From 059fab3ad374847da031d23d7e25ee25e888bce0 Mon Sep 17 00:00:00 2001 From: Ian Jackson Date: Sun, 19 Jul 2009 13:52:56 +0100 Subject: [PATCH] Adjust colours early so that we don't have to keep all RGBs in RAM --- pctb/common.h | 1 - pctb/convert.c | 17 +-- pctb/convert.h | 33 ++++- pctb/pages.c | 39 ++++-- pctb/structure.c | 328 +++++++++++++++++++++++++++++------------------ pctb/structure.h | 7 +- pctb/x.gdb | 2 +- 7 files changed, 262 insertions(+), 165 deletions(-) diff --git a/pctb/common.h b/pctb/common.h index 1fb7e11..9d06fab 100644 --- a/pctb/common.h +++ b/pctb/common.h @@ -53,7 +53,6 @@ typedef struct { int w,h; - const struct RgbImage *rgb; char d[]; } CanonImage; diff --git a/pctb/convert.c b/pctb/convert.c index 4fdb063..1f13850 100644 --- a/pctb/convert.c +++ b/pctb/convert.c @@ -36,28 +36,12 @@ const char *get_vardir(void) { return "."; } const char *get_libdir(void) { return "."; } -enum mode { - mf_findwindow= 00001, - mf_screenshot= 00010, - mf_readscreenshot= 00020, - mf_analyse= 00100, - mfm_special= 07000, - - mode_findwindow= 00001, - mode_screenshot= 00011, - mode_analyse= 00120, - mode_showcharset= 01000, - - mode_all= 00111, -}; - enum outmodekind { omk_unset, omk_upload, omk_str, omk_raw, omk_none }; static enum outmodekind o_outmode_kind; static const char *o_outmode_str= 0; -static enum mode o_mode= mode_all; static char *o_screenshot_fn; static const char *o_serv_pctb, *o_serv_dict_fetch, *o_serv_dict_submit; @@ -68,6 +52,7 @@ int o_quiet; static pid_t screenshot_compressor=-1; +enum mode o_mode= mode_all; enum flags o_flags= ff_charset_allowedit | ff_dict_fetch|ff_dict_submit|ff_dict_pirate; diff --git a/pctb/convert.h b/pctb/convert.h index 0dc2255..1dcb8a2 100644 --- a/pctb/convert.h +++ b/pctb/convert.h @@ -69,10 +69,21 @@ static inline Rgb ri_rgb(const RgbImage *ri, int x, int y) { /*----- from structure.c -----*/ -void find_structure(CanonImage *im, int *max_relevant_y_r, +typedef struct PageStruct PageStruct; + +void find_structure(const CanonImage *im, + PageStruct **pagestruct_r, /* caller must free() */ + int *max_relevant_y_r, Point *commod_focus_point_r, Point *commod_page_point_r, Point *commod_focuslast_point_r); +void store_current_page(CanonImage *ci /*pointer saved*/, + PageStruct *pagestruct, /*freed*/ + RgbImage *rgb /*freed*/); +void adjust_colours(CanonImage *ci, const RgbImage *rgb); + +void select_page(int page); + Rect find_sunshine_widget(void); void canon_colour_prepare(void); @@ -122,6 +133,22 @@ enum flags { }; extern enum flags o_flags; +enum mode { + mf_findwindow= 00001, + mf_screenshot= 00010, + mf_readscreenshot= 00020, + mf_analyse= 00100, + mfm_special= 07000, + + mode_findwindow= 00001, + mode_screenshot= 00011, + mode_analyse= 00120, + mode_showcharset= 01000, + + mode_all= 00111, +}; +extern enum mode o_mode; + extern const char *o_ocean, *o_pirate; extern int o_quiet; @@ -134,7 +161,9 @@ void take_screenshots(void); void take_one_screenshot(void); #define MAX_PAGES 1000 -extern CanonImage *page_images[MAX_PAGES]; + +extern const CanonImage *page_images[MAX_PAGES]; +extern const RgbImage *page0_rgbimage; extern int npages; extern const char *ocean, *pirate; diff --git a/pctb/pages.c b/pctb/pages.c index 0c0a79c..95a1c46 100644 --- a/pctb/pages.c +++ b/pctb/pages.c @@ -41,9 +41,6 @@ #include #include -CanonImage *page_images[MAX_PAGES]; -int npages; - const char *ocean, *pirate; static XWindowAttributes attr; @@ -355,7 +352,7 @@ static void wait_for_stability(Snapshot **output, " last_input=%f previously=%p `%s'\n", last_input, previously, doing); - double min_interval= 0.025; /*us*/ + double min_interval= 0.025; for (;;) { progress_spinner("%s",doing); @@ -520,21 +517,25 @@ static void set_focus_commodity(void) { debugf("PAGING raise_and_set_focus done.