static int xlsx_verify_sheets (xlsx_info *xinfo, PRN *prn)
{
xmlDocPtr doc = NULL;
xmlNodePtr cur = NULL;
char *checker;
char *ID, *fname;
int i, err;
checker = calloc(xinfo->n_sheets, 1);
if (checker == NULL) {
return E_ALLOC;
}
err = gretl_xml_open_doc_root("xl/_rels/workbook.xml.rels",
"Relationships",
&doc, &cur);
if (!err) {
cur = cur->xmlChildrenNode;
while (cur != NULL) {
if (!xmlStrcmp(cur->name, (XUC) "Relationship")) {
ID = (char *) xmlGetProp(cur, (XUC) "Id");
if ((i = xlsx_match_sheet_id(xinfo, ID)) >= 0) {
fname = (char *) xmlGetProp(cur, (XUC) "Target");
if (fname != NULL) {
if (xlsx_sheet_has_data(fname)) {
checker[i] = 1;
free(xinfo->filenames[i]);
xinfo->filenames[i] = fname;
} else {
free(fname);
}
}
}
free(ID);
}
cur = cur->next;
}
xmlFreeDoc(doc);
}
if (!err) {
int j = 0;
for (i=0; i<xinfo->n_sheets; i++) {
if (checker[j++] == 0) {
fprintf(stderr, "dropping sheet '%s'\n",
xinfo->sheetnames[i]);
xlsx_expunge_sheet(xinfo, i--);
}
}
}
free(checker);
return err;
}
static int xlsx_verify_specific_sheet (xlsx_info *xinfo,
int idx, PRN *prn)
{
xmlDocPtr doc = NULL;
xmlNodePtr cur = NULL;
char *ID, *fname;
int err;
err = gretl_xml_open_doc_root("xl/_rels/workbook.xml.rels",
"Relationships",
&doc, &cur);
if (!err) {
const char *sname = xinfo->sheetnames[idx];
int found = 0;
cur = cur->xmlChildrenNode;
while (cur != NULL && !found) {
if (!xmlStrcmp(cur->name, (XUC) "Relationship")) {
ID = (char *) xmlGetProp(cur, (XUC) "Id");
if (xlsx_match_sheet_id(xinfo, ID) == idx) {
found = 1;
fname = (char *) xmlGetProp(cur, (XUC) "Target");
if (fname == NULL) {
pprintf(prn, "'%s': couldn't find filename\n", sname);
err = E_DATA;
} else if (xlsx_sheet_has_data(fname)) {
pprintf(prn, "'%s' -> %s\n", sname, fname);
free(xinfo->filenames[idx]);
xinfo->filenames[idx] = fname;
} else {
pprintf(prn, "'%s': contains no data\n", sname);
err = E_DATA;
free(fname);
}
}
free(ID);
}
cur = cur->next;
}
xmlFreeDoc(doc);
if (!found) {
pprintf(prn, "'%s': couldn't find file Id\n", sname);
err = E_DATA;
}
if (!err) {
/* record the pre-checked sheet selection */
xinfo->selsheet = idx;
}
}
return err;
}
static int wbook_get_info (const char *fname, const int *list,
char *sheetname, wbook *book,
PRN *prn)
{
xmlDocPtr doc;
xmlNodePtr cur, sub;
char *tmp = NULL;
int got_index = 0;
int err = 0;
err = gretl_xml_open_doc_root(fname, "Workbook",
&doc, &cur);
if (err) {
return err;
}
wbook_init(book, list, sheetname);
/* Now walk the tree */
cur = cur->xmlChildrenNode;
while (cur != NULL && !got_index && !err) {
if (!xmlStrcmp(cur->name, (XUC) "SheetNameIndex")) {
got_index = 1;
sub = cur->xmlChildrenNode;
while (sub != NULL && !err) {
if (!xmlStrcmp(sub->name, (XUC) "SheetName")) {
tmp = (char *) xmlNodeGetContent(sub);
if (tmp != NULL) {
if (wbook_record_name(tmp, book)) {
err = 1;
free(tmp);
}
}
}
sub = sub->next;
}
}
cur = cur->next;
}
xmlFreeDoc(doc);
return err;
}
static int xlsx_workbook_get_sheetnames (xlsx_info *xinfo,
const char *fname)
{
xmlDocPtr doc = NULL;
xmlNodePtr c1, cur = NULL;
char *ID, *sheetname;
int ns = 0, found = 0;
