Improved baseline/current/diff error checking. (#397)

henry2cox · web-flow · commit 5186f7394417 · 2025-03-28T10:14:15.000-04:00
Signed-off-by: Henry Cox &lt;henry.cox@mediatek.com&gt;
diff --git a/bin/genhtml b/bin/genhtml
@@ -3479,7 +3479,7 @@ sub new
 {
     my $class = shift;
     my $self = [{},          # linemap
-                {},          # filemap
+                {},          # filemap: new_filename->old_filename
                 {},          # baseline: filename -> baseline lineno -> text
                 undef,       # diff filename
                 [{}, {}],    # def location
@@ -3497,10 +3497,11 @@ sub load
     my ($self, $path, $info, $buildDirs) = @_;
     $self->_read_udiff($path);
 
+    # find list of soft links in [buildDirs] which may point to files in $info
+    #  we need to keep all the aliases as there may be files in baseline
+    #  which are not in current
     if (@$buildDirs) {
-        # find list of soft links in [buildDirs] which point to files in $info
         my @stack = @$buildDirs;
-        my %alias;
         while (@stack) {
             my $dir = pop(@stack);
             die("unexpected non-directory '$dir'") unless -d $dir;
@@ -3515,7 +3516,10 @@ sub load
                     push(@stack, $path);
                 } elsif (-l $path) {
                     my $l = Cwd::realpath($path);
-                    $alias{$path} = $l if (-f $l);
+                    next if (!-e $l || TraceFile::skipCurrentFile($l));
+                    # may need to turn $l into relative path??
+                    $self->[ALIASES]->{$path} = $l if (-f $l);
+                    #lcovutil::info("add alias '$path' -> '$l'\n");
                     # really, this should be a file...
                     die("unexpected soft link $path to directory")
                         unless -f $l;
@@ -3524,17 +3528,6 @@ sub load
             }
             closedir($dh);
         }
-        # now look through list of filenames in the current .info file, and see
-        #  if any match the soft links found above
-        foreach my $f ($info->files()) {
-            my $path = $f;
-            $path = File::Spec->catfile($main::cwd, $f)
-                unless File::Spec->file_name_is_absolute($f);
-            if (exists($alias{$path})) {
-                #lcovutil::info("found alias $alias{$path} of $f\n");
-                $self->[ALIASES]->{$path} = $alias{$path};
-            }
-        }
     }
     return $self;
 }
@@ -3772,20 +3765,23 @@ sub check_path_consistency
     # That is:  files whose version differs should appear in the diff data
     $self->check_version_match($baseline, $current);
 
