uint16_t json_object(char **in_buf, uint16_t len, char *name){
return put_json(in_buf, len, OBJ_BEGIN, name, 0, 0);
}
uint16_t json_end(char **in_buf, uint16_t len){
return put_json(in_buf, len, END, NULL, 0, 0);
}
void put_json_str (char *str, FILE *fp)
{
char c;
int i = 0;
while ((c = str[i++]) != '\0') put_json(c, fp);
}
uint16_t json_start(char **in_buf, uint16_t len){
return put_json(in_buf, len, ROOT_BEGIN, NULL, 0, 0);
}
int main (int argc, char *argv[])
{
FILE *fpi = stdin, *fpo = stdout, *fpmap;
struct mapper mapper;
struct record record;
struct map *map;
struct field *field;
struct buf *val_buf = buf_new(1024), *xml_buf = NULL;
char *xml_key, *key_buf, prev_seq[9], seq[9], tag[4], ind1, ind2;
int file_args = 0, array = 1, xml = 0, mapfile = 0, count = 0,
first = 0, first_sub = 0, first_char, skip_sub = 0, c, i, j;
tag[3] = '\0';
for (i = 1; i < argc; i++) {
if (!strcmp(argv[i], "--usage") || !strcmp(argv[i], "--help") || !strcmp(argv[i], "-?")) {
usage(argv[0], 0);
} else if (!strcmp(argv[i], "--newline")) {
array = 0;
} else if (!strcmp(argv[i], "--xml")) {
xml = 1;
} else if (!strcmp(argv[i], "--json")) {
xml = 0;
} else if (!strcmp(argv[i], "--xmlkey")) {
if (++i == argc) usage(argv[0], 1);
xml_key = argv[i];
xml_buf = buf_new(2048);
} else if (!strcmp(argv[i], "--mapfile")) {
if (++i == argc) usage(argv[0], 1);
key_buf = malloc(256);
if ((fpmap = fopen(argv[i], "r")) == NULL) {
fprintf(stderr, "Can't open %s for reading\n", argv[i]);
return 1;
}
if (!mapfile) {
mapfile = 1;
mapper.size = mapper.grow = 100;
mapper.used = 0;
mapper.maps = malloc(sizeof(struct map) * mapper.size);
record.size = record.grow = 100;
record.used = 0;
record.fields = malloc(sizeof(struct field) * record.size);
for (j = 0; j < record.size; j++) {
record.fields[j].val = buf_new(1024);
}
}
while ((c = getc(fpmap)) != EOF) {
if (isspace(c)) continue;
if (c == '#') {
while (more(c = getc(fpmap)));
continue;
}
map = mapper.maps + mapper.used;
map->key = NULL;
map->tag[0] = c;
map->tag[1] = getc(fpmap);
map->tag[2] = getc(fpmap);
if (isspace(c = getc(fpmap))) {
map->op = '\0';
} else {
map->op = c;
fscanf(fpmap, "%s", map->subfields);
}
fscanf(fpmap, "%s", key_buf);
for (j = 0; j < mapper.used; j++) {
if (!strcmp(mapper.maps[j].key, key_buf)) {
map->key = mapper.maps[j].key;
break;
}
}
if (!map->key) {
map->key = malloc(strlen(key_buf) + 1);
strcpy(map->key, key_buf);
}
mapper.used++;
if (mapper.used == mapper.size) {
mapper.size += mapper.grow;
mapper.maps = realloc(mapper.maps, sizeof(struct map) * mapper.size);
}
}
free(key_buf);
fclose(fpmap);
} else {
switch (++file_args) {
case 1:
if ((fpi = fopen(argv[i], "r")) == NULL) {
fprintf(stderr, "Can't open %s for reading\n", argv[i]);
return 1;
}
break;
case 2:
if ((fpo = fopen(argv[i], "w")) == NULL) {
