-#include <pam.h>
-#include <stdint.h>
-#include <inttypes.h>
-#include <assert.h>
-#include <string.h>
-#include <stdlib.h>
-#define eassert assert
-#define debug stdout
+#include "ocr.h"
typedef struct {
unsigned long rgb; /* on screen */
debug_flush();
}
-typedef uint32_t Pixcol;
-#define PSPIXCOL(priscan) priscan##32
-
-typedef struct {
- Pixcol col;
- struct OCRDatabaseNode *then;
-} OCRDatabaseLink;
-
-#define MAXGLYPHCHRS 3
-
-typedef struct OCRDatabaseNode {
- char s[MAXGLYPHCHRS+1]; /* null-terminated; "" means no match here */
- int nlinks, alinks;
- OCRDatabaseLink *links;
-} OCRDatabaseNode;
-
-#define N_OCR_CONTEXTS 2
-static OCRDatabaseNode ocr_contexts[N_OCR_CONTEXTS];
-
-static void load_ocr_database(void) {
- int ctx,nchrs;
- OCRDatabaseNode *current, *additional;
- char chrs[MAXGLYPHCHRS+1];
- Pixcol cv;
- int r,i,j;
-
- FILE *db= fopen("database","r"); eassert(db);
-
- for (;;) {
- r= fscanf(db, "%d %d", &ctx, &nchrs);
- if (r==EOF) break;
- eassert(r==2);
- eassert(ctx>=0 && ctx<N_OCR_CONTEXTS);
- eassert(nchrs>0 && nchrs<=MAXGLYPHCHRS);
-
- for (i=0; i<nchrs; i++) {
- int c;
- r= fscanf(db, "%x", &c); eassert(r==1);
- eassert(c>0 && c<=255);
- chrs[i]= c;
- }
- chrs[nchrs]= 0;
-
- int twidth;
- r= fscanf(db, "%d", &twidth); eassert(r==1);
- current= &ocr_contexts[ctx];
- for (i=0; i<twidth; i++) {
- r= fscanf(db, "%"PSPIXCOL(SCNx), &cv); eassert(r==1);
- for (j=0; j<current->nlinks; j++)
- if (current->links[j].col == cv) {
- current= current->links[j].then;
- goto found_link;
- }
-
- additional= malloc(sizeof(*additional)); eassert(additional);
- additional->s[0]= 0;
- additional->nlinks= additional->alinks= 0;
- additional->links= 0;
- if (current->nlinks==current->alinks) {
- current->alinks++;
- current->alinks<<=1;
- current->links= realloc(current->links,
- sizeof(*current->links) * current->alinks);
- eassert(current->links);
- }
- current->links[current->nlinks].col= cv;
- current->links[current->nlinks].then= additional;
- current->nlinks++;
- current= additional;
-
- found_link:;
- }
-
- eassert(!current->s[0]);
- strcpy(current->s, chrs);
- }
- eassert(!ferror(db));
- eassert(feof(db));
- fclose(db);
-}
-
static void ocr_rectangle(Rect r) {
+ OcrResultGlyph *results, *res;
+
int w= r.br.x - r.tl.x + 1;
int h= r.br.y - r.tl.y + 1;
Pixcol cols[w+1];
}
cols[w]= 0;
- int nspaces=0;
- int ctx=1,i;
- x=0;
-
- for (;;) {
- if (x>w) break;
-
- if (!cols[x]) {
- nspaces++;
- x++;
- if (nspaces>3) ctx=1;
- continue;
- }
-
- OCRDatabaseNode *current=0, *lastmatch=0;
- int startx=x;
- int afterlastmatchx=-1;
- current= &ocr_contexts[ctx];
- for (;;) {
- if (x>w) break;
- Pixcol cv= cols[x];
- for (i=0; i<current->nlinks; i++)
- if (current->links[i].col == cv)
- goto found;
- /* not found */
- break;
- found:
- x++;
- current= current->links[i].then;
- if (current->s[0]) { lastmatch=current; afterlastmatchx=x; }
- }
-
- if (!lastmatch) {
- int x2;
- for (x2=x+1; x2<w && cols[x2]; x2++);
