Add error handling for missing or invalid Resources and invalid or duplicate ann…

…otations in page objects - Repair invalid or missing Resources in page object trees with warnings - Remove invalid Annots arrays with warnings - Warn about duplicate annotations - Update test cases and output to reflect new error handling. - Improve robustness for annotation and resource validation.

Add error handling for missing or invalid Resources and invalid or duplicate ann…
…otations in page objects - Repair invalid or missing Resources in page object trees with warnings - Remove invalid Annots arrays with warnings - Warn about duplicate annotations - Update test cases and output to reflect new error handling. - Improve robustness for annotation and resource validation.
m-holger
1 parent 5c7af348
Showing 18 changed files with 61 additions and 22 deletions
libqpdf/QPDFPageDocumentHelper.cc
libqpdf/QPDF_pages.cc
qpdf/qpdf.testcov
qpdf/qtest/form-xobject.test
qpdf/qtest/many-nulls.test
qpdf/qtest/page-errors.test
qpdf/qtest/qpdf/annotation-no-resources-warn.out
qpdf/qtest/qpdf/issue-449.out
qpdf/qtest/qpdf/many-nulls.out
qpdf/qtest/qpdf/overlay-no-resources.out
qpdf/qtest/qpdf/overlay-no-resources.pdf
qpdf/qtest/qpdf/page-inherit-mediabox-out.pdf
qpdf/qtest/qpdf/page-inherit-mediabox.out
qpdf/qtest/qpdf/page-inherit-mediabox.pdf
qpdf/qtest/qpdf/page-missing-mediabox-out.pdf
qpdf/qtest/qpdf/page-missing-mediabox.out
qpdf/qtest/qpdf/page-no-content.out
qpdf/qtest/qpdf/page-with-no-resources.out
@@ -66,7 +66,7 @@ QPDFPageDocumentHelper::flattenAnnotations(int required_flags, int forbidden_fla
     for (auto& ph: getAllPages()) {
         QPDFObjectHandle resources = ph.getAttribute("/Resources", true);
         if (!resources.isDictionary()) {
-            QTC::TC("qpdf", "QPDFPageDocumentHelper flatten resources missing or invalid");
+            // As of #1521, this should be impossible unless a user inserted an invalid page.
             resources = ph.getObjectHandle().replaceKeyAndGetNew(
                 "/Resources", QPDFObjectHandle::newDictionary());
         }
@@ -158,10 +158,38 @@ QPDF::getAllPagesInternal(
                     QPDFObjectHandle::newArray(QPDFObjectHandle::Rectangle(0, 0, 612, 792)));
                 ++errors;
             }
-            if (!resources && !kid.getKey("/Resources").isDictionary()) {
-                // Consider adding an information message
-                ++errors;
+            if (!resources) {
+                auto res = kid.getKey("/Resources");
+
+                if (!res.isDictionary()) {
+                    ++errors;
+                    kid.warn(
+                        "kid " + std::to_string(i) +
+                        " (from 0) Resources is missing or invalid; repairing");
+                    kid.replaceKey("/Resources", QPDFObjectHandle::newDictionary());
+                }
             }
+            auto annots = kid.getKey("/Annots");
+            if (!annots.null()) {
+                if (!annots.isArray()) {
+                    kid.warn(
+                        "kid " + std::to_string(i) + " (from 0) Annots is not an array; removing");
+                    kid.removeKey("/Annots");
+                    ++errors;
+                } else {
+                    QPDFObjGen::set seen_annots;
+                    for (auto& annot: annots.as_array()) {
+                        if (!seen_annots.add(annot)) {
+                            kid.warn(
+                                "kid " + std::to_string(i) +
+                                " (from 0) Annots has duplicate entry for annotation " +
+                                annot.id_gen().unparse(' '));
+                            ++errors;
+                        }
+                    }
+                }
+            }
+
             if (!seen.add(kid)) {
                 // Make a copy of the page. This does the same as shallowCopyPage in
                 // QPDFPageObjectHelper.
@@ -674,7 +674,6 @@ QPDFPageObjectHelper copied fallback 0
 QPDFPageObjectHelper used fallback without copying 0
 QPDF skipping cache for known unchecked object 0
 QPDF fix dangling triggered xref reconstruction 0
-QPDFPageDocumentHelper flatten resources missing or invalid 0
 QPDF recover xref stream 0
 QPDFJob misplaced page range 0
 QPDFJob duplicated range 0
@@ -96,7 +96,7 @@ $td-&gt;runtest(&quot;overlay on page with no resources&quot;,
              {$td->COMMAND =>
                   "qpdf --deterministic-id page-with-no-resources.pdf" .
