[exdoc] Refactor.

Avoid globals, use proper data structures instead of strings, and
produce some warnings for documentation errors.

1 files changed, 329 insertions, 188 deletions

diff --git a/docs/exdoc b/docs/exdoc
index 76da62004af..64aa85fdf44 100644
--- a/docs/exdoc
+++ b/docs/exdoc
@@ -3,84 +3,255 @@
 use warnings;
 use strict;
 
-my $sourcedir = '';
-my $dir = '';
-my $html = 0;
-my $css = '';
-my @files = ();
-my $filecount = 0;
-my @files_content = ();
-my @apis = ();
-my %deprecated = ();
-my %prototype = ();
-my %arguments = ();
-my %returns = ();
-my %bodies = ();
-
-if ($ARGV[0] eq "-h") {
-    $sourcedir = $ARGV[1];
-    $dir = $sourcedir;
-    $html = 1;
-    shift @ARGV;
-    shift @ARGV;
+use Getopt::Long;
+use Pod::Usage;
+
+# Options
+my $HELP = 0;
+my $SOURCE_DIR = '';
+my $TARGET_DIR = '';
+my $WARNINGS = 0;
+
+GetOptions(
+    "help" => \$HELP,
+    "html|h=s" => \$SOURCE_DIR,
+    "target|t=s" => \$TARGET_DIR,
+    "warnings|W" => \$WARNINGS,
+) or pod2usage(1);
+
+pod2usage(0) if $HELP;
+
+exdoc();
+
+#
+# Main entry point.
+#
+sub exdoc {
+    my %templates = ();
+    my %docs = ();
+    my $stylesheet = load_stylesheet($SOURCE_DIR);
+    load_templates($SOURCE_DIR, \%templates);
+    process_source_files(\%docs);
+    merge(\%docs, \%templates, \$stylesheet);
+}
+
+#
+# Load CSS stylesheet.
+#
+sub load_stylesheet {
+    my ($dir_path) = @_;
+    my $file_path = "$dir_path/api-style.css";
+    open (my $file, '<', $file_path) or die "Could not open $file_path";
+    local $/;
+    my $contents = <$file>;
+    close $file;
+    return $contents;
+}
+
+#
+# Load HTML templates.
+#
+sub load_templates {
+    my ($dir_path, $templates) = @_;
+    opendir (my $dir, "$dir_path/sources/") or die "Could not open $dir_path";
+    while (my $file_name = readdir ($dir)) {
+        next if $file_name !~ /mono-api-.*\.html$/;
+        open (my $file, "$dir_path/sources/$file_name") or die "Could not open $file_name";
+        my $contents = '';
+        my @api = ();
+        while (<$file>) {
+            $contents .= $_;
+            if (/name="api:(.*?)"/) {
+                s/.*name="api:(\w+?)".*/$1/;
+                push @api, $_;
+            }
+        }
+        close $file;
+        $templates->{$file_name}->{contents} = $contents;
+        $templates->{$file_name}->{api} = \@api;
+    }
+    closedir $dir;
 }
-open (FILE, "$dir/api-style.css" || die "Did not find $dir/api-style.css");
-while (<FILE>) {
-    $css = $css . $_;
+
+#
+# Extract documentation from all source files.
+#
+sub process_source_files {
+    my ($docs) = @_;
+    for my $file_path (@ARGV) {
+        process_source_file($file_path, $docs);
+    }
 }
 
-if ($ARGV[0] eq "-t") {
-    $dir = $ARGV[1];
-    shift @ARGV;
+#
+# Extract documentation from a single source file.
+#
+sub process_source_file {
+    my ($file_path, $docs) = @_;
+    open (my $file, '<', $file_path) or die "Could not open $file_path";
+    while (<$file>) {
+        next if (!/\/\*\* *\n/);
+        process_function($file, $file_path, $docs);
+    }
+    close $file;
 }
 
