void
go_xml_node_set_cstr (xmlNodePtr node, char const *name, char const *val)
{
if (name)
xmlSetProp (node, CC2XML (name), CC2XML (val));
else
xmlNodeSetContent (node, CC2XML (val));
}
/*
* Handles incomplete html fragments as may occur on the clipboard,
* e.g. a <td> without <tr> and <table> in front of it.
*/
static void
html_search_for_tables (htmlNodePtr cur, htmlDocPtr doc,
WorkbookView *wb_view, GnmHtmlTableCtxt *tc)
{
htmlNodePtr ptr;
if (cur == NULL) {
xmlGenericError(xmlGenericErrorContext,
"htmlNodeDumpFormatOutput : node == NULL\n");
return;
}
if (cur->type != XML_ELEMENT_NODE)
return;
if (xmlStrEqual (cur->name, CC2XML ("table"))) {
html_read_table (cur, doc, wb_view, tc);
} else if (starts_inferred_table (cur) || starts_inferred_row (cur)) {
htmlNodePtr tnode = xmlNewNode (NULL, "table");
/* Link in a table node */
xmlAddPrevSibling (cur, tnode);
if (starts_inferred_row (cur)) {
htmlNodePtr rnode = xmlNewNode (NULL, "tr");
/* Link in a row node */
xmlAddChild (tnode, rnode);
/* Make following elements children of the row node,
* until we meet one which isn't legal in a row. */
while ((ptr = tnode->next) != NULL) {
if (ends_inferred_row (ptr))
break;
xmlUnlinkNode (ptr);
xmlAddChild (rnode, ptr);
}
}
/* Make following elements children of the row node,
* until we meet one which isn't legal in a table. */
while ((ptr = tnode->next) != NULL) {
if (ends_inferred_table (ptr))
break;
xmlUnlinkNode (ptr);
xmlAddChild (tnode, ptr);
}
html_read_table (tnode, doc, wb_view, tc);
} else {
for (ptr = cur->children; ptr != NULL ; ptr = ptr->next) {
html_search_for_tables (ptr, doc, wb_view, tc);
/* ptr may now have been pushed down in the tree,
* if so, ptr->next is not the right pointer to
* follow */
while (ptr->parent != cur)
ptr = ptr->parent;
}
}
}
/**
* go_xml_node_get_cstr: (skip)
* @node: #xmlNodePtr
* @name: attribute name
* Get an xmlChar * value for a node carried as an attibute
* result must be xmlFree
*
* Returns: (transfer full): the attribute value
*/
xmlChar *
go_xml_node_get_cstr (xmlNodePtr node, char const *name)
{
if (name != NULL)
return xmlGetProp (node, CC2XML (name));
/* in libxml1 <foo/> would return NULL
* in libxml2 <foo/> would return ""
*/
if (node->xmlChildrenNode != NULL)
return xmlNodeGetContent (node);
return NULL;
}
static gboolean
is_elt_type (htmlNodePtr ptr, char const ** types)
{
char const **p;
gboolean ret = FALSE;
for (p = types; *p; p++)
if (xmlStrEqual (ptr->name, CC2XML ((*p)))) {
ret = TRUE;
break;
}
return ret;
}
static void
html_read_table (htmlNodePtr cur, htmlDocPtr doc, WorkbookView *wb_view,
GnmHtmlTableCtxt *tc)
{
Workbook *wb;
htmlNodePtr ptr, ptr2;
g_return_if_fail (cur != NULL);
g_return_if_fail (wb_view != NULL);
wb = wb_view_get_workbook (wb_view);
for (ptr = cur->children; ptr != NULL ; ptr = ptr->next) {
if (ptr->type != XML_ELEMENT_NODE)
continue;
if (xmlStrEqual (ptr->name, CC2XML ("caption"))) {
xmlBufferPtr buf;
buf = xmlBufferCreate ();
for (ptr2 = ptr->children; ptr2 != NULL ; ptr2 = ptr2->next) {