- printf("UNKNOWN x=%d ctx=%d %d..%d\n",x, ctx, startx,x2);
- x++;
- } else {
- printf("OUTPUT x=%d `%s'\n", x, lastmatch->s);
- x= afterlastmatchx;
- ctx= 0;
- }
- }
+ results= ocr(w,h,cols);
+ printf("YES! \"");
+ for (res=0; res->s; res++)
+ printf("%s",res->s);
+ printf("\"\n");
+ eassert(!ferror(stdout));
+ eassert(!fflush(stdout));
}
int main(void) {
Rect thisr, entryr;
int tryrect, colno;
- load_ocr_database();
+ ocr_init();
load_image_and_canonify();
find_structure();
--- /dev/null
+/*
+ */
+
+#include "ocr.h"
+
+typedef struct {
+ Pixcol col;
+ struct OCRDatabaseNode *then;
+} OCRDatabaseLink;
+
+#define MAXGLYPHCHRS 3
+
+typedef struct OCRDatabaseNode {
+ char s[MAXGLYPHCHRS+1]; /* null-terminated; "" means no match here */
+ int nlinks, alinks;
+ OCRDatabaseLink *links;
+} OCRDatabaseNode;
+
+#define N_OCR_CONTEXTS 2
+
+static OCRDatabaseNode ocr_contexts[N_OCR_CONTEXTS];
+static FILE *db;
+static OcrResultGlyph *results;
+static int aresults, nresults;
+
+static FILE *resolver;
+static pid_t resolver_pid;
+
+static void ocr_readdb(void) {
+ int ctx,nchrs;
+ OCRDatabaseNode *current, *additional;
+ char chrs[MAXGLYPHCHRS+1];
+ Pixcol cv;
+ int r,i,j;
+
+ assert(!db);
+ db= fopen("database","r"); eassert(db);
+
+ for (;;) {
+ r= fscanf(db, "%d %d", &ctx, &nchrs);
+ if (r==EOF) break;
+ eassert(r==2);
+ eassert(ctx>=0 && ctx<N_OCR_CONTEXTS);
+ eassert(nchrs>0 && nchrs<=MAXGLYPHCHRS);
+
+ for (i=0; i<nchrs; i++) {
+ int c;
+ r= fscanf(db, "%x", &c); eassert(r==1);
+ eassert(c>0 && c<=255);
+ chrs[i]= c;
+ }
+ chrs[nchrs]= 0;
+
+ int twidth;
+ r= fscanf(db, "%d", &twidth); eassert(r==1);
+ current= &ocr_contexts[ctx];
+ for (i=0; i<twidth; i++) {
+ r= fscanf(db, "%"PSPIXCOL(SCNx), &cv); eassert(r==1);
+ for (j=0; j<current->nlinks; j++)
+ if (current->links[j].col == cv) {
+ current= current->links[j].then;
+ goto found_link;
+ }
+
+ additional= malloc(sizeof(*additional)); eassert(additional);
+ additional->s[0]= 0;
+ additional->nlinks= additional->alinks= 0;
+ additional->links= 0;
+ if (current->nlinks==current->alinks) {
+ current->alinks++;
+ current->alinks<<=1;
+ current->links= realloc(current->links,
+ sizeof(*current->links) * current->alinks);
+ eassert(current->links);
+ }
+ current->links[current->nlinks].col= cv;
+ current->links[current->nlinks].then= additional;
+ current->nlinks++;
+ current= additional;
+
+ found_link:;
+ }
+
+ eassert(!current->s[0]);
+ strcpy(current->s, chrs);
+ }
+ eassert(!ferror(db));
+ eassert(feof(db));
+}
+
+static void callout_unknown(int w, int h, Pixcol cols[], int unk_l, int unk_r,
+ const OcrResultGlyph *sofar, int nsofar) {
+ int pfd[2], c, r,i, x,y;
+ const OcrResultGlyph *s;
+ const char *p;
+ Pixcol pv;
+
+ if (!resolver) {
+ r= pipe(pfd); eassert(!r);
+ resolver_pid= fork();
+ eassert(resolver_pid!=-1);
+ if (!resolver_pid) {
+ r= dup2(pfd[0],0); eassert(!r);
+ r= close(pfd[1]); eassert(!r);
+ execlp("./show-thing.tcl", "./show-thing.tcl",(char*)0);
+ eassert(!"execlp failed");
+ }
+ r= close(pfd[0]); eassert(!r);