-if ($html) {
-    opendir (D, "$sourcedir/sources/") || die "Can not open $dir";
-    while (my $n = readdir (D)) {
-        if ($n =~ /mono-api-.*\.html$/) {
-            open (IN, "$sourcedir/sources/$n") || die "Can not open $n";
-            $files[$filecount] = $n;
-            $files_content[$filecount] = '';
-            while (<IN>) {
-                $files_content[$filecount] .= $_;
-                if (/name="api:(.*?)"/) {
-                    $_ =~ s/.*name="api:(\w+?)".*/$1/;
-                    $apis[$filecount] .= "$_";
+#
+# Extract documentation from a single function.
+#
+sub process_function {
+
+    my ($file, $file_path, $docs) = @_;
+
+    my $PARAMETER_SECTION = 0;
+    my $BODY_SECTION = 1;
+    my $RETURN_SECTION = 2;
+    my $section = $PARAMETER_SECTION;
+
+    my $name = do {
+        $_ = <$file>;
+        chomp;
+        s/^ \* //;
+        s/:$//;
+        $_
+    };
+
+    # Ignore irrelevant functions, and those with the wrong doc format.
+    return if $name !~ /^mono_\w+$/;
+
+    my $deprecated;
+    my @parameters = ();
+    my $body = '';
+    my $returns = '';
+    my $prototype = '';
+
+    while (<$file>) {
+
+        # We've reached the last line in the documentation block.
+        if (/^ \*\*?\//) {
+
+            # Grab function prototype.
+            while (<$file>) {
+                $prototype .= $_;
+                last if /\{/;
+            }
+
+            # Clean up prototype.
+            $prototype = do {
+                $_ = $prototype;
+                # Strip braces and trailing whitespace.
+                s/{//;
+                s/ +$//;
+                # Turn "Type * xxx" into "Type* xxx"
+                s/^(\w+)\W+\*/$1*/;
+                $_;
+            };
+
+            # Process formatting within sections.
+            for my $parameter (@parameters) {
+                process_formatting(\$parameter->{description});
+            }
+            process_formatting(\$returns);
+            process_formatting(\$body);
+            $body =~ s/\n/ /g;
+
+            if (exists($docs->{body}->{$name})) {
+                my $origin = $docs->{origin}->{$name};
+                if ($WARNINGS) {
+                    warn
+                      "$file_path:$.: Redundant documentation for $name\n",
+                      "$origin->{file}:$origin->{line}: Previously defined here\n";
                 }
             }
-            $filecount++;
-            close IN;
+            $docs->{origin}->{$name} = { file => $file_path, line => $. };
+            $docs->{body}->{$name} = $body;
+            $docs->{parameters}->{$name} = \@parameters;
+            $docs->{deprecated}->{$name} = $deprecated if defined $deprecated;
+            $docs->{return}->{$name} = $returns;
+            $docs->{prototype}->{$name} = $prototype;
+            last;
+
+        }
+
+        # Strip newlines and asterisk prefix.
+        chomp;
+        s/^ +\*//;
+
+        # Replace blank lines with paragraph breaks.
+        $_ = '<p>' if /^\s*$/;
+
+        if ($section == $PARAMETER_SECTION) {
+            if (/\s*(\w+):(.*)/) {
+                if ($1 eq 'deprecated') {
+                    $deprecated = $2;
+                } else {
+                    push @parameters, { name => $1, description => $2 };
+                }
+            } else {
+                $body = "\t$_\n";
+                $section = $BODY_SECTION;
+            }
+        } elsif ($section == $BODY_SECTION) {
+            if (/Returns?:/) {
+                s/Returns?://;
+                $returns = "\t$_\n";
+                $section = $RETURN_SECTION;
+            } else {
+                $body .= "\n\t$_";
+            }
+        } elsif ($section == $RETURN_SECTION) {
+            $returns .= "\n\t$_";
+        } else {
+            die "Invalid section $section\n";
         }
     }
 }
 
-while (<ARGV>) {
-    if (/\/\*\* *\n/) {
-        &process_doc;
-    } else {
-        #print "IGNORING: $_";
-    }
+#
+# Substitute formatting within documentation text.
+#
+sub process_formatting {
+    my ($content) = @_;
+    $_ = $$content;
+
+    # Constants
+    s{NULL}{<code>NULL</code>}g;
+    s{TRUE}{<code>TRUE</code>}g;
+    s{FALSE}{<code>FALSE</code>}g;
+
+    # Parameters
+    s{@(\w+)}{<i>$1</i>}g;
+
+    # Code
+    s{#(\w+)}{<code>$1</code>}g;
+    s{\`([:.\w\*]+)\`}{<code>$1</code>}g;
+
+    $$content = $_;
 }
 
