void parse_urls(const char *filename, const url_list_t *elem)
{
TidyDoc tdoc;
int err;
FILE *outfile = NULL;
tdoc = tidyCreate();
tidyOptSetBool(tdoc, TidyForceOutput, yes);
tidyOptSetBool(tdoc, TidyMark, no);
tidyOptSetBool(tdoc, TidyHideEndTags, yes);
tidyOptSetBool(tdoc, TidyDropEmptyParas, no);
tidyOptSetBool(tdoc, TidyJoinStyles, no);
tidyOptSetBool(tdoc, TidyPreserveEntities, yes);
tidyOptSetInt(tdoc, TidyMergeDivs, no);
tidyOptSetInt(tdoc, TidyMergeSpans, no);
tidyOptSetInt(tdoc, TidyWrapLen, 4096);
tidyOptSetValue(tdoc, TidyCharEncoding, "utf8");
tidySetReportFilter(tdoc, filter_cb);
err = tidyParseFile(tdoc, filename);
if (err >= 0)
err = tidyCleanAndRepair(tdoc);
if (err >= 0) {
outfile = option_values.save_relative_links && !option_values.disable_save_tree
? fopen(filename, "w") : NULL;
parse_html(tdoc, tidyGetRoot(tdoc), elem, 1, outfile);
if (outfile)
fclose(outfile);
}
tidyRelease(tdoc);
}
/*
* Append a line of text to the buffer
*/
void html_text_buffer_append(GtkTextView *text_view, char *txt, int ignore)
{
gchar *text = convert_to_utf8(txt);
GtkTextIter iter;
GtkTextMark *insert_mark;
GdkRectangle iter_loc;
GdkRectangle visible_rect;
GtkTextBuffer *buffer = gtk_text_view_get_buffer(text_view);
if (strcasestr(text, "<br>")) {
char *c = text;
while ((c = strchr(text, '\n')) != 0)
*c = ' ';
while ((c = strchr(text, '\r')) != 0)
*c = ' ';
} else if (strchr(text, '\r')) {
char *c = text;
if (strchr(text, '\n')) {
while ((c = strchr(c, '\r')) != 0)
*c = ' ';
} else {
while ((c = strchr(c, '\r')) != 0)
*c = '\n';
}
}
gtk_text_buffer_get_end_iter(buffer, &iter);
insert_mark = gtk_text_buffer_get_mark(buffer, "real_end_mark");
if (insert_mark) {
GtkTextIter del;
gtk_text_buffer_get_iter_at_mark(buffer, &del, insert_mark);
gtk_text_buffer_delete(buffer, &del, &iter);
gtk_text_buffer_get_end_iter(buffer, &iter);
}
else
insert_mark = gtk_text_buffer_create_mark(buffer,
"real_end_mark", &iter, TRUE);
/* Decide first if we want to scroll the text to the end or not */
gtk_text_view_get_iter_location(text_view, &iter, &iter_loc);
gtk_text_view_get_visible_rect(text_view, &visible_rect);
gtk_text_buffer_insert(buffer, &iter, text, -1);
parse_html(text_view, *insert_mark, ignore);
if (iter_loc.y <= visible_rect.y + visible_rect.height) {
GtkTextMark *end_mark;
gtk_text_buffer_get_end_iter(buffer, &iter);
end_mark = gtk_text_buffer_create_mark(buffer, NULL, &iter,
TRUE);
gtk_text_view_scroll_mark_onscreen(text_view, end_mark);
gtk_text_buffer_delete_mark(buffer, end_mark);
}
if (!(ignore & HTML_IGNORE_END))
gtk_text_buffer_delete_mark(buffer, insert_mark);
g_free(text);
}
void format_table(unsigned char *attr, unsigned char *html, unsigned char *eof, unsigned char **end, void *f)
{
struct part *p = f;
int border, cellsp, vcellpd, cellpd, align;
int frame, rules, width, wf;
struct rgb bgcolor;
struct table *t;
char *al;
int cye;
int x;
int i;
/*int llm = last_link_to_move;*/
struct s_e *bad_html;
int bad_html_n;
struct node *n, *nn;
int cpd_pass, cpd_width, cpd_last;
/*if (!p->data) {
debug("nested tables not supported");
return;
}*/
