*~
t
+core
+
pctb/*.o
pctb/t.*
pctb/u.*
+
+pctb/#*#.*
pctb/convert
pctb/x-manip-window
all: convert x-manip-window
-CONVERT_OBJS= convert.o ocr.o pages.o
+CONVERT_OBJS= convert.o ocr.o pages.o structure.o
convert: $(CONVERT_OBJS) -lnetpbm -lXtst -lX11
-$(CONVERT_OBJS): ocr.h
+$(CONVERT_OBJS): ocr.h convert.h structure.h
x-manip-window: -lXtst -lX11
-#include "ocr.h"
+#include "convert.h"
-static CanonImage *cim;
void debug_flush(void) {
eassert(!fflush(debug));
eassert(!ferror(debug));
}
-typedef struct {
- int x, y;
-} Point;
-typedef struct { /* both inclusive */
- Point tl;
- Point br;
-} Rect;
-
-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 START_MAIN {200,200}
-#define MIN_COLUMNS 6
-#define INTERESTING_COLUMNS 6
-#define TEXT_COLUMNS 2
-#define MAX_COLUMNS 7
-
-static Rect mainr = { START_MAIN,START_MAIN };
-static int commbasey, comminty;
-static int colrightx[INTERESTING_COLUMNS];
-static int text_h;
-static OcrReader *rd;
-
-const CanonColourInfo canoncolourinfos[]= {
- { 0x475A5E, '*' }, /* edge */
- { 0x2C5F7A, '*' }, /* edge just under box heading shadow */
- { 0x7D9094, '+' }, /* interbox */
-
- { 0xBDC5BF, ' ' }, /* background - pale Sugar cane, etc. */
- { 0xADB5AF, ' ' }, /* background - dark */
- { 0xC7E1C3, ' ' }, /* background - pale Swill, etc. */
- { 0xB5CFB1, ' ' }, /* background - dark */
- { 0xD6CEB0, ' ' }, /* background - pale Madder, etc. */
- { 0xC8C0A2, ' ' }, /* background - dark */
- { 0xE0E1D3, ' ' }, /* background - pale Lorandite, etc. */
- { 0xD0D1C3, ' ' }, /* background - dark */
- { 0xE5E6C1, ' ' }, /* background - pale Cloth */
- { 0xD7D8B3, ' ' }, /* background - dark */
- { 0xEDDED9, ' ' }, /* background - pale Dye */
- { 0xDACBC6, ' ' }, /* background - dark */
- { 0xD3DEDF, ' ' }, /* background - pale Paint */
- { 0xC5D0D1, ' ' }, /* background - dark */
- { 0xDCD1CF, ' ' }, /* background - pale Enamel */
- { 0xCEC3C1, ' ' }, /* background - dark */
- { 0xF3F6F5, ' ' }, /* background - pale fruit */
- { 0xE2E7E5, ' ' }, /* background - dark */
-
- { 0x000000, 'o' }, /* foreground */
- { 0xD4B356, ' ' }, /* background (cursor) */
- { 0xFFFFFF, 'o' }, /* foreground (cursor) */
-
- { 0x5B93BF, '_' }, /* selector dropdown background */
- { 0xD7C94F, 'X' }, /* selector dropdown foreground */
- { 0,0 }
-};
-
-static void require_rectangle(int tlx, int tly, int brx, int bry,
- const char *ok) {
- int x,y;
- for (x=tlx; x<=brx; x++)
- for (y=tly; y<=bry; y++) {
- int c= get(x,y);
- assert(strchr(ok,c));
- }
-}
-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);
-}
+const char *get_vardir(void) { return "."; }
-static void debug_rect(const char *what, int whati, Rect rr) {
-#ifdef DEBUG_RECTANGLES
- int y,r,w;
- fprintf(debug, "%s %d: %d,%d..%d,%d:\n", what, whati,
- rr.tl.x,rr.tl.y, rr.br.x,rr.br.y);
- w= rr.br.x - rr.tl.x + 1;
