Add PDFJS scripts and template

bbd365ab · Lyza Danger Gardner · Lyza Gardner · 18c74abf · bbd365ab · bbd365ab
Commit bbd365ab authored Aug 21, 2020 by Lyza Danger Gardner Committed by Lyza Gardner Aug 25, 2020
Showing with 557 additions and 0 deletions

pdfjs-viewer.mustache dev-server/templates/pdfjs-viewer.mustache +423 -0

generate-pdfjs-template.py scripts/generate-pdfjs-template.py +78 -0

update-pdfjs scripts/update-pdfjs +56 -0

No files found.
--- a/dev-server/templates/pdfjs-viewer.mustache
+++ b/dev-server/templates/pdfjs-viewer.mustache
--- a/scripts/generate-pdfjs-template.py
+++ b/scripts/generate-pdfjs-template.py
+#!/usr/bin/env python
+
+# This is a helper used by `update-pdfjs` to update the Mustache template for
+# serving PDFs with PDFJS with the local dev server.
+
+import os
+import sys
+
+# Header to insert at the top of the generated PDF.js viewer template
+FILE_HEADER = """
+<!-- AUTO-GENERATED BY {}. DO NOT EDIT. -->
+""".format(
+    sys.argv[0]
+)
+
+# Header to insert after the original `<title>` tag in the PDF viewer HTML
+# mustache template.
+#
+# This header is responsible for:
+#
+# - Adding a `<base>` tag so that relative URLs in the pre-built viewer HTML
+#   resolve to the right URL.
+# - Injecting custom PDF.js viewer configuration
+# - Injecting the Hypothesis client entry point and configuration
+#
+# The header needs to be inserted after the `<title>` tag so we can override it,
+# but before any relative asset links which will be affected by the `<base>`
+# tag.
+#
+HYPOTHESIS_HEADER = """
+<!-- Begin Hypothesis modifications -->
+<base href="/scripts/pdfjs-2/web/">
+
+<title>via Hypothesis</title>
+{#
+  It's worth noting that this link tag is *not* currently used by the
+  Hypothesis client to determine the URL of this page. For consistency with
+  how these pages are served on via, however, we serve it with the PDF.js
+  viewer application.
+-#}
+<link rel="canonical" href="{{{ documentUrl }}}"/>
+<script>
+window.DOCUMENT_URL = '{{{documentUrl}}}';
+window.PDF_URL = '{{{ url }}}';
+window.CLIENT_URL = '{{{clientUrl}}}'.replace('{current_host}', document.location.hostname);
+</script>
+
+<script src="/scripts/pdfjs-init.js"></script>
+
+{# Configure Hypothesis client. #}
+<script>
+window.hypothesisConfig = function() {
+  return {
+    openSidebar: true,
+  };
+};
+</script>
+<!-- End Hypothesis modifications -->
+"""
+
+
+def insert_after(str_, search_str, insert_str):
+    return str_.replace(search_str, search_str + insert_str)
+
+
+input_file_path = sys.argv[1]
+output_file_path = sys.argv[2]
+
+input_file = open(input_file_path, "r")
+output_file = open(output_file_path, "w")
+base_dir = os.path.dirname(input_file_path)
+
+viewer_html = input_file.read()
+viewer_html = insert_after(viewer_html, "<!DOCTYPE html>", FILE_HEADER)
+viewer_html = insert_after(
+    viewer_html, "</title>", HYPOTHESIS_HEADER.replace("$BASEDIR", base_dir)
+)
+output_file.write(viewer_html)
\ No newline at end of file
--- a/scripts/update-pdfjs
+++ b/scripts/update-pdfjs
+#!/bin/sh
+
+set -eu
+
+# This script fetches the latest build of PDF.js from the viewer demo
+# page. It also regenerates the Mustache-templated HTML viewer for use
+# by the local devserver.
+#
+# See https://github.com/mozilla/pdf.js/wiki/Setup-pdf.js-in-a-website#from-examples
+#
+# To update PDF.js to the latest version:
+#
+# 1. Create a new branch and run this script.
+# 2. Re-start `make dev` and verify that PDFs in the `dev-server/documents/pdf`
+#    directory work correctly.
+# 3. Commit the changes to the `static/` and `templates/` directories inside of
+#    `dev-server`
+#
+# If you make changes to the PDF.js viewer template generator only and want to
+# regenerate the HTML template, run this script with the `--no-download` flag.
+
+DEST_DIR=dev-server/static/scripts/pdfjs-2
+PREFIX=pdf.js-gh-pages
+COMPONENTS="$PREFIX/build $PREFIX/web $PREFIX/LICENSE"
+
+download_pdfjs=1
+for arg in "$@"
+do
+  if [ "$arg" = "--no-download" ]; then
+    download_pdfjs=""
+  fi
+done
+
+# Download the latest version of the PDF.js library and viewer.
+if [ $download_pdfjs ]; then
+  rm -rf $DEST_DIR
+  mkdir -p $DEST_DIR
+
+  # Get the latest build of the viewer
+  curl -L https://github.com/mozilla/pdf.js/archive/gh-pages.tar.gz \
+    | tar -xz --directory $DEST_DIR --strip-components=1 $COMPONENTS
+
+  # Remove example content from viewer
+  rm $DEST_DIR/web/*.pdf
+
+  # Remove the check that the PDF being loaded is from the same origin as the
+  # viewer.
+  sed -i '' -e 's/HOSTED_VIEWER_ORIGINS.includes(viewerOrigin)/true \/* Hypothesis *\//' $DEST_DIR/web/viewer.js
+fi
+
+# Generate the PDF viewer mustache (HTML) template.
+BASEDIR=$(dirname $0)
+python \
+  $BASEDIR/generate-pdfjs-template.py \
+  $DEST_DIR/web/viewer.html \
+  dev-server/templates/pdfjs-viewer.mustache
\ No newline at end of file