New --show-charset mode

[ypp-sc-tools.db-test.git] / pctb / ocr.c

diff --git a/pctb/ocr.c b/pctb/ocr.c
index 6448ee44d5e72bb275e17459a2863e3be68d7d2c..f174c68117c8d31e29da62fe94f153882eb8d14f 100644 (file)
--- a/pctb/ocr.c
+++ b/pctb/ocr.c
@@ -36,7 +36,7 @@ typedef struct {
 typedef struct DatabaseNode {
   char *str;
   int nlinks, alinks;
-  unsigned match:1, defined:1, endsword:1;
+  unsigned match:1, defined:1, endsword:1, local:1;
   DatabaseLink *links;
 } DatabaseNode;
 
@@ -99,7 +99,7 @@ static void cleardb_node(DatabaseNode *n) {
     cleardb_node(n->links[i].then);
 }
 
-static void readdb1(OcrReader *rd, const char *which);
+static void readdb1(OcrReader *rd, const char *which, int local);
 
 static void readdb(OcrReader *rd) {
   int ctxi;
@@ -107,11 +107,11 @@ static void readdb(OcrReader *rd) {
   for (ctxi=0; ctxi<NCONTEXTS; ctxi++)
     cleardb_node(&rd->contexts[ctxi]);
 
-  readdb1(rd, "master");
-  readdb1(rd, "local");
+  readdb1(rd, "master", 0);
+  readdb1(rd, "local",  1);
 }
 
-static void readdb1(OcrReader *rd, const char *which) {
+static void readdb1(OcrReader *rd, const char *which, int local) {
   int nchrs;
   DatabaseNode *current, *additional;
   char chrs[100];
@@ -197,6 +197,7 @@ static void readdb1(OcrReader *rd, const char *which) {
       current->str= 0;
       current->defined= 1;
       current->match= 0;
+      current->local= local;
 
       if (nchrs) {
        current->str= mmalloc(nchrs+1);
@@ -483,3 +484,119 @@ OcrReader *ocr_init(int h) {
   readdb(rd);
   return rd;
 }
+
+/*---------- character set dump ----------*/
+
+static void show_recurse(const DatabaseNode *t, int *count,
+                        const DatabaseNode **store_ary) {
+  if (t->defined) {
+    if (store_ary) store_ary[*count]= t;
+    (*count)++;
+  }
+  int l;
+  for (l=0; l<t->nlinks; l++)
+    show_recurse(t->links[l].then, count,store_ary);
+}
+
+static int show_char_compar(const void *av, const void *bv) {
+  const DatabaseNode *const *ap= av;  const DatabaseNode *a= *ap;
+  const DatabaseNode *const *bp= bv;  const DatabaseNode *b= *bp;
+  return strcmp(a->str, b->str) ?:
+    ((int)a->match -    (int)b->match) ?:
+    ((int)a->endsword - (int)b->endsword) ?:
+    ((int)a->local -    (int)b->local) ?:
+    0;
+}
+
+void ocr_showcharsets(void) {
+  DIR *d;
+  struct dirent *de;
+  char found[32];
+  pcre *fnpat;
+  int matchvec[10];
+  char hbuf[10];
+  const char *pcre_err;
+  int pcre_erroffset;
+
+  memset(found,0,sizeof(found));
+  
+  fnpat= pcre_compile("\\#(?:master|local)\\-char([1-9]\\d{0,2})\\#\\.txt$",
+                     PCRE_ANCHORED|PCRE_DOLLAR_ENDONLY,
+                     &pcre_err,&pcre_erroffset, 0);
+  debugf("pcre_compile %p %s\n",fnpat,pcre_err);
+  assert(fnpat);
+
+  sysassert( d= opendir(get_vardir()) );
+  for (;;) {
+    errno=0; de= readdir(d);  if (!de) break;
+
+    int rer= pcre_exec(fnpat,0, de->d_name,strlen(de->d_name), 0,0,
+                      matchvec,ARRAYSIZE(matchvec));
+    debugf("pcre_exec `%s' => %d\n", de->d_name,rer);
+
+    if (rer==PCRE_ERROR_NOMATCH || rer==PCRE_ERROR_BADUTF8) continue;
+    assert(rer==2);
+
+    rer= pcre_copy_substring(de->d_name,matchvec,rer, 1, hbuf,sizeof(hbuf));
+    debugf("pcre_copy_substring => %d\n", rer);
+    assert(rer>0);
+
+    int h= atoi(hbuf);
+    if (h >= ARRAYSIZE(found)) continue;
+    
+    found[h]= 1;
+  }
+
+  int h;
+  for (h=0; h<ARRAYSIZE(found); h++) {
+    if (!found[h]) continue;
+
+    OcrReader *rd= ocr_init(h); /* we leak this but never mind */
+    progress("");
+
+    int ctxi;
+    for (ctxi=0; ctxi<NCONTEXTS; ctxi++) {
+      int nchars= 0;
+      show_recurse(&rd->contexts[ctxi], &nchars, 0);
+      const DatabaseNode **chars= mmalloc(sizeof(*chars) * nchars);
+      int chari= 0;
+      show_recurse(&rd->contexts[ctxi], &chari, chars);
+      assert(chari==nchars);
+      qsort(chars, nchars, sizeof(*chars), show_char_compar);
+
+      int local;
+      for (local=0; local<2; local++) {
+       printf("%2d %-6s %-6s ", h, context_names[ctxi],
+              local?"local":"master");
+       for (chari=0; chari<nchars; chari++) {
+         const DatabaseNode *t= chars[chari];
+         static const char accept[]=
+           "abcdefghijklmnopqrstuvwxyz"
+           "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
+           "0123456789" "'!&-+.,>";
+
+         if (t->local != local) continue;
+
+         if (!t->match)
+           printf(" [nomatch]");
+         else if (!t->endsword && strspn(t->str, accept) == strlen(t->str))
+           printf(" %s",t->str);
+         else {
+           printf(" \"");
+           char *p= t->str;
+           int c;
+           while ((c=*p++)) {
+             if (c=='"' || c=='\\') printf("\\%c",c);
+             else if (c>=' ' && c<=126) putchar(c);
+             else printf("\\x%02x", (unsigned char)c);
+           }
+           if (t->endsword) putchar(' ');
+           putchar('"');
+         }
+       }
+       putchar('\n');
+      }
+      free(chars);
+    }
+  }
+}