- for (y=rr.tl.y; y<=rr.br.y; y++) {
- fprintf(debug, "%4d%*s|", y, rr.tl.x,"");
- r= fwrite(cim->d + y*cim->w + rr.tl.x, 1, w, debug);
- eassert(r==w);
- fputc('|',debug);
- fputc('\n',debug);
- }
-#endif
- debug_flush();
-}
-#define WALK_UNTIL(point,coord,increm,last,edge) \
- for (;;) { \
- if ((point).coord == (last)+(increm)) break; \
- if (get_p((point)) == (edge)) { (point).coord -= (increm); break; } \
- (point).coord += (increm); \
+static enum { mode_all=03, mode_screenshot=01, mode_analyse=02 }
+ o_mode= mode_all;
+static char *o_screenshots_fn;
+static int o_single_page;
+
+FILE *screenshots_file;
+
+int main(int argc, char **argv) {
+ const char *arg;
+ int r;
+
+ while ((arg=*++argv)) {
+ if (!strcmp(arg,"--screenshots-only"))
+ o_mode= mode_screenshot;
+ else if (!strcmp(arg,"--analyse-only"))
+ o_mode= mode_analyse;
+ else if (!strcmp(arg,"--single-page"))
+ o_single_page= 1;
+ else if (!strcmp(arg,"--screenshots-file"))
+ eassert( o_screenshots_fn= *++argv );
+ else if (!strcmp(arg,"--window-id")) {
+ char *ep;
+ eassert((arg=*++argv));
+ unsigned long windowid= strtoul(arg,&ep,0);
+ eassert(!*ep);
+ set_yppclient_window(windowid);
+ } else
+ eassert(!"bad option");
}
-
-#define WALK_UNTIL_MUST(point,coord,increm,last,edge) \
- do { \
- WALK_UNTIL(point,coord,increm,last,edge); \
- eassert((point).coord != (last)+(increm)); \
- } while(0)
-
-void find_structure(CanonImage *im) {
- cim= im;
- Rect whole = { {0,0}, {cim->w-1,cim->h-1} };
-
- 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; \
- } \
- } while(0)
-
- debug_rect("mainr",0, mainr);
-
- 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);
-
- Point up = START_MAIN;
- WALK_UNTIL_MUST(up, y,-1, mainr.tl.y, '+');
-
- Point down = START_MAIN;
- down.y++;
- WALK_UNTIL_MUST(down, y,+1, mainr.br.y, '+');
-
-#ifdef DEBUG_RECTANGLES
- 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);
+ if (!o_screenshots_fn) {
+ r= asprintf(&o_screenshots_fn,"%s/#pages#.ppm",get_vardir());
+ eassert(r>=0); eassert(o_screenshots_fn);
}
-#endif
-
- commbasey= up.y;
- comminty= down.y - up.y + 2;
-
- Point across= { mainr.tl.x, commbasey };
- int colno=0;
- for (;;) {
- eassert(get_p(across) != '+');
- WALK_UNTIL(across, x,+1, mainr.br.x, '+');
- eassert(colno < MAX_COLUMNS);
- int colrx= across.x;
- if (colrx > mainr.br.x) colrx= mainr.br.x;
- if (colno < INTERESTING_COLUMNS)
- colrightx[colno]= colrx;
-
- colno++;
-
- if (across.x >= mainr.br.x-1)
- break;
- across.x++;
- require_rectangle(across.x,mainr.tl.y, across.x,mainr.br.y, "+");
- across.x++;
+ if (o_mode & mode_screenshot) {
+ screenshot_startup();
+ find_yppclient_window();
+ screenshots_file= fopen(o_screenshots_fn, "w"); eassert(screenshots_file);
+ if (o_single_page)
+ take_one_screenshot();
+ else
+ take_screenshots();
+ } else {
