Merge pull request #1577 from m-holger/global

Refactor QPDFParser

Merge pull request #1577 from m-holger/global
Refactor QPDFParser
m-holger · GitHub
2 parents b0ee84e5 9abd73ff
Showing 18 changed files with 285 additions and 254 deletions
libqpdf/CMakeLists.txt
libqpdf/QPDFParser.cc
libqpdf/QPDF_objects.cc
libqpdf/global.cc
libqpdf/qpdf/QPDFObject_private.hh
libqpdf/qpdf/QPDFParser.hh
libqpdf/qpdf/QPDF_private.hh
libqpdf/qpdf/global_private.hh
qpdf/qpdf.testcov
qpdf/qtest/qpdf/issue-100.out
qpdf/qtest/qpdf/issue-101.out
qpdf/qtest/qpdf/issue-118.out
qpdf/qtest/qpdf/issue-150.out
qpdf/qtest/qpdf/issue-1503.out
qpdf/qtest/qpdf/issue-335a.out
qpdf/qtest/qpdf/issue-51.out
qpdf/qtest/qpdf/issue-99.out
qpdf/qtest/qpdf/parse-object.out
@@ -100,6 +100,7 @@ set(libqpdf_SOURCES
   ResourceFinder.cc
   SecureRandomDataProvider.cc
   SF_FlateLzwDecode.cc
+  global.cc
   qpdf-c.cc
   qpdfjob-c.cc
   qpdflogger-c.cc)
@@ -15,6 +15,8 @@ using namespace qpdf;
  
 using ObjectPtr = std::shared_ptr<QPDFObject>;
  
+static uint32_t const& max_nesting{global::Limits::objects_max_nesting()};
+
 // The ParseGuard class allows QPDFParser to detect re-entrant parsing. It also provides
 // special access to allow the parser to create unresolved objects and dangling references.
 class QPDF::Doc::ParseGuard
@@ -49,16 +51,18 @@ QPDFObjectHandle
 QPDFParser::parse(InputSource& input, std::string const& object_description, QPDF* context)
 {
     qpdf::Tokenizer tokenizer;
-    bool empty = false;
-    return QPDFParser(
-               input,
-               make_description(input.getName(), object_description),
-               object_description,
-               tokenizer,
-               nullptr,
-               context,
-               false)
-        .parse(empty, false);
+    if (auto result = QPDFParser(
+                          input,
+                          make_description(input.getName(), object_description),
+                          object_description,
+                          tokenizer,
+                          nullptr,
+                          context,
+                          false)
+                          .parse()) {
+        return result;
+    }
+    return {QPDFObject::create<QPDF_Null>()};
 }
  
 QPDFObjectHandle
@@ -68,19 +72,24 @@ QPDFParser::parse_content(
     qpdf::Tokenizer& tokenizer,
     QPDF* context)
 {
-    bool empty = false;
-    return QPDFParser(
-               input,
-               std::move(sp_description),
-               "content",
-               tokenizer,
-               nullptr,
-               context,
-               true,
-               0,
-               0,
-               context && context->doc().reconstructed_xref())
-        .parse(empty, true);
+    static const std::string content("content"); // GCC12 - make constexpr
+    auto p = QPDFParser(
+        input,
+        std::move(sp_description),
+        content,
+        tokenizer,
+        nullptr,
+        context,
+        true,
+        0,
+        0,
+        context && context->doc().reconstructed_xref());
+    auto result = p.parse(true);
+    if (result || p.empty_) {
+        // In content stream mode, leave object uninitialized to indicate EOF
+        return result;
+    }
+    return {QPDFObject::create<QPDF_Null>()};
 }
  
 QPDFObjectHandle
@@ -92,18 +101,25 @@ QPDFParser::parse(
     QPDFObjectHandle::StringDecrypter* decrypter,
     QPDF* context)
 {
-    return QPDFParser(
-               input,
-               make_description(input.getName(), object_description),
-               object_description,
-               *tokenizer.m,
-               decrypter,
-               context,
-               false)
-        .parse(empty, false);
+    // ABI: This parse overload is only used by the deprecated QPDFObjectHandle::parse. It is the
+    // only user of the 'empty' member. When removing this overload also remove 'empty'.
+    auto p = QPDFParser(
+        input,
+        make_description(input.getName(), object_description),
+        object_description,
+        *tokenizer.m,
+        decrypter,
+        context,
+        false);
+    auto result = p.parse();
+    empty = p.empty_;
+    if (result) {
+        return result;
+    }
+    return {QPDFObject::create<QPDF_Null>()};
 }
  
-std::pair<QPDFObjectHandle, bool>
+QPDFObjectHandle
 QPDFParser::parse(
     InputSource& input,
     std::string const& object_description,
@@ -112,54 +128,65 @@ QPDFParser::parse(
     QPDF& context,
     bool sanity_checks)
 {
-    bool empty{false};
-    auto result = QPDFParser(
-                      input,
-                      make_description(input.getName(), object_description),
-                      object_description,
-                      tokenizer,
-                      decrypter,
-                      &context,
-                      true,
-                      0,
-                      0,
-                      sanity_checks)
-                      .parse(empty, false);
-    return {result, empty};
+    return QPDFParser(
+               input,
+               make_description(input.getName(), object_description),
+               object_description,
+               tokenizer,
+               decrypter,
+               &context,
+               true,
+               0,
+               0,
+               sanity_checks)
+        .parse();
 }
  
-std::pair<QPDFObjectHandle, bool>
+QPDFObjectHandle
 QPDFParser::parse(
     is::OffsetBuffer& input, int stream_id, int obj_id, qpdf::Tokenizer& tokenizer, QPDF& context)
 {
-    bool empty{false};
-    auto result = QPDFParser(
-                      input,
-                      std::make_shared<QPDFObject::Description>(
-                          QPDFObject::ObjStreamDescr(stream_id, obj_id)),
-                      "",
-                      tokenizer,
-                      nullptr,
-                      &context,
-                      true,
-                      stream_id,
-                      obj_id)
-                      .parse(empty, false);
-    return {result, empty};
+    return QPDFParser(
+               input,
+               std::make_shared<QPDFObject::Description>(
+                   QPDFObject::ObjStreamDescr(stream_id, obj_id)),
+               "",
+               tokenizer,
+               nullptr,
+               &context,
+               true,
+               stream_id,
+               obj_id)
+        .parse();
+}
+
+QPDFObjectHandle
+QPDFParser::parse(bool content_stream)
+{
+    try {
+        return parse_first(content_stream);
+    } catch (Error&) {
+        return {};
+    } catch (QPDFExc& e) {
+        throw e;
+    } catch (std::logic_error& e) {
+        throw e;
+    } catch (std::exception& e) {
+        warn("treating object as null because of error during parsing : "s + e.what());
+        return {};
+    }
 }
  
 QPDFObjectHandle
-QPDFParser::parse(bool& empty, bool content_stream)
+QPDFParser::parse_first(bool content_stream)
 {
     // This method must take care not to resolve any objects. Don't check the type of any object
     // without first ensuring that it is a direct object. Otherwise, doing so may have the side
     // effect of reading the object and changing the file pointer. If you do this, it will cause a
     // logic error to be thrown from QPDF::inParse().
  
