Create src/suffixes_dafsa.c with DAFSA C array

list

@@ -1 +1 @@
-Subproject commit e801df4a56ac8c7519d349ad5125433206930d6e
+Subproject commit c749cdfe6847c7c299045d160d379117caf47bd3

src/Makefile.am

@@ -37,3 +37,4 @@ endif
 # PSL_FILE can be set by ./configure --with-psl-file=[PATH]
 suffixes.c: $(PSL_FILE) psl2c$(EXEEXT)
 	./psl2c$(EXEEXT) "$(PSL_FILE)" suffixes.c
+	./psl2c$(EXEEXT) --dafsa "$(PSL_FILE)" suffixes_dafsa.c

src/psl2c.c

@@ -161,8 +161,9 @@ static void _print_psl_entries(FILE *fpout, const _psl_vector_t *v, const char *
 	fprintf(fpout, "};\n");
 }
 
-static void _print_psl_entries_dafsa(FILE *fpout, const _psl_vector_t *v, const char *varname)
+static void _print_psl_entries_dafsa(FILE *fpout, const _psl_vector_t *v)
 {
+	FILE *fp;
 	int it;
 
 #ifdef BUILTIN_GENERATOR_LIBICU
@@ -182,12 +183,30 @@ static void _print_psl_entries_dafsa(FILE *fpout, const _psl_vector_t *v, const
 	fprintf(fpout, "/* automatically generated by psl2c (without punycode support) */\n");
 #endif
 
-	for (it = 0; it < v->cur; it++) {
-		_psl_entry_t *e = _vector_get(v, it);
+	if ((fp = fopen("in.tmp", "w"))) {
+		for (it = 0; it < v->cur; it++) {
+			_psl_entry_t *e = _vector_get(v, it);
+			unsigned char *s = (unsigned char *)e->label_buf;
 
+			/* search for non-ASCII label and skip it */
+			while (*s && *s < 128) s++;
+			if (*s) continue;
 
-		fprintf(fpout, "\t{ \"%s\", NULL, %hd, %d, %d },\n",
-			e->label_buf, e->length, (int) e->nlabels, (int) e->flags);
+			fprintf(fp, "%s, %d\n", e->label_buf, (int) e->flags);
+		}
+
+		fclose(fp);
+	}
+
+	system("../tools/make_dafsa.py in.tmp out.tmp");
+
+	if ((fp = fopen("out.tmp", "r"))) {
+		char buf[256];
+
+		while (fgets(buf, sizeof(buf), fp))
+			fputs(buf, fpout);
+
+		fclose(fp);
 	}
 }
 
@@ -267,10 +286,10 @@ int main(int argc, const char **argv)
 		return 5;
 	}
 */
-	if ((fpout = fopen(argv[2], "w"))) {
+	if ((fpout = fopen(argv[argpos + 1], "w"))) {
 		FILE *pp;
 		struct stat st;
-		size_t cmdsize = 16 + strlen(argv[1]);
+		size_t cmdsize = 16 + strlen(argv[argpos]);
 		char *cmd = alloca(cmdsize), checksum[64] = "";
 		char *abs_srcfile;
 		const char *source_date_epoch = NULL;
@@ -281,18 +300,18 @@ int main(int argc, const char **argv)
 #endif
 
 		if (dafsa)
-			_print_psl_entries(fpout, psl->suffixes, "suffixes");
+			_print_psl_entries_dafsa(fpout, psl->suffixes);
 		else
-			_print_psl_entries_dafsa(fpout, psl->suffixes, "suffixes_dafsa");
+			_print_psl_entries(fpout, psl->suffixes, "suffixes");
 
-		snprintf(cmd, cmdsize, "sha1sum %s", argv[1]);
+		snprintf(cmd, cmdsize, "sha1sum %s", argv[argpos]);
 		if ((pp = popen(cmd, "r"))) {
 			if (fscanf(pp, "%63[0-9a-zA-Z]", checksum) < 1)
 				*checksum = 0;
 			pclose(pp);
 		}
 