fprintf(stderr, "Can't open %s for writing\n", argv[i]);
return 1;
}
break;
}
}
}
if (xml) {
fputs("<?xml version=\"1.0\" encoding=\"UTF-8\"?><collection xmlns=\"http://www.loc.gov/MARC21/slim\">", fpo);
} else if (array) {
fputs("[", fpo);
}
while ((c = getc(fpi)) != EOF) {
if (isspace(c)) continue;
seq[0] = c;
for (i = 1; i < 9; ++i) seq[i] = getc(fpi);
getc(fpi);
for (i = 0; i < 3; ++i) tag[i] = getc(fpi);
ind1 = getc(fpi);
ind2 = getc(fpi);
for (i = 0; i < 3; ++i) getc(fpi);
val_buf->used = 0;
while (more(c = getc(fpi))) buf_add(val_buf, c);
buf_add(val_buf, '\0');
if (!count || strncmp(seq, prev_seq, 9)) {
if (count++) {
if (xml_buf) {
buf_add_str(xml_buf, "</record>");
buf_add(xml_buf, '\0');
}
if (mapfile) {
fputs("{", fpo);
for (i = 0; i < record.used; i++) {
field = record.fields + i;
if (i) putc(',', fpo);
buf_add(field->val, '\0');
fprintf(fpo, "\"%s\":[%s]", field->key, field->val->str);
}
if (xml_buf) {
if (i) putc(',', fpo);
fprintf(fpo, "\"%s\":\"", xml_key);
put_json_str(xml_buf->str, fpo);
putc('\"', fpo);
}
fputs(array ? "},\n" : "}\n", fpo);
record.used = 0;
} else if (xml) {
fputs("</record>\n", fpo);
} else {
fputs(array ? "]},\n" : "]}\n", fpo);
}
if (xml_buf) {
xml_buf->used = 0;
}
}
}
for (i = 0; i < 9; ++i) prev_seq[i] = seq[i];
if (mapfile) {
for (i = 0; i < mapper.used; i++) {
map = mapper.maps + i;
if ((map->tag[0] == '*' || map->tag[0] == tag[0]) &&
(map->tag[1] == '*' || map->tag[1] == tag[1]) &&
(map->tag[2] == '*' || map->tag[2] == tag[2])) {
field = NULL;
for (j = 0; j < record.used; j++) {
field = record.fields + j;
if (field->key == map->key) {
break;
}
field = NULL;
}
if (!field) {
field = record.fields + record.used;
field->key = map->key;
field->val->used = 0;
record.used++;
}
skip_sub = 0;
first_char = 1;
for (j = 0; j < val_buf->used-1; j++) {
if (val_buf->str[j] == '$' && val_buf->str[j+1] == '$') {
j+= 2;
if ( (map->op == '+' && strchr(map->subfields, val_buf->str[j]) == NULL) ||
(map->op == '-' && strchr(map->subfields, val_buf->str[j]) != NULL) ) {
skip_sub = 1;
continue;
}
skip_sub = 0;
j++;
if (j == val_buf->used-1) break;
}
if (!skip_sub) {
if (first_char) {
if (field->val->used) buf_add(field->val, ',');
buf_add(field->val, '"');
first_char = 0;
}
if (val_buf->str[j] == '"') buf_add(field->val, '\\');
if (val_buf->str[j] == '\\') buf_add(field->val, '\\');
buf_add(field->val, val_buf->str[j]);
}
}
if (!first_char) buf_add(field->val, '"');
}
}
}
if (!strncmp(tag, "FMT", 3)) continue;
if (!strncmp(tag, "LDR", 3)) {
if (xml_buf) {
buf_add_str(xml_buf, "<record><leader>");
buf_add_str(xml_buf, val_buf->str);
buf_add_str(xml_buf, "</leader>");
} else if (xml) {
fprintf(fpo, "<record><leader>%s</leader>", val_buf->str);
} else if (!mapfile) {
