Add QPDF::getXRefTable()

7 files changed, 142 insertions, 1 deletions

diff --git a/include/qpdf/QPDF.hh b/include/qpdf/QPDF.hh
index 736e1676..e5ea08ce 100644
--- a/include/qpdf/QPDF.hh
+++ b/include/qpdf/QPDF.hh
@@ -239,6 +239,8 @@ class QPDF
     QPDFObjectHandle getTrailer();
     QPDF_DLL
     QPDFObjectHandle getRoot();
+    QPDF_DLL
+    std::map<QPDFObjGen, QPDFXRefEntry> getXRefTable();
 
     // Install this object handle as an indirect object and return an
     // indirect reference to it.
@@ -1400,6 +1402,7 @@ class QPDF
         bool fixed_dangling_refs;
         bool immediate_copy_from;
         bool in_parse;
+        bool parsed;
 
         // Linearization data
         qpdf_offset_t first_xref_item_offset; // actual value from file
diff --git a/libqpdf/QPDF.cc b/libqpdf/QPDF.cc
index 6c0c700a..864ac2d5 100644
--- a/libqpdf/QPDF.cc
+++ b/libqpdf/QPDF.cc
@@ -155,6 +155,7 @@ QPDF::Members::Members() :
     fixed_dangling_refs(false),
     immediate_copy_from(false),
     in_parse(false),
+    parsed(false),
     first_xref_item_offset(0),
     uncompressed_after_compressed(false)
 {
@@ -431,6 +432,7 @@ QPDF::parse(char const* password)
 
     initializeEncryption();
     findAttachmentStreams();
+    this->m->parsed = true;
 }
 
 void
@@ -2620,6 +2622,17 @@ QPDF::getRoot()
     return root;
 }
 
+std::map<QPDFObjGen, QPDFXRefEntry>
+QPDF::getXRefTable()
+{
+    if (! this->m->parsed)
+    {
+        throw std::logic_error("QPDF::getXRefTable called before parsing.");
+    }
+
+    return this->m->xref_table;
+}
+
 void
 QPDF::getObjectStreamData(std::map<int, int>& omap)
 {
diff --git a/qpdf/build.mk b/qpdf/build.mk
index 87038c79..204dbd3c 100644
--- a/qpdf/build.mk
+++ b/qpdf/build.mk
@@ -6,7 +6,8 @@ BINS_qpdf = \
     test_pdf_doc_encoding \
     test_pdf_unicode \
     test_tokenizer \
-    test_unicode_filenames
+    test_unicode_filenames \
+    test_xref
 CBINS_qpdf = qpdf-ctest
 
 TARGETS_qpdf = $(foreach B,$(BINS_qpdf) $(CBINS_qpdf),qpdf/$(OUTPUT_DIR)/$(call binname,$(B)))
diff --git a/qpdf/qtest/qpdf.test b/qpdf/qtest/qpdf.test
index d5caa9ad..c4ddce9e 100644
--- a/qpdf/qtest/qpdf.test
+++ b/qpdf/qtest/qpdf.test
@@ -4019,6 +4019,23 @@ foreach my $i (qw(preserve disable generate))
 
