Created attachment 137181 [details] [review]
Add type annotation to poppler_document_new_from_data()

In Poppler-0.18.gir poppler_document_new_from_data() is defined as following:

      <constructor name="new_from_data"
                   c:identifier="poppler_document_new_from_data"
                   throws="1">
        <doc xml:space="preserve">Creates a new #PopplerDocument.  If %NULL is returned, then @error will be
set. Possible errors include those in the #POPPLER_ERROR and #G_FILE_ERROR
domains.</doc>
        <return-value transfer-ownership="full">
          <doc xml:space="preserve">A newly created #PopplerDocument, or %NULL</doc>
          <type name="Document" c:type="PopplerDocument*"/>
        </return-value>
        <parameters>
          <parameter name="data" transfer-ownership="none">
            <doc xml:space="preserve">the pdf data contained in a char array</doc>
            <type name="utf8" c:type="char*"/>
          </parameter>
          <parameter name="length" transfer-ownership="none">
            <doc xml:space="preserve">the length of #data</doc>
            <type name="gint" c:type="int"/>
          </parameter>
          <parameter name="password"
                     transfer-ownership="none"
                     nullable="1"
                     allow-none="1">
            <doc xml:space="preserve">password to unlock the file with, or %NULL</doc>
            <type name="utf8" c:type="const char*"/>
          </parameter>
        </parameters>
      </constructor>


This definition validates data as UTF-8 string when read a PDF data using poppler_document_new_from_data() via gobject-introspection.
In general, PDF data is not UTF-8 string. So we should treat parameter data as binary.

I wrote tiny patch and confirm that the poppler_document_new_from_data() definition is updated after applied this patch.
However, I'm not familiar with gobject-introspection in honest.