-    ParseGuard pg(context);
-    empty = false;
+    QPDF::Doc::ParseGuard pg(context);
     start = input.tell();
-
     if (!tokenizer.nextToken(input, object_description)) {
         warn(tokenizer.getErrorMessage());
     }
@@ -168,31 +195,27 @@ QPDFParser::parse(bool&amp; empty, bool content_stream)
     case QPDFTokenizer::tt_eof:
         if (content_stream) {
             // In content stream mode, leave object uninitialized to indicate EOF
+            empty_ = true;
             return {};
         }
-        QTC::TC("qpdf", "QPDFParser eof in parse");
         warn("unexpected EOF");
-        return {QPDFObject::create<QPDF_Null>()};
+        return {};
  
     case QPDFTokenizer::tt_bad:
-        QTC::TC("qpdf", "QPDFParser bad token in parse");
-        return {QPDFObject::create<QPDF_Null>()};
+        return {};
  
     case QPDFTokenizer::tt_brace_open:
     case QPDFTokenizer::tt_brace_close:
-        QTC::TC("qpdf", "QPDFParser bad brace");
         warn("treating unexpected brace token as null");
-        return {QPDFObject::create<QPDF_Null>()};
+        return {};
  
     case QPDFTokenizer::tt_array_close:
-        QTC::TC("qpdf", "QPDFParser bad array close");
         warn("treating unexpected array close token as null");
-        return {QPDFObject::create<QPDF_Null>()};
+        return {};
  
     case QPDFTokenizer::tt_dict_close:
-        QTC::TC("qpdf", "QPDFParser bad dictionary close");
         warn("unexpected dictionary close token");
-        return {QPDFObject::create<QPDF_Null>()};
+        return {};
  
     case QPDFTokenizer::tt_array_open:
     case QPDFTokenizer::tt_dict_open:
@@ -224,13 +247,17 @@ QPDFParser::parse(bool&amp; empty, bool content_stream)
             if (content_stream) {
                 return withDescription<QPDF_Operator>(value);
             } else if (value == "endobj") {
-                // We just saw endobj without having read anything.  Treat this as a null and do
-                // not move the input source's offset.
+                // We just saw endobj without having read anything. Nothing in the PDF spec appears
+                // to allow empty objects, but they have been encountered in actual PDF files and
+                // Adobe Reader appears to ignore them. Treat this as a null and do not move the
+                // input source's offset.
+                empty_ = true;
                 input.seek(input.getLastOffset(), SEEK_SET);
-                empty = true;
-                return {QPDFObject::create<QPDF_Null>()};
+                if (!content_stream) {
+                    warn("empty object treated as null");
+                }
+                return {};
             } else {
-                QTC::TC("qpdf", "QPDFParser treat word as string");
                 warn("unknown token while reading object; treating as string");
                 return withDescription<QPDF_String>(value);
             }
@@ -247,7 +274,7 @@ QPDFParser::parse(bool&amp; empty, bool content_stream)
  
     default:
         warn("treating unknown token type as null while reading object");
-        return {QPDFObject::create<QPDF_Null>()};
+        return {};
     }
 }
  
@@ -283,19 +310,19 @@ QPDFParser::parseRemainder(bool content_stream)
             } else if (
                 int_count >= 2 && tokenizer.getType() == QPDFTokenizer::tt_word &&
                 tokenizer.getValue() == "R") {
-                if (context == nullptr) {
-                    QTC::TC("qpdf", "QPDFParser indirect without context");
+                if (!context) {
                     throw std::logic_error(
-                        "QPDFParser::parse called without context on an object "
-                        "with indirect references");
+                        "QPDFParser::parse called without context on an object with indirect "
+                        "references");
                 }
                 auto id = QIntC::to_int(int_buffer[(int_count - 1) % 2]);
                 auto gen = QIntC::to_int(int_buffer[(int_count) % 2]);
                 if (!(id < 1 || gen < 0 || gen >= 65535)) {
                     add(ParseGuard::getObject(context, id, gen, parse_pdf));
                 } else {
-                    QTC::TC("qpdf", "QPDFParser invalid objgen");
-                    addNull();
+                    add_bad_null(
+                        "treating bad indirect reference (" + std::to_string(id) + " " +
+                        std::to_string(gen) + " R) as null");
                 }
                 int_count = 0;
                 continue;
@@ -317,34 +344,20 @@ QPDFParser::parseRemainder(bool content_stream)
                 // In content stream mode, leave object uninitialized to indicate EOF
                 return {};
             }
-            QTC::TC("qpdf", "QPDFParser eof in parseRemainder");
             warn("unexpected EOF");
-            return {QPDFObject::create<QPDF_Null>()};
+            return {};
  
         case QPDFTokenizer::tt_bad:
-            QTC::TC("qpdf", "QPDFParser bad token in parseRemainder");
-            if (tooManyBadTokens()) {
-                return {QPDFObject::create<QPDF_Null>()};
-            }
+            check_too_many_bad_tokens();
             addNull();
             continue;
  
         case QPDFTokenizer::tt_brace_open:
         case QPDFTokenizer::tt_brace_close:
-            QTC::TC("qpdf", "QPDFParser bad brace in parseRemainder");
-            warn("treating unexpected brace token as null");
-            if (tooManyBadTokens()) {
-                return {QPDFObject::create<QPDF_Null>()};
-            }
-            addNull();
+            add_bad_null("treating unexpected brace token as null");
             continue;
  
         case QPDFTokenizer::tt_array_close:
-            if ((bad_count || sanity_checks) && !max_bad_count) {
-                // Trigger warning.
-                (void)tooManyBadTokens();
-                return {QPDFObject::create<QPDF_Null>()};
-            }
             if (frame->state == st_array) {
                 auto object = frame->null_count > 100
                     ? QPDFObject::create<QPDF_Array>(std::move(frame->olist), true)
@@ -361,33 +374,22 @@ QPDFParser::parseRemainder(bool content_stream)
                 frame = &stack.back();
                 add(std::move(object));
             } else {
-                QTC::TC("qpdf", "QPDFParser bad array close in parseRemainder");
                 if (sanity_checks) {
                     // During sanity checks, assume nesting of containers is corrupt and object is
                     // unusable.
                     warn("unexpected array close token; giving up on reading object");
-                    return {QPDFObject::create<QPDF_Null>()};
+                    return {};
                 }
-                warn("treating unexpected array close token as null");
-                if (tooManyBadTokens()) {
-                    return {QPDFObject::create<QPDF_Null>()};
-                }
-                addNull();
+                add_bad_null("treating unexpected array close token as null");
             }
             continue;
  
