From 2be04e240deaac14500e079c956611e2c1d0c6f3 Mon Sep 17 00:00:00 2001
From: Alejandro Colomar <alx@kernel.org>
Date: Fri, 3 Nov 2023 17:41:22 +0100
Subject: bin/grepc: Optimize

Call pcre2grep(1) only once, which allows removing named pipes.  This is
an important optimization, and will also allow accepting several file
names in the command line (in a future commit).  The source code is also
significantly simplified.

It has a caveat: this single call to pcre2grep(1) will consume more
resources, and will crash on certain input files.  The workaround is to
restrict the search to just one or a few types of code.  Document this
in the manual page.  This caveat only applies to very specific files,
which so far I've only found in the Linux kernel source tree.
---
 bin/grepc              | 286 +++++++++----------------------------------------
 share/man/man1/grepc.1 |   7 ++
 2 files changed, 57 insertions(+), 236 deletions(-)

diff --git a/bin/grepc b/bin/grepc
index 05eeee4..1a7b519 100755
--- a/bin/grepc
+++ b/bin/grepc
@@ -1,4 +1,4 @@
-#!/bin/sh
+#!/bin/bash
 
 
 # Defaults:
@@ -183,26 +183,60 @@ grepc_parse_cmd()
 }
 
 
-grepc_helper()
+grepc_e()		{ echo '(?s)^([\w[]+[\w\s]*)?\benum\b[ \t]*([\w \t[\]]|::)*\n*([ \t]*){[^}]*^[ \t]*'"$1"'\b\s*[=,].*?^\3}.*?;'; }
+grepc_fp()		{ echo '(?s)^[\w[](?:[\w\s\(,\)[\]*]|::)+[\w\s\)*\]]\s+\**\(?'"$1"'\)?\s*(\((?:[\w\s,[\]*]|::|(?1))*(?:\.\.\.)?\))(?:[\w\s\(,\)[\]]|::)*;'; }
+grepc_fd()		{ echo '(?s)^[\w[](?:[\w\s\(,\)[\]*]|::)*[\w\s\)*\]]\s+\**\(?'"$1"'\)?\s*(\((?:[\w\s,[\]*]|::|(?1))*(?:\.\.\.)?\))[ \t]*\n([ \t]*){.*?^\2}'; }
+grepc_fgd_libm()	{ grepc_fd "M_DECL_FUNC \(__$1\)"; }
+grepc_fgp_libio()	{ grepc_fp "_IO_$1"; }
+grepc_fgd_libio()	{ grepc_fd "_IO_$1"; }
+grepc_fgp()		{ grepc_fgp_libio "$1"; }
+grepc_fsp()		{ echo '(?s)^asmlinkage\s+[\w\s]+\**sys_'"$1"'\s*\(.*?\)'; }
+grepc_fsd()		{ echo '(?s)^(COMPAT_)?SYSCALL_DEFINE.\('"$1"'\b.*?^}'; }
+grepc_mf()		{ echo '(?s)^[ \t]*#\s*define\s[\s\\]*'"$1"'\(.*?[^\\]$'; }
+grepc_mo()		{ echo '(?s)^[ \t]*#\s*define\s[\s\\]*'"$1"'\b(?!\().*?(?<!\\)$'; }
+grepc_t_braced()	{ echo '(?s)^([\w[]([\w\s\(,\)[\]*]|::)*[\w\s\)*\]]\s+)?\b(struct|union|enum)\b([\w \t[\]]|::)+\b'"$1"'\b[ \t]*\n*([ \t]*){.*?^\5}.*?;'; }
+grepc_t_td_simple()	{ echo '(?s)^[ \t]*typedef\s+[^{};]+'"$1"';'; }
+grepc_t_td_braced()	{ echo '(?s)^[ \t]*typedef\s+(struct|union|enum)\b(?:(?!\W'"$1"'\W)([\w \t[\]]|::))*\n*([ \t]*){(?:(?!^\3?}).)*?^\3}\s*'"$1"'(\[[\w\(,\)]\])*;'; }
