Add warnings for some additional page tree repair

8 files changed, 34 insertions, 7 deletions

diff --git a/libqpdf/QPDF_pages.cc b/libqpdf/QPDF_pages.cc
index e8416b2b..d43267d2 100644
--- a/libqpdf/QPDF_pages.cc
+++ b/libqpdf/QPDF_pages.cc
@@ -110,12 +110,19 @@ QPDF::getAllPagesInternal(
             QPDFObjectHandle kid = kids.getArrayItem(i);
             if (!kid.isIndirect()) {
                 QTC::TC("qpdf", "QPDF handle direct page object");
+                cur_node.warnIfPossible(
+                    "kid " + QUtil::int_to_string(i) +
+                    " (from 0) is direct; converting to indirect");
                 kid = makeIndirectObject(kid);
                 kids.setArrayItem(i, kid);
             } else if (seen.count(kid.getObjGen())) {
                 // Make a copy of the page. This does the same as
                 // shallowCopyPage in QPDFPageObjectHelper.
                 QTC::TC("qpdf", "QPDF resolve duplicated page object");
+                cur_node.warnIfPossible(
+                    "kid " + QUtil::int_to_string(i) +
+                    " (from 0) appears more than once in the pages tree;"
+                    " creating a new page object as a copy");
                 kid = makeIndirectObject(QPDFObjectHandle(kid).shallowCopy());
                 kids.setArrayItem(i, kid);
             }
diff --git a/qpdf/qtest/pages-tree.test b/qpdf/qtest/pages-tree.test
index 4ad260ef..e40d6de1 100644
--- a/qpdf/qtest/pages-tree.test
+++ b/qpdf/qtest/pages-tree.test
@@ -14,13 +14,14 @@ cleanup();
 
 my $td = new TestDriver('pages-tree');
 
-my $n_tests = 11;
+my $n_tests = 13;
 
 $td->runtest("linearize duplicated pages",
              {$td->COMMAND =>
                   "qpdf --static-id --linearize" .
                   " page_api_2.pdf a.pdf"},
-             {$td->STRING => "", $td->EXIT_STATUS => 0},
+             {$td->FILE => "duplicate-page-warning.out",
+                  $td->EXIT_STATUS => 3},
              $td->NORMALIZE_NEWLINES);
 $td->runtest("compare files",
              {$td->FILE => "a.pdf"},
@@ -29,7 +30,8 @@ $td->runtest("extract duplicated pages",
              {$td->COMMAND =>
                   "qpdf --static-id page_api_2.pdf" .
                   " --pages . -- a.pdf"},
-             {$td->STRING => "", $td->EXIT_STATUS => 0},
+             {$td->FILE => "duplicate-page-warning.out",
+                  $td->EXIT_STATUS => 3},
              $td->NORMALIZE_NEWLINES);
 $td->runtest("compare files",
              {$td->FILE => "a.pdf"},
@@ -37,7 +39,7 @@ $td->runtest("compare files",
 $td->runtest("direct pages",
              {$td->COMMAND =>
                   "qpdf --static-id direct-pages.pdf --pages . -- a.pdf"},
-             {$td->STRING => "", $td->EXIT_STATUS => 0},
+             {$td->FILE => "direct-page-warning.out", $td->EXIT_STATUS => 3},
              $td->NORMALIZE_NEWLINES);
 $td->runtest("check output",
              {$td->FILE => "a.pdf"},
@@ -45,7 +47,7 @@ $td->runtest("check output",
 $td->runtest("show direct pages",
              {$td->COMMAND =>
                   "qpdf --show-pages direct-pages.pdf"},
-             {$td->FILE => "direct-pages.out", $td->EXIT_STATUS => 0},
+             {$td->FILE => "direct-pages.out", $td->EXIT_STATUS => 3},
              $td->NORMALIZE_NEWLINES);
 