         case QPDFTokenizer::tt_dict_close:
-            if ((bad_count || sanity_checks) && !max_bad_count) {
-                // Trigger warning.
-                (void)tooManyBadTokens();
-                return {QPDFObject::create<QPDF_Null>()};
-            }
             if (frame->state <= st_dictionary_value) {
                 // Attempt to recover more or less gracefully from invalid dictionaries.
                 auto& dict = frame->dict;
  
                 if (frame->state == st_dictionary_value) {
-                    QTC::TC("qpdf", "QPDFParser no val for last key");
                     warn(
                         frame->offset,
                         "dictionary ended prematurely; using null as value for last key");
@@ -426,22 +428,17 @@ QPDFParser::parseRemainder(bool content_stream)
                     // During sanity checks, assume nesting of containers is corrupt and object is
                     // unusable.
                     warn("unexpected dictionary close token; giving up on reading object");
-                    return {QPDFObject::create<QPDF_Null>()};
+                    return {};
                 }
-                warn("unexpected dictionary close token");
-                if (tooManyBadTokens()) {
-                    return {QPDFObject::create<QPDF_Null>()};
-                }
-                addNull();
+                add_bad_null("unexpected dictionary close token");
             }
             continue;
  
         case QPDFTokenizer::tt_array_open:
         case QPDFTokenizer::tt_dict_open:
-            if (stack.size() > 499) {
-                QTC::TC("qpdf", "QPDFParser too deep");
+            if (stack.size() > max_nesting) {
                 warn("ignoring excessively deeply nested data structure");
-                return {QPDFObject::create<QPDF_Null>()};
+                return {};
             } else {
                 b_contents = false;
                 stack.emplace_back(
@@ -499,22 +496,15 @@ QPDFParser::parseRemainder(bool content_stream)
                     warn(
                         "unexpected 'endobj' or 'endstream' while reading object; giving up on "
                         "reading object");
-                    return {QPDFObject::create<QPDF_Null>()};
+                    return {};
                 }
  
-                warn("unknown token while reading object; treating as null");
-                if (tooManyBadTokens()) {
-                    return {QPDFObject::create<QPDF_Null>()};
-                }
-                addNull();
+                add_bad_null("unknown token while reading object; treating as null");
                 continue;
             }
  
-            QTC::TC("qpdf", "QPDFParser treat word as string in parseRemainder");
             warn("unknown token while reading object; treating as string");
-            if (tooManyBadTokens()) {
-                return {QPDFObject::create<QPDF_Null>()};
-            }
+            check_too_many_bad_tokens();
             addScalar<QPDF_String>(tokenizer.getValue());
  
             continue;
@@ -538,11 +528,7 @@ QPDFParser::parseRemainder(bool content_stream)
             continue;
  
         default:
-            warn("treating unknown token type as null while reading object");
-            if (tooManyBadTokens()) {
-                return {QPDFObject::create<QPDF_Null>()};
-            }
-            addNull();
+            add_bad_null("treating unknown token type as null while reading object");
         }
     }
 }
@@ -581,6 +567,14 @@ QPDFParser::addNull()
 }
  
 void
+QPDFParser::add_bad_null(std::string const& msg)
+{
+    warn(msg);
+    check_too_many_bad_tokens();
+    addNull();
+}
+
+void
 QPDFParser::addInt(int count)
 {
     auto obj = QPDFObject::create<QPDF_Integer>(int_buffer[count % 2]);
@@ -592,12 +586,12 @@ template &lt;typename T, typename... Args&gt;
 void
 QPDFParser::addScalar(Args&&... args)
 {
-    if ((bad_count || sanity_checks) &&
-        (frame->olist.size() > 5'000 || frame->dict.size() > 5'000)) {
+    auto limit = Limits::objects_max_container_size(bad_count || sanity_checks);
+    if (frame->olist.size() > limit || frame->dict.size() > limit) {
         // Stop adding scalars. We are going to abort when the close token or a bad token is
         // encountered.
         max_bad_count = 0;
-        return;
+        check_too_many_bad_tokens(); // always throws Error()
     }
     auto obj = QPDFObject::create<T>(std::forward<Args>(args)...);
     obj->setDescription(context, description, input.getLastOffset());
@@ -647,34 +641,34 @@ QPDFParser::fixMissingKeys()
     }
 }
  
-bool
-QPDFParser::tooManyBadTokens()
+void
+QPDFParser::check_too_many_bad_tokens()
 {
-    if (frame->olist.size() > 5'000 || frame->dict.size() > 5'000) {
+    auto limit = Limits::objects_max_container_size(bad_count || sanity_checks);
+    if (frame->olist.size() > limit || frame->dict.size() > limit) {
         if (bad_count) {
             warn(
-                "encountered errors while parsing an array or dictionary with more than 5000 "
-                "elements; giving up on reading object");
-            return true;
+                "encountered errors while parsing an array or dictionary with more than " +
+                std::to_string(limit) + " elements; giving up on reading object");
+            throw Error();
         }
         warn(
-            "encountered an array or dictionary with more than 5000 elements during xref recovery; "
-            "giving up on reading object");
+            "encountered an array or dictionary with more than " + std::to_string(limit) +
+            " elements during xref recovery; giving up on reading object");
     }
     if (max_bad_count && --max_bad_count > 0 && good_count > 4) {
         good_count = 0;
         bad_count = 1;
-        return false;
+        return;
     }
     if (++bad_count > 5 ||
         (frame->state != st_array && QIntC::to_size(max_bad_count) < frame->olist.size())) {
         // Give up after 5 errors in close proximity or if the number of missing dictionary keys
         // exceeds the remaining number of allowable total errors.
         warn("too many errors; giving up on reading object");
-        return true;
+        throw Error();
     }
     good_count = 0;
-    return false;
 }
  
 void
@@ -693,7 +687,6 @@ QPDFParser::warn(QPDFExc const&amp; e) const
 void
 QPDFParser::warnDuplicateKey()
 {
-    QTC::TC("qpdf", "QPDFParser duplicate dict key");
     warn(
         frame->offset,
         "dictionary has duplicated key " + frame->key + "; last occurrence overrides earlier ones");
@@ -1233,13 +1233,9 @@ QPDFObjectHandle
 Objects::readTrailer()
 {
     qpdf_offset_t offset = m->file->tell();
-    auto [object, empty] =
+    auto object =
         QPDFParser::parse(*m->file, "trailer", m->tokenizer, nullptr, qpdf, m->reconstructed_xref);
-    if (empty) {
-        // Nothing in the PDF spec appears to allow empty objects, but they have been encountered in
-        // actual PDF files and Adobe Reader appears to ignore them.
-        warn(damagedPDF("trailer", "empty object treated as null"));
-    } else if (object.isDictionary() && m->objects.readToken(*m->file).isWord("stream")) {
+    if (object.isDictionary() && m->objects.readToken(*m->file).isWord("stream")) {
         warn(damagedPDF("trailer", m->file->tell(), "stream keyword found in trailer"));
     }
     // Override last_offset so that it points to the beginning of the object we just read
@@ -1255,19 +1251,15 @@ Objects::readObject(std::string const&amp; description, QPDFObjGen og)
  