\n"); } -static CanonImage *convert_page(const Snapshot *sn) { +static CanonImage *convert_page(const Snapshot *sn, RgbImage **rgb_r) { CanonImage *im; + RgbImage *rgb; fwrite_ppmraw(screenshot_file, sn); const unsigned char *pixel= sn->data; - CANONICALISE_IMAGE(im, sn->w, sn->h, { + CANONICALISE_IMAGE(im, sn->w, sn->h, rgb, { r= *pixel++; g= *pixel++; b= *pixel++; }); - + sysassert(!ferror(screenshot_file)); sysassert(!fflush(screenshot_file)); + if (rgb_r) *rgb_r= rgb; + else free(rgb); + return im; } @@ -546,8 +547,8 @@ static void prepare_ypp_client(void) { raise_and_get_details(); wait_for_stability(¤t,0,0, "checking current YPP client screen..."); - test= convert_page(current); - find_structure(test, &max_relevant_y, + test= convert_page(current,0); + find_structure(test,0, &max_relevant_y, &commod_focus_point, &commod_page_point, &commod_focuslast_point); @@ -577,6 +578,16 @@ static void prepare_ypp_client(void) { free_snapshot(&status); } +static void convert_store_page(Snapshot *current) { + RgbImage *rgb; + CanonImage *ci; + PageStruct *pstruct; + + ci= convert_page(current,&rgb); + find_structure(ci,&pstruct, 0,0,0,0); + store_current_page(ci,pstruct,rgb); +} + void take_screenshots(void) { Snapshot *current=0, *last=0; @@ -589,19 +600,19 @@ void take_screenshots(void) { /* now to actually page down */ for (;;) { - debugf("paging page %d\n",npages); + debugf("page %d paging\n",npages); if (!(npages < MAX_PAGES)) fatal("Paging down seems to generate too many pages - max is %d.", MAX_PAGES); - - page_images[npages]= convert_page(current); + + convert_store_page(current); free_snapshot(&last); last=current; current=0; debugf("PAGING page %d converted\n",npages); wait_for_stability(¤t,last, 0, - "collecting screenshot of page %d...", + "page %d collecting...", npages+1); if (npages && /* first pagedown doesn't do much */ @@ -628,7 +639,7 @@ void take_one_screenshot(void) { prepare_ypp_client(); wait_for_stability(¤t,0,0, "taking screenshot..."); - page_images[0]= convert_page(current); + convert_store_page(current); npages= 1; progress_log("collected single screenshot."); } diff --git a/pctb/structure.c b/pctb/structure.c index ad95cb5..8ab74ee 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} @@ -40,17 +35,37 @@ DEBUG_DEFINE_DEBUGF(struct) #define TEXT_COLUMNS 2 #define MAX_COLUMNS 7 -static Rect mainr; -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 */ char c; /* canonical */ @@ -119,6 +134,9 @@ void canon_colour_prepare(void) { } } +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, @@ -142,7 +160,7 @@ static void mustfail2(void) { } #define MUST(x, ifnot) do{ \ - if (!(x)) { \ + if (__builtin_expect(!