+grepc_t_td_func()	{ echo '(?s)^[ \t]*typedef\s+[^{};]+\(\**'"$1"'\)\s*\([^{};]+;'; }
+grepc_ue()		{ echo '(?s)^([\w[]+[\w\s]*)?\benum\b([\w \t[\]]|::)*\n*([ \t]*){[^}]*^\s*\w+[\w\s[\]=]*'"$1"'.*?^\3}.*?;'; }
+grepc_uf_def()		{ echo '(?s)^[\w[]([\w\s\(,\)[\]*]|::)*[\w\s\)*\]]\s+\**\w+\s*\(([\w\s\(,\)[\]*]|::)+?(\.\.\.)?\)[ \t]*\n*([ \t]*){(?:(?!^\4?}).)*'"$1"'.*?^\4}'; }
+grepc_linux_use_func_syscall_def() { echo '(?s)^(COMPAT_)?SYSCALL_DEFINE.\(\w+\b(?:(?!^}).)*'"$1"'.?^}'; }
+grepc_uf_linux_def()	{ grepc_linux_use_func_syscall_def "$1"; }
+grepc_um()		{ echo '(?s)^[ \t]*#\s*define\s[\s\\]*\w+\b(\([^\)]*\))?(?:(?![^\\]$).)*'"$1"'.*?[^\\]$'; }
+grepc_ut_su()		{ echo '(?s)^(?!^[ \t]*typedef\b)([\w[]([\w\s\(,\)[\]*]|::)*[\w\s\)*\]]\s+)?\b(struct|union)\b([\w \t[\]]|::)*\w+[ \t]*\n*([ \t]*){(?:(?!^\5?}).)*?'"$1"'.*?^\5}.*?;'; }
+grepc_ut_td_simple()	{ echo '(?s)^[ \t]*typedef\s+[^{};]*'"$1"'[^{};]+;'; }
+grepc_ut_td_su()	{ echo '(?s)^[ \t]*typedef\s+(struct|union)\b([\w \t[\]]|::)*\n*([ \t]*){(?:(?!^\3?}|^\s*typedef).)*'"$1"'(?:(?!^\3?}|^\s*typedef).)*^\3}\s*\w+;'; }
+
+
+grepc_patterns()
 {
-	grep -${iflag}zP -- "$1" \
-	| grep -${iflag}zP -- "$2" \
-	| head -c-1 \
-	| pcre2grep -${iflag}${lflag}HMn --label="$file" -- "$3" \
-	| if [ "$kflag" = 'no' ]; then
-		sed -E 's/^[^: ]+:[0-9]+:/\n\n&\n/';
-	else
-		cat;
-	fi;
+	test $t_e = yes && grepc_e "$1";
+	test $t_fp = yes && grepc_fp "$1";
+	test $t_fd = yes && grepc_fd "$1";
+	test $t_fsp = yes && grepc_fsp "$1";
+	test $t_fsd = yes && grepc_fsd "$1";
+	test $t_fgp = yes && grepc_fgp "$1";
+	test $t_fgd_libm = yes && grepc_fgd_libm "$1";
+	test $t_fgd_libio = yes && grepc_fgd_libio "$1";
+	test $t_mf = yes && grepc_mf "$1";
+	test $t_mo = yes && grepc_mo "$1";
+	test $t_t_braced = yes && grepc_t_braced "$1";
+	test $t_t_td_simple = yes && grepc_t_td_simple "$1";
+	test $t_t_td_braced = yes && grepc_t_td_braced "$1";
+	test $t_t_td_func = yes && grepc_t_td_func "$1";
+	test $t_ue = yes && grepc_ue "$1";
+	test $t_uf_def = yes && grepc_uf_def "$1";
+	test $t_uf_linux_def = yes && grepc_uf_linux_def "$1";
+	test $t_um = yes && grepc_um "$1";
+	test $t_ut_su = yes && grepc_ut_su "$1";
+	test $t_ut_td_simple = yes && grepc_ut_td_simple "$1";
+	test $t_ut_td_su = yes && grepc_ut_td_su "$1";
 }
 
 
-grepc_helper_use()
+grepc_search()
 {
-	grep -${iflag}zP -- "$1" \
-	| grep -${iflag}zP -- "$2" \
-	| head -c-1 \