+ screenshots_file= fopen(o_screenshots_fn, "r"); eassert(screenshots_file);
+ read_screenshots();
}
- eassert(colno >= MIN_COLUMNS);
-
- text_h = comminty - 1;
-}
-
-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; }
- if (rr->tl.y > mainr.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)
- require_rectangle(rr->tl.x,rr->br.y+1, rr->br.x,rr->br.y+1, "+");
-
- rr->tl.x= mainr.tl.x;
- rr->br.x= mainr.br.x;
-}
-
-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");
-}
-
-CanonImage *alloc_canon_image(int w, int h) {
- CanonImage *im= malloc(sizeof(CanonImage) + w*h);
- eassert(im);
- im->w= w;
- im->h= h;
- memset(im->d,'?',w*h);
- return im;
-}
-
-CanonImage *file_read_image(FILE *f) {
- struct pam inpam;
- unsigned char rgb_buf[3];
- CanonImage *im;
-
- pnm_readpaminit(f, &inpam, sizeof(inpam));
- eassert(inpam.maxval == 255);
- eassert(inpam.bytes_per_sample == 1);
-
- CANONICALISE_IMAGE(im, inpam.width, inpam.height, {
- r= fread(&rgb_buf,1,3,f); eassert(r==3);
-
- rgb=
- ((unsigned long)rgb_buf[0]<<16) |
- ((unsigned long)rgb_buf[1]<<8) |
- (rgb_buf[2]);
- });
-
- return im;
-}
-
-static void load_image_and_canonify(void) {
- cim= file_read_image(stdin);
-}
-
-static void ocr_rectangle(Rect r, const OcrCellType ct) {
- OcrResultGlyph *results, *res;
-
- int w= r.br.x - r.tl.x + 1;
- 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) {
- switch (get(x+r.tl.x, y+r.tl.y)) {
- case ' ': break;
- case 'o': cx |= rv; break;
- default: eassert(!"wrong pixel");
- }
- }
- cols[x]= cx;
- }
- cols[w]= 0;
-
- results= ocr(rd,ct,w,cols);
- printf("YES! \"");
- for (res=results; res->s; res++)
- printf("%s",res->s);
- printf("\"\n");
- eassert(!ferror(stdout));
- eassert(!fflush(stdout));
-}
-
-int main_test(void) {
- Rect thisr, entryr;
- int tryrect, colno;
-
- load_image_and_canonify();
- find_structure(cim);
- rd= ocr_init(text_h);
-
- for (tryrect= +cim->h; tryrect >= -cim->h; tryrect--) {
- find_commodity(tryrect, &thisr);
- if (thisr.tl.x < 0)
- continue;
- debug_rect("commod",tryrect, thisr);
-
- for (colno=0; colno<MIN_COLUMNS; colno++) {
- find_table_entry(thisr,colno,&entryr);
- ocr_rectangle(entryr,
- colno<TEXT_COLUMNS
- ? &ocr_celltype_text
- : &ocr_celltype_number);
- }
+ if (o_mode & mode_analyse) {
+ analyse();
+ //output_tsv();
}
return 0;
}
-
-const char *get_vardir(void) { return "."; }
--- /dev/null
+#ifndef CONVERT_H
+#define CONVERT_H
+
+
+#include "ocr.h"
+
+#include <pam.h>
+#include <time.h>
+#include <limits.h>
+
+#include <sys/time.h>
+
+
+/*----- from structure.c -----*/
+
+void find_structure(CanonImage *im);
+CanonImage *file_read_image_ppm(FILE *f);
+void read_screenshots(void);
+void analyse(void);
+
+/*----- from convert.c -----*/
+
+extern FILE *screenshots_file;
+
+/*----- from pages.c -----*/
+
+void screenshot_startup(void);
+void set_yppclient_window(unsigned long wul);