(x), 0)) { \ mustfail1(__FILE__,__LINE__,#x); \ ifnot; \ mustfail2(); \ @@ -255,12 +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)) { @@ -276,44 +297,44 @@ void find_structure(CanonImage *im, int *max_relevant_y_r, } Point mainr_tl= START_MAIN; - mainr.tl= mainr_tl; - WALK_UNTIL_MUST(mainr.tl, y,-1, whole.tl.y, ' '); - mainr.br= mainr.tl; + 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(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, '*'); + 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= s.mr.tl; \ csb_p.x++; csb_p.y++; \ - csb_p.xy= mainr.tlbr.xy; \ + 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); Rect updown= {START_MAIN,START_MAIN}; const int chkw= 100; @@ -321,46 +342,47 @@ void find_structure(CanonImage *im, int *max_relevant_y_r, updown.br.y++; debug_rect("updown",__LINE__,updown); - ADJUST_BOX(updown, "+", >=,chkw, mainr.tl.y, MUST, tl,y,-1); + 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; - ADJUST_BOX(updown, "+*",>=,chkw, mainr.tl.y-1, MUST, tl,y,-1); + ADJUST_BOX(updown, "+*",>=,chkw, s.mr.tl.y-1, MUST, tl,y,-1); debug_rect("updown",__LINE__,updown); - commbasey= updown.tl.y + 1; - comminty= updown.br.y - updown.tl.y; + s.commbasey= updown.tl.y + 1; + s.comminty= updown.br.y - updown.tl.y; - Rect across= {{ mainr.tl.x - 1, commbasey }, - { mainr.tl.x, commbasey + comminty-2 }}; + Rect across= {{ s.mr.tl.x - 1, s.commbasey }, + { s.mr.tl.x, s.commbasey + s.comminty-2 }}; int colno=0; for (;;) { #define LIMIT_QUITEQ(cond,mp) { if (!(cond)) break; } debug_rect("across",colno*1000000+__LINE__, across); - ADJUST_BOX(across, "+",>=,comminty-1, mainr.br.x, LIMIT_QUITEQ, br,x,+1); + 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(mainr);MI(commbasey); ); + MUST( colno < MAX_COLUMNS, + MI(colno);MR(across);MR(s.mr);MI(s.commbasey); ); int colrx= across.br.x-1; - if (colrx >= mainr.br.x) colrx= mainr.br.x; + 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.br.x >= mainr.br.x) + if (across.br.x >= s.mr.br.x) break; - REQUIRE_RECTANGLE(across.br.x,mainr.tl.y, across.br.x,mainr.br.y, "+"); + 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);MR(across); ); + MUST( colno >= MIN_COLUMNS, MI(colno);MR(s.mr);MR(across); ); const int pagerh= 6; - Rect pager= {{ mainr.br.x, mainr.br.y - (pagerh-1) }, - { mainr.br.x + 1, mainr.br.y }}; + 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); @@ -372,29 +394,35 @@ void find_structure(CanonImage *im, int *max_relevant_y_r, 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, mainr.tl.y,LIMIT_QUITEQ, tl,y,-1); + 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= mainr.tl; + *commod_focus_point_r= s.mr.tl; commod_focus_point_r->x += 10; - commod_focus_point_r->y += comminty/3; + commod_focus_point_r->y += s.comminty/3; } if (commod_focuslast_point_r) { - *commod_focuslast_point_r= mainr.br; + *commod_focuslast_point_r= s.mr.br; commod_focuslast_point_r->x -= 10; - commod_focuslast_point_r->y -= comminty/3; + 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; @@ -463,6 +491,10 @@ static void file_read_image_ppm(FILE *f) { struct pam inpam; unsigned char rgb_buf[3]; CanonImage *im; + RgbImage *rgb; + PageStruct *pstruct; + + progress("page %d reading...",npages); pnm_readpaminit(f, &inpam, sizeof(inpam)); if (!(inpam.maxval == 255 && @@ -470,7 +502,7 @@ 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, { + CANONICALISE_IMAGE(im, inpam.width, inpam.height, rgb, { int rr= fread(&rgb_buf,1,3,f); sysassert(!ferror(f)); if (rr!