysassert(!ferror(tsv_output));
+ sysassert(!fflush(tsv_output));
+ }
+ }
+ progress("Commodity table scan complete.");
+}
+
+//static Rect islandnamer;
+
+DEBUG_DEFINE_SOME_DEBUGF(structcolon,colondebugf)
+
+Rect find_sunshine_widget(void) {
+ Rect sunshiner;
+
+ sunshiner.tl.x= cim->w - 1034 + 885;
+ sunshiner.br.x= cim->w - 1034 + 1020;
+ sunshiner.tl.y= 227;
+ sunshiner.br.y= 228;
+
+ ADJUST_BOX(sunshiner,"o*",30, 100,MUST, tl,y,-1);
+ ADJUST_BOX(sunshiner,"o*",30, 100,MUST, br,y,+1);
+ debug_rect("sunshiner",0, sunshiner);
+
+ MUST(sunshiner.br.y - sunshiner.tl.y > 20, MR(sunshiner));
+ sunshiner.br.y--;
+
+ ADJUST_BOX(sunshiner,"o",20, (cim->w - 1034 + 700), MUST, tl,x,-1);
+ ADJUST_BOX(sunshiner,"o",20, cim->w, MUST, br,x,+1);
+ debug_rect("sunshiner",1, sunshiner);
+ return sunshiner;
+}
+
+void find_islandname(RgbImage *ri) {
+ find_structure(page_images[0], 0);
+
+ Rect sunshiner= find_sunshine_widget();
+ char sunshine[MAXIMGIDENT], archisland[MAXIMGIDENT];
+
+ const unsigned char *srcp;
+ unsigned char *destp, *endp;
+ for (srcp=page0_rgbimage->data, destp=ri->data,
+ endp= ri->data + 3 * ri->w * ri->h;
+ destp < endp;
+ srcp++, destp++) {
+ unsigned char c= *srcp & 0xf0;
+ *destp= c | (c>>4);
+ }
+
+ identify_rgbimage(ri, sunshiner, sunshine, "sunshine widget");
+
+ if (!memcmp(sunshine,"Vessel ",5)) {
+ Rect islandnamer;
+
+ islandnamer.tl.x= cim->w - 1034 + 885;
+ islandnamer.br.x= cim->w - 1034 + 1020;
+ islandnamer.tl.y= 128;
+ islandnamer.br.y= 156;
+
+ ADJUST_BOX(islandnamer,"o",5, 0, MUST, tl,y,+1);
+ ADJUST_BOX(islandnamer,"o",5, cim->h, MUST, br,y,-1);
+
+ ADJUST_BOX(islandnamer,"o",1, 0, MUST, tl,x,+1);
+ ADJUST_BOX(islandnamer,"o",1, cim->w, MUST, br,x,-1);
+
+ debug_rect("islandnamer",0, islandnamer);
+// int larger_islandnamebry= islandnamer.tl.y + 25;
+// MUST(islandnamer.br.y < larger_islandnamebry,
+// MR(islandnamer);MI(larger_islandnamebry));
+// islandnamer.br.y = larger_islandnamebry;
+ debug_rect("islandnamer",1, islandnamer);
+
+ int x,y,i;
+ 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;
+ }
+ }
+ }
+
+ identify_rgbimage(ri, islandnamer, archisland, "island");
+ } else {
+ Rect islandnamer;
+
+ islandnamer.tl.x= (sunshiner.tl.x + sunshiner.br.x) / 2;
+ islandnamer.tl.y= sunshiner.tl.y + 100;
+ islandnamer.br= islandnamer.tl;
+ debug_rect("islandnamer",__LINE__, islandnamer);
+
+ WALK_UNTIL_MUST(islandnamer.tl,y, -1, sunshiner.br.y, 'H');
+ WALK_UNTIL_MUST(islandnamer.tl,x, -1, 0, 'o');
+ WALK_UNTIL_MUST(islandnamer.br,x, +1, cim->w, 'o');
+ debug_rect("islandnamer",__LINE__, islandnamer);
+
+ require_rectangle_r(islandnamer, "O*", __LINE__);
+
+ int rw= RECT_W(islandnamer);
+ ADJUST_BOX(islandnamer,"O",rw-4, cim->h, MUST,br,y,+1);
+ debug_rect("islandnamer",__LINE__, islandnamer);
+
+ islandnamer.br.y += 2;
+ ADJUST_BOX(islandnamer,"O",1, cim->h, MUST,br,y,+1);
+ debug_rect("islandnamer",__LINE__, islandnamer);
+
+ islandnamer.tl.y= islandnamer.br.y-1;
+ islandnamer.br.y= islandnamer.br.y+1;
+ ADJUST_BOX(islandnamer,"*",rw, cim->h, MUST,br,y,+1);
+ ADJUST_BOX(islandnamer,"O",1, cim->w, MUST,tl,x,+1);
+ debug_rect("islandnamer",__LINE__, islandnamer);
+
+ MUST( RECT_H(islandnamer) <= 31, MR(islandnamer));
+
+ Point p;
+ int nspaces=1, might_be_colon=0;
+ uint32_t colon_pattern= 0;
+ p.y=-1;
+
+ for (p.x=islandnamer.br.x; p.x>islandnamer.tl.x; p.x--) {
+ colondebugf("structcolon: x=%4d nsp=%2d mbc=%d cp=%08"PRIx32" ",
+ p.x, nspaces, might_be_colon, colon_pattern);
+
+ uint32_t pattern=0;
+ int runs[32], nruns=0;
+ runs[0]=0; runs[1]=0;
+
+ for (p.y=islandnamer.tl.y; p.y<=islandnamer.br.y; p.y++) {
+ pattern <<= 1;
+ switch (get_p(p)) {
+ case 'O': runs[nruns]++; pattern |= 1u; break;
+ case '*': if (runs[nruns]) { nruns++; runs[nruns]=0; } break;
+ default: abort();
+ }
+ }
+
+ colondebugf(" pat=%08"PRIx32" nruns=%d runs[]={%d,%d..} ",
+ pattern, nruns, runs[0],runs[1]);
+
+ if (!pattern) {
+ if (might_be_colon)
+ /* omg it _is_ a colon */
+ goto colon_found;
+ nspaces++;
+ might_be_colon=0;
+ } else {
+ if (nruns==2 && runs[1]==runs[0]) {
+ if (!nspaces) {
+ if (pattern==colon_pattern)
+ goto ok_might_be_colon;
+ } else if (nspaces>=2) {
+ colon_pattern= pattern;
+ might_be_colon=1;
+ goto ok_might_be_colon;
+ }
+ }
+ might_be_colon=0;
+ ok_might_be_colon:
+ nspaces= 0;
+ }
+ colondebugf(" nsp=%2d mbc=%d\n", nspaces, might_be_colon);
}
+ MUST(!"colon found", MP(p);MR(islandnamer) );
+
+ colon_found:
+ colondebugf(" found\n");
+ islandnamer.br.x= p.x;
+
+ identify_rgbimage(ri, islandnamer, archisland, "island");
}
+
+ char *delim= strstr(archisland," - ");
+ assert(delim);
+ archipelago= masprintf("%.*s", (int)(delim-archisland), archisland);
+ island= masprintf("%s", delim+3);
+
+ free(ri);
}
~yarrgweb / ypp-sc-tools.db-test.git / blobdiff