htmlNodeDump (buf, doc, ptr2);
}
if (buf->use > 0) {
char *name;
name = g_strndup (CXML2C (buf->content), buf->use);
tc->sheet = html_get_sheet (name, wb);
g_free (name);
}
xmlBufferFree (buf);
} else if (xmlStrEqual (ptr->name, CC2XML ("thead")) ||
xmlStrEqual (ptr->name, CC2XML ("tfoot")) ||
xmlStrEqual (ptr->name, CC2XML ("tbody"))) {
html_read_rows (ptr, doc, wb, tc);
} else if (xmlStrEqual (ptr->name, CC2XML ("tr"))) {
html_read_rows (cur, doc, wb, tc);
break;
}
}
}
static void
html_read_rows (htmlNodePtr cur, htmlDocPtr doc, Workbook *wb,
GnmHtmlTableCtxt *tc)
{
htmlNodePtr ptr;
for (ptr = cur->children; ptr != NULL ; ptr = ptr->next) {
if (ptr->type != XML_ELEMENT_NODE)
continue;
if (xmlStrEqual (ptr->name, CC2XML ("tr"))) {
tc->row++;
if (tc->sheet == NULL)
tc->sheet = html_get_sheet (NULL, wb);
html_read_row (ptr, doc, tc);
}
}
}
gboolean
go_xml_node_get_gocolor (xmlNodePtr node, char const *name, GOColor *res)
{
xmlChar *color;
int r, g, b;
color = xmlGetProp (node, CC2XML (name));
if (color == NULL)
return FALSE;
if (sscanf (CXML2C (color), "%X:%X:%X", &r, &g, &b) == 3) {
r >>= 8;
g >>= 8;
b >>= 8;
*res = GO_COLOR_FROM_RGBA (r,g,b,0xff);
xmlFree (color);
return TRUE;
}
static void
html_read_row (htmlNodePtr cur, htmlDocPtr doc, GnmHtmlTableCtxt *tc)
{
htmlNodePtr ptr;
int col = -1;
for (ptr = cur->children; ptr != NULL ; ptr = ptr->next) {
if (xmlStrEqual (ptr->name, CC2XML ("td")) ||
xmlStrEqual (ptr->name, CC2XML ("th"))) {
GString *buf;
xmlBufferPtr a_buf;
xmlAttrPtr props;
int colspan = 1;
int rowspan = 1;
GnmCellPos pos;
GnmStyle *mstyle;
GSList *hrefs = NULL;
GnmHLink *lnk = NULL;
/* Check whether we need to skip merges from above */
pos.row = tc->row;
pos.col = col + 1;
while (gnm_sheet_merge_contains_pos (tc->sheet, &pos)) {
col++;
pos.col++;
}
/* Do we span across multiple rows or cols? */
props = ptr->properties;
while (props) {
if (xmlStrEqual (props->name, CC2XML ("colspan")) && props->children)
colspan = atoi (CXML2C (props->children->content));
if (xmlStrEqual (props->name, CC2XML ("rowspan")) && props->children)
rowspan = atoi (CXML2C (props->children->content));
props = props->next;
}
if (colspan < 1)
colspan = 1;
if (rowspan < 1)
rowspan = 1;
/* Let's figure out the content of the cell */
buf = g_string_new (NULL);
a_buf = xmlBufferCreate ();
mstyle = gnm_style_new_default ();
if (xmlStrEqual (ptr->name, CC2XML ("th")))
gnm_style_set_font_bold (mstyle, TRUE);
html_read_content (ptr, buf, mstyle, a_buf,
&hrefs, TRUE, doc, tc);
if (g_slist_length (hrefs) >= 1 &&
buf->len > 0) {
/* One hyperlink, and text to make it
* visible */
char *url;
xmlBufferPtr h_buf = xmlBufferCreate ();
hrefs = g_slist_reverse (hrefs);
htmlNodeDump (
h_buf, doc, (htmlNodePtr)hrefs->data);
url = g_strndup (
CXML2C (h_buf->content), h_buf->use);
if (strncmp (url, "mailto:",
strlen ("mailto:")) == 0)
lnk = gnm_hlink_new (
gnm_hlink_email_get_type (),
tc->sheet);
else
lnk = gnm_hlink_new (
gnm_hlink_url_get_type (),
