SAS7BDAT parser: Faster string parsing (#47404)

Author: jonashaag

doc/source/whatsnew/v1.5.0.rst

@@ -799,6 +799,7 @@ Performance improvements
 - Performance improvement in :class:`BusinessHour` ``str`` and ``repr`` (:issue:`44764`)
 - Performance improvement in datetime arrays string formatting when one of the default strftime formats ``"%Y-%m-%d %H:%M:%S"`` or ``"%Y-%m-%d %H:%M:%S.%f"`` is used. (:issue:`44764`)
 - Performance improvement in :meth:`Series.to_sql` and :meth:`DataFrame.to_sql` (:class:`SQLiteTable`) when processing time arrays. (:issue:`44764`)
+- Performance improvements to :func:`read_sas` (:issue:`47403`, :issue:`47404`, :issue:`47405`)
 -
 
 .. ---------------------------------------------------------------------------

pandas/io/sas/sas.pyx

@@ -424,8 +424,11 @@ cdef class Parser:
                 jb += 1
             elif column_types[j] == column_type_string:
                 # string
-                string_chunk[js, current_row] = np.array(source[start:(
-                    start + lngt)]).tobytes().rstrip(b"\x00 ")
+                # Skip trailing whitespace. This is equivalent to calling
+                # .rstrip(b"\x00 ") but without Python call overhead.
+                while lngt > 0 and source[start+lngt-1] in b"\x00 ":
+                    lngt -= 1
+                string_chunk[js, current_row] = (&source[start])[:lngt]
                 js += 1
 
         self.current_row_on_page_index += 1