-    my %diffMap;
-    my %diffBaseMap;
+    my %diffMap;        # current_filename -> where_used
+    my %diffBaseMap;    # current_directory -> []
+                        # for every 'current' filename in the udiff file
     foreach my $f ($self->files()) {
-        $diffMap{$f} = 0;
+        $diffMap{$f} = 0;    # this file not used in baseline or current (yet)
         my $b = File::Basename::basename($f);
         $diffBaseMap{$b} = [[], {}]
             unless exists($diffBaseMap{$b});
         push(@{$diffBaseMap{$b}->[0]}, $f);
     }
+    # foreach unchanged file in udiff data
     foreach my $f (keys %{$self->[UNCHANGED]}) {
         # unchanged in baseline and current
         $diffMap{$f} = 3;
     }
     my %missed;
+    # for each file in 'current' info:
     foreach my $curr ($current->files()) {
         my $b = File::Basename::basename($curr);
         if ($self->containsFile($curr)) {
@@ -3796,10 +3792,12 @@ sub check_path_consistency
             $diffBaseMap{$b}->[1]->{$alias} = 0
                 unless exists($diffBaseMap{$b}->[1]->{$alias});
             ++$diffBaseMap{$b}->[1]->{$alias};
-        } else {
+        } elsif (!exists($self->[UNCHANGED]->{$self->findName($curr)})) {
             $missed{$curr} = 1;       # in current but not in diff
+            #lcovutil::info("missed curr: $curr\n");
         }
     }
+    # for each file in 'baseline' info:
     foreach my $base ($baseline->files()) {
         my $b = File::Basename::basename($base);
         if ($self->containsFile($base)) {
@@ -3810,8 +3808,9 @@ sub check_path_consistency
             $diffBaseMap{$b}->[1]->{$alias} = 0
                 unless exists($diffBaseMap{$b}->[1]->{$alias});
             ++$diffBaseMap{$b}->[1]->{$alias};
-        } else {
+        } elsif (!exists($self->[UNCHANGED]->{$self->findName($base)})) {
             # in baseline but not in diff
+            #lcovutil::info("missed base: $base\n");
             if (exists($missed{$base})) {
                 $missed{$base} |= 2;
             } else {
@@ -3822,45 +3821,46 @@ sub check_path_consistency
     my $ok = 1;
     foreach my $f (sort keys(%missed)) {
         my $b = File::Basename::basename($f);
-        if (exists($diffBaseMap{$b})) {
-            # this basename is in the diff file and didn't match any other
-            #   trace filename entry (i.e., same filename in more than one
-            #   source code directory) - then warn about possible pathname issue
-            my ($diffFiles, $sourceFiles) = @{$diffBaseMap{$b}};
-            # find the files which appear in the 'diff' list which have the
-            #   same basename and were not matched - those might be candidates
-            my @unused;
-            for my $d (@$diffFiles) {
-                # my $location = $self->[DIFF_FILENAME] . ':' . $self->[LOCATION]->[NEW]->{$d};
-                push(@unused, $d)
-                    unless exists($sourceFiles->{$d});
-            }
-            if (scalar(@unused)) {
-                my $type;
-
-                # my $baseData = $baseline->data($f);
-                # my $baseLocation = join(":", ${$baseData->location()});
-                # my $currData = $current->data($f);
-                # my $currLocation = join(":", ${$currData->location()});
-
-                if (2 == $missed{$f}) {
-                    $type = "baseline";
-                } elsif (1 == $missed{$f}) {
-                    $type = "current";
-                } else {
-                    $type = "both baseline and current";
-                }
-                my $single = 1 == scalar(@unused);
-                # @todo could print line numbers in baseline, current .info files and
-                #   in diff file ..
-                if (lcovutil::warn_once($lcovutil::ERROR_MISMATCH, $f)) {
-                    my $suffix =
-                        $main::elide_path_mismatch ? ' (elided)' :
-                        lcovutil::explain_once(
-                        'elide-path-mismatch',
-                        '  Perhaps see "--elide-path-mismatch", "--substitute" and "--build-directory" options in \'man '
-                            . $lcovutil::tool_name . '\'');
-                    lcovutil::ignorable_warning(
+        next unless exists($diffBaseMap{$b});
+
+        # this basename is in the diff file and didn't match any other
+        #   trace filename entry (i.e., same filename in more than one
+        #   source code directory) - then warn about possible pathname issue
+        my ($diffFiles, $sourceFiles) = @{$diffBaseMap{$b}};
+        # find the files which appear in the 'diff' list which have the
+        #   same basename and were not matched - those might be candidates
+        my @unused;
+        for my $d (@$diffFiles) {
+            # my $location = $self->[DIFF_FILENAME] . ':' . $self->[LOCATION]->[NEW]->{$d};
+            push(@unused, $d)
+                unless exists($sourceFiles->{$d});
+        }
+        next unless @unused;
+
+        # my $baseData = $baseline->data($f);
+        # my $baseLocation = join(":", ${$baseData->location()});
+        # my $currData = $current->data($f);
+        # my $currLocation = join(":", ${$currData->location()});
+
+        my $type;
+        if (2 == $missed{$f}) {
+            $type = "baseline";
+        } elsif (1 == $missed{$f}) {
+            $type = "current";
+        } else {
+            $type = "both baseline and current";
+        }
+        my $single = 1 == scalar(@unused);
+        # @todo could print line numbers in baseline, current .info files and
+        #   in diff file ..
+        if (lcovutil::warn_once($lcovutil::ERROR_MISMATCH, $f)) {
+            my $suffix =
+                $main::elide_path_mismatch ? ' (elided)' :
+                lcovutil::explain_once(
+                'elide-path-mismatch',
+                '  Perhaps see "--elide-path-mismatch", "--substitute" and "--build-directory" options in \'man '
+                    . $lcovutil::tool_name . '\'');
+            lcovutil::ignorable_warning(
                                        $lcovutil::ERROR_MISMATCH,
                                        "source file '$f' (in $type .info file" .
                                            ($missed{$f} == 3 ? "s" : "") .
@@ -3874,16 +3874,14 @@ sub check_path_consistency
                                            ($single ? "  " : "\n\t") .
                                            'Possible pathname mismatch?' .
                                            $suffix);
-                }
-                if ($main::elide_path_mismatch &&
-                    $missed{$f} == 3 &&
-                    $single) {
-                    $self->[FILEMAP]->{$f} = $f;
-                    $self->[LINEMAP]->{$f} = $self->[LINEMAP]->{$unused[0]};
-                } else {
-                    $ok = 0;
-                }
-            }
+        }
+        if ($main::elide_path_mismatch &&
+            $missed{$f} == 3 &&
+            $single) {
+            $self->[FILEMAP]->{$f} = $f;
+            $self->[LINEMAP]->{$f} = $self->[LINEMAP]->{$unused[0]};
+        } else {
+            $ok = 0;
         }
     }
     return $ok;
diff --git a/scripts/gitdiff b/scripts/gitdiff
@@ -2,6 +2,7 @@
 
