From 33839117e10bba8283971166f0cd2ea19aefd6c5 Mon Sep 17 00:00:00 2001
From: rofl0r <retnyg@gmx.net>
Date: Tue, 16 Oct 2012 04:57:36 +0200
Subject: [PATCH] factor out parser code into a standalone, reusable module

... and add a demo app using it. this one was used to find
C files that define functions that are specified in headers
in an attempt to find and disable libc replacement functions
in gnulib and libiberty.
---
 CParser-test.pl    |  15 +++++
 CParser.pm         | 185 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 find-in-headers.pl |  29 +++++++++
 3 files changed, 229 insertions(+)
 create mode 100755 CParser-test.pl
 create mode 100644 CParser.pm
 create mode 100755 find-in-headers.pl

diff --git a/CParser-test.pl b/CParser-test.pl
new file mode 100755
index 0000000..1772ae7
--- /dev/null
+++ b/CParser-test.pl
@@ -0,0 +1,15 @@
+#!/usr/bin/env perl
+use strict;
+use warnings;
+
+use File::Basename; 
+use Cwd 'abs_path';
+use lib dirname(abs_path($0));
+use CParser;
+use Data::Dump qw(dump);
+
+my $file = $ARGV[0] or die ("need filename");
+
+my $p = CParser->new($file);
+$p->parse();
+dump($p);
\ No newline at end of file
diff --git a/CParser.pm b/CParser.pm
new file mode 100644
index 0000000..395c266
--- /dev/null
+++ b/CParser.pm
@@ -0,0 +1,185 @@
+package CParser;
+
+use strict;
+use warnings;
+
+sub file_ext {
+	my $x = shift;
+	my $l = length($x);
+	$l-- while($l && substr($x, $l, 1) ne ".");
+	return substr($x, $l) if($l);
+	return "";
+}
+
+
+sub new {
+	my ($pkg, @args) = @_;
+	die("need C file name as arg") if(scalar(@args) != 1);
+	my $isheader = file_ext($args[0]) eq ".h";
+	my $self = {
+		cfile => $args[0],
+		isheader => $isheader,
+		printerrors => 0,
+
+		openbraces => 0,
+		incomment => 0,
+		line => "",
+
+		includes => [],
+		statics => [],
+		extern => [],
+		typedefs_macros => [],
+		funcs => undef,
+	};
+	bless $self, $pkg;
+}
+
+sub addfunc {
+	my ($self, $funcname, $code) = @_;
+	$self->{funcs}->{$funcname} = $code;
+}
+
+sub handlesub {
+	my $self = shift;
+	my $_ = shift;
+	my $name = "";
+	my $wasstatic = 0;
+	while(!$name) {
+		my $x = 0;
+		$x++ while(substr($_, $x, 1) !~ /\s/);
+		my $word = substr($_, 0, $x);
+		my $extern = 1 if($word eq "extern");
+		if($word eq "static" || $word eq "inline") {
+			$_ = substr($_, $x);
+			s/^\s+//;
+			$wasstatic = 1 unless $extern;
+			next;
+		} else {
+			if(/(.*?)([\w_]+)\s*\(([\w\s_,\*\[\]\.\(\)]*?)\)\s*\{/
+				|| ($self->{isheader} &&
+					/(.*?)([\w_]+)\s*\((.*?)\)\s*;/)
+			) {
+				$name = $2;
+				my $decl = $1 . $name . "(" . $3 . ");";
+				push @{$self->{statics}}, $decl if($wasstatic);
+				#print $name , "\n" if $wasstatic;
+				$self->addfunc($name, $_);
+				#print "function $name\n$_";
+			} elsif($self->{isheader}) {
+				return;
+			} else {
+				warn "ERROR: $_\n" if $self->{printerrors};
+				return;
+			}
+		}
+	}
+}
+
+sub scanbraces {
+	my $self = shift;
+	my $shit = shift;
+	my @chars = split //, $shit;
+	for my $c(@chars) {
+		if ($c eq "{") {
+			$self->{openbraces}++;
+		} elsif($c eq "}") {
+			$self->{openbraces}--;
+		}
+	}
+}
+
+sub parseline {
+	my $self = shift;
+	$_ = shift;
+	#printf "parse line: %s, is header: %d\n", $_, $self->{isheader};