-	| pcre2grep -${iflag}${lflag}HMn --label="$file" -- "$3" \
+	pcre2grep -${iflag}${lflag}HMn --label="$file" -f <(grepc_patterns "$identifier") \
 	| if [ "$kflag" = 'no' ]; then
 		sed -E 's/^[^: ]+:[0-9]+:/\n\n&\n/';
 	else
@@ -216,226 +250,6 @@ grepc_helper_use()
 }
 
 
-grepc_e()
-{
-	grepc_helper \
-	  '\benum\b' \
-	  "^[ \t]*$1\b\s*[,=]" \
-	  '(?s)^([\w[]+[\w\s]*)?\benum\b[ \t]*([\w \t[\]]|::)*\n*([ \t]*){[^}]*^[ \t]*'"$1"'\b\s*[=,].*?^\3}.*?;';
-}
-
-
-grepc_fp()
-{
-	grepc_helper \
-	  "(\($1\)|\b$1)\s*\(" \
-	  '.' \
-	  '(?s)^[\w[](?:[\w\s\(,\)[\]*]|::)+[\w\s\)*\]]\s+\**\(?'"$1"'\)?\s*(\((?:[\w\s,[\]*]|::|(?1))*(?:\.\.\.)?\))(?:[\w\s\(,\)[\]]|::)*;';
-}
-
-
-grepc_fd()
-{
-	grepc_helper \
-	  "(\($1\)|\b$1)\s*\(" \
-	  '.' \
-	  '(?s)^[\w[](?:[\w\s\(,\)[\]*]|::)*[\w\s\)*\]]\s+\**\(?'"$1"'\)?\s*(\((?:[\w\s,[\]*]|::|(?1))*(?:\.\.\.)?\))[ \t]*\n([ \t]*){.*?^\2}';
-}
-
-
-grepc_fgd_libm()  { grepc_fd "M_DECL_FUNC \(__$1\)"; }
-grepc_fgp_libio() { grepc_fp "_IO_$1"; }
-grepc_fgd_libio() { grepc_fd "_IO_$1"; }
-grepc_fgp()       { grepc_fgp_libio "$1"; }
-
-
-grepc_fsp()
-{
-	grepc_helper \
-	  "^asmlinkage\s+[\w\s]+\**sys_$1\s*\(" \
-	  '.' \
-	  '(?s)^asmlinkage\s+[\w\s]+\**'"sys_$1"'\s*\(.*?\)';
-}
-
-
-grepc_fsd()
-{
-	grepc_helper \
-	  "SYSCALL_DEFINE.\($1\b" \
-	  '.' \
-	  '(?s)^(COMPAT_)?SYSCALL_DEFINE.\('"$1"'\b.*?^}';
-}
-
-
-grepc_mf()
-{
-	grepc_helper \
-	  "#\s*define\s+$1\(" \
-	  '.' \
-	  '(?s)^[ \t]*#\s*define\s[\s\\]*'"$1"'\(.*?[^\\]$';
-}
-
-
-grepc_mo()
-{
-	grepc_helper \
-	  "#\s*define\s+$1\b(?!\()" \
-	  '.' \
-	  '(?s)^[ \t]*#\s*define\s[\s\\]*'"$1"'\b(?!\().*?(?<!\\)$';
-}
-
-
-grepc_t_braced()
-{
-	grepc_helper \
-	  '\b(struct|union|enum)\b([\w \t[\]]|::)+\b'"$1"'\b' \
-	  '.' \
-	  '(?s)^([\w[]([\w\s\(,\)[\]*]|::)*[\w\s\)*\]]\s+)?\b(struct|union|enum)\b([\w \t[\]]|::)+\b'"$1"'\b[ \t]*\n*([ \t]*){.*?^\5}.*?;';
-}
-
-
-grepc_t_td_simple()
-{
-	grepc_helper \
-	  '^[ \t]*typedef\s' \
-	  "\b$1;" \
-	  '(?s)^[ \t]*typedef\s+[^{};]+'"$1"';';
-}
-
-
-grepc_t_td_braced()
-{
-	grepc_helper \
-	  '^[ \t]*typedef\s+(struct|union|enum)\b[^;]*$' \
-	  "^[ \t]*}\s*$1(\[[\w\(,\)]\])*;" \
-	  '(?s)^[ \t]*typedef\s+(struct|union|enum)\b(?:(?!\W'"$1"'\W)([\w \t[\]]|::))*\n*([ \t]*){(?:(?!^\3?}).)*?^\3}\s*'"$1"'(\[[\w\(,\)]\])*;';
-}
-
-
-grepc_t_td_func()
-{
-	grepc_helper \
-	  '^[ \t]*typedef\s' \
-	  "\(\**$1\)\s*\(" \
-	  '(?s)^[ \t]*typedef\s+[^{};]+\(\**'"$1"'\)\s*\([^{};]+;';
-}
-
-
-grepc_ue()