 # Json mode for direct and duplicated pages illustrates that the
@@ -62,8 +64,12 @@ foreach my $f (qw(page_api_2 direct-pages))
     $td->runtest("json for $f (with pages)",
                  {$td->COMMAND =>
                       "qpdf --json=latest $f.pdf" .
-                      " --json-key=objects --json-key=pages"},
-                 {$td->FILE => "$f-json-pages.out", $td->EXIT_STATUS => 0},
+                      " --json-key=objects --json-key=pages 2>warn.tmpout"},
+                 {$td->FILE => "$f-json-pages.out", $td->EXIT_STATUS => 3},
+                 $td->NORMALIZE_NEWLINES);
+    $td->runtest("check page warnings",
+                 {$td->FILE => "warn.tmpout"},
+                 {$td->FILE => "$f-json-warning.out"},
                  $td->NORMALIZE_NEWLINES);
 }
 
diff --git a/qpdf/qtest/qpdf/direct-page-warning.out b/qpdf/qtest/qpdf/direct-page-warning.out
new file mode 100644
index 00000000..253dbb8c
--- /dev/null
+++ b/qpdf/qtest/qpdf/direct-page-warning.out
@@ -0,0 +1,3 @@
+WARNING: direct-pages.pdf, object 2 0 at offset 143: kid 0 (from 0) is direct; converting to indirect
+WARNING: direct-pages.pdf, object 2 0 at offset 143: kid 1 (from 0) is direct; converting to indirect
+qpdf: operation succeeded with warnings; resulting file may have some problems
diff --git a/qpdf/qtest/qpdf/direct-pages-json-warning.out b/qpdf/qtest/qpdf/direct-pages-json-warning.out
new file mode 100644
index 00000000..2427e4bb
--- /dev/null
+++ b/qpdf/qtest/qpdf/direct-pages-json-warning.out
@@ -0,0 +1,3 @@
+WARNING: direct-pages.pdf, object 2 0 at offset 143: kid 0 (from 0) is direct; converting to indirect
+WARNING: direct-pages.pdf, object 2 0 at offset 143: kid 1 (from 0) is direct; converting to indirect
+qpdf: operation succeeded with warnings
diff --git a/qpdf/qtest/qpdf/direct-pages.out b/qpdf/qtest/qpdf/direct-pages.out
index 4b4c3c47..5dd9d1af 100644
--- a/qpdf/qtest/qpdf/direct-pages.out
+++ b/qpdf/qtest/qpdf/direct-pages.out
@@ -1,6 +1,9 @@
+WARNING: direct-pages.pdf, object 2 0 at offset 143: kid 0 (from 0) is direct; converting to indirect
+WARNING: direct-pages.pdf, object 2 0 at offset 143: kid 1 (from 0) is direct; converting to indirect
 page 1: 7 0 R
   content:
     3 0 R
 page 2: 8 0 R
   content:
     3 0 R
+qpdf: operation succeeded with warnings
diff --git a/qpdf/qtest/qpdf/duplicate-page-warning.out b/qpdf/qtest/qpdf/duplicate-page-warning.out
new file mode 100644
index 00000000..6e751811
--- /dev/null
+++ b/qpdf/qtest/qpdf/duplicate-page-warning.out
@@ -0,0 +1,2 @@
+WARNING: page_api_2.pdf, object 3 0 at offset 175: kid 1 (from 0) appears more than once in the pages tree; creating a new page object as a copy
+qpdf: operation succeeded with warnings; resulting file may have some problems
diff --git a/qpdf/qtest/qpdf/page_api_2-json-warning.out b/qpdf/qtest/qpdf/page_api_2-json-warning.out
new file mode 100644
index 00000000..274d6bfe
--- /dev/null
+++ b/qpdf/qtest/qpdf/page_api_2-json-warning.out
@@ -0,0 +1,2 @@
+WARNING: page_api_2.pdf, object 3 0 at offset 175: kid 1 (from 0) appears more than once in the pages tree; creating a new page object as a copy
+qpdf: operation succeeded with warnings
diff --git a/qpdf/qtest/qpdf/page_api_2.out b/qpdf/qtest/qpdf/page_api_2.out
index 30e9b9c3..f874486d 100644
--- a/qpdf/qtest/qpdf/page_api_2.out
+++ b/qpdf/qtest/qpdf/page_api_2.out
@@ -1 +1,2 @@
+WARNING: page_api_2.pdf, object 3 0 at offset 175: kid 1 (from 0) appears more than once in the pages tree; creating a new page object as a copy
 test 17 done