[CI] More title normalization tweaks

Author: chalin

package.json

@@ -44,7 +44,7 @@
     "all": "bash -c 'x=0; for c in \"$@\"; do npm run $c || x=$((x+1)); done; ((!x)) || (echo \"ERROR: some scripts failed!\" && exit 1)' -",
     "build:preview": "set -x && npm run _build -- --minify",
     "build:production": "npm run _hugo -- --minify",
-    "build": "npm run _build",
+    "build": "npm run _build --",
     "cd:public": "cd public &&",
     "check:expired": "find content -name '*.md' | xargs ./scripts/list-expired.pl",
     "check:filenames": "test -z \"$(npm run -s _ls-bad-filenames)\" || npm run -s _filename-error",

scripts/content-modules/normalize-titles.pl

@@ -1,7 +1,6 @@
 #!/usr/bin/perl -w -i
 #
 # DRAFT script used to normalize semconv doc-page tiles and add Hugo front matter
-#
 
 $^W = 1;
 
@@ -30,43 +29,28 @@
 my $beforeTitle = '';
 
 sub toTitleCase($) {
-    my $str = shift;
+    my ($str) = @_;
 
-    my @mixedCaseWords; # mixed-case or ALLCAPS
-    while ($str =~ /\b([a-z]?[A-Z][A-Z0-9]+|[A-Z]\w*[A-Z]\w*)\b/g) {
-      push @mixedCaseWords, $1;
-    }
+    # Capitalize non-mixedcase words
+    $str =~ s/(^|\s)([a-zA-Z])([a-z][a-z0-9]*)\b/$1\u$2$3/g;
 
-    $str =~ s/(\w+)/\u\L$1/g;
+    # Revert / lowercase articles etc.
+    $str =~ s/\b(A|And|As|By|For|In|On)\b/\L$1/g;
 
-    foreach my $word (@mixedCaseWords) {
-        my $lc_word = lc($word);
-        $str =~ s/\b$lc_word\b/$word/ig;
-    }
-    $str =~ s/\b(A|And|As|By|For|In|On|\.Js)\b/\L$1/g;
     return $str;
 }
 
-my @specialWords = qw(Core); # for .NET
-
 sub toSentenceCase($) {
-    my $str = shift;
+    my ($str) = @_;
 
-    my @mixedCaseWords = @specialWords; # mixed-case or ALLCAPS
-    while ($str =~ /\b([a-z]?[A-Z][A-Z0-9]+|[A-Z]\w*[A-Z]\w*)\b/g) {
-        push @mixedCaseWords, $1;
-    }
+    # Lowercase non-mixedcase words
+    $str =~ s/\b([A-Z]?[a-z][a-z0-9]*)\b/\l$1/g;
 
-    $str = lc $str;
+    # Capitalize the first word unless it is mixed case
+    $str =~ s/^([a-z][a-z0-9]*)\b/\u$1/;
 
-    # Replace words with their mixed-case or ALL CAPS versions
-    foreach my $word (@mixedCaseWords) {
-        my $lc_word = lc($word);
-        $str =~ s/\b\Q$lc_word\E\b/$word/g;
-    }
-
-    # Capitalize the first letter of the string
-    $str =~ s/^(\s*\w)/\u$1/;
+    # Handle exception
+    $str =~ s/(ASP.NET) core/$1 Core/;
 
     return $str;
 }
@@ -75,7 +59,7 @@ ()
   my $frontMatter = '';
   if ($frontMatterFromFile) {
     # printf STDOUT "> $file has front matter:\n$frontMatterFromFile\n"; # if $gD;
-    $frontMatterFromFile = '' unless $frontMatterFromFile =~ /aliases|cSpell|cascade/i;
+    $frontMatterFromFile = '' unless $frontMatterFromFile =~ /auto_gen|aliases/i;
     # printf STDOUT "> $file\n" if $ARGV =~ /\/system\b/;
   }
   $linkTitle = $title;
@@ -110,6 +94,7 @@ ()
   $linkTitle = 'HTTP' if $linkTitle =~ /^HTTP Client and Server/i;
   $linkTitle = 'SQL' if $title =~ /SQL Databases$/i;
   $linkTitle = 'System use cases' if $title =~ /System .*?General Use Cases/i;
+  $linkTitle = $1 if $title =~ /GenAI (\w+)$/i;
 
