\n") {
- $mode = 'tr';
- } else {
- # ignore...
- }
- } elsif ($mode eq 'tr') {
- if ($_ =~ m!^ | <(?:code|span) data-x="([^"]+)">[^<]*(?: \(in [^\)]+\))?;?\n$!os) {
- $attributes{$1} = 1;
- $mode = 'index-in';
- } else {
- # ignore...
- }
- } elsif ($mode eq 'index-in') {
- if ($_ =~ m!^ <(?:code|span) data-x="([^"]+)">[^<]*(?: \(in [^\)]+\))?;?\n$!os) {
- $attributes{$1} = 1;
- } elsif ($_ =~ m@^ | (.+?)(?:)?\n$@os) {
- local $" = ', ';
- my $description = $1;
- my $altdescription = $2;
- foreach my $key (keys %attributes) {
- foreach my $entry (@{$instances{$key}}) {
- my $line = $entry->{line};
- if ($entry->{special} eq 'global') {
- $$line =~ s!(\.\n)$!: $description<\!--SPECIAL-->$1!os;
- $$line =~ s!<\!--SPECIAL-->: *([^ ])!; \l$1!os;
- } elsif ($entry->{special} eq 'alt') {
- die "$key wants alt description but we have none" unless defined $altdescription;
- $$line =~ s!(\n)$! — $altdescription$1!os;
- } else {
- $$line =~ s!(\n)$! — $description$1!os;
- }
- }
- }
- %attributes = ();
- $mode = 'index';
- } else {
- die "$.: unexpected line in index: $_";
- }
- } else {
- # ignore
- }
- }
-}
-
-foreach (@lines) {
- $$_ =~ s///gos;
- print $$_;
-}
diff --git a/.pre-process-index-generator.pl b/.pre-process-index-generator.pl
deleted file mode 100644
index d495d8d..0000000
--- a/.pre-process-index-generator.pl
+++ /dev/null
@@ -1,58 +0,0 @@
-#!/usr/bin/perl -w
-use strict;
-
-my $seenInsertionPoint = 0;
-my $after = '';
-
-my %definitions;
-my $inpre = 0;
-while (<>) {
- $inpre = 1 if /]*)( id="?([^ ">]*)"?)?[^>]*>([^<:]*)?<\/(span|dfn|a)>/os) {
- my $partial = $1;
- my $id;
- my $name;
- if ($partial) {
- if ($4) {
- ($id, $name) = ($4, $5);
- } else {
- die "partial interface entry for $5 is missing an id (required for interface index)";
- }
- $definitions{$name} = { } unless defined $definitions{$name};
- $definitions{$name}{partial} = [] unless exists $definitions{$name}{partial};
- push @{$definitions{$name}{partial}}, $id;
- } else {
- $name = $5;
- $definitions{$name} = { } unless defined $definitions{$name};
- die "duplicate interface definitions for $name" if exists $definitions{$name}{primary};
- }
- }
- $inpre = 0 if /<\/pre>/os;
- if (/^INSERT INTERFACES HERE\n?$/os) {
- $seenInsertionPoint = 1;
- } else {
- if ($seenInsertionPoint) {
- $after .= $_;
- } else {
- print $_;
- }
- }
-}
-
-die unless $seenInsertionPoint;
-
-print " \n";
-print $after;
diff --git a/.pre-process-main.pl b/.pre-process-main.pl
deleted file mode 100644
index 023ffce..0000000
--- a/.pre-process-main.pl
+++ /dev/null
@@ -1,94 +0,0 @@
-#!/usr/bin/perl -w
-# -d:DProf
-use strict;
-use File::Path;
-use File::Basename;
-use Time::HiRes qw(time);
-
-$| = 1;
-
-my $verbose = $ARGV[0] && '--verbose' eq "$ARGV[0]";
-sub report($) { print STDERR $_[0] if $verbose; }
-
-report "Loading...";
-my @lines = ;
-
-report "\rPreprocessing...";
-
-# monitoring
-my $lineNumber = 0;
-my $lastProgress = -1;
-my $maxProgress = 20;
-my $start = time;
-
-# actual work
-my $currentRepresents = '';
-my $represents = {};
-
-while (@lines) {
- my $line = shift @lines;
- $lineNumber += 1;
- my $progress = int($maxProgress * $lineNumber / (scalar @lines + $lineNumber));
- if ($progress != $lastProgress) {
- my $rate = 0;
- my $elapsed = (time - $start);
- if ($elapsed > 0) {
- $rate = $lineNumber / $elapsed;
- }
- report sprintf "\rParsing... [" . ('#' x $progress) . (' ' x ($maxProgress - $progress)) . "] %5d lines per second", $rate;
- $lastProgress = $progress;
- }
-
- if ($line =~ m|^(.*)(.*)\n$|os) {
- unshift @lines, split("\n", $1 . `cat $ENV{'HTML_CACHE'}/$2` . $3);
- next;
- } elsif ($line =~ m!^( *)(]*>(?:]*>)?)EXAMPLE (offline/|workers/|canvas/)((?:[-a-z0-9]+/){1,2}[-a-z0-9]+.[-a-z0-9]+)((?:)?