table_level++;
memcpy(&bgcolor, &par_format.bgcolor, sizeof(struct rgb));
get_bgcolor(attr, &bgcolor);
if ((border = get_num(attr, "border")) == -1) border = has_attr(attr, "border") || has_attr(attr, "rules") || has_attr(attr, "frame");
/*if (!border) border = 1;*/
if ((cellsp = get_num(attr, "cellspacing")) == -1) cellsp = 1;
if ((cellpd = get_num(attr, "cellpadding")) == -1) {
vcellpd = 0;
cellpd = !!border;
} else {
vcellpd = cellpd >= HTML_CHAR_HEIGHT / 2 + 1;
cellpd = cellpd >= HTML_CHAR_WIDTH / 2 + 1;
}
if (!border) cellsp = 0;
else if (!cellsp) cellsp = 1;
if (border > 2) border = 2;
if (cellsp > 2) cellsp = 2;
align = par_format.align;
if (align == AL_NO || align == AL_BLOCK) align = AL_LEFT;
if ((al = get_attr_val(attr, "align"))) {
if (!strcasecmp(al, "left")) align = AL_LEFT;
if (!strcasecmp(al, "center")) align = AL_CENTER;
if (!strcasecmp(al, "right")) align = AL_RIGHT;
mem_free(al);
}
frame = F_BOX;
if ((al = get_attr_val(attr, "frame"))) {
if (!strcasecmp(al, "void")) frame = F_VOID;
if (!strcasecmp(al, "above")) frame = F_ABOVE;
if (!strcasecmp(al, "below")) frame = F_BELOW;
if (!strcasecmp(al, "hsides")) frame = F_HSIDES;
if (!strcasecmp(al, "vsides")) frame = F_VSIDES;
if (!strcasecmp(al, "lhs")) frame = F_LHS;
if (!strcasecmp(al, "rhs")) frame = F_RHS;
if (!strcasecmp(al, "box")) frame = F_BOX;
if (!strcasecmp(al, "border")) frame = F_BOX;
mem_free(al);
}
rules = border ? R_ALL : R_NONE;
if ((al = get_attr_val(attr, "rules"))) {
if (!strcasecmp(al, "none")) rules = R_NONE;
if (!strcasecmp(al, "groups")) rules = R_GROUPS;
if (!strcasecmp(al, "rows")) rules = R_ROWS;
if (!strcasecmp(al, "cols")) rules = R_COLS;
if (!strcasecmp(al, "all")) rules = R_ALL;
mem_free(al);
}
if (!border) frame = F_VOID;
wf = 0;
if ((width = get_width(attr, "width", p->data || p->xp)) == -1) {
width = par_format.width - par_format.leftmargin - par_format.rightmargin;
if (width < 0) width = 0;
wf = 1;
}
if (!(t = parse_table(html, eof, end, &bgcolor, p->data || p->xp, &bad_html, &bad_html_n))) {
mem_free(bad_html);
goto ret0;
}
for (i = 0; i < bad_html_n; i++) {
while (bad_html[i].s < bad_html[i].e && WHITECHAR(*bad_html[i].s)) bad_html[i].s++;
while (bad_html[i].s < bad_html[i].e && WHITECHAR(bad_html[i].e[-1])) bad_html[i].e--;
if (bad_html[i].s < bad_html[i].e) parse_html(bad_html[i].s, bad_html[i].e, put_chars_f, line_break_f, special_f, p, NULL);
}
mem_free(bad_html);
html_stack_dup();
html_top.dontkill = 1;
par_format.align = AL_LEFT;
t->p = p;
t->border = border;
t->cellpd = cellpd;
t->vcellpd = vcellpd;
t->cellsp = cellsp;
t->frame = frame;
t->rules = rules;
t->width = width;
t->wf = wf;
cpd_pass = 0;
cpd_last = t->cellpd;
cpd_width = 0; /* not needed, but let the warning go away */
again:
get_cell_widths(t);
if (get_column_widths(t)) goto ret2;
get_table_width(t);
if (!p->data && !p->xp) {
if (!wf && t->max_t > width) t->max_t = width;
if (t->max_t < t->min_t) t->max_t = t->min_t;
if (t->max_t + par_format.leftmargin + par_format.rightmargin > p->xmax) p->xmax = t->max_t + par_format.leftmargin + par_format.rightmargin;