   # Missing an `s` in "Semantic Convention"?
   if ($title =~ /^Semantic Convention\b/i and $title !~ /Groups$/i) {
@@ -137,53 +122,68 @@ ()
     $linkTitle =~ s/Semantic Conventions? Stability //i;
   }
 
-  if ($linkTitle and $linkTitle ne $title) {
-    $linkTitle = toSentenceCase($linkTitle) unless $linkTitle =~ /^gRPC/;
-    if ($frontMatterFromFile =~ /linkTitle: /) {
-      $frontMatterFromFile =~ s/^(linkTitle: ).*$/$1$linkTitle/m;
-    } else {
-      $frontMatter .= "linkTitle: $linkTitle\n"
+  $frontMatter .= $frontMatterFromFile if $frontMatterFromFile;
+
+  if ($linkTitle && $linkTitle ne $title) {
+    $linkTitle = toSentenceCase($linkTitle); # unless $linkTitle =~ /^gRPC/;
+    if ($frontMatter !~ /linkTitle: /) {
+      $frontMatter .= "linkTitle: $linkTitle\n";
+    } elsif ($frontMatter !~ /^auto_gen:/m) {
+      $frontMatter =~ s/^(linkTitle: ).*$/$1$linkTitle/m;
     }
   }
 
-  $frontMatter .= $frontMatterFromFile if $frontMatterFromFile;
-
   if ($ARGV =~ /docs\/(.*?)(README|_index).md$/) {
     $frontMatter .= "path_base_for_github_subdir:\n";
     $frontMatter .= "  from: tmp/semconv/docs/$1_index.md\n";
     $frontMatter .= "  to: $1README.md\n";
   }
-  $frontMatter .= "weight: -1\n" if $title eq 'General Semantic Conventions';
 
   return $frontMatter;
 }
 
 sub printTitleAndFrontMatter() {
-  my $frontMatter;
-
-
-  # if ($ARGV =~ /docs\/(README|_index)/) {
-  #   print STDOUT "> $ARGV\n  > frontMatterFromFile: $frontMatterFromFile\n";
-  #   print STDOUT "  > title: $title\n";
-  #   print STDOUT "    > linkTitle: $linkTitle\n";
-  # }
-
-  if ($frontMatterFromFile && $frontMatterFromFile =~ /auto_gen:\s*false/) {
-    $frontMatter = $frontMatterFromFile;
-  } else {
-    $frontMatter = computeTitleAndFrontMatter();
-  }
+  my $frontMatter = computeTitleAndFrontMatter();
 
   if ($frontMatter) {
     $frontMatter = "<!--- Hugo front matter used to generate the website version of this page:\n" . $frontMatter;
     $frontMatter .= "--->\n";
     print "$frontMatter\n";
   }
+
   print $beforeTitle if $beforeTitle;
   $title = toTitleCase($title);
   print "# $title\n"
 }
 
+sub gatherFrontMatter() {
+  my $autoGenValues = 'false|below';
+  my $autoGenDirective = '';
+  my $autoGenSkip = 0;
+
+  while(<>) {
+    last if /^--->/;
+    next if $autoGenSkip;
+
+    my ($keyWord, $autoGenDirective) = /^(auto.?gen): ([^\#]+)/;
+    if ($keyWord) {
+      # print STDOUT ">> $ARGV:\n$frontMatterFromFile";
+      if ($keyWord ne 'auto_gen') {
+        warn "$ARGV: WARN: misspelled keyword, should be 'auto_gen' not '$keyWord'\n";
+      } elsif (!$autoGenDirective or $autoGenDirective !~ /^($autoGenValues)/) {
+        warn "$ARGV: WARN: missing or unrecognized 'auto_gen' value, should match '$autoGenValues', not $autoGenDirective\n";
+      } elsif ($autoGenDirective =~ /^below/) {
+        $autoGenSkip = 1;
+        # print STDOUT ">>>> skipping\n";
+      } else {
+        # print STDOUT ">> wa?\n";
+      }
+    }
+
+    $frontMatterFromFile .= $_;
+  }
+}
+
 # main
 
 my $titleRegexStr = '^#\s+(.*)';
@@ -200,10 +200,7 @@ ()
     $beforeTitle = '';
     $linkTitle = '';
     if (/^<!--- Hugo/) {
-        while(<>) {
-          last if /^--->/;
-          $frontMatterFromFile .= $_;
-        }
+        gatherFrontMatter();
         next;
     }
   }