     StringDecrypter decrypter{&qpdf, og};
     StringDecrypter* decrypter_ptr = m->encp->encrypted ? &decrypter : nullptr;
-    auto [object, empty] = QPDFParser::parse(
+    auto object = QPDFParser::parse(
         *m->file,
         m->last_object_description,
         m->tokenizer,
         decrypter_ptr,
         qpdf,
         m->reconstructed_xref || m->in_read_xref_stream);
-    ;
-    if (empty) {
-        // Nothing in the PDF spec appears to allow empty objects, but they have been encountered in
-        // actual PDF files and Adobe Reader appears to ignore them.
-        warn(damagedPDF(*m->file, m->file->getLastOffset(), "empty object treated as null"));
-        return object;
+    if (!object) {
+        return {};
     }
     auto token = readToken(*m->file);
     if (object.isDictionary() && token.isWord("stream")) {
@@ -1366,24 +1358,6 @@ Objects::validateStreamLineEnd(QPDFObjectHandle&amp; object, QPDFObjGen og, qpdf_off
     }
 }
  
-QPDFObjectHandle
-Objects::readObjectInStream(is::OffsetBuffer& input, int stream_id, int obj_id)
-{
-    auto [object, empty] = QPDFParser::parse(input, stream_id, obj_id, m->tokenizer, qpdf);
-    if (empty) {
-        // Nothing in the PDF spec appears to allow empty objects, but they have been encountered in
-        // actual PDF files and Adobe Reader appears to ignore them.
-        warn(QPDFExc(
-            qpdf_e_damaged_pdf,
-            m->file->getName() + " object stream " + std::to_string(stream_id),
-            +"object " + std::to_string(obj_id) + " 0, offset " +
-                std::to_string(input.getLastOffset()),
-            0,
-            "empty object treated as null"));
-    }
-    return object;
-}
-
 bool
 Objects ::findEndstream()
 {
@@ -1536,25 +1510,25 @@ Objects::readObjectAtOffset(
         return;
     }
  
-    QPDFObjectHandle oh = readObject(description, og);
+    if (auto oh = readObject(description, og)) {
+        // Determine the end offset of this object before and after white space.  We use these
+        // numbers to validate linearization hint tables.  Offsets and lengths of objects may imply
+        // the end of an object to be anywhere between these values.
+        qpdf_offset_t end_before_space = m->file->tell();
  
-    // Determine the end offset of this object before and after white space.  We use these
-    // numbers to validate linearization hint tables.  Offsets and lengths of objects may imply
-    // the end of an object to be anywhere between these values.
-    qpdf_offset_t end_before_space = m->file->tell();
-
-    // skip over spaces
-    while (true) {
-        char ch;
-        if (!m->file->read(&ch, 1)) {
-            throw damagedPDF(m->file->tell(), "EOF after endobj");
-        }
-        if (!isspace(static_cast<unsigned char>(ch))) {
-            m->file->seek(-1, SEEK_CUR);
-            break;
+        // skip over spaces
+        while (true) {
+            char ch;
+            if (!m->file->read(&ch, 1)) {
+                throw damagedPDF(m->file->tell(), "EOF after endobj");
+            }
+            if (!isspace(static_cast<unsigned char>(ch))) {
+                m->file->seek(-1, SEEK_CUR);
+                break;
+            }
         }
+        m->objects.updateCache(og, oh.obj_sp(), end_before_space, m->file->tell());
     }
-    m->objects.updateCache(og, oh.obj_sp(), end_before_space, m->file->tell());
 }
  
 QPDFObjectHandle
@@ -1564,7 +1538,7 @@ Objects::readObjectAtOffset(
     auto og = read_object_start(offset);
     auto oh = readObject(description, og);
  
-    if (!m->objects.isUnresolved(og)) {
+    if (!oh || !m->objects.isUnresolved(og)) {
         return oh;
     }
  
@@ -1804,8 +1778,9 @@ Objects::resolveObjectsInStream(int obj_stream_number)
         if (entry != m->xref_table.end() && entry->second.getType() == 2 &&
             entry->second.getObjStreamNumber() == obj_stream_number) {
             is::OffsetBuffer in("", {b_start + obj_offset, obj_size}, obj_offset);
-            auto oh = readObjectInStream(in, obj_stream_number, obj_id);
-            updateCache(og, oh.obj_sp(), end_before_space, end_after_space);
+            if (auto oh = QPDFParser::parse(in, obj_stream_number, obj_id, m->tokenizer, qpdf)) {
+                updateCache(og, oh.obj_sp(), end_before_space, end_after_space);
+            }
         } else {
             QTC::TC("qpdf", "QPDF not caching overridden objstm object");
         }
+#include <qpdf/global_private.hh>
+
+using namespace qpdf;
+
+global::Limits global::Limits::l;
@@ -5,11 +5,12 @@
 // include/qpdf/QPDFObject.hh. See comments there for an explanation.
  
 #include <qpdf/Constants.h>
+#include <qpdf/Types.h>
+
 #include <qpdf/JSON.hh>
 #include <qpdf/JSON_writer.hh>
 #include <qpdf/QPDF.hh>
 #include <qpdf/QPDFObjGen.hh>
-#include <qpdf/Types.h>
  
 #include <map>
 #include <memory>
@@ -5,13 +5,24 @@
 #include <qpdf/QPDFObjectHandle_private.hh>
 #include <qpdf/QPDFObject_private.hh>
 #include <qpdf/QPDFTokenizer_private.hh>
+#include <qpdf/global_private.hh>
  
 #include <memory>
 #include <string>
  
+using namespace qpdf;
+using namespace qpdf::global;
+
 class QPDFParser
 {
   public:
+    class Error: public std::exception
+    {
+      public:
+        Error() = default;
+        virtual ~Error() noexcept = default;
+    };
+
     static QPDFObjectHandle
     parse(InputSource& input, std::string const& object_description, QPDF* context);
  
