require existing session for imgsrv

Dockerfile

@@ -75,7 +75,7 @@ RUN mkdir -p /l/local/bin
 RUN ln -s /usr/bin/unzip /l/local/bin/unzip
 RUN ln -s /usr/bin/convert /l/local/bin/convert
 RUN ln -s /usr/bin/plackup /l/local/bin/plackup
-RUN /bin/bash -c 'for cmd in pamflip jpegtopnm tifftopnm bmptopnm pngtopam ppmmake pamcomp pnmscalefixed pamscale pnmrotate pnmpad pamtotiff pnmtotiff pnmtojpeg pamrgbatopng ppmtopgm pnmtopng; do ln -s /usr/bin/$cmd /l/local/bin; done'
+RUN /bin/bash -c 'for cmd in pamflip jpegtopnm tifftopnm bmptopnm pngtopam ppmmake pamcomp pnmscalefixed pamscale pnmrotate pnmpad pamtotiff pnmtotiff pnmtojpeg pamrgbatopng ppmtopgm pnmtopng pamthreshold; do ln -s /usr/bin/$cmd /l/local/bin; done'
 
 RUN mkdir /htapps/babel/cache
 RUN chmod 4777 /htapps/babel/cache

imgsrv/lib/SRV/Image.pm

@@ -87,10 +87,15 @@ sub run {
     $self->_validate_params($env);
 
     my $C = $$env{'psgix.context'};
+
     my $mdpItem = $C->get_object('MdpItem');
     my $gId = $mdpItem->GetId();
     my $restricted = $$env{'psgix.restricted'};
 
+    # require an existing session
+    my $ses = $C->get_object('Session');
+    if ( $$ses{is_new} ) { $restricted = 1; }
+
     # now we deal with extracting
     my $cache_dir = SRV::Utils::get_cachedir();
     my $logfile = SRV::Utils::get_logfile();
@@ -240,7 +245,8 @@ sub call_core {
         Plack::Util::set_io_path($fh, Cwd::realpath($$output{filename}));
     }
 
-    my $res = $req->new_response(200);
+    my $status = ( $$output{restricted} ) ? 403 : 200;
+    my $res = $req->new_response($status);
     $res->content_type($$output{mimetype});
     $res->header('X-HathiTrust-ImageSize' => $$output{metadata}{width} . "x" . $$output{metadata}{height});
 

imgsrv/lib/SRV/Volume/Base.pm

@@ -477,17 +477,27 @@ sub _authorize {
 
     unless ( defined $self->restricted ) {
 
+
         my $C = $$env{'psgix.context'};
         my $mdpItem = $C->get_object('MdpItem');
         my $ar = $C->get_object('Access::Rights');
         my $gId = $mdpItem->GetId();
+
+        # require a valid session
+        my $ses = $C->get_object('Session');
+
+        # limit to users in an existing session
+        if ( $$ses{is_new} ) { 
+          $self->restricted(1);
+        } else {
 
-        my $final_access_status = $ar->assert_final_access_status($C, $gId);
-        my $download_access_status = $ar->get_single_page_PDF_access_status($C, $gId);
+          my $final_access_status = $ar->assert_final_access_status($C, $gId);
+          my $download_access_status = $ar->get_single_page_PDF_access_status($C, $gId);
 
-        my $restricted = ! ( ( $final_access_status eq 'allow' ) && ( $download_access_status eq 'allow' ) );
-
-        $self->restricted($restricted);
+          my $restricted = ! ( ( $final_access_status eq 'allow' ) && ( $download_access_status eq 'allow' ) );
+
+          $self->restricted($restricted);
+        }
     }
 }
 

imgsrv/lib/SRV/Volume/HTML.pm

@@ -65,12 +65,19 @@ sub run {
         return { contents => "<div></div>", mimetype => 'text/html' };
     }
 
+    # limit to users in a current session
+    my $ses = $C->get_object('Session');
+    if ( $$ses{is_new} ) { 
+      $self->restricted(1);
+    }
+
     my $restricted = $self->restricted;
     unless ( defined $restricted ) {
         # $restricted = $C->get_object('Access::Rights')->assert_final_access_status($C, $gId) ne 'allow';
         $restricted = $$env{'psgix.restricted'};
     }
 
+
     # now we deal with extracting
     my $cache_dir = SRV::Utils::get_cachedir();
 
@@ -180,7 +187,8 @@ sub call_core {
     my $max_age = 86400;    # 1 day = 60 * 60 * 24
     my $cache_control = qq{max-age=$max_age};
 