if (t->min_t + par_format.leftmargin + par_format.rightmargin > p->x) p->x = t->min_t + par_format.leftmargin + par_format.rightmargin;
goto ret2;
}
if (!cpd_pass && t->min_t > width && t->cellpd) {
t->cellpd = 0;
cpd_pass = 1;
cpd_width = t->min_t;
goto again;
}
if (cpd_pass == 1 && t->min_t > cpd_width) {
t->cellpd = cpd_last;
cpd_pass = 2;
goto again;
}
/*debug("%d %d %d", t->min_t, t->max_t, width);*/
if (t->min_t >= width) distribute_widths(t, t->min_t);
else if (t->max_t < width && wf) distribute_widths(t, t->max_t);
else distribute_widths(t, width);
if (!p->data && p->xp == 1) {
int ww = t->rw + par_format.leftmargin + par_format.rightmargin;
if (ww > par_format.width) ww = par_format.width;
if (ww < t->rw) ww = t->rw;
if (ww > p->x) p->x = ww;
p->cy += t->rh;
goto ret2;
}
#ifdef HTML_TABLE_2ND_PASS
check_table_widths(t);
#endif
x = par_format.leftmargin;
if (align == AL_CENTER) x = (par_format.width + par_format.leftmargin - par_format.rightmargin - t->rw) / 2;
if (align == AL_RIGHT) x = par_format.width - par_format.rightmargin - t->rw;
if (x + t->rw > par_format.width) x = par_format.width - t->rw;
if (x < 0) x = 0;
/*display_table(t, x, p->cy, &cye);*/
get_table_heights(t);
if (!p->data) {
if (t->rw + par_format.leftmargin + par_format.rightmargin > p->x) p->x = t->rw + par_format.leftmargin + par_format.rightmargin;
p->cy += t->rh;
goto ret2;
}
n = p->data->nodes.next;
n->yw = p->yp - n->y + p->cy;
display_complicated_table(t, x, p->cy, &cye);
display_table_frames(t, x, p->cy);
nn = mem_alloc(sizeof(struct node));
nn->x = n->x;
nn->y = p->yp + cye;
nn->xw = n->xw;
add_to_list(p->data->nodes, nn);
/*sdbg(p->data);*/
/*for (y = p->cy; y < cye; y++) {
last_link_to_move = llm;
align_line(p, y);
}*/
/*if (p->cy + t->rh != cye) internal("size does not match; 1:%d, 2:%d", p->cy + t->rh, cye);*/
p->cy = cye;
p->cx = -1;
ret2:
p->link_num = t->link_num;
if (p->cy > p->y) p->y = p->cy;
/*ret1:*/
free_table(t);
kill_html_stack_item(&html_top);
ret0:
/*ret:*/
table_level--;
if (!table_level) free_table_cache();
}
void do_format(char *start, char *end, struct part *part, unsigned char *head)
{
parse_html(start, end, (int (*)(void *, unsigned char *, int)) put_chars_conv, (void (*)(void *)) line_break, (void *(*)(void *, int, ...)) html_special, part, head);
/*if ((part->y -= line_breax) < 0) part->y = 0;*/
}
static void parse_html(TidyDoc tdoc, TidyNode tnod, const url_list_t *elem, int indent, FILE *outfile)
{
TidyNode child;
TidyAttr attr;
TidyAttrId attr_id = TidyAttr_UNKNOWN;
TidyNodeType node_type;
TidyTagId node_id;
ctmbstr name;
char *url, *relative_url = NULL;
int found = 0;
int get_html_link = (!option_values.depth || elem->level < option_values.depth);
int get_int_html_link = (!option_values.depth || elem->level < option_values.depth+1);
int get_ext_depends = ((!option_values.depth || elem->level < option_values.depth+1)
&& !option_values.no_html_dependencies);
for (child = tidyGetChild(tnod); child; child = tidyGetNext(child)) {
node_type = tidyNodeGetType(child);
switch (node_type) {
case TidyNode_Start:
case TidyNode_StartEnd:
node_id = tidyNodeGetId(child);
if (get_html_link && (node_id == TidyTag_A || node_id == TidyTag_AREA || node_id == TidyTag_MAP)) {
found = 1;
attr_id = TidyAttr_HREF;
}
else if (get_int_html_link && (node_id == TidyTag_FRAME || node_id == TidyTag_IFRAME)) {
found = 1;
attr_id = TidyAttr_SRC;
}
else if (get_ext_depends) {
if (node_id == TidyTag_LINK) {
found = 1;
attr_id = TidyAttr_HREF;
}
else if (node_id == TidyTag_IMG || node_id == TidyTag_SCRIPT) {
found = 1;
attr_id = TidyAttr_SRC;
}
else {
found = 0;
attr_id = TidyAttr_UNKNOWN;
}
}
else {
found = 0;
attr_id = TidyAttr_UNKNOWN;
}
if (found && (attr = tidyAttrGetById(child, attr_id)) != NULL) {
url = (char *) tidyAttrValue(attr);
string_free(relative_url);
if (url && *url)
add_new_url_and_check(elem, url, outfile ? &relative_url : NULL);
}
if (outfile && (name = tidyNodeGetName(child)) != NULL) {
fprintf(outfile, "%*.*s%s", indent, indent, "<", name);
for (attr = tidyAttrFirst(child); attr; attr = tidyAttrNext(attr)) {
fprintf(outfile, " %s", tidyAttrName(attr));
if (relative_url && (tidyAttrGetId(attr) == attr_id))
fprintf(outfile, "=\"%s\"", relative_url);
else if (tidyAttrValue(attr))
fprintf(outfile, "=\"%s\"", tidyAttrValue(attr) ? tidyAttrValue(attr) : "");
else
fprintf(outfile, "=\"\"");
}
string_free(relative_url);
if (node_type == TidyNode_StartEnd)
fprintf(outfile, "/>\n");
else {
fprintf(outfile, ">\n");
parse_html(tdoc, child, elem, indent + 1, outfile);
fprintf(outfile, "%*.*s%s>\n", indent + 1, indent + 1, "</", name);
}
}
else {
string_free(relative_url);
parse_html(tdoc, child, elem, indent + 1, outfile);
}
break;
case TidyNode_End:
if (outfile) {
if ((name = tidyNodeGetName(child)) != NULL)
fprintf(outfile, "%*.*s/%s>\n", indent, indent, "<", name);
}
break;
case TidyNode_Text:
if (outfile) {
TidyBuffer buf;
TidyTagId parent_node_id = tidyNodeGetId(tnod);
tidyBufInit(&buf);
if (parent_node_id == TidyTag_SCRIPT || parent_node_id == TidyTag_STYLE)
tidyNodeGetValue(tdoc, child, &buf);
else
tidyNodeGetText(tdoc, child, &buf);
if (buf.bp)
fprintf(outfile, "%s", (char *)buf.bp);
tidyBufFree(&buf);
}
break;
case TidyNode_Comment:
if (outfile) {
TidyBuffer buf;
tidyBufInit(&buf);
tidyNodeGetValue(tdoc, child, &buf);
if (buf.bp)
fprintf(outfile, "<!--%s-->\n", (char *)buf.bp);
tidyBufFree(&buf);
}
break;
case TidyNode_CDATA:
if (outfile) {
TidyBuffer buf;
tidyBufInit(&buf);
tidyNodeGetValue(tdoc, child, &buf);
if (buf.bp)
fprintf(outfile, "<![CDATA[%s]]>\n", (char *)buf.bp);
tidyBufFree(&buf);
}
break;
case TidyNode_DocType:
if (outfile) {
int pub = 0;
fprintf(outfile, "<!DOCTYPE %s", tidyNodeGetName(child));
for (attr = tidyAttrFirst(child); attr; attr = tidyAttrNext(attr)) {
if (!pub) {
fprintf(outfile, " %s", tidyAttrName(attr));
if (!string_casecmp(tidyAttrName(attr), "PUBLIC"))
pub = 1;
}
if (tidyAttrValue(attr))
fprintf(outfile, " \"%s\"", tidyAttrValue(attr));
}
fprintf(outfile, ">\n");
}
break;
default:
if (outfile) {
TidyBuffer buf;
tidyBufInit(&buf);
tidyNodeGetValue(tdoc, child, &buf);
if (buf.bp)
fprintf(outfile, "%s", (char *)buf.bp);
tidyBufFree(&buf);
}
break;
}
}
}