+void find_yppclient_window(void);
+void take_screenshots(void);
+void take_one_screenshot(void);
+
+#define MAX_PAGES 100
+extern CanonImage *page_images[MAX_PAGES];
+extern int npages;
+
+
+#endif /*CONVERT_H*/
+/*
+ * ocr.c forms a mostly-self-contained bit
+ * so we put its declarations in this separate file
+ */
+
#ifndef OCR_H
#define OCR_H
-#define DEBUG_RECTANGLES
+// #define DEBUG_RECTANGLES
// #define DEBUG_OCR
-
#define _GNU_SOURCE
-#include <pam.h>
-#include <stdint.h>
-#include <inttypes.h>
-#include <assert.h>
#include <string.h>
-#include <stdlib.h>
#include <stdio.h>
+#include <errno.h>
+#include <assert.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <unistd.h>
#include <stdarg.h>
-#include <time.h>
-#include <limits.h>
+#include <inttypes.h>
#include <sys/types.h>
#include <sys/wait.h>
-#include <sys/time.h>
typedef struct {
char d[];
} CanonImage;
-
typedef uint32_t Pixcol;
#define PSPIXCOL(priscan) priscan##32
unsigned ctxmap; /* match context index */
} OcrResultGlyph;
+
typedef const struct OcrCellTypeInfo *OcrCellType;
extern const struct OcrCellTypeInfo ocr_celltype_text;
extern const struct OcrCellTypeInfo ocr_celltype_number;
+
typedef struct OcrReader OcrReader;
OcrReader *ocr_init(int h);
* array is valid until next call to ocr()
*/
-void debug_flush(void);
-void find_structure(CanonImage *im);
+/*----- debugging arrangements, rather contingent -----*/
+
+void debug_flush(void);
#define eassert assert
-#define debug stdout
+#define debug stderr
const char *get_vardir(void);
-CanonImage *file_read_image(FILE *f);
-int main_test(void);
-
-#define MAX_PAGES 100
-extern CanonImage *page_images[MAX_PAGES];
-extern int npages;
-
-
-typedef struct {
- unsigned long rgb; /* on screen */
- char c; /* canonical */
-} CanonColourInfo;
-
-extern const CanonColourInfo canoncolourinfos[];
-
-CanonImage *alloc_canon_image(int w, int h);
-
-#ifdef DEBUG_RECTANGLES
-# define CANIMG_DEBUG_RECTANGLE_1LINE(im,w,h) \
- fprintf(debug, "%4d ",y); \
- r= fwrite(im->d + y*w, 1,w, debug); \
- eassert(r==w); \
- fputc('\n',debug);
-#else
-# define CANIMG_DEBUG_RECTANGLE_1LINE(im,y,h) /* nothing */
-#endif
-
-#define CANONICALISE_IMAGE(im,w,h, COMPUTE_RGB) do{ \
- /* compute_rgb should be a number of statements, or \
- * a block, which assigns to \
- * unsigned long rgb; \
- * given the values of \
- * int x,y; \
- * all of which are anamorphic. Result is stored in im. \
- */ \
- (im)= alloc_canon_image((w), (h)); \
- \
- int x,y,r; \
- for (y=0; y<(h); y++) { \
- for (x=0; x<(w); x++) { \
- const CanonColourInfo *cci; \
- unsigned long rgb; \
- COMPUTE_RGB; \
- for (cci=canoncolourinfos; cci->c; cci++) \
- if (cci->rgb == rgb) { \
- (im)->d[y*(w) + x]= cci->c; \
- break; \
- } \
- } \
- CANIMG_DEBUG_RECTANGLE_1LINE((im),(w),(h)) \
- } \
- debug_flush(); \
- }while(0)
