int aresults, nresults;
};
-static FILE *resolver;
-static pid_t resolver_pid;
-static int resolver_done;
-
DEBUG_DEFINE_DEBUGF(ocr)
-#define dbassert(x) \
- ((x) ? (void)0 : \
- fatal("Error in character set database.\n" \
- " Requirement not met: %s:%d: %s", __FILE__,__LINE__, #x))
-
-static void fgetsline(FILE *f, char *lbuf, size_t lbufsz) {
- errno=0;
- char *s= fgets(lbuf,lbufsz,f);
- sysassert(!ferror(f));
- dbassert(!feof(f));
- assert(s);
- int l= strlen(lbuf);
- dbassert(l>0); dbassert(lbuf[--l]='\n');
- lbuf[l]= 0;
-}
-#define FGETSLINE(f,buf) (fgetsline(f,buf,sizeof(buf)))
+#define FGETSLINE (dbfile_getsline(lbuf,sizeof(lbuf),__FILE__,__LINE__))
static void cleardb_node(DatabaseNode *n) {
int i;
DatabaseNode *current, *additional;
char chrs[MAXGLYPHCHRS+1];
Pixcol cv;
- int r,j,ctxi;
+ int j,ctxi;
int h, endsword;
char lbuf[100];
- FILE *db;
for (ctxi=0; ctxi<NCONTEXTS; ctxi++)
cleardb_node(&rd->contexts[ctxi]);
asprintf(&dbfname,"%s/charset-%d.txt",get_vardir(),rd->h);
sysassert(dbfname);
- db= fopen(dbfname,"r");
- free(dbfname);
- if (!db) {
- sysassert(errno==ENOENT);
- return;
- }
+ if (!dbfile_open(dbfname))
+ goto x;
- FGETSLINE(db,lbuf);
+ FGETSLINE;
dbassert(!strcmp(lbuf,"# ypp-sc-tools pctb font v1"));
- r= fscanf(db, "%d", &h);
- dbassert(r==1);
+ dbassert( dbfile_scanf("%d", &h) == 1);
dbassert(h==rd->h);
for (;;) {
- FGETSLINE(db,lbuf);
- if (!lbuf || lbuf[0]=='#') continue;
+ FGETSLINE;
+ if (!lbuf[0] || lbuf[0]=='#') continue;
if (!strcmp(lbuf,".")) break;
for (ctxi=0; ctxi<NCONTEXTS; ctxi++)
if (!strcmp(lbuf,context_names[ctxi]))
goto found_ctx;
/* not found, just skip */
- for (;;) { FGETSLINE(db,lbuf); if (!lbuf[0]) break; }
+ for (;;) { FGETSLINE; if (!lbuf[0]) break; }
continue;
found_ctx:
for (nchrs=0;;) {
- int c= fgetc(db); sysassert(!ferror(db)); dbassert(c!=EOF);
+ int c= fgetc(dbfile); sysassert(!ferror(dbfile)); dbassert(c!=EOF);
if (c=='\n') { dbassert(nchrs); break; }
dbassert(nchrs<MAXGLYPHCHRS);
if (c=='\\') {
unsigned cr;
- c= fgetc(db); sysassert(!ferror(db)); dbassert(c=='x');
- r= fscanf(db, "%2x", &cr); sysassert(!ferror(db)); dbassert(r==1);
+ c= fgetc(dbfile); sysassert(!ferror(dbfile)); dbassert(c=='x');
+ dbassert( dbfile_scanf("%2x", &cr) == 1);
assert(cr>0 && cr<=255);
c= cr;
}
current= &rd->contexts[ctxi];
for (;;) {
- FGETSLINE(db,lbuf);
+ FGETSLINE;
if (!lbuf[0]) { dbassert(current != &rd->contexts[ctxi]); break; }
char *ep;
cv= strtoul(lbuf,&ep,16); dbassert(!*ep);
strcpy(current->s, chrs);
current->endsword= endsword;
}
- sysassert(!ferror(db));
- sysassert(!fclose(db));
+ x:
+ dbfile_close();
+ free(dbfname);
}
-static void cu_pr_ctxmap(unsigned ctxmap) {
+static void cu_pr_ctxmap(FILE *resolver, unsigned ctxmap) {
fprintf(resolver,"{");
const char *spc="";
int ctxi;
static void callout_unknown(OcrReader *rd, int w, Pixcol cols[],
int unk_l, int unk_r, unsigned unk_ctxmap) {
- int jobpipe[2],donepipe[2], c,i, x,y;
+ int c,i, x,y;
const OcrResultGlyph *s;
const char *p;
- char cb;
Pixcol pv;
- if (!o_resolver)
+ FILE *resolver= resolve_start();
+ if (!resolver)
fatal("OCR failed - unrecognised characters or ligatures.\n"
"Character set database needs to be updated or augmented.\n"
"See README.charset.\n");
- if (!resolver) {
- sysassert(! pipe(jobpipe) );
- sysassert(! pipe(donepipe) );
- resolver_pid= fork();
- sysassert(resolver_pid!=-1);
- if (!resolver_pid) {
- sysassert( dup2(jobpipe[0],0) ==0 );
- sysassert(! close(jobpipe[1]) );
- sysassert(! close(donepipe[0]) );
- /* we know donepipe[1] is >= 4 and we have dealt with all the others
- * so we aren't in any danger of overwriting some other fd 4: */
- sysassert( dup2(donepipe[1],4) ==4 );
- execlp(o_resolver, o_resolver,
- DEBUGP(callout) ? "--debug" : "--noop-arg",
- "--automatic-1",
- (char*)0);
- sysassert(!"execlp ocr-resolver failed");
- }
- sysassert(! close(jobpipe[0]) );
- sysassert(! close(donepipe[1]) );
- resolver= fdopen(jobpipe[1],"w"); sysassert(resolver);
- resolver_done= donepipe[0];
- }
- fprintf(resolver,"%d %d ",unk_l,unk_r);
- cu_pr_ctxmap(unk_ctxmap);
+ fprintf(resolver,
+ "char\n"
+ "%d %d ",unk_l,unk_r);
+ cu_pr_ctxmap(resolver,unk_ctxmap);
for (i=0, s=rd->results; i<rd->nresults; i++, s++) {
if (!strcmp(s->s," ")) continue;
fprintf(resolver," %d %d ",s->l,s->r);
- cu_pr_ctxmap(s->ctxmap);
+ cu_pr_ctxmap(resolver,s->ctxmap);
fprintf(resolver," ");
for (p=s->s; (c= *p); p++) {
if (c=='\\') fprintf(resolver,"\\%c",c);
fputs("\",\n",resolver);
}
fputs("};\n",resolver);
- sysassert(!ferror(resolver));
- sysassert(!fflush(resolver));
-
- sysassert(resolver);
-
- int r;
- for (;;) {
- r= read(resolver_done,&cb,1);
- if (r==-1) { sysassert(errno==EINTR); continue; }
- break;
- }
-
- if (r==0) {
- waitpid_check_exitstatus(resolver_pid, "character resolver");
- fclose(resolver);
- close(resolver_done);
- resolver= 0;
- } else {
- assert(r==1);
- sysassert(cb==0);
- }
+ resolve_finish();
readdb(rd);
}