Allow arbitrary whitespace, not just newline, after xref

Fixes #27.

ChangeLog

@@ -1,3 +1,8 @@
+2013-12-14  Jay Berkenbilt  <ejb@ql.org>
+
+	* Allow anyspace rather than just newline to follow xref header.
+	This allows qpdf to read a wider range of damaged files.
+
 2013-11-29  Jay Berkenbilt  <ejb@ql.org>
 
 	* If NO_GET_ENVIRONMENT is #defined, for Windows only,

libqpdf/QPDF.cc

@@ -487,7 +487,7 @@ QPDF::read_xref(qpdf_offset_t xref_offset)
 qpdf_offset_t
 QPDF::read_xrefTable(qpdf_offset_t xref_offset)
 {
-    PCRE xref_first_re("^\\s*(\\d+)\\s+(\\d+)");
+    PCRE xref_first_re("^\\s*(\\d+)\\s+(\\d+)\\s*");
     PCRE xref_entry_re("(?s:(^\\d{10}) (\\d{5}) ([fn])[ \r\n]{2}$)");
 
     std::vector<QPDFObjGen> deleted_items;
@@ -496,7 +496,10 @@ QPDF::read_xrefTable(qpdf_offset_t xref_offset)
     bool done = false;
     while (! done)
     {
-	std::string line = this->file->readLine(50);
+        char linebuf[51];
+        memset(linebuf, 0, sizeof(linebuf));
+        this->file->read(linebuf, sizeof(linebuf) - 1);
+	std::string line = linebuf;
 	PCRE::Match m1 = xref_first_re.match(line.c_str());
 	if (! m1)
 	{
@@ -505,6 +508,8 @@ QPDF::read_xrefTable(qpdf_offset_t xref_offset)
 			  "xref table", this->file->getLastOffset(),
 			  "xref syntax invalid");
 	}
+        file->seek(this->file->getLastOffset() + m1.getMatch(0).length(),
+                   SEEK_SET);
 	int obj = atoi(m1.getMatch(1).c_str());
 	int num = atoi(m1.getMatch(2).c_str());
 	static int const xref_entry_size = 20;

qpdf/qtest/qpdf.test

@@ -464,6 +464,7 @@ $td->runtest("object with zero offset",
              {$td->COMMAND => "qpdf --check zero-offset.pdf"},
              {$td->FILE => "zero-offset.out", $td->EXIT_STATUS => 3},
              $td->NORMALIZE_NEWLINES);
+# leading-junk also has a space instead of a newline after xref
 $td->runtest("check file with leading junk",
              {$td->COMMAND => "qpdf --check leading-junk.pdf"},
              {$td->FILE => "leading-junk.out", $td->EXIT_STATUS => 0},