-		if (stat(argv[1], &st) != 0)
+		if (stat(argv[argpos], &st) != 0)
 			st.st_mtime = 0;
 		fprintf(fpout, "static time_t _psl_file_time = %lu;\n", st.st_mtime);
 
@@ -307,22 +326,22 @@ int main(int argc, const char **argv)
 
 		/* We need an absolute path here, else psl_builtin_outdated() won't work reliable */
 		/* Caveat: symbolic links are resolved by realpath() */
-		if ((abs_srcfile = realpath(argv[1], NULL))) {
+		if ((abs_srcfile = realpath(argv[argpos], NULL))) {
 			fprintf(fpout, "static const char _psl_filename[] = \"%s\";\n", abs_srcfile);
 			free(abs_srcfile);
 		} else
-			fprintf(fpout, "static const char _psl_filename[] = \"%s\";\n", argv[1]);
+			fprintf(fpout, "static const char _psl_filename[] = \"%s\";\n", argv[argpos]);
 
 		if (fclose(fpout) != 0)
 			ret = 4;
 	} else {
-		fprintf(stderr, "Failed to write open '%s'\n", argv[2]);
+		fprintf(stderr, "Failed to write open '%s'\n", argv[argpos + 1]);
 		ret = 3;
 	}
 
 	psl_free(psl);
 #else
-	if ((fpout = fopen(argv[2], "w"))) {
+	if ((fpout = fopen(argv[argpos + 1], "w"))) {
 		fprintf(fpout, "static _psl_entry_t suffixes[1];\n");
 		fprintf(fpout, "static time_t _psl_file_time;\n");
 		fprintf(fpout, "static time_t _psl_compile_time;\n");
@@ -335,7 +354,7 @@ int main(int argc, const char **argv)
 		if (fclose(fpout) != 0)
 			ret = 4;
 	} else {
-		fprintf(stderr, "Failed to write open '%s'\n", argv[2]);
+		fprintf(stderr, "Failed to write open '%s'\n", argv[argpos + 1]);
 		ret = 3;
 	}
 #endif /* GENERATE_BUILTIN_DATA */

tools/make_dafsa.py

@@ -442,16 +442,16 @@ def parse_gperf(infile):
   """Parses gperf file and extract strings and return code"""
   lines = [line.strip() for line in infile]
   # Extract strings after the first '%%' and before the second '%%'.
-  begin = lines.index('%%') + 1
-  end = lines.index('%%', begin)
-  lines = lines[begin:end]
+  #begin = lines.index('%%') + 1
+  #end = lines.index('%%', begin)
+  #lines = lines[begin:end]
   for line in lines:
     if line[-3:-1] != ', ':
       raise InputError('Expected "domainname, <digit>", found "%s"' % line)
     # Technically the DAFSA format could support return values in range [0-31],
     # but the values below are the only with a defined meaning.
-    if line[-1] not in '0124':
-      raise InputError('Expected value to be one of {0,1,2,4}, found "%s"' %
+    if line[-1] not in '01245':
+      raise InputError('Expected value to be one of {0,1,2,4,5}, found "%s"' %
                        line[-1])
   return [line[:-3] + line[-1] for line in lines]
 
@@ -460,8 +460,12 @@ def main():
   if len(sys.argv) != 3:
     print('usage: %s infile outfile' % sys.argv[0])
     return 1
-  with open(sys.argv[1], 'r') as infile, open(sys.argv[2], 'w') as outfile:
-    outfile.write(words_to_cxx(parse_gperf(infile)))
+  if sys.argv[1] == '-':
+    with open(sys.argv[2], 'w') as outfile:
+      outfile.write(words_to_cxx(parse_gperf(sys.stdin)))
+  else:
+    with open(sys.argv[1], 'r') as infile, open(sys.argv[2], 'w') as outfile:
+      outfile.write(words_to_cxx(parse_gperf(infile)))
   return 0