[pkgsrc/trunk]: pkgsrc/pkgtools/pkglint/files - Improved the parsers for shel...

[rillig <rillig%pkgsrc.org@localhost>]

details:   https://anonhg.NetBSD.org/pkgsrc/rev/3533bf8d45a6
branches:  trunk
changeset: 506126:3533bf8d45a6
user:      rillig <rillig%pkgsrc.org@localhost>
date:      Thu Jan 12 13:01:47 2006 +0000

description:
- Improved the parsers for shell words and shell commands. Parsing errors
  are only shown in -Wdebug mode, as the diagnostic messages are too
  cryptic.
- Added detection for the $(...) replacement for backticks, as the Solaris
  /bin/sh cannot handle them.

diffstat:

 pkgtools/pkglint/files/pkglint.pl |  52 +++++++++++++++++++++++++-------------
 1 files changed, 34 insertions(+), 18 deletions(-)

diffs (119 lines):

diff -r 4800ad7f9a7b -r 3533bf8d45a6 pkgtools/pkglint/files/pkglint.pl
--- a/pkgtools/pkglint/files/pkglint.pl Thu Jan 12 12:59:15 2006 +0000
+++ b/pkgtools/pkglint/files/pkglint.pl Thu Jan 12 13:01:47 2006 +0000
@@ -1,5 +1,5 @@
 #! @PERL@ -w
-# $NetBSD: pkglint.pl,v 1.466 2006/01/12 10:39:30 rillig Exp $
+# $NetBSD: pkglint.pl,v 1.467 2006/01/12 13:01:47 rillig Exp $
 #
 
 # pkglint - static analyzer and checker for pkgsrc packages
@@ -1019,12 +1019,15 @@
 # it cannot parse all kinds of shell programs, but this pattern will
 # catch all shell programs that are portable enough to be used in pkgsrc.
 my $regex_shellword            =  qr"\s*(
-       (?:     '[^']*'
-       |       \"(?:\\.|[^\"\\])*\"
-       |       \`[^\`]*\`
-       |       \\.
-       |       \$\{[^{}]+\}
-       |       [^\(\)'\"\\\s;&\|<>\#]
+       (?:     '[^']*'                 # single quoted string
+       |       \"(?:\\.|[^\"\\])*\"    # double quoted string
+       |       \`[^\`]*\`              # backticks string
+       |       \\.                     # any escaped character
+       |       \$\{[^{}]+\}            # make(1) variable
+       |       \$\$[0-9A-Z_a-z]+       # shell variable
+       |       \$\$\{[0-9A-Z_a-z]+\}   # shell variable in braces
+       |       \$\$\(                  # POSIX-style backticks replacement
+       |       [^\(\)'\"\\\s;&\|<>\#\`\$] # non-special character
        )+ | ;;? | &&? | \|\|? | \( | \) | <<? | >>? | \#.*)"sx;
 my $regex_varname              = qr"[-*+.0-9A-Z_a-z{}\[]+";
 
@@ -2003,20 +2006,26 @@
                        }
 
                } elsif ($state == SWST_PLAIN) {
-                       if ($rest =~ s/^[!&\(\)*+,\-.\/0-9:;<=>?\@A-Z\[\]_a-z|~]+//) {
+                       if ($rest =~ s/^[!\%&\(\)*+,\-.\/0-9:;<=>?\@A-Z\[\]_a-z{|}~]+//) {
                        } elsif ($rest =~ s/^\'//) {
                                $state = SWST_SQUOT;
                        } elsif ($rest =~ s/^\"//) {
                                $state = SWST_DQUOT;
                        } elsif ($rest =~ s/^\`//) {
                                $state = SWST_BACKT;
-                       } elsif ($rest =~ s/^\\[ "'\(\)*;]//) {
+                       } elsif ($rest =~ s/^\\[ !"#'\(\)*;^{}]//) {
                        } elsif ($rest =~ s/^\$\$([0-9A-Z_a-z]+)//
                            || $rest =~ s/^\$\$\{([0-9A-Z_a-z]+)\}//) {
                                my ($shvarname) = ($1);
-                               if ($opt_warn_quoting and $check_quoting) {
+                               if ($opt_warn_quoting && $check_quoting) {
                                        $line->log_warning("Unquoted shell variable \"${shvarname}\".");
                                }
+                       } elsif ($rest =~ s/\$\$\(/(/) {
+                               $line->log_warning("Unportable subshell call via \$(...\).");
+                               $line->explain(
+                                       "The Solaris /bin/sh does not know this way to execute a command in a",
+                                       "subshell. Please use backticks (\`...\`) as a replacement.");
+
                        } else {
                                last;
                        }
@@ -2034,17 +2043,17 @@
                        if ($rest =~ s/^\"//) {
                                $state = SWST_PLAIN;
                        } elsif ($rest =~ s/^[^\$"\\\`]//) {
-                       } elsif ($rest =~ s/^\\[\\\"\`\$]//) {
+                       } elsif ($rest =~ s/^\\(?:[\\\"\`]|\$\$)//) {
                        } elsif ($rest =~ s/^\$\$\{([0-9A-Za-z_]+)\}//
                            || $rest =~ s/^\$\$([0-9A-Za-z_]+)//) {
                                my ($varname) = ($1);
                                $line->log_debug("[checkline_mk_shellword] Found double-quoted variable ${varname}.");
-                       } elsif ($rest =~ s/^(\\[\(\)n])//) {
-                               my ($paren) = ($1);
-                               $line->log_warning("Please use \"\\${paren}\" instead of \"${paren}\".");
+                       } elsif ($rest =~ s/^\\([\(\)*.0-9n])//) {
+                               my ($char) = ($1);
+                               $line->log_warning("Please use \"\\\\${char}\" instead of \"\\${char}\".");
                                $line->explain(
                                        "Although the current code may work, it is not good style to rely on",
-                                       "the shell passing \"\\${paren}\" exactly as is, and not discarding the",
+                                       "the shell passing \"\\${char}\" exactly as is, and not discarding the",
                                        "backslash. Alternatively you can use single quotes instead of double",
                                        "quotes.");
                        } else {
@@ -2055,6 +2064,13 @@
                        if ($rest =~ s/^\`//) {
                                $state = SWST_PLAIN;
                        } elsif ($rest =~ s/^[^\\\"\'\`\$]+//) {
+                       } elsif ($rest =~ s/^\$\$\{([0-9A-Za-z_]+)\}//
+                           || $rest =~ s/^\$\$([0-9A-Za-z_]+)//) {
+                               my ($shvarname) = ($1);
+                               if ($opt_warn_quoting && $check_quoting) {
+                                       $line->log_warning("Unquoted shell variable \$${shvarname}.");
+                               }
+
                        } else {
                                last;
                        }
@@ -2220,8 +2236,8 @@
                        };
        }
 
-       if ($rest !~ qr"^\s*$") {
-               $line->log_warning("Invalid shell word \"${text}\".");
+       if ($rest ne "") {
+               $opt_warn_debug && $line->log_error("[checkline_mk_shelltext] " . scst_statename->[$state] . ": rest=${rest}");
        }
 }
 
@@ -2661,7 +2677,7 @@
                        }
 
                        if ($rest !~ qr"^\s*$") {
-                               $line->log_warning("Invalid shell word \"${value}\" at the end.");
+                               $opt_warn_debug and $line->log_warning("Invalid shell word \"${value}\" at the end.");
                        }
 
                } else {