+	#print "PL: length line: ". length($line) . "\n";
+	return unless defined $_;
+	return if $_ eq "";
+	if(/^\s*#\s*(\w+)/) {
+		my $kw = $1;
+		if($kw eq "if" || $kw eq "ifdef" || $kw eq "elif" || $kw eq "else" || $kw eq "endif") {
+			push @{$self->{typedefs_macros}}, $_;
+			return;
+		}
+#		$self->{line} = "" if $self->{isheader};
+#		return;
+	}
+	if($_ =~ /extern \"C\"/) {
+#		$self->{line} = "";
+		return;
+	}
+	#$self->{line} .= $_ . "\n" if(!$self->{openbraces} || $self->{line} ne "");
+	#printf "%d\n", $self->{openbraces};
+	$self->{line} .= $_ . "\n" if(!$self->{openbraces} || $self->{line} ne "");
+	$self->scanbraces($_) unless $self->{line} =~ /^\s*#define/;
+
+	#print "$_ , line is $self->{line}\n";
+	
+	if($self->{line} ne "" && !$self->{openbraces}) {
+		#print "A $self->{line}\n";
+		if($self->{line} =~ /^\s*#/) {
+			if($self->{line} =~ /\\\s*$/) {
+
+			} else {
+				push @{$self->{typedefs_macros}}, $self->{line};
+				$self->{line} = ""
+			}
+		} elsif($self->{line} =~ /([;\}]{1})\s*\n*$/) {
+			if($1 eq ";") {
+				#print $self->{line};
+				if ($self->{line} =~ /=/ || $self->{line} =~ /^\s*static[\s\n]+/) {
+					#print "extern!\n";
+					$self->{line} =~ s/^\s*static\s*//;
+					push @{$self->{extern}}, $self->{line};
+				} elsif($self->{isheader}) {
+					$self->handlesub($self->{line});
+				} else {
+					push @{$self->{typedefs_macros}}, $self->{line};
+				}
+				$self->{line} = "";
+				return;
+			}
+			$self->handlesub($self->{line});
+			$self->{line} = "";
+		} 
+	} #elsif($self->{isheader} && !$self->{openbraces} && $self->{line} eq "" &&
+	#	/(extern\w){0, 1}
+}
+
+sub parse {
+
+	my $self = shift;
+
+	my $f;
+	open($f, "<", $self->{cfile});
+
+	while(<$f>) {
+		
+	#	print;
+		chomp;
+	#	print "$openbraces, $incomment\n";
+		if (/^\s*#\s*include\s+[<\"]{1}[\w_\-\/\.]+[>\"]{1}/) {
+			push @{$self->{includes}}, $_;
+		} else {
+			next if(/^\s*$/); #skip empty lines
+			next if(/^\s*\/\//); #skip one line comments.
+			# normal source code line.
+			if (!$self->{incomment} && /(.*?)\/\*(.*?)$/) {
+				$self->parseline($1);
+				my $rest = $2;
+				$self->{incomment} = 1 unless $rest =~ /\*\//;
+			} elsif($self->{incomment}) {
+				if(/\*\/(.*?)$/) {
+					$self->parseline($2);
+					$self->{incomment} = 0;
+				}
+			} else {
+				$self->parseline($_);
+			}
+		}
+	}
+	close $f;
+
+}
+
+1;
\ No newline at end of file
diff --git a/find-in-headers.pl b/find-in-headers.pl
new file mode 100755
index 0000000..d9e575e
--- /dev/null
+++ b/find-in-headers.pl
@@ -0,0 +1,29 @@
+#!/usr/bin/env perl
+use strict;
+use warnings;
+
+use File::Basename; 
+use Cwd 'abs_path';
+use lib dirname(abs_path($0));
+use CParser;
+use Data::Dump qw(dump);
+
+my $file = $ARGV[0] or die ("need filename");
+my $dir = $ARGV[1] or die ("need dir with headers as second arg");
+
+my $p = CParser->new($file);
+$p->parse();
+
+my @hdrs=`find $dir -name '*.h'`;
+for(@hdrs) {
+	chomp;
+	#print "processing $_\n";
+	#next unless /unistd/; 
+	my $h = CParser->new($_);
+	$h->parse();
+	#dump $h;
+	#exit;
+	for(keys %{$p->{funcs}}) {
+		print "$_ defined in $file and $h->{cfile}\n" if(defined($h->{funcs}->{$_}));
+	}
+}
-- 
2.11.4.GIT