/usr/share/systemtap/tapset
Edit: /usr/share/systemtap/tapset/regex.stp (4107B)
// Regular expression subexpression tapset
// Copyright (C) 2017 Serhei Makarov
// Copyright (C) 2013 Red Hat, Inc.
//
// This file is part of systemtap, and is free software. You can
// redistribute it and/or modify it under the terms of the GNU General
// Public License (GPL); either version 2, or (at your option) any
// later version.
%{
#define STAP_NEED_CONTEXT_SUBEXPRESSION 1
%}
/**
* sfunction matched_str - Return the last matched string.
*
* Description: returns the string matched by the last successful
* use of the =~ regexp matching operator. Returns an error if the
* last use of =~ led to a failed match.
*/
function matched_str:string() { return matched(0) }
/**
* sfunction matched - Return a given matched subexpression.
*
* @n: index to the subexpression to return. 0 corresponds to the
* entire regular expression.
*
* Description: returns the content of the n'th subexpression of the
* last successful use of the =~ regex matching operator. Returns an
* empty string if the n'th subexpression was not matched (e.g. due to
* alternation). Throws an error if the last use of =~ was a failed
* match, or if fewer than n subexpressions are present in the
* original regexp.
*/
function matched:string(n:long)
%{ /* pure */ /* unprivileged */ /* pragma:tagged_dfa */
int start_ix, end_ix; // indices into tag buffer
int start, end, length; // actual coordinate values
if (!CONTEXT->last_match.result) {
snprintf(CONTEXT->error_buffer, sizeof(CONTEXT->error_buffer),
"Attempted to get subexpression %lld from failed match", (long long) STAP_ARG_n);
CONTEXT->last_error = CONTEXT->error_buffer;
}
start_ix = 2 * STAP_ARG_n, end_ix = start_ix + 1;
if (end_ix >= CONTEXT->last_match.num_final_tags) {
snprintf(CONTEXT->error_buffer, sizeof(CONTEXT->error_buffer),
"Attempted to get nonexistent subexpression %lld", (long long) STAP_ARG_n);
CONTEXT->last_error = CONTEXT->error_buffer;
}
start = CONTEXT->last_match.tag_vals[start_ix];
end = CONTEXT->last_match.tag_vals[end_ix];
// _stp_printf ("**DEBUG** Extracted subexpression #%lld:(%d,%d) from %d to %d\n", STAP_ARG_n, start_ix, end_ix, start, end);
if (start < 0 || end < 0) {
// If indices are negative, the group was not matched. Return empty string:
start = end = 0;
// snprintf(CONTEXT->error_buffer, sizeof(CONTEXT->error_buffer),
// "Unknown coordinates for subexpression %lld", STAP_ARG_n);
// CONTEXT->last_error = CONTEXT->error_buffer;
}
if (start > end) {
// This should not happen.
snprintf(CONTEXT->error_buffer, sizeof(CONTEXT->error_buffer),
"BUG: inverted coordinates for subexpression %lld", (long long) STAP_ARG_n);
CONTEXT->last_error = CONTEXT->error_buffer;
}
length = end - start;
// TODOXXX assert (start <= strlen(matched_str)) ??
// XXX: Must add 1 to length to account for NUL byte in strlcpy().
strlcpy(STAP_RETVALUE, CONTEXT->last_match.matched_str + start, length + 1);
%}
/**
* sfunction ngroups - Number of subexpressions in the last match.
*
* Description: returns the number of subexpressions from the
* last successful use of the =~ regex matching operator.
*
* Note that this number includes subexpressions which are present in
* the regex but did not match any string; for example, given the
* regex "a|(b)", the subexpressions will count the group for (b)
* regardless of whether it matched a string or not. Throws an error
* if the last use of =~ was a failed match.
*/
function ngroups:long()
%{ /* pure */ /* unprivileged */ /* pragma:tagged_dfa */
if (!CONTEXT->last_match.result) {
snprintf(CONTEXT->error_buffer, sizeof(CONTEXT->error_buffer),
"Attempted to get subexpression count from failed match");
CONTEXT->last_error = CONTEXT->error_buffer;
}
STAP_RETVALUE = CONTEXT->last_match.num_final_tags / 2;
%}
// XXX: perhaps implement matched_start, matched_end to get indices?
// XXX: some kind of find-replace functionality?
// XXX: some kind of splitting / multiple-match functionality?