-
-
#endif /*OCR_H*/
/*
*/
-#include "ocr.h"
+#include "structure.h"
#include <X11/Xlib.h>
#include <X11/extensions/XTest.h>
return XKeysymToKeycode(disp,sym);
}
-#if 0
-static void check_exitstatus(int st) {
- eassert(WIFEXITED(st));
- eassert(!WEXITSTATUS(st));
-}
-
-static void check_pclose(FILE *f, char *cmd) {
- int r;
- eassert(!ferror(f));
- r= fgetc(f); eassert(r==EOF); eassert(feof(f));
- r= pclose(f); eassert(r>=0); check_exitstatus(r);
- free(cmd);
-}
-
-static CanonImage *screenshot_now(void) {
- char *cmd;
- CanonImage *ci;
- int r;
-
- r= asprintf(&cmd, "xwd -silent -id 0x%lx | xwdtopnm", (unsigned long)id);
- eassert(r>=0);
- FILE *f= popen(cmd,"r"); eassert(f);
- ci= file_read_image(f);
- check_pclose(f, cmd);
- return ci;
-}
-#endif
-
-static void screenshot_startup(void) {
+void screenshot_startup(void) {
int r;
disp= XOpenDisplay(0); eassert(disp);
r= gettimeofday(&tv_startup,0); eassert(!r);
}
-#if 0
-static CanonImage *single_page(void) {
- int r;
- r= XRaiseWindow(disp, id); eassert(r);
- r= XSync(disp, False); eassert(r);
- return screenshot_now();
-}
-#endif
-
/*---------- pager ----------*/
typedef XImage Snapshot;
-//static size_t snapshot_alloc= 1024;
static double last_input;
static const double min_update_allowance= 0.25;
fprintf(stderr,"PAGING raise_and_set_focus done.\n");
}
+#define SAMPLEMASK 0xfful
+
typedef struct {
- unsigned long mask;
int lshift, rshift;
} ShMask;
-static void compute_shift_mask(ShMask *sm, int targshift,
- unsigned long ximage_mask) {
- unsigned long below;
-
+static void compute_shift_mask(ShMask *sm, unsigned long ximage_mask) {
sm->lshift= 0;
sm->rshift= 0;
- sm->mask= 0xfful << targshift;
- below= ~0ul << targshift;
for (;;) {
- if (ximage_mask < sm->mask) {
+ if (ximage_mask <= (SAMPLEMASK>>1)) {
sm->lshift++; ximage_mask <<= 1;
- } else if ((ximage_mask & ~below) > sm->mask) {
+ } else if (ximage_mask > SAMPLEMASK) {
sm->rshift++; ximage_mask >>= 1;
} else {
break;
ShMask shiftmasks[3];
CanonImage *im;
-#define COMPUTE_SHIFT_MASK(ix, targshift, rgb) \
- compute_shift_mask(&shiftmasks[ix], targshift, sn->rgb##_mask)
- COMPUTE_SHIFT_MASK(0, 16, red);
- COMPUTE_SHIFT_MASK(1, 8, green);
- COMPUTE_SHIFT_MASK(2, 0, blue);
+ fprintf(screenshots_file,
+ "P6\n"
+ "%d %d\n"
+ "255\n", sn->width, sn->height);
+
+#define COMPUTE_SHIFT_MASK(ix, rgb) \
+ compute_shift_mask(&shiftmasks[ix], sn->rgb##_mask)
+ COMPUTE_SHIFT_MASK(0, red);
+ COMPUTE_SHIFT_MASK(1, green);
+ COMPUTE_SHIFT_MASK(2, blue);
CANONICALISE_IMAGE(im, sn->width, sn->height, {
long xrgb= XGetPixel(sn, x, y);
int i;
rgb= 0;
- for (i=0; i<3; i++)
- rgb |= ((xrgb << shiftmasks[i].lshift)
- >> shiftmasks[i].rshift) & shiftmasks[i].mask;
+ for (i=0; i<3; i++, rgb <<= 8) {
+ unsigned long sample=
+ ((xrgb << shiftmasks[i].lshift)
+ >> shiftmasks[i].rshift) & SAMPLEMASK;