-if ($html) {
+#
+# Merge templates with stylesheet and documentation extracted from sources.
+#
+sub merge {
+    my ($docs, $templates, $stylesheet) = @_;
     my $last = '';
-    for (my $f = 0; $f < $filecount; $f++) {
-        my $name = $files[$f];
-        open (OUT, "> $dir/html/$name") || die "Can not create $dir/html/$name";
+    for my $name (keys %$templates) {
+        open (my $output_file, '>', "$TARGET_DIR/html/$name")
+          or die "Could not create $TARGET_DIR/html/$name";
         print "Merging: $name\n";
-        print OUT <<EOF;
+        print $output_file <<EOF;
 <?xml version="1.0" encoding="utf-8"?>
 <html xmlns="http://www.w3.org/1999/xhtml">
 <head>
     <title>$name</title>
     <style type="text/css">
-$css
+$stylesheet
    </style>
 </head>
 <body>
 <div class="mapi-docs">
 EOF
-        my @a = split (/\n/, $files_content[$f]);
+        my @a = split (/\n/, $templates->{$name}->{contents});
         my $strike = '';
         my $strikeextra = '';
         my $api_shown = 0;
@@ -88,19 +259,19 @@ EOF
             my $line = $a[$ai];
             if (my ($api, $caption) = ($line =~ /<h4><a name=\"api:(\w+)\">(\w+)<\/a><\/h4>/)) {
                 if ($api_shown == 1) {
-                    print OUT "</div> <!-- class=mapi -->\n\n";
-                    if ($deprecated{$api}) {
+                    print $output_file "</div> <!-- class=mapi -->\n\n";
+                    if ($docs->{deprecated}->{$api}) {
                         $strike = "mapi-strike";
-                        $strikeextra = "</div><br><div class='mapi-deprecated'><b>Deprecated:</b> " . $deprecated{$api};
+                        $strikeextra = "</div><br><div class='mapi-deprecated'><b>Deprecated:</b> " . $docs->{deprecated}->{$api};
                     } else {
                         $strike = "";
                         $strikeextra = "";
                     }
                 }
                 $api_shown = 1;
-                my $proto = $prototype{$api} // $api;
+                my $proto = $docs->{prototype}->{$api} // $api;
 
-                print OUT <<EOF;
+                print $output_file <<EOF;
 <a name="api:$api"></a>
 <div class="mapi">
     <div class="mapi-entry $strike"><code>$api$strikeextra</code></div>
@@ -113,61 +284,64 @@ EOF
             <div class="mapi-prototype">$proto</div>
             <p>
 EOF
-                if (exists ($arguments{$api})) {
-                    my $ppars = $arguments{$api};
-                    if ($ppars !~ /^[ \t]+$/) {
-                        print OUT "            <div class=\"mapi-section\">Parameters</div>\n";
-                        print OUT "            <table class=\"mapi-parameters\"><tbody>".${arguments{$api}}."</tbody></table>";
+                if (exists ($docs->{parameters}->{$api})) {
+                    my $ppars = $docs->{parameters}->{$api};
+                    if (@$ppars) {
+                        print $output_file
+                          "            <div class=\"mapi-section\">Parameters</div>\n",
+                          "            <table class=\"mapi-parameters\"><tbody>",
+                          render_parameters($ppars),
+                          "</tbody></table>";
                     }
                 }
 
-                &opt_print ("Return value", $returns{$api}, 0);
-                &opt_print ("Description", $bodies{$api}, 0);
-                print OUT "        </div><!--mapi-description-->\n    </div><!--height container-->\n";
+                opt_print ($output_file, "Return value", $docs->{return}->{$api});
+                opt_print ($output_file, "Description", $docs->{body}->{$api});
+                print $output_file "        </div><!--mapi-description-->\n    </div><!--height container-->\n";
             } else {
                 if ($line =~ /\@API_IDX\@/) {
-                    my $apis_toc = &create_toc ($apis[$f]);
+                    my $apis_toc = create_toc ($docs, $templates->{$name}->{api});
                     $line =~ s/\@API_IDX\@/$apis_toc/;
                 }
                 if ($line =~ /^<h4/) {
-                    print OUT "</div>\n";
+                    print $output_file "</div>\n";
                     $api_shown = 0;
                 }
                 if ($line =~ /`/) {
                 }
-                print OUT "$line\n";
+                print $output_file "$line\n";
             }
         }
-        print OUT <<EOF;
-   </div>
-</body>
-</html>
-EOF
-        close OUT;
-        system ("$ENV{runtimedir}/mono-wrapper convert.exe $dir/html/$name $dir/html/x-$name");
-
-
-        # clean up the mess that AgilityPack does, it CDATAs our CSS
-        open HACK, "$dir/html/x-$name" || die "Could not open $dir/html/x-$name";
-        open HACKOUT, ">$dir/deploy/$name" || die "Could not open output";
+        print $output_file
+          "   </div>",
+          "</body>",
+          "</html>";
+        close $output_file;
+        system ("$ENV{runtimedir}/mono-wrapper convert.exe $TARGET_DIR/html/$name $TARGET_DIR/html/x-$name");
+
+        # Clean up the mess that AgilityPack makes (it CDATAs our CSS).
+        open (my $hack_input, '<', "$TARGET_DIR/html/x-$name")
+          or die "Could not open $TARGET_DIR/html/x-$name";
+        open (my $hack_output, '>', "$TARGET_DIR/deploy/$name")
+          or die "Could not open output";
 