 show_ntests();
 # ----------
+$td->notify("--- Get XRef Table ---");
+$n_tests += 2;
+
+$td->runtest("without object streams",
+	     {$td->COMMAND => "test_xref minimal.pdf"},
+	     {$td->FILE => "minimal-xref.out",
+	      $td->EXIT_STATUS => 0},
+	     $td->NORMALIZE_NEWLINES);
+
+$td->runtest("with object streams",
+	     {$td->COMMAND => "test_xref digitally-signed.pdf"},
+	     {$td->FILE => "digitally-signed-xref.out",
+	      $td->EXIT_STATUS => 0},
+	     $td->NORMALIZE_NEWLINES);
+
+show_ntests();
+# ----------
 $td->notify("--- Large File Tests ---");
 my $nlarge = 1;
 if (defined $large_file_test_path)
diff --git a/qpdf/qtest/qpdf/digitally-signed-xref.out b/qpdf/qtest/qpdf/digitally-signed-xref.out
new file mode 100644
index 00000000..46271ea5
--- /dev/null
+++ b/qpdf/qtest/qpdf/digitally-signed-xref.out
@@ -0,0 +1,30 @@
+1/0, uncompressed, offset = 64450 (0xfbc2)
+2/0, uncompressed, offset = 65177 (0xfe99)
+3/0, uncompressed, offset = 68242 (0x10a92)
+4/0, uncompressed, offset = 68383 (0x10b1f)
+5/0, uncompressed, offset = 68538 (0x10bba)
+6/0, compressed, stream number = 1, stream index = 0
+7/0, compressed, stream number = 1, stream index = 1
+8/0, compressed, stream number = 1, stream index = 2
+9/0, compressed, stream number = 3, stream index = 0
+10/0, compressed, stream number = 4, stream index = 0
+11/0, uncompressed, offset = 16 (0x10)
+12/0, uncompressed, offset = 649 (0x289)
+13/0, uncompressed, offset = 726 (0x2d6)
+14/0, uncompressed, offset = 900 (0x384)
+15/0, uncompressed, offset = 1049 (0x419)
+16/0, uncompressed, offset = 1256 (0x4e8)
+17/0, uncompressed, offset = 1382 (0x566)
+18/0, uncompressed, offset = 2411 (0x96b)
+19/0, uncompressed, offset = 57705 (0xe169)
+20/0, uncompressed, offset = 59412 (0xe814)
+21/0, uncompressed, offset = 59506 (0xe872)
+22/0, uncompressed, offset = 116 (0x74)
+23/0, compressed, stream number = 19, stream index = 0
+24/0, compressed, stream number = 19, stream index = 1
+25/0, compressed, stream number = 19, stream index = 2
+26/0, compressed, stream number = 19, stream index = 3
+27/0, compressed, stream number = 19, stream index = 4
+28/0, compressed, stream number = 19, stream index = 5
+29/0, compressed, stream number = 19, stream index = 6
+30/0, uncompressed, offset = 471 (0x1d7)
diff --git a/qpdf/qtest/qpdf/minimal-xref.out b/qpdf/qtest/qpdf/minimal-xref.out
new file mode 100644
index 00000000..55c0d57a
--- /dev/null
+++ b/qpdf/qtest/qpdf/minimal-xref.out
@@ -0,0 +1,6 @@
+1/0, uncompressed, offset = 9 (0x9)
+2/0, uncompressed, offset = 63 (0x3f)
+3/0, uncompressed, offset = 135 (0x87)
+4/0, uncompressed, offset = 307 (0x133)
+5/0, uncompressed, offset = 403 (0x193)
+6/0, uncompressed, offset = 438 (0x1b6)
diff --git a/qpdf/test_xref.cc b/qpdf/test_xref.cc
new file mode 100644
index 00000000..9e04f3f7
--- /dev/null
+++ b/qpdf/test_xref.cc
@@ -0,0 +1,71 @@
+#include <qpdf/QPDF.hh>
+#include <qpdf/QPDFObjGen.hh>
+#include <qpdf/QPDFXRefEntry.hh>
+
+#include <iostream>
+#include <map>
+#include <string>
+#include <cstdlib>
+
+int main(int argc, char *argv[])
+{
+    if (argc != 2)
+    {
+        std::cerr << "usage: test_xref INPUT.pdf" << std::endl;
+        std::exit(2);
+    }
+
+    try
+    {
+        QPDF qpdf;
+        qpdf.processFile(argv[1]);
+
+        std::map<QPDFObjGen, QPDFXRefEntry> xref
+            = qpdf.getXRefTable();
+
+        for (std::map<QPDFObjGen, QPDFXRefEntry>::iterator iter = xref.begin();
+             iter != xref.end(); ++iter)
+        {
+            std::cout
+                << iter->first.getObj() << "/" << iter->first.getGen()
+                << ", ";
+            switch (iter->second.getType())
+            {
+              case 0:
+                std::cout
+                    << "free entry"
+                    << std::endl;
+                break;
+              case 1:
+                std::cout
+                    << "uncompressed, offset = "
+                    << iter->second.getOffset()
+                    << " (0x"
+                    << std::hex << iter->second.getOffset() << std::dec
+                    << ")"
+                    << std::endl;
+                break;
+              case 2:
+                std::cout
+                    << "compressed, stream number = "
+                    << iter->second.getObjStreamNumber()
+                    << ", stream index = "
+                    << iter->second.getObjStreamIndex()
+                    << std::endl;
+                break;
+              default:
+                std::cerr
+                    << "unknown"
+                    << std::endl;
+                std::exit(2);
+            }
+        }
+    }
+    catch (std::exception& e)
+    {
+        std::cerr << e.what() << std::endl;
+        std::exit(2);
+    }
+
+    return 0;
+}