tc->sheet);
gnm_hlink_set_target (lnk, url);
gnm_style_set_hlink (mstyle, lnk);
gnm_style_set_font_uline (mstyle,
UNDERLINE_SINGLE);
gnm_style_set_font_color (mstyle,
gnm_color_new_go (GO_COLOR_BLUE));
g_free (url);
xmlBufferFree (h_buf);
}
if (g_slist_length (hrefs) > 1 || buf->len <= 0) {
/* Multiple links,
* or no text to give hyperlink style,
* so put them in a comment */
GSList *l;
for (l = hrefs; l != NULL; l = l->next) {
htmlNodeDump (a_buf, doc,
(htmlNodePtr)l->data);
xmlBufferAdd (a_buf, CC2XML ("\n"),
-1);
}
}
g_slist_free (hrefs);
if (buf->len > 0) {
GnmCell *cell = sheet_cell_fetch (tc->sheet, col + 1, tc->row);
sheet_style_set_pos (tc->sheet, col + 1, tc->row, mstyle);
gnm_cell_set_text (cell, buf->str);
} else
gnm_style_unref (mstyle);
if (a_buf->use > 0) {
char *name;
name = g_strndup (CXML2C (a_buf->content), a_buf->use);
cell_set_comment (tc->sheet, &pos, NULL, name, NULL);
g_free (name);
}
g_string_free (buf, TRUE);
xmlBufferFree (a_buf);
/* If necessary create the merge */
if (colspan > 1 || rowspan > 1) {
GnmRange range;
GnmRange *r = ⦥
range_init (r, col + 1, tc->row, col + colspan, tc->row + rowspan - 1);
gnm_sheet_merge_add (tc->sheet, r, FALSE, NULL);
}
col += colspan;
}
}
}
static void
html_read_content (htmlNodePtr cur, GString *buf, GnmStyle *mstyle,
xmlBufferPtr a_buf, GSList **hrefs, gboolean first,
htmlDocPtr doc, GnmHtmlTableCtxt *tc)
{
htmlNodePtr ptr;
for (ptr = cur->children; ptr != NULL ; ptr = ptr->next) {
if (ptr->type == XML_TEXT_NODE) {
if (g_utf8_validate (ptr->content, -1, NULL))
html_append_text (buf, ptr->content);
else
g_string_append (buf, _("[Warning: Invalid text string has been removed.]"));
} else if (ptr->type == XML_ELEMENT_NODE) {
if (first) {
if (xmlStrEqual (ptr->name, CC2XML ("i"))
|| xmlStrEqual (ptr->name, CC2XML ("em")))
gnm_style_set_font_italic (mstyle, TRUE);
if (xmlStrEqual (ptr->name, CC2XML ("b")))
gnm_style_set_font_bold (mstyle, TRUE);
}
if (xmlStrEqual (ptr->name, CC2XML ("a"))) {
xmlAttrPtr props;
props = ptr->properties;
while (props) {
if (xmlStrEqual (props->name, CC2XML ("href")) && props->children) {
*hrefs = g_slist_prepend (
*hrefs, props->children);
}
props = props->next;
}
}
if (xmlStrEqual (ptr->name, CC2XML ("img"))) {
xmlAttrPtr props;
props = ptr->properties;
while (props) {
if (xmlStrEqual (props->name, CC2XML ("src")) && props->children) {
htmlNodeDump (a_buf, doc, props->children);
xmlBufferAdd (a_buf, CC2XML ("\n"), -1);
}
props = props->next;
}
}
if (xmlStrEqual (ptr->name, CC2XML ("table"))) {
Sheet *last_sheet = tc->sheet;
int last_row = tc->row;
tc->sheet = NULL;
tc->row = -1;
html_read_table (ptr, doc, tc->wb_view, tc);
if (tc->sheet) {
g_string_append_printf (buf, _("[see sheet %s]"), tc->sheet->name_quoted);
xmlBufferAdd (a_buf, CC2XML (_("The original html file is\n"
"using nested tables.")), -1);
}
tc->sheet = last_sheet;
tc->row = last_row;
} else
html_read_content
(ptr, buf, mstyle, a_buf, hrefs, first, doc, tc);
}
first = FALSE;
}
}