@@ -30,8 +41,8 @@ class QPDFParser
         QPDFObjectHandle::StringDecrypter* decrypter,
         QPDF* context);
  
-    // For use by QPDF. Return parsed object and whether it is empty.
-    static std::pair<QPDFObjectHandle, bool> parse(
+    // For use by QPDF.
+    static QPDFObjectHandle parse(
         InputSource& input,
         std::string const& object_description,
         qpdf::Tokenizer& tokenizer,
@@ -39,7 +50,7 @@ class QPDFParser
         QPDF& context,
         bool sanity_checks);
  
-    static std::pair<QPDFObjectHandle, bool> parse(
+    static QPDFObjectHandle parse(
         qpdf::is::OffsetBuffer& input,
         int stream_id,
         int obj_id,
@@ -101,14 +112,16 @@ class QPDFParser
         int null_count{0};
     };
  
-    QPDFObjectHandle parse(bool& empty, bool content_stream);
+    QPDFObjectHandle parse(bool content_stream = false);
+    QPDFObjectHandle parse_first(bool content_stream);
     QPDFObjectHandle parseRemainder(bool content_stream);
     void add(std::shared_ptr<QPDFObject>&& obj);
     void addNull();
+    void add_bad_null(std::string const& msg);
     void addInt(int count);
     template <typename T, typename... Args>
     void addScalar(Args&&... args);
-    bool tooManyBadTokens();
+    void check_too_many_bad_tokens();
     void warnDuplicateKey();
     void fixMissingKeys();
     void warn(qpdf_offset_t offset, std::string const& msg) const;
@@ -136,7 +149,7 @@ class QPDFParser
     // it only gets incremented or reset when a bad token is encountered.
     int bad_count{0};
     // Number of bad tokens (remaining) before giving up.
-    int max_bad_count{15};
+    uint32_t max_bad_count{Limits::objects_max_errors()};
     // Number of good tokens since last bad token. Irrelevant if bad_count == 0.
     int good_count{0};
     // Start offset including any leading whitespace.
@@ -145,6 +158,7 @@ class QPDFParser
     int int_count{0};
     long long int_buffer[2]{0, 0};
     qpdf_offset_t last_offset_buffer[2]{0, 0};
+    bool empty_{false};
 };
  
 #endif // QPDFPARSER_HH
@@ -1039,7 +1039,6 @@ class QPDF::Doc::Objects: Common
     QPDFObjectHandle readObject(std::string const& description, QPDFObjGen og);
     void readStream(QPDFObjectHandle& object, QPDFObjGen og, qpdf_offset_t offset);
     void validateStreamLineEnd(QPDFObjectHandle& object, QPDFObjGen og, qpdf_offset_t offset);
-    QPDFObjectHandle readObjectInStream(qpdf::is::OffsetBuffer& input, int stream_id, int obj_id);
     size_t recoverStreamLength(
         std::shared_ptr<InputSource> input, QPDFObjGen og, qpdf_offset_t stream_offset);
  