-    my $res = $req->new_response(200);
+    my $status = ( $self->restricted ) ? 403 : 200;
+    my $res = $req->new_response($status);
     $res->content_type($$target{mimetype} . ";charset=utf-8");
 
     my $contents = encode_utf8($$target{contents});

imgsrv/lib/SRV/Volume/Image/Bundle.pm

@@ -92,12 +92,11 @@ sub _authorize {
 
     $self->SUPER::_authorize($env);
     unless ( $self->restricted ) {
-        # technically the user has access but we need to 
-        # limit resources for bundling to users in a current session
-        # unless you're using XYZZY=1 on the command line
         my $C = $$env{'psgix.context'};
+        # limit to users in a current session
         my $ses = $C->get_object('Session');
-        if ( $$ses{is_new} && ! $ENV{XYZZY} ) { $self->restricted(1); }
+        if ( $$ses{is_new} ) { $self->restricted(1); }
+        # limit TIFF bundling to 10 pages
         elsif ( $self->format eq 'image/tiff' && $self->total_pages > 10 ) {
             $self->restricted(1);
         }

mdp-lib/Institutions.pm

@@ -329,7 +329,7 @@ sub get_idp_list {
         $seen{$$hash_ref{inst_id}} = 1;
 
         my $host = $ENV{'HTTP_HOST'} || 'localhost';
-        my $idp_url = $hash_ref->{template};
+        my $idp_url = $hash_ref->{template} || 'https://default.invalid';
         $idp_url =~ s,___HOST___,$host,;
         $idp_url =~ s,&,&,;
 

pt/web/firebird/tests/imgsrv_download.spec.js

@@ -11,7 +11,7 @@ test.describe('imgsrv download', () => {
   test('download whole item pdf, full resolution', async ({ request, page }) => {
     var currentTime = new Date().getTime();
 
-    const initialResponse = await request.get(
+    const initialResponse = await page.context().request.get(
       'http://apache:8080/cgi/imgsrv/download/pdf?id=test.pd_open&callback=tunnelCallback&_=' + currentTime
     );
     const initialBody = await initialResponse.text();
@@ -33,7 +33,7 @@ test.describe('imgsrv download', () => {
     let done = false;
 
     while (done == false) {
-      const callbackResponse = await request.get('http://apache:8080' + callbackUrl);
+      const callbackResponse = await page.context().request.get('http://apache:8080' + callbackUrl);
       const callbackJson = await callbackResponse.json();
 
       if (callbackJson.status == 'DONE') {
@@ -47,7 +47,7 @@ test.describe('imgsrv download', () => {
       }
     }
 
-    const downloadResponse = await request.get('http://apache:8080' + downloadUrl);
+    const downloadResponse = await page.context().request.get('http://apache:8080' + downloadUrl);
     const downloadHeaders = downloadResponse.headers();
     const downloadBody = await downloadResponse.text();
 
@@ -57,10 +57,10 @@ test.describe('imgsrv download', () => {
     expect(downloadBody.length).toBeGreaterThan(512 * 1024);
   });
 
-  // test('download epub', async ({ request, page }) => {
+//   test('download epub', async ({ request, page }) => {
 //     var currentTime = new Date().getTime();
 // 
-//     const initialResponse = await request.get(
+//     const initialResponse = await page.context().request.get(
 //       'http://apache:8080/cgi/imgsrv/download/epub?id=test.pd_open&callback=tunnelCallback&_=' + currentTime
 //     );
 //     const initialBody = await initialResponse.text();
@@ -79,7 +79,7 @@ test.describe('imgsrv download', () => {
 //     let done = false;
 // 
 //     while (done == false) {
-//       const callbackResponse = await request.get('http://apache:8080' + callbackUrl);
+//       const callbackResponse = await page.context().request.get('http://apache:8080' + callbackUrl);
 //       const callbackJson = await callbackResponse.json();
 // 
 //       if (callbackJson.status == 'DONE') {
@@ -93,7 +93,7 @@ test.describe('imgsrv download', () => {
 //       }
 //     }
 // 
-//     const downloadResponse = await request.get('http://apache:8080' + downloadUrl);
+//     const downloadResponse = await page.context().request.get('http://apache:8080' + downloadUrl);
 //     const downloadHeaders = downloadResponse.headers();
 //     const downloadBody = await downloadResponse.text();
 // 
@@ -106,7 +106,7 @@ test.describe('imgsrv download', () => {
   test('download single tiff current page, full resolution', async ({ request, page }) => {
     // no callback tunnel on single tiff
 
-    const downloadResponse = await request.get(
+    const downloadResponse = await page.context().request.get(
       'http://apache:8080/cgi/imgsrv/image?id=test.pd_open&attachment=1&tracker=D1&format=image/tiff&size=full&seq=1'
     );
     const downloadHeaders = downloadResponse.headers();
@@ -121,7 +121,7 @@ test.describe('imgsrv download', () => {
   test('download single page jpeg, high resolution', async ({ request, page }) => {
     //no callback tunnel on single pages
 