) *\n$!os) {
- my $indent = $1;
- my $starttags = $2;
- my $folder = $3;
- my $example = $4;
- my $endtags = $5;
-
- my $data;
- my $fh;
-
- open($fh, "<:encoding(UTF-8)", "$ENV{'HTML_SOURCE'}/demos/$folder$example")
- or die "\rCannot open $ENV{'HTML_SOURCE'}/demos/$folder$example";
- while (<$fh>) {
- $data .= $_;
- }
- close $fh;
-
- $data =~ s/&/&/gos;
- $data =~ s/The ([^<]+) element represents (.*) *\n$|os) {
- $represents->{$1} = "\u$2";
- } elsif ($line =~ m|^ *The ([^<]+) element represents ?(.*)\n$|os) {
- $currentRepresents = $1;
- $represents->{$currentRepresents} = "\u$2";
- } elsif ($currentRepresents) {
- if ($line =~ m|^ *(.*) *\n$|os) {
- $represents->{$currentRepresents} .= " $1";
- $currentRepresents = '';
- } elsif ($line =~ m|^ *(.+?) *\n$|os) {
- $represents->{$currentRepresents} .= " $1";
- } else {
- die "missed end of <$currentRepresents> intro.\n";
- }
- }
- $line =~ s|| if (exists $represents->{$1}) { $represents->{$1} } else { die "\nUnknown element <$1> used in REPRESENTS pragma.\n" }|gose;
-
- # This seems to be necessary due to the file substitutions, for some reason.
- $line = normalizeNewlines($line);
-
- print "$line";
-}
-report "\n";
-
-sub normalizeNewlines {
- $_ = shift;
- chomp;
- return "$_\n";
-}
diff --git a/.pre-process-tag-omission.pl b/.pre-process-tag-omission.pl
deleted file mode 100644
index 691f852..0000000
--- a/.pre-process-tag-omission.pl
+++ /dev/null
@@ -1,89 +0,0 @@
-#!/usr/bin/perl -w
-use strict;
-
-my $state = undef;
-
-my $mode = 'bored';
-my $current = '';
-my %voids = ();
-my %insertionPoints = ();
-my @lines = ();
-
-sub pushLine {
- my($element, $text) = @_;
- die unless exists $insertionPoints{$element};
- my $line = $insertionPoints{$element};
- if ($$line eq '') {
- $$line .= " Tag omission in text/html:\n";
- }
- $text =~ s!<(/?)p>!<${1}dd>!g;
- $text =~ s!may!can!g;
- $$line .= " $text";
-}
-
-while (defined($_ = <>)) {
- if ($mode eq 'bored') {
- if ($_ eq " Optional tags\n") {
- $mode = 'optionals';
- } elsif ($_ eq " Void elements\n") {
- $mode = 'voids';
- } elsif ($_ =~ m!([^<]+) elements?!
- || $_ =~ m!id="the-[^-]+-element"[^>]*>The ]*>([^<]+)!) {
- $current = $1;
- $mode = 'element';
- }
- } elsif ($mode eq 'element') {
- if ($_ eq " Content attributes:\n") {
- my $line = '';
- push(@lines, \$line);
- $insertionPoints{$current} = \$line;
- $mode = 'bored';
- } else {
- # ignore...
- }
- } elsif ($mode eq 'voids') {
- if ($_ =~ m!([^<]+)!g) {
- $voids{$1} = 1;
- }
- }
- } elsif ($mode eq 'optionals') {
- if ($_ =~ m!An? ([^<]+)!) {
- $current = $1;
- pushLine($current, $_);
- $mode = 'optionals-in';
- } elsif ($_ =~ m!!) {
- $mode = 'optionals';
- }
- pushLine($current, $_);
- } elsif ($mode eq 'done') {
- # ignore...
- } else {
- die 'unknown mode';
- }
- my $line = "$_";
- push(@lines, \$line);
-}
-
-foreach (keys %insertionPoints) {
- my $line = $insertionPoints{$_};
- if ($$line eq '') {
- if ($voids{$_}) {
- pushLine($_, " No end tag. \n");
- } else {
- pushLine($_, " Neither tag is omissible. \n");
- }
- }
-}
-
-foreach (@lines) {
- print $$_;
-}
diff --git a/Dockerfile b/Dockerfile
index 5dd646b..773335e 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -20,5 +20,4 @@ RUN pipx install bs-highlighter
COPY . /whatwg/html-build/
ENV SKIP_BUILD_UPDATE_CHECK true
-ENV PROCESS_WITH_RUST true
ENTRYPOINT ["bash", "/whatwg/html-build/build.sh"]
diff --git a/README.md b/README.md
index ead701b..7c620ab 100644
--- a/README.md
+++ b/README.md
@@ -28,7 +28,7 @@ Local builds will be fastest, but require installing a lot of prerequisites. Usi
To build locally, you'll need the following commands installed on your system:
-- `curl`, `grep`, `perl`, `unzip`
+- `curl`, `grep`, `perl`, `unzip`, `cargo`
Optionally, for faster builds, you can install [Wattsi](https://github.com/whatwg/wattsi). If you don't bother with that, we will use the [build server](https://github.com/whatwg/build.whatwg.org), which requires an internet connection.