         my $line = 0;
         my $doprint = 0;
-        while (<HACK>) {
-            print HACKOUT $last if ($doprint);
+        while (<$hack_input>) {
+            print $hack_output $last if ($doprint);
             $line++;
             s/^\/\/<!\[CDATA\[//;
             s/^\/\/\]\]>\/\///;
 
-            # Remove the junk <span> wrapper generated by AgilityPack
+            # Remove the junk <span> wrapper generated by AgilityPack.
             if ($line==1) {
                 s/<span>//;
             }
             if (/<style type/) {
-                # Replace the CSS in the XHTML output with the original CSS
-                print HACKOUT $_;
-                print HACKOUT $css;
-                while (<HACK>) {
+                # Replace the CSS in the XHTML output with the original CSS.
+                print $hack_output $_;
+                print $hack_output $$stylesheet;
+                while (<$hack_input>) {
                     last if (/<\/style>/);
                 }
             }
@@ -175,106 +349,23 @@ EOF
             $doprint = 1;
         }
         if (!($last =~ /span/)) {
-            print HACKOUT $last;
+            print $hack_output $last;
         }
-
-        # system ("cp.exe $dir/html/$name $dir/deploy/$name");
-    }
-}
-
-sub process_doc {
-    my $doc = "";
-    my $func = <>;
-    chop $func;
-    $func =~ s/^ \* //;
-    $func =~ s/:$//;
-    print "Function: $func\n" if (!$html);
-    my $args = "";
-    my $inbody = 0;
-    my $returns = "";
-    my $body = "";
-    my $deprecated = 0;
-    # Process arguments
-    while (<>) {
-        s/NULL/<code>NULL<\/code>/g;
-        s/TRUE/<code>TRUE<\/code>/g;
-        s/FALSE/<code>FALSE<\/code>/g;
-        if (/^ \*\*?\//) {
-            $body =~ s/\@(\w+)/<i>$1<\/i>/g;
-            $returns =~ s/\@(\w+)/<i>$1<\/i>/g;
-            $args =~ s/\@(\w+)/<i>$1<\/i>/g;
-
-            $body =~ s/#(\w+)/<code>$1<\/code>/g;
-            $returns =~ s/#(\w+)/<code>$1<\/code>/g;
-            $args =~ s/#(\w+)/<code>$1<\/code>/g;
-
-            $returns =~ s/\`([:.\w\*]+)\`/<code>$1<\/code>/g;
-            $args =~ s/\`([:.\w\*]+)\`/<code>$1<\/code>/g;
-            $body =~ s/\`([:.\w\*]+)\`/<code>$1<\/code>/g;
-
-            $body =~ s/\n/ /;
-            $bodies{$func} = $body;
-            $arguments{$func} = $args;
-            $deprecated{$func} = $deprecated;
-            $returns{$func} = $returns;
-            my $proto = "";
-            while (<>) {
-                $proto .= $_;
-                last if (/\{/);
-            }
-            $proto =~ s/{//;
-            # clean it up a little, remove newlines, empty space at end
-            $proto =~ s/ +$//;
-            # Turn "Type * xxx" into "Type* xxx"
-            $proto =~ s/^(\w+)\W+\*/$1\*/;
-            $prototype{$func} = $proto;
-            return;
-        }
-        chop;
-        s/^\ \*//;
-        $_ = "<p>" if (/^\s*$/);
-
-        if ($inbody == 0) {
-            if (/\s*(\w+):(.*)/) {
-                if ($1 eq "deprecated") {
-                    $deprecated = $2;
-                } else {
-                    # $args .= "<dt><i>$1:</i></dt><dd>$2</dd>";
-                    $args .= "<tr><td><i>$1</i><td>$2</td></td></tr>";
-                }
-            } else {
-
-                $body = "\t$_\n";
-
-                $inbody = 1;
-            }
-        } elsif ($inbody == 1) {
-            if (/Returns?:/) {
-                s/Returns?://;
-                $returns = "\t$_\n";
-                $inbody = 2;
-            } else {
-                $body .= "\n\t$_";
-            }
-        } else {
-            $returns .= "\n\t$_";
-        }
-
+        # system ("cp.exe $TARGET_DIR/html/$name $TARGET_DIR/deploy/$name");
     }
 }
 