int err;
err = gretl_xml_open_doc_root(fname, "workbook",
&doc, &cur);
if (!err) {
cur = cur->xmlChildrenNode;
while (cur != NULL && !found) {
if (!xmlStrcmp(cur->name, (XUC) "sheets")) {
c1 = cur->xmlChildrenNode;
while (c1 != NULL) {
if (!xmlStrcmp(c1->name, (XUC) "sheet")) {
ID = (char *) xmlGetProp(c1, (XUC) "id");
sheetname = (char *) xmlGetProp(c1, (XUC) "name");
if (ID != NULL && sheetname != NULL) {
strings_array_add(&xinfo->sheetnames,
&xinfo->n_sheets,
sheetname);
strings_array_add(&xinfo->filenames,
&ns, ID);
}
free(ID);
free(sheetname);
}
c1 = c1->next;
}
found = 1;
}
cur = cur->next;
}
xmlFreeDoc(doc);
}
return err;
}
gretl_bundle *gretl_bundle_read_from_file (const char *fname,
int from_dotdir,
int *err)
{
xmlDocPtr doc = NULL;
xmlNodePtr cur = NULL;
char fullname[FILENAME_MAX];
gretl_bundle *b;
b = gretl_bundle_new();
if (b == NULL) {
*err = E_ALLOC;
return NULL;
}
if (from_dotdir) {
build_path(fullname, gretl_dotdir(), fname, NULL);
} else {
strcpy(fullname, fname);
}
*err = gretl_xml_open_doc_root(fullname, "gretl-bundle", &doc, &cur);
if (!*err) {
gretl_push_c_numeric_locale();
cur = cur->xmlChildrenNode;
*err = load_bundled_items(b, cur, doc);
gretl_pop_c_numeric_locale();
xmlFreeDoc(doc);
}
if (*err) {
gretl_bundle_destroy(b);
b = NULL;
}
return b;
}
static int xlsx_sheet_has_data (const char *fname)
{
xmlDocPtr doc = NULL;
xmlNodePtr c1, cur = NULL;
gchar *fullname;
int err, ret = 0;
fullname = g_strdup_printf("xl%c%s", SLASH, fname);
err = gretl_xml_open_doc_root(fullname, "worksheet",
&doc, &cur);
if (!err) {
cur = cur->xmlChildrenNode;
while (cur != NULL && ret == 0) {
if (!xmlStrcmp(cur->name, (XUC) "sheetData")) {
c1 = cur->xmlChildrenNode;
while (c1 != NULL && ret == 0) {
if (!xmlStrcmp(c1->name, (XUC) "row")) {
ret = 1;
}
c1 = c1->next;
}
}
cur = cur->next;
}
xmlFreeDoc(doc);
}
if (!ret) {
fprintf(stderr, "%s: contains no data\n", fname);
}
g_free(fullname);
return ret;
}
static int xlsx_read_shared_strings (xlsx_info *xinfo, PRN *prn)
{
xmlDocPtr doc = NULL;
xmlNodePtr cur = NULL;
xmlNodePtr val;
char *tmp;
int i, n = 0;
int err = 0;
err = gretl_xml_open_doc_root(xinfo->stringsfile, "sst",
&doc, &cur);
if (err) {
pprintf(prn, "Couldn't find shared strings table\n");
pprintf(prn, "%s", gretl_errmsg_get());
return err;
}
tmp = (char *) xmlGetProp(cur, (XUC) "uniqueCount");
if (tmp == NULL) {
tmp = (char *) xmlGetProp(cur, (XUC) "count");
}
if (tmp == NULL) {
pprintf(prn, "didn't get sst count\n");
err = E_DATA;
} else {
n = atoi(tmp);
if (n <= 0) {
pprintf(prn, "didn't get valid sst count\n");
err = E_DATA;
}
free(tmp);
}
if (!err) {
xinfo->strings = strings_array_new(n);
if (xinfo->strings == NULL) {
err = E_ALLOC;
}
}
cur = cur->xmlChildrenNode;
/* The strings in an <sst> are mostly set up as
<si><t>XXX</t></si>
<si><t>YYY</t></si> ...
But there are also weird cases where junk is interposed
and the structure becomes
<si><r>...<t>XXX</t></r><r>...<t>YYY</t></r></si> ...
That is, an <si> element may contain more than one <r>
element, which embeds a <t> along with formatting crap.