-    const downloadResponse = await request.get(
+    const downloadResponse = await page.context().request.get(
       'http://apache:8080/cgi/imgsrv/image?id=test.pd_open&attachment=1&tracker=D1&format=image/jpeg&size=ppi:300&seq=2'
     );
     const downloadHeaders = downloadResponse.headers();
@@ -135,7 +135,7 @@ test.describe('imgsrv download', () => {
   test('download selected pages jpeg, full resolution', async ({ request, page }) => {
     //no callback tunnel on non-tiff selections <11 pages
 
-    const downloadResponse = await request.get(
+    const downloadResponse = await page.context().request.get(
       'http://apache:8080/cgi/imgsrv/image?id=test.pd_open&attachment=1&tracker=D1&format=image/jpeg&target_ppi=0&seq=1&seq=2'
     );
     const downloadHeaders = downloadResponse.headers();
@@ -149,7 +149,7 @@ test.describe('imgsrv download', () => {
     expect(downloadBody.length).toBeGreaterThan(1);
   });
   test('download pdf with bogus seq', async ({ request, page }) => {
-    const initialResponse = await request.get(
+    const initialResponse = await page.context().request.get(
       'http://apache:8080/cgi/imgsrv/download/pdf?id=test.pd_open&attachment=1&tracker=D1&seq=mashed_potatoes'
     );
     expect(initialResponse.status()).toEqual(200);
@@ -172,7 +172,7 @@ test.describe('imgsrv download', () => {
     expect(downloadBody.length).toBeGreaterThan(1);
   });
   test('download single selected page txt', async ({ request, page }) => {
-    const downloadResponse = await request.get(
+    const downloadResponse = await page.context().request.get(
       'http://apache:8080/cgi/imgsrv/download/plaintext?id=test.pd_open&attachment=1&tracker=D5&seq=2'
     );
     const downloadHeaders = downloadResponse.headers();

pt/web/firebird/tests/imgsrv_session.spec.js

@@ -0,0 +1,70 @@
+import { expect, test } from '@playwright/test';
+
+test.describe('imgsrv requires session', () => {
+
+  test.describe('without session', () => {
+    test('download full pdf', async ({ request }) => {
+      var currentTime = new Date().getTime();
+
+      const response = await request.get(
+        'http://apache:8080/cgi/imgsrv/download/pdf?id=test.pd_open&callback=tunnelCallback&_=' + currentTime
+      );
+
+      expect(response.status()).toEqual(403)
+    });
+
+    test('download single page pdf', async ({ request }) => {
+      const response = await request.get(
+        'http://apache:8080/cgi/imgsrv/download/image?id=test.pd_open&attachment=1&tracker=D1&format=image%2Ftiff&target_ppi=0&seq=2'
+      );
+      expect(response.status()).toEqual(403);
+    });
+
+    test('page image, full resolution', async ({ request }) => {
+      const response = await request.get(
+        'http://apache:8080/cgi/imgsrv/image?id=test.pd_open&size=full&seq=1'
+      );
+
+      expect(response.status()).toEqual(403);
+    });
+
+    test('page image, default resolution', async ({ request }) => {
+      const response = await request.get(
+        'http://apache:8080/cgi/imgsrv/image?id=test.pd_open&seq=1'
+      );
+
+      expect(response.status()).toEqual(403);
+    });
+
+    test('page ocr', async ({ request }) => {
+      const response = await request.get(
+        'http://apache:8080/cgi/imgsrv/html?id=test.pd_open&seq=1'
+      );
+
+      expect(response.status()).toEqual(403);
+    });
+  });
+
+  test.describe('with session', () => { 
+    // make sure we have appropriate session cookies etc before calling imgsrv
+    test.beforeEach(async ({ page }) => {
+      await page.goto('/cgi/pt?id=test.pd_open');
+      //accept the cookie banner before each test
+      await page.getByRole('button', { name: 'Allow all cookies' }).click();
+    });
+
+    test('page image', async ({ request, page }) => {
+      const response = await page.context().request.get(
+        'http://apache:8080/cgi/imgsrv/image?id=test.pd_open&seq=1'
+      );
+      expect(response.status()).toEqual(200);
+    });
+
+    test('ocr', async ({ request, page }) => {
+      const response = await page.context().request.get(
+        'http://apache:8080/cgi/imgsrv/html?id=test.pd_open&seq=1'
+      );
+      expect(response.status()).toEqual(200);
+    });
+  });
+});