+ resolver= fdopen(pfd[1],"w"); eassert(resolver);
+ }
+ fprintf(resolver,"%d %d",unk_l,unk_r);
+ for (i=0, s=sofar; i<nsofar; i++, s++) {
+ fprintf(resolver," %d %d %d ",s->l,s->r,s->ctx);
+ for (p=s->s; (c= *p); p++) {
+ if (c=='\\') fprintf(resolver,"\\%c",c);
+ else if (c>=33 && c<=126) fputc(c,resolver);
+ else fprintf(resolver,"\\x%02x",(unsigned char)c);
+ }
+ }
+ fputc('\n',resolver);
+
+ fprintf(resolver,
+ "/* XPM */\n"
+ "static char *t[] = {\n"
+ "/* columns rows colors chars-per-pixel */\n"
+ "\"%d %d 2 1\",\n"
+ "\" c black\",\n"
+ "\"o c white\",\n",
+ w,h);
+ for (y=0, pv=1; y<h; y++, pv<<=1) {
+ fputc('"',resolver);
+ for (x=0; x<w; x++)
+ fputc(cols[x] & pv ? 'o' : ' ', resolver);
+ fputs("\",\n",resolver);
+ }
+ fputs("};\n",resolver);
+ eassert(!ferror(resolver));
+ eassert(!fflush(resolver));
+
+ for (;;) {
+ eassert(resolver);
+ pid_t pid= waitpid(resolver_pid, &r, WUNTRACED);
+ if (pid==-1) { eassert(errno==EINTR); continue; }
+ eassert(pid==resolver_pid);
+ if (WIFEXITED(r)) {
+ eassert(!WEXITSTATUS(r));
+ fclose(resolver);
+ resolver= 0;
+ } else if (WIFSTOPPED(r)) {
+ r= kill(resolver_pid,SIGCONT);
+ eassert(!r);
+ } else if (WIFSIGNALED(r)) {
+ eassert(!"resolver child died due to signal");
+ } else {
+ eassert(!"weird wait status");
+ }
+ struct stat stab, fstab;
+ r= stat("database",&stab); eassert(!r);
+ r= fstat(fileno(db),&fstab); eassert(!r);
+ if (stab.st_ino != fstab.st_ino ||
+ stab.st_dev != fstab.st_dev)
+ break;
+ }
+ fclose(db);
+ db= 0;
+ ocr_readdb();
+}
+
+static void add_result(const char *s, int l, int r, int ctx) {
+ if (nresults >= aresults) {
+ aresults++; aresults<<=1;
+ results= realloc(results,sizeof(*results)*aresults);
+ eassert(results);
+ }
+ results[nresults].s= s;
+ results[nresults].l= l;
+ results[nresults].r= r;
+ results[nresults].ctx= ctx;
+ nresults++;
+}
+
+OcrResultGlyph *ocr(int w, int h, Pixcol cols[]) {
+ int nspaces=0;
+ int ctx=1,i, x;
+
+ nresults=0;
+ assert(db);
+
+ restart:
+ x=0;
+ for (;;) {
+ /* skip spaces */
+ if (x>=w) break;
+
+ if (!cols[x]) {
+ nspaces++;
+ x++;
+ if (nspaces>3) ctx=1;
+ continue;
+ }
+
+ /* find character */
+ OCRDatabaseNode *current=0, *bestmatch=0;
+ int lx=x;
+ int bestmatch_rx=-1;
+ current= &ocr_contexts[ctx];
+ for (;;) {
+ if (x>w) break;
+ Pixcol cv= cols[x];
+ for (i=0; i<current->nlinks; i++)
+ if (current->links[i].col == cv)
+ goto found;
+ /* not found */
+ break;
+ found:
+ current= current->links[i].then;
+ if (current->s[0]) { bestmatch=current; bestmatch_rx=x; }
+ x++;
+ }
+
+ if (!bestmatch) {
+ add_result(bestmatch->s, lx, bestmatch_rx, ctx);
+ x= bestmatch_rx+1;
+ ctx= 0;
+ } else {
+
+ int rx;
+ for (rx=lx+1; rx<w && cols[rx]; rx++);
+ printf("UNKNOWN x=%d ctx=%d %d..%d\n",x, ctx, lx,rx);
+ callout_unknown(w,h,cols, lx,rx, results,nresults);
+ goto restart;
+ }
+ }
+ add_result(0,-1,-1,0);
+ return results;
+}
+
+void ocr_init(void) {
+ ocr_readdb();
+}
~yarrgweb / ypp-sc-tools.main.git / commitdiff