*/
i = 0;
while (cur != NULL && !err) {
if (!xmlStrcmp(cur->name, (XUC) "si")) {
int gotstr = 0;
val = cur->xmlChildrenNode;
while (val != NULL && !err && !gotstr) {
if (!xmlStrcmp(val->name, (XUC) "t")) {
/* got a regular <t> element */
tmp = (char *) xmlNodeGetContent(val);
if (tmp == NULL) {
pprintf(prn, "failed reading string %d\n", i);
err = E_DATA;
} else {
xinfo->strings[i++] = tmp;
gotstr = 1;
}
} else if (!xmlStrcmp(val->name, (XUC) "r")) {
/* hunt for <t> inside an <r> element */
xmlNodePtr sub = val->xmlChildrenNode;
while (sub != NULL && !err && i < n) {
if (!xmlStrcmp(sub->name, (XUC) "t")) {
tmp = (char *) xmlNodeGetContent(sub);
if (tmp == NULL) {
pprintf(prn, "failed reading string %d\n", i);
err = E_DATA;
} else {
xinfo->strings[i++] = tmp;
gotstr = 1;
}
}
sub = sub->next;
}
}
val = val->next;
}
}
if (i == n) {
break;
}
cur = cur->next;
}
if (!err && i < n) {
pprintf(prn, "expected %d shared strings but only found %d\n",
n, i);
err = E_DATA;
}
if (!err) {
xinfo->n_strings = i;
} else if (xinfo->strings != NULL) {
strings_array_free(xinfo->strings, n);
xinfo->strings = NULL;
}
xmlFreeDoc(doc);
return err;
}
static int xlsx_read_worksheet (xlsx_info *xinfo, PRN *prn)
{
xmlDocPtr doc = NULL;
xmlNodePtr data_node = NULL;
xmlNodePtr cur = NULL;
xmlNodePtr c1;
int gotdata = 0;
int err = 0;
sprintf(xinfo->sheetfile, "xl%c%s", SLASH,
xinfo->filenames[xinfo->selsheet]);
#if XDEBUG
fprintf(stderr, "xlsx_read_worksheet: sheetnum=%d, name='%s'\n",
xinfo->selsheet, xinfo->filenames[xinfo->selsheet]);
#endif
sprintf(xinfo->stringsfile, "xl%csharedStrings.xml", SLASH);
err = gretl_xml_open_doc_root(xinfo->sheetfile, "worksheet",
&doc, &cur);
if (err) {
pprintf(prn, "didn't get worksheet\n");
pprintf(prn, "%s", gretl_errmsg_get());
return err;
}
/* walk the tree, first pass */
cur = cur->xmlChildrenNode;
while (cur != NULL && !err && !gotdata) {
if (!xmlStrcmp(cur->name, (XUC) "sheetData")) {
data_node = c1 = cur->xmlChildrenNode;
while (c1 != NULL && !err) {
if (!xmlStrcmp(c1->name, (XUC) "row")) {
err = xlsx_read_row(c1, xinfo, prn);
}
c1 = c1->next;
}
gotdata = 1;
}
cur = cur->next;
}
#if XDEBUG
if (!err) {
pprintf(prn, "Max row = %d, max col = %d\n", xinfo->maxrow,
xinfo->maxcol);
pprintf(prn, "Accessed %d shared strings\n", xinfo->n_strings);
}
#endif
if (!err && xinfo->dset == NULL) {
err = xlsx_check_dimensions(xinfo, prn);
if (!err) {
gretl_push_c_numeric_locale();
c1 = data_node;
while (c1 != NULL && !err) {
if (!xmlStrcmp(c1->name, (XUC) "row")) {
err = xlsx_read_row(c1, xinfo, prn);
}
c1 = c1->next;
}
gretl_pop_c_numeric_locale();
}
}
xmlFreeDoc(doc);
return err;
}
static int wsheet_get_data (const char *fname, wsheet *sheet,
int *obscol, PRN *prn)
{
xmlDocPtr doc;
xmlNodePtr cur, sub;
char *tmp = NULL;
int got_sheet = 0;
int err;
err = gretl_xml_open_doc_root(fname, "Workbook", &doc, &cur);
if (err) {
return err;
}
cur = cur->xmlChildrenNode;
/* Now walk the tree */
while (!err && cur != NULL && !got_sheet) {
if (!xmlStrcmp(cur->name, (XUC) "Sheets")) {
int sheetcount = 0;
sub = cur->xmlChildrenNode;
while (sub != NULL && !got_sheet && !err) {
if (!xmlStrcmp(sub->name, (XUC) "Sheet")) {
xmlNodePtr snode = sub->xmlChildrenNode;
while (snode != NULL && !err) {
if (!xmlStrcmp(snode->name, (XUC) "Name")) {
sheetcount++;
tmp = (char *) xmlNodeGetContent(snode);
if (tmp) {
tailstrip(tmp);
if (!strcmp(tmp, sheet->name) &&
sheetcount == sheet->ID + 1) {
got_sheet = 1;
}
free(tmp);
}
} else if (got_sheet && !xmlStrcmp(snode->name, (XUC) "Cells")) {
err = wsheet_get_real_size_etc(snode, sheet, obscol);
if (!err) {
err = wsheet_parse_cells(snode, sheet, *obscol, prn);
}
}
snode = snode->next;
}
}
sub = sub->next;
}
}
cur = cur->next;
}
xmlFreeDoc(doc);
if (!got_sheet) {
fprintf(stderr, "'%s': couldn't find the requested sheet\n", sheet->name);
err = 1;
}
return err;
}
