Merge pull request #1392 from m-holger/i1335

Refine recovery from missing startxref (fixes #1335)

Merge pull request #1392 from m-holger/i1335
Refine recovery from missing startxref (fixes #1335)
m-holger · GitHub
2 parents ab48d664 7927241d
Showing 4 changed files with 29 additions and 4 deletions
include/qpdf/QPDF.hh
libqpdf/QPDF.cc
qpdf/qpdf.testcov
qpdf/qtest/qpdf/recover-xref-stream.out
@@ -762,7 +762,7 @@ class QPDF
     void setTrailer(QPDFObjectHandle obj);
     void read_xref(qpdf_offset_t offset);
     bool resolveXRefTable();
-    void reconstruct_xref(QPDFExc& e);
+    void reconstruct_xref(QPDFExc& e, bool found_startxref = true);
     bool parse_xrefFirst(std::string const& line, int& obj, int& num, int& bytes);
     bool read_xrefEntry(qpdf_offset_t& f1, int& f2, char& type);
     bool read_bad_xrefEntry(qpdf_offset_t& f1, int& f2, char& type);
@@ -471,7 +471,7 @@ QPDF::parse(char const* password)
         }
     } catch (QPDFExc& e) {
         if (m->attempt_recovery) {
-            reconstruct_xref(e);
+            reconstruct_xref(e, xref_offset > 0);
             QTC::TC("qpdf", "QPDF reconstructed xref table");
         } else {
             throw;
@@ -531,7 +531,7 @@ QPDF::setTrailer(QPDFObjectHandle obj)
 }
 void
-QPDF::reconstruct_xref(QPDFExc& e)
+QPDF::reconstruct_xref(QPDFExc& e, bool found_startxref)
 {
     if (m->reconstructed_xref) {
         // Avoid xref reconstruction infinite loops. This is getting very hard to reproduce because
@@ -569,6 +569,7 @@ QPDF::reconstruct_xref(QPDFExc&amp; e)
     std::vector<std::tuple<int, int, qpdf_offset_t>> found_objects;
     std::vector<qpdf_offset_t> trailers;
+    std::vector<qpdf_offset_t> startxrefs;
     m->file->seek(0, SEEK_END);
     qpdf_offset_t eof = m->file->tell();
@@ -594,11 +595,34 @@ QPDF::reconstruct_xref(QPDFExc&amp; e)
             m->file->seek(pos, SEEK_SET);
         } else if (!m->trailer && t1.isWord("trailer")) {
             trailers.emplace_back(m->file->tell());
+        } else if (!found_startxref && t1.isWord("startxref")) {
+            startxrefs.emplace_back(m->file->tell());
         }
         check_warnings();
         m->file->findAndSkipNextEOL();
     }
+    if (!found_startxref && !startxrefs.empty() && !found_objects.empty() &&
+        startxrefs.back() > std::get<2>(found_objects.back())) {
+        try {
+            m->file->seek(startxrefs.back(), SEEK_SET);
+            if (auto offset = QUtil::string_to_ll(readToken(*m->file).getValue().data())) {
+                read_xref(offset);
+                if (getRoot().getKey("/Pages").isDictionary()) {
+                    QTC::TC("qpdf", "QPDF startxref more than 1024 before end");
+                    warn(
+                        damagedPDF("", 0, "startxref was more than 1024 bytes before end of file"));
+                    initializeEncryption();
+                    m->parsed = true;
+                    m->reconstructed_xref = false;
+                    return;
+                }
+            }
+        } catch (...) {
+            // ok, bad luck. Do recovery.
+        }
+    }
+
     auto rend = found_objects.rend();
     for (auto it = found_objects.rbegin(); it != rend; it++) {
         auto [obj, gen, token_start] = *it;
@@ -53,6 +53,7 @@ QPDF xref gen &gt; 0 1
 QPDF xref size mismatch 0
 QPDF not a pdf file 0
 QPDF can't find startxref 0
+QPDF startxref more than 1024 before end 0
 QPDF invalid xref 0
 QPDF invalid xref entry 0
 QPDF missing trailer 0
 WARNING: recover-xref-stream.pdf: file is damaged
 WARNING: recover-xref-stream.pdf: can't find startxref
 WARNING: recover-xref-stream.pdf: Attempting to reconstruct cross-reference table
-WARNING: recover-xref-stream.pdf: reported number of objects (14) is not one plus the highest object number (15)
+WARNING: recover-xref-stream.pdf: startxref was more than 1024 bytes before end of file
 qpdf: operation succeeded with warnings; resulting file may have some problems