Set text extraction mode to "physical layout".
[read/write property] physicalLayoutMode([out, retval] VARIANT_BOOL *physLayout) physicalLayoutMode([in] VARIANT_BOOL physLayout)
Setting this property to true sets the text extraction mode to "physical layout". XpdfText will maintain, as much as possible, the original text layout, including columns, hyphenation, etc.
pdf.physicalLayoutMode = True