+ rgb |= sample;
+ fputc(sample, screenshots_file);
+ }
});
+ eassert(!fflush(screenshots_file));
+
return im;
}
-static void read_pages(void) {
+void take_screenshots(void) {
Snapshot *current=0, *last=0;
CanonImage *test;
fprintf(stderr,"PAGING all done.\n");
}
-int main(int argc, char **argv) {
- screenshot_startup();
-
- id= strtoul(*++argv,0,0);
+void take_one_screenshot(void) {
+ Snapshot *current=0;
+
+ raise_and_get_details();
+ sync_after_input();
+ snapshot(¤t);
+ page_images[0]= convert_page(current);
+ npages= 1;
+}
- read_pages();
- return 0;
+void set_yppclient_window(unsigned long wul) {
+ id= wul;
}
+void find_yppclient_window(void) {
+ if (id) return;
+ eassert(!"finding client window NYI");
+}
--- /dev/null
+/*
+ */
+
+#include "structure.h"
+
+
+typedef struct {
+ int x, y;
+} Point;
+
+typedef struct { /* both inclusive */
+ Point tl;
+ Point br;
+} Rect;
+
+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 START_MAIN {200,200}
+#define MIN_COLUMNS 6
+#define INTERESTING_COLUMNS 7
+#define TEXT_COLUMNS 2
+#define MAX_COLUMNS 7
+
+static Rect mainr = { START_MAIN,START_MAIN };
+static int commbasey, comminty;
+static int colrightx[INTERESTING_COLUMNS];
+static int text_h=-1, columns=-1;
+static OcrReader *rd;
+
+const CanonColourInfo canoncolourinfos[]= {
+ { 0x475A5E, '*' }, /* edge */
+ { 0x2C5F7A, '*' }, /* edge just under box heading shadow */
+ { 0x7D9094, '+' }, /* interbox */
+
+ { 0xBDC5BF, ' ' }, /* background - pale Sugar cane, etc. */
+ { 0xADB5AF, ' ' }, /* background - dark */
+ { 0xC7E1C3, ' ' }, /* background - pale Swill, etc. */
+ { 0xB5CFB1, ' ' }, /* background - dark */
+ { 0xD6CEB0, ' ' }, /* background - pale Madder, etc. */
+ { 0xC8C0A2, ' ' }, /* background - dark */
+ { 0xE0E1D3, ' ' }, /* background - pale Lorandite, etc. */
+ { 0xD0D1C3, ' ' }, /* background - dark */
+ { 0xE5E6C1, ' ' }, /* background - pale Cloth */
+ { 0xD7D8B3, ' ' }, /* background - dark */
+ { 0xEDDED9, ' ' }, /* background - pale Dye */
+ { 0xDACBC6, ' ' }, /* background - dark */
+ { 0xD3DEDF, ' ' }, /* background - pale Paint */
+ { 0xC5D0D1, ' ' }, /* background - dark */
+ { 0xDCD1CF, ' ' }, /* background - pale Enamel */
+ { 0xCEC3C1, ' ' }, /* background - dark */
+ { 0xF3F6F5, ' ' }, /* background - pale fruit */
+ { 0xE2E7E5, ' ' }, /* background - dark */
+
+ { 0x000000, 'o' }, /* foreground */
+ { 0xD4B356, ' ' }, /* background (cursor) */
+ { 0xFFFFFF, 'o' }, /* foreground (cursor) */
+
+ { 0x5B93BF, '_' }, /* selector dropdown background */
+ { 0xD7C94F, 'X' }, /* selector dropdown foreground */
+ { 0,0 }
+};
+
+static void require_rectangle(int tlx, int tly, int brx, int bry,
+ const char *ok) {
+ int x,y;
+ for (x=tlx; x<=brx; x++)
+ for (y=tly; y<=bry; y++) {
+ int c= get(x,y);
+ assert(strchr(ok,c));
+ }
+}