 use Getopt::Long;
 use strict;
+use Cwd /realpath/;
 
 my $verbose = 0;
 
@@ -39,18 +40,20 @@ my @include_patterns;
 my $suppress_unchanged;
 my $prefix = '';
 my $ignore_whitespace;
+my $repo = '.';    # cwd
 
 if (!GetOptions("exclude=s"    => \@exclude_patterns,
                 "include=s"    => \@include_patterns,
                 'no-unchanged' => \$suppress_unchanged,
                 'prefix=s'     => \$prefix,
                 'b|blank'      => \$ignore_whitespace,
+                'repo=s'       => \$repo,
                 'verbose+'     => \$verbose) ||
     (2 != scalar(@ARGV) &&
         3 != scalar(@ARGV))
 ) {
     print(STDERR
-            "usage: [(--exclude|include) regexp[,regexp]] [-b] [dir] base_changelist current_changelist\n'exclude' wins if both exclude and include would match.\n"
+            "usage: [(--exclude|include) regexp[,regexp]] [--repo repo] [-b] [dir] base_changelist current_changelist\n'exclude' wins if both exclude and include would match.\n"
     );
     exit(1);
 }
@@ -71,15 +74,16 @@ if ($verbose) {
 }
 
 $prefix .= '/' if $prefix;
-my $cmd;
 if (3 == scalar(@ARGV)) {
     my $dir = shift @ARGV;
     push(@include_patterns, "$dir");
 }
 my $base_sha    = shift @ARGV;
 my $current_sha = shift @ARGV;
 
-$cmd = "git diff $ignore_whitespace $base_sha $current_sha";
+$repo = Cwd::realpath($repo);
+
+my $cmd = "cd $repo ; git diff $ignore_whitespace $base_sha $current_sha";
 
 open(HANDLE, "-|", $cmd) or
     die("failed to exec git diff: $!");
@@ -104,8 +108,10 @@ while (<HANDLE>) {
         $includeCurrentFile =
             (include_me($fileA, \@include_patterns, \@exclude_patterns) ||
              include_me($fileB, \@include_patterns, \@exclude_patterns));
-        $allFiles{$fileB} = $fileA
-            if ($includeCurrentFile);
+        if ($includeCurrentFile) {
+            $allFiles{$fileB} = $fileA;
+            $line =~ s/($fileA|$fileB)/$1/g;
+        }
     }
 
     printf("%s\n", $line)
@@ -119,19 +125,17 @@ if (0 != $?) {
 
 exit 0 if defined($suppress_unchanged);
 
-my $p = `git rev-parse --show-prefix`;
-chomp $p;
-
 # now find the list of files in the current SHA - we can process that
 #  to find the list of all current files which were unchanged since the
 #  baseline SHA
 die("failed to exec git ls-tree")
-    unless (open(HANDLE, "-|", "git ls-tree -r --name-only $current_sha"));
+    unless (
+      open(HANDLE, "-|", "cd $repo ; git ls-tree -r --name-only $current_sha"));
 
 while (<HANDLE>) {
     chomp($_);
     s/\r//g;
-    my $filename = $p . $_;
+    my $filename = $repo . '/' . $_;
     my $path     = $prefix . $filename;
     if (!exists($allFiles{$path}) &&
         include_me($path, \@include_patterns, \@exclude_patterns)) {