fprintf(fpo, "{\"leader\":\"%s\",\"fields\":[", val_buf->str);
}
first = 1;
continue;
}
if (first) {
first = 0;
} else if (!xml_buf && !xml && !mapfile) {
putc(',', fpo);
}
if (tag[0] == '0' && tag[1] == '0') {
if (xml_buf) {
buf_add_str(xml_buf, "<controlfield tag=\"");
for (i = 0; i < 3; i++) buf_add(xml_buf, tag[i]);
buf_add_str(xml_buf, "\">");
buf_add_xml_str(xml_buf, val_buf->str);
buf_add_str(xml_buf, "</controlfield>");
} else if (xml) {
fprintf(fpo, "<controlfield tag=\"%3s\">", tag);
for (i = 0; i < val_buf->used-1; i++) put_xml(val_buf->str[i], fpo);
fputs("</controlfield>", fpo);
} else if (!mapfile) {
fprintf(fpo, "{\"%3s\":\"", tag);
for (i = 0; i < val_buf->used-1; i++) put_json(val_buf->str[i], fpo);
fputs("\"}", fpo);
}
} else {
if (xml_buf) {
buf_add_str(xml_buf, "<datafield tag=\"");
for (i = 0; i < 3; i++) buf_add(xml_buf, tag[i]);
buf_add_str(xml_buf, "\" ind1=\"");
buf_add(xml_buf, ind1);
buf_add_str(xml_buf, "\" ind2=\"");
buf_add(xml_buf, ind2);
buf_add_str(xml_buf, "\">");
} else if (xml) {
fprintf(fpo, "<datafield tag=\"%3s\" ind1=\"%c\" ind2=\"%c\">", tag, ind1, ind2);
} else if (!mapfile) {
fprintf(fpo, "{\"%3s\":{\"ind1\":\"%c\",\"ind2\":\"%c\",\"subfields\":[", tag, ind1, ind2);
}
first_sub = 1;
for (i = 0; i < val_buf->used-1; i++) {
if (val_buf->str[i] == '$' && val_buf->str[i+1] == '$') {
i+= 2;
if (first_sub) {
first_sub = 0;
} else {
if (xml_buf) {
buf_add_str(xml_buf, "</subfield>");
} else if (xml) {
fputs("</subfield>", fpo);
} else if (!mapfile) {
fputs("\"},", fpo);
}
}
if (xml_buf) {
buf_add_str(xml_buf, "<subfield code=\"");
buf_add(xml_buf, val_buf->str[i]);
buf_add_str(xml_buf, "\">");
} else if (xml) {
fprintf(fpo, "<subfield code=\"%c\">", val_buf->str[i]);
} else if (!mapfile) {
fprintf(fpo, "{\"%c\":\"", val_buf->str[i]);
}
continue;
}
if (xml_buf) {
buf_add_xml(xml_buf, val_buf->str[i]);
} else if (xml) {
put_xml(val_buf->str[i], fpo);
} else if (!mapfile) {
put_json(val_buf->str[i], fpo);
}
}
if (!first_sub) {
if (xml_buf) {
buf_add_str(xml_buf, "</subfield>");
} else if (xml) {
fputs("</subfield>", fpo);
} else if (!mapfile) {
fputs("\"}", fpo);
}
}
if (xml_buf) {
buf_add_str(xml_buf, "</datafield>");
} else if (xml) {
fputs("</datafield>", fpo);
} else if (!mapfile) {
fputs("]", fpo);
}
}
}
if (count && mapfile) {
fputs("{", fpo);
for (i = 0; i < record.used; i++) {
field = record.fields + i;
if (i) putc(',', fpo);
buf_add(field->val, '\0');
fprintf(fpo, "\"%s\":[%s]", field->key, field->val->str);
}
if (xml_buf) {
if (i) putc(',', fpo);
fprintf(fpo, "\"%s\":\"", xml_key);
put_json_str(xml_buf->str, fpo);
putc('\"', fpo);
}
}
if (xml) {
if (count) fputs("</record>", fpo);
fputs("</collection>", fpo);
} else {
if (count) fputs("}", fpo);
if (array) fputs("]", fpo);
}
fclose(fpi);
fclose(fpo);
return 0;
}