+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 debug_rect(const char *what, int whati, Rect rr) {
+#ifdef DEBUG_RECTANGLES
+ int y,r,w;
+ fprintf(debug, "%s %d: %d,%d..%d,%d:\n", what, whati,
+ rr.tl.x,rr.tl.y, rr.br.x,rr.br.y);
+ w= rr.br.x - rr.tl.x + 1;
+ for (y=rr.tl.y; y<=rr.br.y; y++) {
+ fprintf(debug, "%4d%*s|", y, rr.tl.x,"");
+ r= fwrite(cim->d + y*cim->w + rr.tl.x, 1, w, debug);
+ eassert(r==w);
+ fputc('|',debug);
+ fputc('\n',debug);
+ }
+#endif
+ debug_flush();
+}
+
+#define WALK_UNTIL(point,coord,increm,last,edge) \
+ for (;;) { \
+ if ((point).coord == (last)+(increm)) break; \
+ if (get_p((point)) == (edge)) { (point).coord -= (increm); break; } \
+ (point).coord += (increm); \
+ }
+
+#define WALK_UNTIL_MUST(point,coord,increm,last,edge) \
+ do { \
+ WALK_UNTIL(point,coord,increm,last,edge); \
+ eassert((point).coord != (last)+(increm)); \
+ } while(0)
+
+void find_structure(CanonImage *im) {
+ cim= im;
+
+ Rect whole = { {0,0}, {cim->w-1,cim->h-1} };
+
+ 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; \
+ } \
+ } while(0)
+
+ debug_rect("mainr",0, mainr);
+
+ 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);
+
+ Point up = START_MAIN;
+ WALK_UNTIL_MUST(up, y,-1, mainr.tl.y, '+');
+
+ Point down = START_MAIN;
+ down.y++;
+ WALK_UNTIL_MUST(down, y,+1, mainr.br.y, '+');
+
+#ifdef DEBUG_RECTANGLES
+ 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);
+ }
+#endif
+
+ commbasey= up.y;
+ comminty= down.y - up.y + 2;
+
+ Point across= { mainr.tl.x, commbasey };
+ int colno=0;
+ for (;;) {
+ eassert(get_p(across) != '+');
+ WALK_UNTIL(across, x,+1, mainr.br.x, '+');
+ eassert(colno < MAX_COLUMNS);
+ int colrx= across.x;
+ if (colrx > mainr.br.x) colrx= mainr.br.x;
+ if (colno < INTERESTING_COLUMNS)
+ colrightx[colno]= colrx;
+
+ colno++;
+
+ if (across.x >= mainr.br.x-1)
+ break;
+
+ across.x++;
+ require_rectangle(across.x,mainr.tl.y, across.x,mainr.br.y, "+");
+ across.x++;
+ }
+ eassert(colno >= MIN_COLUMNS);
+
+#define SET_ONCE(var,val) do{ \
+ int v= (val); \
+ if ((var)==-1) (var)= v; \
+ else eassert((var) == v); \
+ }while(0)
+
+ SET_ONCE(columns, colno);
+ SET_ONCE(text_h, comminty - 1);
+}
+
+CanonImage *alloc_canon_image(int w, int h) {
+ CanonImage *im= malloc(sizeof(CanonImage) + w*h);
+ eassert(im);
+ im->w= w;
+ im->h= h;
+ memset(im->d,'?',w*h);
+ return im;
+}
+
+CanonImage *file_read_image_ppm(FILE *f) {
+ struct pam inpam;
+ unsigned char rgb_buf[3];
+ CanonImage *im;
+
+ pnm_readpaminit(f, &inpam, sizeof(inpam));
+ eassert(inpam.maxval == 255);
+ eassert(inpam.bytes_per_sample == 1);
+
+ CANONICALISE_IMAGE(im, inpam.width, inpam.height, {
+ int r= fread(&rgb_buf,1,3,f); eassert(r==3);
+
+ rgb=
+ ((unsigned long)rgb_buf[0]<<16) |
+ ((unsigned long)rgb_buf[1]<<8) |
+ (rgb_buf[2]);
+ });
+
+ return im;
+}
+
+void read_screenshots(void) {
+ int c;
