More data extraction

This commit is contained in:
Roberto Alsina 2023-07-11 20:49:33 -03:00
parent a4f98c56bc
commit 179ee55826

View File

@ -18,27 +18,17 @@
%{ shortcodes[c_sc].name.start = mark-start; %{ shortcodes[c_sc].name.start = mark-start;
shortcodes[c_sc].name.len = p-mark; shortcodes[c_sc].name.len = p-mark;
shortcodes[c_sc].matching = 0; shortcodes[c_sc].matching = 0;
str_copyb(&new_name, mark, p-mark);
}; };
argname = alpha+ argname = alpha+
> mark > mark
% {str_cats(&output, "A "); % {};
str_catb(&output, mark, p-mark);
str_cats(&output, "\n");
};
qvalue = ('"' [^"]* '"') qvalue = ('"' [^"]* '"')
> mark > mark
% {str_cats(&output, "V "); % {};
str_catb(&output, mark+1, p-mark-2);
str_cats(&output, "\n");
};
value = alnum+ value = alnum+
> mark > mark
% {str_cats(&output, "V "); % {};
str_catb(&output, mark, p-mark);
str_cats(&output, "\n");
};
arg = ((argname '=')? (value|qvalue)); arg = ((argname '=')? (value|qvalue));
@ -56,7 +46,6 @@
shortcodes[c_sc].start = p-start-1; shortcodes[c_sc].start = p-start-1;
} }
@ { @ {
str_copy(&open_name, &new_name);
shortcodes[c_sc].len = p-start-shortcodes[c_sc].start+1; shortcodes[c_sc].len = p-start-shortcodes[c_sc].start+1;
data_mark = p+1; data_mark = p+1;
c_sc++; c_sc++;
@ -64,23 +53,36 @@
closing_shortcode = (start spc '/' name spc end) closing_shortcode = (start spc '/' name spc end)
> { > {
// Starting a shortcode, close data // Starting a shortcode, close data (if previous shortcode is opening)
shortcodes[c_sc-1].data.start = data_mark-start; // If it's closing, then previous text can't be data.
shortcodes[c_sc-1].data.len = p-data_mark-1; if (shortcodes[c_sc-1].matching == 0) {
str_copyb(&sc, start+shortcodes[c_sc-1].data.start, shortcodes[c_sc-1].data.len); shortcodes[c_sc-1].data.start = data_mark-start;
shortcodes[c_sc-1].data.len = p-data_mark-1;
}
}; };
matched_shortcode = (shortcode any* closing_shortcode) matched_shortcode = (shortcode any* closing_shortcode)
@ { @ {
shortcodes[c_sc-1].matching = 1; shortcodes[c_sc-1].matching = 1;
shortcodes[c_sc-1].len = p-start-shortcodes[c_sc-1].start + 1; shortcodes[c_sc-1].len = p-start-shortcodes[c_sc-1].start + 1;
shortcodes[c_sc].name.start = 0; if (
if (str_cmp(&open_name, 0, &new_name, 0) != 0) { shortcodes[c_sc-1].name.len != shortcodes[c_sc].name.len ||
strncmp(
start + shortcodes[c_sc-1].name.start,
start + shortcodes[c_sc].name.start,
shortcodes[c_sc-1].name.len) !=0)
{
printf("Mismatched tags!\n");
str_copyb(&sc,start + shortcodes[c_sc-1].name.start, shortcodes[c_sc-1].name.len);
printf("opened: %s\n", sc.s);
str_copyb(&sc,start + shortcodes[c_sc].name.start, shortcodes[c_sc].name.len);
printf("closed: %s\n", sc.s);
// Closing the wrong code // Closing the wrong code
// TODO error reporting // TODO error reporting
return; return;
} }
str_truncate(&open_name,0); // Reuse this shortcode entry for next one
shortcodes[c_sc].name.start = 0;
}; };
main := (any* (shortcode | matched_shortcode))*; main := (any* (shortcode | matched_shortcode))*;
@ -110,11 +112,6 @@ void parse(char *input) {
char *start = input; char *start = input;
char *p = input; char *p = input;
char *pe = p + strlen(input); char *pe = p + strlen(input);
str open_name, new_name, output, data;
str_init(&open_name);
str_init(&new_name);
str_init(&output);
str_init(&data);
shortcode shortcodes[1000]; shortcode shortcodes[1000];
int c_sc = 0; int c_sc = 0;
@ -143,7 +140,8 @@ void parse(char *input) {
} }
int main(int argc, char **argv) { int main(int argc, char **argv) {
parse("{{< c1 arg2 >}}foobar{{% /c1%}}{{% sarasa %}}"); parse(
"bbb{{% sarasa %}}ccc{{< c1 arg2 >}}foobar{{% /c1%}}aaa{{% sarasa %}}");
// if (output == 0) { // if (output == 0) {
// printf("parse error\n"); // printf("parse error\n");
// return 1; // return 1;