+
+#ifndef GLOBAL_PRIVATE_HH
+#define GLOBAL_PRIVATE_HH
+
+#include <qpdf/Constants.h>
+
+#include <cstdint>
+#include <limits>
+
+namespace qpdf
+{
+    namespace global
+    {
+        class Limits
+        {
+          public:
+            Limits(Limits const&) = delete;
+            Limits(Limits&&) = delete;
+            Limits& operator=(Limits const&) = delete;
+            Limits& operator=(Limits&&) = delete;
+
+            static uint32_t const&
+            objects_max_nesting()
+            {
+                return l.objects_max_nesting_;
+            }
+
+            static uint32_t const&
+            objects_max_errors()
+            {
+                return l.objects_max_errors_;
+            }
+
+            static uint32_t const&
+            objects_max_container_size(bool damaged)
+            {
+                return damaged ? l.objects_max_container_size_damaged_
+                               : l.objects_max_container_size_;
+            }
+
+          private:
+            Limits() = default;
+            ~Limits() = default;
+
+            static Limits l;
+
+            uint32_t objects_max_nesting_{499};
+            uint32_t objects_max_errors_{15};
+            uint32_t objects_max_container_size_{std::numeric_limits<uint32_t>::max()};
+            uint32_t objects_max_container_size_damaged_{5'000};
+        };
+
+    } // namespace global
+
+} // namespace qpdf
+
+#endif // GLOBAL_PRIVATE_HH
@@ -27,11 +27,6 @@ main QTest stream 0
 QPDF lin write nshared_total > nshared_first_page 1
 QPDFWriter encrypted hint stream 0
 QPDF xref gen > 0 1
-QPDFParser bad brace 0
-QPDFParser bad brace in parseRemainder 0
-QPDFParser bad array close 0
-QPDFParser bad array close in parseRemainder 0
-QPDFParser bad dictionary close 0
 QPDFTokenizer bad ) 0
 QPDFTokenizer bad > 0
 QPDFTokenizer bad hexstring character 0
@@ -123,7 +118,6 @@ QPDF_Stream provider length not provided 0
 QPDF_Stream unknown stream length 0
 QPDF replaceReserved 0
 QPDFWriter copy use_aes 1
-QPDFParser indirect without context 0
 QPDFObjectHandle trailing data in parse 0
 QPDFTokenizer EOF reading token 0
 QPDFTokenizer EOF reading appendable token 0
@@ -145,11 +139,7 @@ QPDFJob pages range omitted in middle 0
 QPDFWriter standard deterministic ID 1
 QPDFWriter linearized deterministic ID 1
 qpdf-c called qpdf_set_deterministic_ID 0
-QPDFParser invalid objgen 0
-QPDFParser treat word as string 0
-QPDFParser treat word as string in parseRemainder 0
 QPDFParser found fake 1
-QPDFParser no val for last key 0
 QPDFObjectHandle errors in parsecontent 0
 QPDFJob split-pages %d 0
 QPDFJob split-pages .pdf 0
@@ -168,10 +158,6 @@ Pl_QPDFTokenizer found ID 0
 QPDFObjectHandle coalesce called on stream 0
 QPDFObjectHandle coalesce provide stream data 0
 QPDF_Stream bad token at end during normalize 0
-QPDFParser bad token in parse 0
-QPDFParser bad token in parseRemainder 0
-QPDFParser eof in parse 0
-QPDFParser eof in parseRemainder 0
 QPDFObjectHandle boolean returning false 0
 QPDFObjectHandle real returning 0.0 0
 QPDFObjectHandle operator returning fake value 0
@@ -189,7 +175,6 @@ QPDFObjectHandle dictionary ignoring replaceKey 0
 QPDFObjectHandle numeric non-numeric 0
 QPDFObjectHandle erase array bounds 0
 qpdf-c called qpdf_check_pdf 0
-QPDFParser too deep 0
 QPDFFormFieldObjectHelper TU present 0
 QPDFFormFieldObjectHelper TM present 0
 QPDFFormFieldObjectHelper TU absent 0
@@ -252,7 +237,6 @@ QPDFJob image optimize bits per component 0
 QPDF eof skipping spaces before xref 1
 QPDF_encryption user matches owner V < 5 0
 QPDF_encryption same password 1
-QPDFParser duplicate dict key 0
 QPDFWriter no encryption sig contents 0
 QPDFPageObjectHelper colorspace lookup 0
 QPDFPageObjectHelper filter form xobject 0
@@ -7,7 +7,6 @@ WARNING: issue-100.pdf (trailer, offset 950): recovered trailer has no /Root ent
 WARNING: issue-100.pdf (trailer, offset 488): stream keyword found in trailer
 WARNING: issue-100.pdf (trailer, offset 418): recovered trailer has no /Root entry
 WARNING: issue-100.pdf (object 1 0, offset 83): unexpected dictionary close token
-WARNING: issue-100.pdf (object 1 0, offset 87): expected endobj
 WARNING: issue-100.pdf (object 5 0, offset 268): unknown token while reading object; treating as null
 WARNING: issue-100.pdf (object 5 0, offset 286): unknown token while reading object; treating as null
 WARNING: issue-100.pdf (object 5 0, offset 289): unknown token while reading object; treating as null
@@ -15,9 +14,7 @@ WARNING: issue-100.pdf (object 5 0, offset 294): unknown token while reading obj
 WARNING: issue-100.pdf (object 5 0, offset 297): unknown token while reading object; treating as null
 WARNING: issue-100.pdf (object 5 0, offset 304): unknown token while reading object; treating as null
 WARNING: issue-100.pdf (object 5 0, offset 304): too many errors; giving up on reading object
-WARNING: issue-100.pdf (object 5 0, offset 308): expected endobj
 WARNING: issue-100.pdf (object 8 0, offset 107): invalid character ()) in hexstring
-WARNING: issue-100.pdf (object 8 0, offset 109): expected endobj
 WARNING: issue-100.pdf (object 9 0, offset 527): unknown token while reading object; treating as string
 WARNING: issue-100.pdf (object 9 0, offset 529): expected endobj
 WARNING: issue-100.pdf (object 10 0, offset 573): expected endobj
@@ -8,7 +8,6 @@ WARNING: issue-101.pdf (object 11 0, offset 637): unknown token while reading ob
 WARNING: issue-101.pdf (object 11 0, offset 639): unknown token while reading object; treating as null
 WARNING: issue-101.pdf (object 11 0, offset 644): unknown token while reading object; treating as null
 WARNING: issue-101.pdf (object 11 0, offset 644): too many errors; giving up on reading object
-WARNING: issue-101.pdf (object 11 0, offset 647): expected endobj
 WARNING: issue-101.pdf (trailer, offset 4433): recovered trailer has no /Root entry
 WARNING: issue-101.pdf (trailer, offset 4183): stream keyword found in trailer
 WARNING: issue-101.pdf (trailer, offset 4113): recovered trailer has no /Root entry
@@ -31,6 +30,8 @@ WARNING: issue-101.pdf (trailer, offset 1508): stream keyword found in trailer
 WARNING: issue-101.pdf (trailer, offset 1438): recovered trailer has no /Root entry
 WARNING: issue-101.pdf (object 2 0, offset 244): unknown token while reading object; treating as null
 WARNING: issue-101.pdf (object 2 0, offset 29): dictionary has duplicated key /Parent; last occurrence overrides earlier ones
+WARNING: issue-101.pdf (object 2 0, offset 333): treating bad indirect reference (0 0 R) as null
+WARNING: issue-101.pdf (object 5 0, offset 1247): treating bad indirect reference (0 0 R) as null
 WARNING: issue-101.pdf (object 5 0, offset 1242): dictionary ended prematurely; using null as value for last key
 WARNING: issue-101.pdf (object 5 0, offset 1242): expected dictionary keys but found non-name objects; ignoring
 WARNING: issue-101.pdf (object 7 0, offset 3855): unknown token while reading object; treating as null
@@ -40,9 +41,7 @@ WARNING: issue-101.pdf (object 7 0, offset 3866): unknown token while reading ob
 WARNING: issue-101.pdf (object 7 0, offset 3873): unknown token while reading object; treating as null
 WARNING: issue-101.pdf (object 7 0, offset 3879): unknown token while reading object; treating as null