+ while ((c= fgetc(screenshots_file) != EOF)) {
+ ungetc(c, screenshots_file);
+
+ eassert(npages < MAX_PAGES);
+ page_images[npages++]= file_read_image_ppm(screenshots_file);
+ }
+}
+
+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; }
+ if (rr->tl.y > mainr.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)
+ require_rectangle(rr->tl.x,rr->br.y+1, rr->br.x,rr->br.y+1, "+");
+
+ rr->tl.x= mainr.tl.x;
+ rr->br.x= mainr.br.x;
+}
+
+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");
+}
+
+static void ocr_rectangle(Rect r, const OcrCellType ct) {
+ OcrResultGlyph *results, *res;
+
+ int w= r.br.x - r.tl.x + 1;
+ 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) {
+ switch (get(x+r.tl.x, y+r.tl.y)) {
+ case ' ': break;
+ case 'o': cx |= rv; break;
+ default: eassert(!"wrong pixel");
+ }
+ }
+ cols[x]= cx;
+ }
+ cols[w]= 0;
+
+ results= ocr(rd,ct,w,cols);
+ for (res=results; res->s; res++)
+ printf("%s",res->s);
+}
+
+void analyse(void) {
+ Rect thisr, entryr;
+ int page, tryrect, colno;
+
+ for (page=0; page<npages; page++) {
+ find_structure(page_images[page]);
+
+ if (!rd)
+ rd= ocr_init(text_h);
+
+ for (tryrect= +cim->h; 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; colno<columns; colno++) {
+ find_table_entry(thisr,colno,&entryr);
+ fputs(tab, stdout);
+ ocr_rectangle(entryr,
+ colno<TEXT_COLUMNS
+ ? &ocr_celltype_text
+ : &ocr_celltype_number);
+ tab= "\t";
+ }
+ fputs("\n", stdout);
+ eassert(!ferror(stdout));
+ eassert(!fflush(stdout));
+ }
+ }
+}
--- /dev/null
+#ifndef STRUCTURE_H
+#define STRUCTURE_H
+
+
+#include "convert.h"
+
+
+typedef struct {
+ unsigned long rgb; /* on screen */
+ char c; /* canonical */
+} CanonColourInfo;
+
+extern const CanonColourInfo canoncolourinfos[];
+
+CanonImage *alloc_canon_image(int w, int h);
+
+#ifdef DEBUG_RECTANGLES
+# define CANIMG_DEBUG_RECTANGLE_1LINE(im,w,h) \
+ fprintf(debug, "%4d ",y); \
+ int r= fwrite(im->d + y*w, 1,w, debug); \
+ eassert(r==w); \
+ fputc('\n',debug);
+#else
+# define CANIMG_DEBUG_RECTANGLE_1LINE(im,y,h) /* nothing */
+#endif
+
+#define CANONICALISE_IMAGE(im,w,h, COMPUTE_RGB) do{ \
+ /* compute_rgb should be a number of statements, or \
+ * a block, which assigns to \
+ * unsigned long rgb; \
+ * given the values of \
+ * int x,y; \
+ * all of which are anamorphic. Result is stored in im. \
+ * The COMPUTE_RGB is executed exactly once for \
+ * each pixel in reading order. \
+ */ \
+ (im)= alloc_canon_image((w), (h)); \
+ \
+ int x,y; \
+ for (y=0; y<(h); y++) { \
+ for (x=0; x<(w); x++) { \
+ const CanonColourInfo *cci; \
+ unsigned long rgb; \
+ COMPUTE_RGB; \
+ for (cci=canoncolourinfos; cci->c; cci++) \
+ if (cci->rgb == rgb) { \
+ (im)->d[y*(w) + x]= cci->c; \
+ break; \
+ } \
+ } \
+ CANIMG_DEBUG_RECTANGLE_1LINE((im),(w),(h)) \
+ } \
+ debug_flush(); \
+ }while(0)
+
+
+#endif /*STRUCTURE_H*/