diff --git a/build.sh b/build.sh
index 6b2eccf..366bb29 100755
--- a/build.sh
+++ b/build.sh
@@ -33,7 +33,6 @@ HTML_CACHE=${HTML_CACHE:-$DIR/.cache}
HTML_TEMP=${HTML_TEMP:-$DIR/.temp}
HTML_OUTPUT=${HTML_OUTPUT:-$DIR/output}
HTML_GIT_CLONE_OPTIONS=${HTML_GIT_CLONE_OPTIONS:-"--depth=2"}
-PROCESS_WITH_RUST=${PROCESS_WITH_RUST:-false}
# These are used by child scripts, and so we export them
export HTML_CACHE
@@ -640,23 +639,13 @@ function processSource {
BUILD_TYPE="$2"
cp -p entities/out/entities.inc "$HTML_CACHE"
cp -p entities/out/entities-dtd.url "$HTML_CACHE"
- if [[ $PROCESS_WITH_RUST == "true" ]]; then
- if hash html-build 2>/dev/null; then
- html-build <"$HTML_SOURCE/$SOURCE_LOCATION" >"$HTML_TEMP/source-whatwg-complete"
- else
- CARGO_ARGS=( --release )
- $VERBOSE && CARGO_ARGS+=( --verbose )
- $QUIET && CARGO_ARGS+=( --quiet )
- cargo run "${CARGO_ARGS[@]}" <"$HTML_SOURCE/$SOURCE_LOCATION" >"$HTML_TEMP/source-whatwg-complete"
- fi
+ if hash html-build 2>/dev/null; then
+ html-build <"$HTML_SOURCE/$SOURCE_LOCATION" >"$HTML_TEMP/source-whatwg-complete"
else
- if $VERBOSE; then
- perl .pre-process-main.pl --verbose < "$HTML_SOURCE/$SOURCE_LOCATION" > "$HTML_TEMP/source-expanded-1"
- else
- perl .pre-process-main.pl < "$HTML_SOURCE/$SOURCE_LOCATION" > "$HTML_TEMP/source-expanded-1"
- fi
- perl .pre-process-annotate-attributes.pl < "$HTML_TEMP/source-expanded-1" > "$HTML_TEMP/source-expanded-2" # this one could be merged
- perl .pre-process-tag-omission.pl < "$HTML_TEMP/source-expanded-2" | perl .pre-process-index-generator.pl > "$HTML_TEMP/source-whatwg-complete" # this one could be merged
+ CARGO_ARGS=( --release )
+ $VERBOSE && CARGO_ARGS+=( --verbose )
+ $QUIET && CARGO_ARGS+=( --quiet )
+ cargo run "${CARGO_ARGS[@]}" <"$HTML_SOURCE/$SOURCE_LOCATION" >"$HTML_TEMP/source-whatwg-complete"
fi
runWattsi "$HTML_TEMP/source-whatwg-complete" "$HTML_TEMP/wattsi-output" "$HIGHLIGHT_SERVER_URL"
diff --git a/ci-build/docker-build.sh b/ci-build/docker-build.sh
index c01403c..957b94b 100755
--- a/ci-build/docker-build.sh
+++ b/ci-build/docker-build.sh
@@ -13,12 +13,11 @@ function main {
# We want the image to contain:
# * All of the important stuff from the top-level (html-build) directory
# * But, the Dockerfile from this (ci-build) directory
- # And in particular it should *not* contain the top-level Dockerfile, non-.pl dotfiles, .git/, and
+ # And in particular it should *not* contain the top-level Dockerfile, dotfiles, .git/, and
# any html/ and output/ directories that might be hanging around from local testing.
cp "$here/Dockerfile" "$TMP_DIR"
cd "$here/.."
cp -r !(.*|html|output|Dockerfile) "$TMP_DIR"
- cp .*.pl "$TMP_DIR"
cd "$TMP_DIR"
trap cleanTemp EXIT
diff --git a/ci-build/inside-container.sh b/ci-build/inside-container.sh
index 292da8b..efdd85e 100644
--- a/ci-build/inside-container.sh
+++ b/ci-build/inside-container.sh
@@ -6,7 +6,7 @@ cd "$(dirname "$0")/../.."
PDF_SERVE_PORT=8080
-PROCESS_WITH_RUST=true SKIP_BUILD_UPDATE_CHECK=true ./html-build/build.sh
+SKIP_BUILD_UPDATE_CHECK=true ./html-build/build.sh
echo ""
echo "Running conformance checker..."
|