                   " --overlay minimal.pdf -- a.pdf"},
-             {$td->STRING => "", $td->EXIT_STATUS => 0},
+             {$td->FILE => "page-with-no-resources.out", $td->EXIT_STATUS => 3},
              $td->NORMALIZE_NEWLINES);
 $td->runtest("check overlay with no resources output",
              {$td->COMMAND => "qpdf-test-compare a.pdf overlay-no-resources.pdf"},
@@ -27,7 +27,7 @@ $td-&gt;runtest(&quot;compare output&quot;,
              {$td->FILE => "many-nulls.pdf", $td->EXIT_STATUS => 0});
 $td->runtest("run check file",
              {$td->COMMAND => "qpdf --check a.pdf"},
-             {$td->FILE => "many-nulls.out", $td->EXIT_STATUS => 0},
+             {$td->FILE => "many-nulls.out", $td->EXIT_STATUS => 3},
              $td->NORMALIZE_NEWLINES);
 $td->runtest("copy sparse array",
              {$td->COMMAND => "test_driver 97 many-nulls.pdf"},
@@ -29,7 +29,7 @@ $td-&gt;runtest(&quot;check output&quot;,
              {$td->FILE => "page-missing-mediabox-out.pdf"});
 $td->runtest("handle page with inherited MediaBox",
              {$td->COMMAND => "qpdf --static-id --empty --pages page-inherit-mediabox.pdf -- a.pdf"},
-             {$td->STRING => "", $td->EXIT_STATUS => 0},
+             {$td->FILE => "page-inherit-mediabox.out", $td->EXIT_STATUS => 0},
              $td->NORMALIZE_NEWLINES);
 $td->runtest("check output",
              {$td->COMMAND => "qpdf-test-compare a.pdf page-inherit-mediabox-out.pdf"},
+WARNING: annotation-no-resources.pdf, object 7 0 at offset 1526: kid 0 (from 0) Resources is missing or invalid; repairing
+qpdf: operation succeeded with warnings; resulting file may have some problems
 WARNING: issue-449.pdf, object 3 0 at offset 139: kid 0 (from 0) MediaBox is undefined; setting to letter / ANSI A
+WARNING: issue-449.pdf, object 3 0 at offset 139: kid 0 (from 0) Resources is missing or invalid; repairing
 WARNING: issue-449.pdf, object 4 0 at offset 211: kid 1 (from 0) MediaBox is undefined; setting to letter / ANSI A
+WARNING: issue-449.pdf, object 4 0 at offset 211: kid 1 (from 0) Resources is missing or invalid; repairing
 test 69 done
@@ -2,5 +2,5 @@ checking a.pdf
 PDF Version: 1.5
 File is not encrypted
 File is not linearized
-No syntax or stream encoding errors found; the file may still contain
-errors that qpdf cannot detect
+WARNING: a.pdf object stream 1, object 5 0 at offset 2000188: kid 0 (from 0) Resources is missing or invalid; repairing
+qpdf: operation succeeded with warnings
+WARNING: annotation-no-resources.pdf, object 7 0 at offset 1526: kid 0 (from 0) Resources is missing; repairing
+qpdf: operation succeeded with warnings; resulting file may have some problems
@@ -25,6 +25,8 @@ endobj
     612
     792
   ]
+  /Resources <<
+  >>
   /Type /Pages
 >>
 endobj
@@ -163,22 +165,22 @@ xref
 0000000000 65535 f 
 0000000052 00000 n 
 0000000133 00000 n 
-0000000308 00000 n 
-0000000537 00000 n 
-0000000626 00000 n 
-0000000871 00000 n 
-0000000970 00000 n 
-0000001016 00000 n 
-0000001161 00000 n 
-0000001246 00000 n 
-0000001347 00000 n 
-0000001395 00000 n 
-0000001542 00000 n 
+0000000329 00000 n 
+0000000558 00000 n 
+0000000647 00000 n 
+0000000892 00000 n 
+0000000991 00000 n 
+0000001037 00000 n 
+0000001182 00000 n 
+0000001267 00000 n 
+0000001368 00000 n 
+0000001416 00000 n 
+0000001563 00000 n 
 trailer <<
   /Root 1 0 R
   /Size 14
   /ID [<c302d043194ffe0cc1787745a3d7787f><963eac977ec4dfaf9fbcb48aae925c7a>]
 >>
 startxref
-1578
+1599
 %%EOF
 WARNING: page-no-content.pdf, object 4 0 at offset 288: kid 1 (from 0) MediaBox is undefined; setting to letter / ANSI A
+WARNING: page-no-content.pdf, object 4 0 at offset 288: kid 1 (from 0) Resources is missing or invalid; repairing
 qpdf: operation succeeded with warnings; resulting file may have some problems
 WARNING: page-no-content.pdf, object 4 0 at offset 288: kid 1 (from 0) MediaBox is undefined; setting to letter / ANSI A
+WARNING: page-no-content.pdf, object 4 0 at offset 288: kid 1 (from 0) Resources is missing or invalid; repairing
 page 1: 3 0 R
   content:
     6 0 R
+WARNING: page-with-no-resources.pdf, object 3 0 at offset 133: kid 0 (from 0) Resources is missing or invalid; repairing
+qpdf: operation succeeded with warnings; resulting file may have some problems