 sub create_toc {
-    my ($apis_listed) = @_;
+    my ($docs, $apis_listed) = @_;
     my $type_size = 0;
     my $name_size = 0;
     my ($ret, $xname, $args);
     my $apis_toc = "";
 
-
-    # Try to align things, so compute type size, method size, and arguments
+    # Try to align things; compute type size, method size, and arguments.
     foreach my $line (split /\n/, $apis_listed) {
-        if (exists ($prototype{$line})) {
-            my $p = $prototype{$line};
+        if (exists ($docs->{prototype}->{$line})) {
+            my $p = $docs->{prototype}->{$line};
             if (my ($ret, $xname, $args) = ($p =~ /(.*)\n(\w+)[ \t](.*)/)) {
                 my $tl = length ($ret);
                 my $pl = length ($xname);
@@ -289,13 +380,13 @@ sub create_toc {
 
     foreach my $line (split /\n/, $apis_listed) {
         chomp($line);
-        if (exists($prototype{$line})) {
-            my $p = $prototype{$line};
+        if (exists($docs->{prototype}->{$line})) {
+            my $p = $docs->{prototype}->{$line};
             if (my ($ret, $xname, $args) = ($p =~ /(.*)\n(\w+)[ \t](.*)/)) {
                 $xname = $line if $xname eq "";
                 my $rspace = " " x ($type_size - length ($ret));
                 my $nspace = " " x ($name_size - length ($xname));
-                $args = &format ($args, length ($ret . $rspace . $xname . $nspace), 60);
+                $args = wrap ($args, length ($ret . $rspace . $xname . $nspace), 60);
                 $apis_toc .= "$ret$rspace<a href=\"\#api:$line\">$xname</a>$nspace$args\n";
             }
         }
@@ -303,10 +394,7 @@ sub create_toc {
     return $apis_toc;
 }
 
-#
-# Formats the rest of the arguments in a way that will fit in N columns
-#
-sub format {
+sub wrap {
     my ($args, $size, $limit) = @_;
     my $sret = "";
 
@@ -333,11 +421,64 @@ sub format {
     return $sret;
 }
 
+#
+# Print a section if non-empty.
+#
 sub opt_print {
-    my ($caption, $opttext, $quote) = @_;
+    my ($output, $caption, $opttext) = @_;
+    if (defined($opttext) && $opttext ne '' && $opttext !~ /^[ \t]+$/) {
+        print $output
+          "             <div class=\"mapi-section\">$caption</div>\n",
+          "             <div>$opttext</div>\n";
+    }
+}
 
-    if (defined($opttext) && $opttext !~ /^[ \t]+$/) {
-        print OUT "             <div class=\"mapi-section\">$caption</div>\n";
-        print OUT "             <div>$opttext</div>\n";
+#
+# Render parameter information as table.
+#
+sub render_parameters {
+    my ($parameters) = @_;
+    my $result = '';
+    for my $parameter (@$parameters) {
+        $result .= "<tr><td><i>$parameter->{name}</i></td><td>$parameter->{description}</td></tr>";
     }
+    return $result;
 }
+
+__END__
+
+=head1 NAME
+
+exdoc - Compiles API docs from Mono sources and HTML templates.
+
+=head1 SYNOPSIS
+
+    exdoc [OPTIONS] [FILE...]
+
+=head1 OPTIONS
+
+=over 4
+
+=item B<--help>
+
+Print this help message.
+
+=item B<--html> I<DIR>, B<-h> I<DIR>
+
+Use I<DIR> as the input path for HTML sources.
+
+=item B<--target> I<DIR>, B<-t> I<DIR>
+
+Use I<DIR> as the target path for output.
+
+=item B<--warnings>, B<-W>
+
+Enable warnings about documentation errors.
+
+=back
+
+=head1 DESCRIPTION
+
+Reads HTML templates and C sources, extracting documentation from the sources and splicing it into the templates.
+
+=cut