-{
-	grepc_helper_use \
-	  '\benum\b' \
-	  "\b$1\b" \
-	  '(?s)^([\w[]+[\w\s]*)?\benum\b([\w \t[\]]|::)*\n*([ \t]*){[^}]*^\s*\w+[\w\s[\]=]*'"$1"'.*?^\3}.*?;';
-}
-
-
-grepc_uf_def()
-{
-	grepc_helper_use \
-	  "\b$1\b" \
-	  '.' \
-	  '(?s)^[\w[]([\w\s\(,\)[\]*]|::)*[\w\s\)*\]]\s+\**\w+\s*\(([\w\s\(,\)[\]*]|::)+?(\.\.\.)?\)[ \t]*\n*([ \t]*){(?:(?!^\4?}).)*'"$1"'.*?^\4}';
-}
-
-
-grepc_linux_use_func_syscall_def()
-{
-	grepc_helper_use \
-	  "SYSCALL_DEFINE.\(" \
-	  "\b$1\b" \
-	  '(?s)^(COMPAT_)?SYSCALL_DEFINE.\(\w+\b(?:(?!^}).)*'"$1"'.?^}';
-}
-
-
-grepc_uf_linux_def() { grepc_linux_use_func_syscall_def "$1"; }
-
-
-grepc_um()
-{
-	grepc_helper_use \
-	  "\b$1\b" \
-	  'define' \
-	  '(?s)^[ \t]*#\s*define\s[\s\\]*\w+\b(\([^\)]*\))?(?:(?![^\\]$).)*'"$1"'.*?[^\\]$';
-}
-
-
-grepc_ut_su()
-{
-	grepc_helper_use \
-	  "\b(struct|union)\b" \
-	  "\b$1\b" \
-	  '(?s)^(?!^[ \t]*typedef\b)([\w[]([\w\s\(,\)[\]*]|::)*[\w\s\)*\]]\s+)?\b(struct|union)\b([\w \t[\]]|::)*\w+[ \t]*\n*([ \t]*){(?:(?!^\5?}).)*?'"$1"'.*?^\5}.*?;';
-}
-
-
-grepc_ut_td_simple()
-{
-	grepc_helper_use \
-	  '^[ \t]*typedef\s' \
-	  "\b$1\b" \
-	  '(?s)^[ \t]*typedef\s+[^{};]*'"$1"'[^{};]+;';
-}
-
-
-grepc_ut_td_su()
-{
-	grepc_helper_use \
-	  '^[ \t]*typedef\s+(struct|union)\b[^;]*$' \
-	  "\b$1\b" \
-	  '(?s)^[ \t]*typedef\s+(struct|union)\b([\w \t[\]]|::)*\n*([ \t]*){(?:(?!^\3?}|^\s*typedef).)*'"$1"'(?:(?!^\3?}|^\s*typedef).)*^\3}\s*\w+;';
-}
-
-
-grepc_search()
-{
-	local t="";
-	local f="$(mktemp -u -t grepc.XXXXXX)";
-	local fi="";
-	local fo="";
-
-	t="$t e";
-	t="$t fp fd fsp fsd fgp fgd_libm fgd_libio";
-	t="$t mf mo";
-	t="$t t_braced t_td_simple t_td_braced t_td_func";
-	t="$t ue uf_def uf_linux_def um ut_su ut_td_simple ut_td_su";
-
-	for ti in $t; do
-		fi="$fi $f.$ti.i";
-		fo="$fo $f.$ti.o";
-	done;
-
-	mkfifo -m600 $fi $fo;
-	cat $fo &
-
-	if test $t_e = yes; then grepc_e "$1"; else cat >/dev/null & printf ''; fi <$f.e.i >$f.e.o &
-	if test $t_fp = yes; then grepc_fp "$1"; else cat >/dev/null & printf ''; fi <$f.fp.i >$f.fp.o &
-	if test $t_fd = yes; then grepc_fd "$1"; else cat >/dev/null & printf ''; fi <$f.fd.i >$f.fd.o &
-	if test $t_fsp = yes; then grepc_fsp "$1"; else cat >/dev/null & printf ''; fi <$f.fsp.i >$f.fsp.o &
-	if test $t_fsd = yes; then grepc_fsd "$1"; else cat >/dev/null & printf ''; fi <$f.fsd.i >$f.fsd.o &
-	if test $t_fgp = yes; then grepc_fgp "$1"; else cat >/dev/null & printf ''; fi <$f.fgp.i >$f.fgp.o &
