X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?p=ypp-sc-tools.db-test.git;a=blobdiff_plain;f=pctb%2Fconvert.c;h=a0ecca16703749cd47ef6dbd81c03437e2868a33;hp=b518dbabf6913e202761dfc9b34cd524d77d4d55;hb=1971b4961a061d32a86a5e08a64f1b960f666802;hpb=2f9b824c735e9174d7cc7d8aea829a85d2276d21

diff --git a/pctb/convert.c b/pctb/convert.c
index b518dba..a0ecca1 100644
--- a/pctb/convert.c
+++ b/pctb/convert.c
@@ -3,6 +3,7 @@
 #include <inttypes.h>
 #include <assert.h>
 #include <string.h>
+#include <stdlib.h>
 
 #define eassert assert
 #define debug stdout
@@ -34,8 +35,9 @@ 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 MAX_COLUMNS 7
+#define MAX_COLUMNS         7
 
 static Rect mainr = { START_MAIN,START_MAIN };
 static int commbasey, comminty;
@@ -178,7 +180,7 @@ static void find_structure(void) {
     require_rectangle(across.x,mainr.tl.y, across.x,mainr.br.y, "+");
     across.x++;
   }
-  
+  eassert(colno >= MIN_COLUMNS);
 }		    
 
 static void find_commodity(int offset, Rect *rr) {
@@ -195,6 +197,15 @@ static void find_commodity(int offset, Rect *rr) {
   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 load_image_and_canonify(void) {
   struct pam inpam;
   unsigned char rgb[3];
@@ -231,16 +242,168 @@ static void load_image_and_canonify(void) {
   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) {
+  int w= r.br.x - r.tl.x + 1;
+  int h= r.br.y - r.tl.y + 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<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;
+
+  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;
+    }
+  }
+}
+
 int main(void) {
+  Rect thisr, entryr;
+  int tryrect, colno;
+
+  load_ocr_database();
   load_image_and_canonify();
   find_structure();
 
-  Rect thisr;
-  int tryrect;
   for (tryrect= +height; tryrect >= -height; tryrect--) {
     find_commodity(tryrect, &thisr);
-    if (thisr.tl.x >= 0)
-      debug_rect("commod",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);
+    }
   }
   return 0;
 }