 WARNING: issue-101.pdf (object 7 0, offset 3879): too many errors; giving up on reading object
-WARNING: issue-101.pdf (object 7 0, offset 3888): expected endobj
 WARNING: issue-101.pdf (object 8 0, offset 4067): invalid character ()) in hexstring
-WARNING: issue-101.pdf (object 8 0, offset 4069): expected endobj
 WARNING: issue-101.pdf (object 9 0, offset 2832): unknown token while reading object; treating as string
 WARNING: issue-101.pdf (object 9 0, offset 2834): expected endobj
 qpdf: issue-101.pdf: unable to find trailer dictionary while recovering damaged file
 WARNING: issue-118.pdf: can't find PDF header
+WARNING: issue-118.pdf (xref stream: object 8 0, offset 720): treating bad indirect reference (0 0 R) as null
 WARNING: issue-118.pdf (xref stream, offset 732): self-referential object stream 2
 WARNING: issue-118.pdf (xref stream, offset 732): object stream id 12336 for object 3 is impossibly large
 WARNING: issue-118.pdf (xref stream, offset 732): object stream id 12336 for object 4 is impossibly large
 WARNING: issue-150.pdf: can't find PDF header
+WARNING: issue-150.pdf (xref stream: object 8 0, offset 56): treating object as null because of error during parsing : overflow/underflow converting 9900000000000000000 to 64-bit integer
 WARNING: issue-150.pdf: file is damaged
-WARNING: issue-150.pdf: error reading xref: overflow/underflow converting 9900000000000000000 to 64-bit integer
+WARNING: issue-150.pdf (offset 4): xref not found
 WARNING: issue-150.pdf: Attempting to reconstruct cross-reference table
 WARNING: issue-150.pdf (object 8 0): object has offset 0
 qpdf: issue-150.pdf: unable to find trailer dictionary while recovering damaged file
@@ -6,11 +6,8 @@ WARNING: issue-1503.pdf (object 31 0, offset 813): unknown token while reading o
 WARNING: issue-1503.pdf (object 31 0, offset 851): unknown token while reading object; treating as null
 WARNING: issue-1503.pdf (object 31 0, offset 856): unknown token while reading object; treating as null
 WARNING: issue-1503.pdf (object 31 0, offset 861): unexpected 'endobj' or 'endstream' while reading object; giving up on reading object
-WARNING: issue-1503.pdf (object 31 0, offset 871): expected endobj
 WARNING: issue-1503.pdf (object 38 0, offset 1126): unexpected 'endobj' or 'endstream' while reading object; giving up on reading object
-WARNING: issue-1503.pdf (object 38 0, offset 1133): expected endobj
 WARNING: issue-1503.pdf (object 40 0, offset 1195): unexpected array close token; giving up on reading object
-WARNING: issue-1503.pdf (object 40 0, offset 1198): expected endobj
 WARNING: issue-1503.pdf (object 41 0, offset 1359): stream dictionary lacks /Length key
 WARNING: issue-1503.pdf (object 41 0, offset 1411): attempting to recover stream length
 WARNING: issue-1503.pdf (object 41 0, offset 1411): recovered stream length: 54
@@ -22,9 +19,7 @@ WARNING: issue-1503.pdf (object 44 0, offset 1814): name with stray # will not w
 WARNING: issue-1503.pdf (object 44 0, offset 1821): unknown token while reading object; treating as null
 WARNING: issue-1503.pdf (object 44 0, offset 1826): unknown token while reading object; treating as null
 WARNING: issue-1503.pdf (object 44 0, offset 1826): too many errors; giving up on reading object
-WARNING: issue-1503.pdf (object 44 0, offset 1829): expected endobj
 WARNING: issue-1503.pdf (object 46 0, offset 1923): unexpected array close token; giving up on reading object
-WARNING: issue-1503.pdf (object 46 0, offset 1926): expected endobj
 WARNING: issue-1503.pdf (object 47 0, offset 2087): stream dictionary lacks /Length key
 WARNING: issue-1503.pdf (object 47 0, offset 2139): attempting to recover stream length
 WARNING: issue-1503.pdf (object 47 0, offset 2139): recovered stream length: 54
@@ -59,8 +54,6 @@ WARNING: issue-1503.pdf (object 151 0, offset 3836): unknown token while reading
 WARNING: issue-1503.pdf (object 151 0, offset 3958): unknown token while reading object; treating as null
 WARNING: issue-1503.pdf (object 152 0, offset 4088): parse error while reading object
 WARNING: issue-1503.pdf (object 152 0, offset 4088): unexpected EOF
-WARNING: issue-1503.pdf (object 152 0, offset 4088): expected endobj
-WARNING: issue-1503.pdf (object 152 0, offset 4088): EOF after endobj
 WARNING: issue-1503.pdf (object 155 0, offset 162): unknown token while reading object; treating as null
 WARNING: issue-1503.pdf (object 155 0, offset 342): unknown token while reading object; treating as null
 WARNING: issue-1503.pdf (object 155 0, offset 345): unknown token while reading object; treating as null
@@ -39,12 +39,14 @@ WARNING: issue-335a.pdf (trailer, offset 22844): expected dictionary keys but fo
 WARNING: issue-335a.pdf (trailer, offset 22880): stream keyword found in trailer
 WARNING: issue-335a.pdf (trailer, offset 22840): recovered trailer has no /Root entry
 WARNING: issue-335a.pdf (trailer, offset 22702): unknown token while reading object; treating as null
+WARNING: issue-335a.pdf (trailer, offset 22713): treating bad indirect reference (0 0 R) as null
 WARNING: issue-335a.pdf (trailer, offset 22701): expected dictionary keys but found non-name objects; ignoring
 WARNING: issue-335a.pdf (trailer, offset 22746): stream keyword found in trailer
 WARNING: issue-335a.pdf (trailer, offset 22697): recovered trailer has no /Root entry
 WARNING: issue-335a.pdf (trailer, offset 22687): unknown token while reading object; treating as null
 WARNING: issue-335a.pdf (trailer, offset 22690): unknown token while reading object; treating as null
 WARNING: issue-335a.pdf (trailer, offset 22702): unknown token while reading object; treating as null
+WARNING: issue-335a.pdf (trailer, offset 22713): treating bad indirect reference (0 0 R) as null
 WARNING: issue-335a.pdf (trailer, offset 22701): expected dictionary keys but found non-name objects; ignoring
 WARNING: issue-335a.pdf (trailer, offset 22740): unknown token while reading object; treating as null
 WARNING: issue-335a.pdf (trailer, offset 22748): unknown token while reading object; treating as null
@@ -58,6 +60,7 @@ WARNING: issue-335a.pdf (trailer, offset 22675): unknown token while reading obj
 WARNING: issue-335a.pdf (trailer, offset 22687): unknown token while reading object; treating as null
 WARNING: issue-335a.pdf (trailer, offset 22690): unknown token while reading object; treating as null
 WARNING: issue-335a.pdf (trailer, offset 22702): unknown token while reading object; treating as null
+WARNING: issue-335a.pdf (trailer, offset 22713): treating bad indirect reference (0 0 R) as null
 WARNING: issue-335a.pdf (trailer, offset 22701): expected dictionary keys but found non-name objects; ignoring
 WARNING: issue-335a.pdf (trailer, offset 22740): unknown token while reading object; treating as null
 WARNING: issue-335a.pdf (trailer, offset 22748): unknown token while reading object; treating as null
@@ -66,6 +69,8 @@ WARNING: issue-335a.pdf (trailer, offset 22791): unknown token while reading obj
 WARNING: issue-335a.pdf (trailer, offset 22794): unexpected >
 WARNING: issue-335a.pdf (trailer, offset 22794): too many errors; giving up on reading object
 WARNING: issue-335a.pdf (trailer, offset 22437): unknown token while reading object; treating as null
+WARNING: issue-335a.pdf (trailer, offset 22448): treating bad indirect reference (0 0 R) as null