-	if test $t_fgd_libm = yes; then grepc_fgd_libm "$1"; else cat >/dev/null & printf ''; fi <$f.fgd_libm.i >$f.fgd_libm.o &
-	if test $t_fgd_libio = yes; then grepc_fgd_libio "$1"; else cat >/dev/null & printf ''; fi <$f.fgd_libio.i >$f.fgd_libio.o &
-	if test $t_mf = yes; then grepc_mf "$1"; else cat >/dev/null & printf ''; fi <$f.mf.i >$f.mf.o &
-	if test $t_mo = yes; then grepc_mo "$1"; else cat >/dev/null & printf ''; fi <$f.mo.i >$f.mo.o &
-	if test $t_t_braced = yes; then grepc_t_braced "$1"; else cat >/dev/null & printf ''; fi <$f.t_braced.i >$f.t_braced.o &
-	if test $t_t_td_simple = yes; then grepc_t_td_simple "$1"; else cat >/dev/null & printf ''; fi <$f.t_td_simple.i >$f.t_td_simple.o &
-	if test $t_t_td_braced = yes; then grepc_t_td_braced "$1"; else cat >/dev/null & printf ''; fi <$f.t_td_braced.i >$f.t_td_braced.o &
-	if test $t_t_td_func = yes; then grepc_t_td_func "$1"; else cat >/dev/null & printf ''; fi <$f.t_td_func.i >$f.t_td_func.o &
-	if test $t_ue = yes; then grepc_ue "$1"; else cat >/dev/null & printf ''; fi <$f.ue.i >$f.ue.o &
-	if test $t_uf_def = yes; then grepc_uf_def "$1"; else cat >/dev/null & printf ''; fi <$f.uf_def.i >$f.uf_def.o &
-	if test $t_uf_linux_def = yes; then grepc_uf_linux_def "$1"; else cat >/dev/null & printf ''; fi <$f.uf_linux_def.i >$f.uf_linux_def.o &
-	if test $t_um = yes; then grepc_um "$1"; else cat >/dev/null & printf ''; fi <$f.um.i >$f.um.o &
-	if test $t_ut_su = yes; then grepc_ut_su "$1"; else cat >/dev/null & printf ''; fi <$f.ut_su.i >$f.ut_su.o &
-	if test $t_ut_td_simple = yes; then grepc_ut_td_simple "$1"; else cat >/dev/null & printf ''; fi <$f.ut_td_simple.i >$f.ut_td_simple.o &
-	if test $t_ut_td_su = yes; then grepc_ut_td_su "$1"; else cat >/dev/null & printf ''; fi <$f.ut_td_su.i >$f.ut_td_su.o &
-
-	tee $fi >/dev/null;
-	wait;
-	rm $fi $fo;
-}
-
-
 main()
 {
 	grepc_parse_cmd "$@" </dev/null;
diff --git a/share/man/man1/grepc.1 b/share/man/man1/grepc.1
index 2d9cde5..e80c648 100644
--- a/share/man/man1/grepc.1
+++ b/share/man/man1/grepc.1
@@ -151,6 +151,13 @@ This option can be passed multiple times
 to search for various types of code.
 Default:
 .BR "e f m t" .
+.SH CAVEATS
+In some cases,
+internal calls to
+.MR pcre2grep 1
+may fail after consuming too much resources.
+To solve that,
+restrict the "types of code" of your search.
 .SH EXAMPLES
 .EX
 .RB \(ti/src/nginx/unit$ " grepc nxt_sprintf;"
-- 
cgit v1.2.3