=3) fatal("PNM screenshot(s) file ends unexpectedly"); @@ -485,12 +517,25 @@ 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,rgb); + npages++; +} + +void store_current_page(CanonImage *ci, PageStruct *pstruct, RgbImage *rgb) { + assert(ci==cim); + progress("page %d condensing...",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) { progress("reading screenshot..."); - file_read_image_ppm(screenshot_file); progress_log("read screenshot."); } @@ -508,15 +553,15 @@ 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); } -static double find_aa_density(const RgbImage *ri, Point p, long background, - long foreground, int fg_extra) { +static inline double find_aa_density(const RgbImage *ri, + Point p, long background, + long foreground, int fg_extra) { Rgb here= ri_rgb(ri, p.x, p.y); double alpha[3], alpha_mean=0; @@ -555,37 +600,84 @@ static double find_aa_density(const RgbImage *ri, Point p, long background, 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; - +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; colnobr.x, cell->br.y); - memcpy(chanbg, RI_PIXEL(ri, cell->br.x, cell->br.y), 3); + background= ri_rgb(ri, cell.br.x, cell.br.y); + memcpy(chanbg, RI_PIXEL(ri, cell.br.x, cell.br.y), 3); - FOR_P_RECT(p,*cell) { + FOR_P_RECT(p,cell) { const unsigned char *here_pixel= RI_PIXEL(ri, p.x, p.y); int i; for (i=0; i<3; i++) { @@ -595,10 +687,10 @@ static void find_table_entry(Rect commod, int colno, Rect *cell) { else if (here > chanbg[i]) light_count += (here - chanbg[i])/4 + 1; } } - long total_count= RECT_W(*cell) * RECT_H(*cell) * 3; + 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) @@ -615,60 +707,46 @@ static void find_table_entry(Rect commod, int colno, Rect *cell) { fg_extra= -1; } else { MUST( !"tell light from dark", - MR(*cell);MIL(total_count);MIL(bg_count); + 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, + 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) { + FOR_P_RECT(p,cell) { double alpha= find_aa_density(ri,p,background,foreground,fg_extra); int here_int= floor((AAMAXVAL+1)*alpha); 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; colnorgb->w, - page_images[0]->rgb->h); - memcpy(ri->data, page_images[0]->rgb->data, ri->w * ri->h * 3); + RgbImage *ri= alloc_rgb_image(rgbsrc->w, 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, + for (srcp= rgbsrc->data, destp=ri->data, endp= ri->data + 3 * ri->w * ri->h; destp < endp; srcp++, destp++) { diff --git a/pctb/structure.h b/pctb/structure.h index 84fdcae..73f6571 100644 --- a/pctb/structure.h +++ b/pctb/structure.h @@ -50,7 +50,7 @@ static inline char canon_lookup_colour(unsigned char r, return blues->blue2[b]; } -#define CANONICALISE_IMAGE(im,w,h, COMPUTE_RGB) do{ \ +#define CANONICALISE_IMAGE(im,w,h,rgb_save, COMPUTE_RGB) do{ \ /* compute_rgb should be a number of statements, or \ * a block, which assigns to \ * Rgb rgb; \ @@ -61,15 +61,14 @@ static inline char canon_lookup_colour(unsigned char r, * each pixel in reading order. \ */ \ (im)= alloc_canon_image((w), (h)); \ - RgbImage *rgb_save; \ - (im)->rgb= rgb_save= alloc_rgb_image((w), (h)); \ + (rgb_save)= alloc_rgb_image((w), (h)); \ \ int x,y; \ for (y=0; y<(h); y++) { \ for (x=0; x<(w); x++) { \ unsigned char r,g,b; \ COMPUTE_RGB; \ - CANONIMG_ALSO_STORERGB(rgb_save); \ + CANONIMG_ALSO_STORERGB((rgb_save)); \ (im)->d[y*(w) + x]= canon_lookup_colour(r,g,b); \ } \ if (DEBUGP(rect)) { \ diff --git a/pctb/x.gdb b/pctb/x.gdb index 0f6bf86..614122f 100644 --- a/pctb/x.gdb +++ b/pctb/x.gdb @@ -1,4 +1,4 @@ file ypp-commodities -set args -Drect 2>u --edit-charset --raw-tsv --same --screenshot-file t.ppm >raw.tsv +set args -Drect 2>u --edit-charset --raw-tsv --same >raw.tsv break mustfail2 run -- 2.30.2