+WARNING: issue-335a.pdf (trailer, offset 22471): treating bad indirect reference (20 -1 R) as null
 WARNING: issue-335a.pdf (trailer, offset 22436): expected dictionary keys but found non-name objects; ignoring
 WARNING: issue-335a.pdf (trailer, offset 22482): stream keyword found in trailer
 WARNING: issue-335a.pdf (trailer, offset 22432): recovered trailer has no /Root entry
@@ -98,27 +103,30 @@ WARNING: issue-335a.pdf (trailer, offset 22134): stream keyword found in trailer
 WARNING: issue-335a.pdf (trailer, offset 22083): recovered trailer has no /Root entry
 WARNING: issue-335a.pdf (trailer, offset 22000): invalid character (t) in hexstring
 WARNING: issue-335a.pdf (trailer, offset 21937): unknown token while reading object; treating as null
+WARNING: issue-335a.pdf (trailer, offset 21948): treating bad indirect reference (0 0 R) as null
 WARNING: issue-335a.pdf (trailer, offset 21962): unknown token while reading object; treating as null
 WARNING: issue-335a.pdf (trailer, offset 21991): unknown token while reading object; treating as null
 WARNING: issue-335a.pdf (trailer, offset 22000): invalid character (t) in hexstring
 WARNING: issue-335a.pdf (trailer, offset 22003): unknown token while reading object; treating as null
+WARNING: issue-335a.pdf (trailer, offset 22026): treating bad indirect reference (-4 0 R) as null
 WARNING: issue-335a.pdf (trailer, offset 21936): dictionary has duplicated key /Length; last occurrence overrides earlier ones
 WARNING: issue-335a.pdf (trailer, offset 22028): unexpected >
 WARNING: issue-335a.pdf (trailer, offset 22030): unknown token while reading object; treating as null
-WARNING: issue-335a.pdf (trailer, offset 22038): unknown token while reading object; treating as null
-WARNING: issue-335a.pdf (trailer, offset 22038): too many errors; giving up on reading object
+WARNING: issue-335a.pdf (trailer, offset 22030): too many errors; giving up on reading object
 WARNING: issue-335a.pdf (trailer, offset 21918): unknown token while reading object; treating as null
 WARNING: issue-335a.pdf (trailer, offset 21925): unknown token while reading object; treating as null
 WARNING: issue-335a.pdf (trailer, offset 21937): unknown token while reading object; treating as null
+WARNING: issue-335a.pdf (trailer, offset 21948): treating bad indirect reference (0 0 R) as null
 WARNING: issue-335a.pdf (trailer, offset 21962): unknown token while reading object; treating as null
 WARNING: issue-335a.pdf (trailer, offset 21991): unknown token while reading object; treating as null
 WARNING: issue-335a.pdf (trailer, offset 22000): invalid character (t) in hexstring
 WARNING: issue-335a.pdf (trailer, offset 22003): unknown token while reading object; treating as null
+WARNING: issue-335a.pdf (trailer, offset 22026): treating bad indirect reference (-4 0 R) as null
 WARNING: issue-335a.pdf (trailer, offset 21936): dictionary has duplicated key /Length; last occurrence overrides earlier ones
 WARNING: issue-335a.pdf (trailer, offset 22028): unexpected >
-WARNING: issue-335a.pdf (trailer, offset 22030): unknown token while reading object; treating as null
-WARNING: issue-335a.pdf (trailer, offset 22030): too many errors; giving up on reading object
+WARNING: issue-335a.pdf (trailer, offset 22028): too many errors; giving up on reading object
 WARNING: issue-335a.pdf (trailer, offset 21837): unknown token while reading object; treating as null
+WARNING: issue-335a.pdf (trailer, offset 21861): treating bad indirect reference (0 0 R) as null
 WARNING: issue-335a.pdf (trailer, offset 21850): expected dictionary keys but found non-name objects; ignoring
 WARNING: issue-335a.pdf (trailer, offset 21892): unknown token while reading object; treating as null
 WARNING: issue-335a.pdf (trailer, offset 21900): unknown token while reading object; treating as null
 WARNING: issue-51.pdf: can't find PDF header
 WARNING: issue-51.pdf: reported number of objects (0) is not one plus the highest object number (8)
+WARNING: issue-51.pdf (object 7 0, offset 500): treating bad indirect reference (0 0 R) as null
 WARNING: issue-51.pdf (object 7 0, offset 476): dictionary has duplicated key /0000; last occurrence overrides earlier ones
 WARNING: issue-51.pdf (object 7 0, offset 553): expected endobj
 issue-51.pdf: unable to find page tree
 WARNING: issue-99.pdf: file is damaged
 WARNING: issue-99.pdf (offset 3526): xref not found
 WARNING: issue-99.pdf: Attempting to reconstruct cross-reference table
+WARNING: issue-99.pdf (trailer, offset 4635): treating bad indirect reference (0 0 R) as null
 WARNING: issue-99.pdf (trailer, offset 4613): recovered trailer has no /Root entry
 WARNING: issue-99.pdf (object 1 0, offset 775): unknown token while reading object; treating as null
 WARNING: issue-99.pdf (object 1 0, offset 795): unknown token while reading object; treating as null
@@ -8,15 +9,13 @@ WARNING: issue-99.pdf (object 1 0, offset 815): unknown token while reading obje
 WARNING: issue-99.pdf (object 1 0, offset 835): unknown token while reading object; treating as null
 WARNING: issue-99.pdf (object 1 0, offset 855): unknown token while reading object; treating as null
 WARNING: issue-99.pdf (object 1 0, offset 855): too many errors; giving up on reading object
-WARNING: issue-99.pdf (object 1 0, offset 858): expected endobj
 WARNING: issue-99.pdf (object 2 0, offset 64): expected endobj
 WARNING: issue-99.pdf (object 5 0, offset 2452): unknown token while reading object; treating as string
 WARNING: issue-99.pdf (object 6 0, offset 2506): unexpected array close token; giving up on reading object
-WARNING: issue-99.pdf (object 6 0, offset 2507): expected endobj
+WARNING: issue-99.pdf (object 8 0, offset 4281): treating bad indirect reference (0 0 R) as null
 WARNING: issue-99.pdf (object 10 0, offset 3708): expected dictionary keys but found non-name objects; ignoring
 WARNING: issue-99.pdf (object 11 0, offset 4485): unknown token while reading object; treating as null
 WARNING: issue-99.pdf (object 11 0, offset 4497): unexpected array close token; giving up on reading object
-WARNING: issue-99.pdf (object 11 0, offset 4499): expected endobj
 WARNING: issue-99.pdf: unable to find trailer dictionary while recovering damaged file
 WARNING: object 1 0: Pages tree includes non-dictionary object; ignoring
 WARNING: object 1 0: operation for dictionary attempted on object of type null: ignoring key replacement request
@@ -6,6 +6,9 @@ WARNING: parsed object: treating unexpected brace token as null
 WARNING: parsed object: treating unexpected brace token as null
 WARNING: parsed object: unexpected dictionary close token
 WARNING: bad-parse.qdf (object 7 0, offset 1121): unexpected EOF
-WARNING: bad-parse.qdf (object 7 0, offset 1121): expected endobj
-WARNING: bad-parse.qdf (object 7 0, offset 1121): EOF after endobj
+WARNING: parsed object (offset 5): treating bad indirect reference (0 0 R) as null
+WARNING: parsed object (offset 12): treating bad indirect reference (-1 0 R) as null
+WARNING: parsed object (offset 22): treating bad indirect reference (1 65535 R) as null
+WARNING: parsed object (offset 33): treating bad indirect reference (1 100000 R) as null
+WARNING: parsed object (offset 40): treating bad indirect reference (1 -1 R) as null
 test 31 done