"Not implemented" Exception when using pywin32 to control Adobe Acrobat

Question

I have written a script in python using pywin32 to save pdf files to text that up until recently was working fine. I use similar methods in Excel. The code is below:

def __pdf2Txt(self, pdf, fileformat="com.adobe.acrobat.accesstext"):
    outputLoc = os.path.dirname(pdf)
    outputLoc = os.path.join(outputLoc, os.path.splitext(os.path.basename(pdf))[0] + '.txt')

    try:
        win32com.client.gencache.EnsureModule('{E64169B3-3592-47d2-816E-602C5C13F328}', 0, 1, 1)
        adobe = win32com.client.DispatchEx('AcroExch.App')
        pdDoc = win32com.client.DispatchEx('AcroExch.PDDoc')
        pdDoc.Open(pdf)
        jObject = pdDoc.GetJSObject()
        jObject.SaveAs(outputLoc, "com.adobe.acrobat.accesstext")
    except:
        traceback.print_exc()
        return False
    finally:
        del jObject
        pdDoc.Close()
        del pdDoc
        adobe.Exit()
        del adobe

However this code has suddenly stopped working and I get the following output:

Traceback (most recent call last):
  File "C:\Documents and Settings\ablishen\workspace\HooverKeyCreator\src\HooverKeyCreator.py", line 38, in __pdf2Txt
    jObject.SaveAs(outputLoc, "com.adobe.acrobat.accesstext")
  File "C:\Python27\lib\site-packages\win32com\client\dynamic.py", line 505, in __getattr__
    ret = self._oleobj_.Invoke(retEntry.dispid,0,invoke_type,1)
com_error: (-2147467263, 'Not implemented', None, None)
False

I have similar code written in VB that works correctly so I'm guessing that it has something to do with the COM interfaces not binding to the appropriate functions correctly? (my COM knowledge is patchy).

Answer 1

Blish, this thread holds the key to the solution you are looking for: https://mail.python.org/pipermail/python-win32/2002-March/000260.html

I admit that the post above is not the easiest to find (probably because Google scores it low based on the age of the content?).

Specifically, applying this piece of advice will get things running for you: https://mail.python.org/pipermail/python-win32/2002-March/000265.html

For reference, the complete piece of code that does not require you to manually patch dynamic.py (snippet should run pretty much out of the box):

# gets all files under ROOT_INPUT_PATH with FILE_EXTENSION and tries to extract text from them into ROOT_OUTPUT_PATH with same filename as the input file but with INPUT_FILE_EXTENSION replaced by OUTPUT_FILE_EXTENSION
from win32com.client import Dispatch
from win32com.client.dynamic import ERRORS_BAD_CONTEXT

import winerror

# try importing scandir and if found, use it as it's a few magnitudes of an order faster than stock os.walk
try:
    from scandir import walk
except ImportError:
    from os import walk

import fnmatch

import sys
import os

ROOT_INPUT_PATH = None
ROOT_OUTPUT_PATH = None
INPUT_FILE_EXTENSION = "*.pdf"
OUTPUT_FILE_EXTENSION = ".txt"

def acrobat_extract_text(f_path, f_path_out, f_basename, f_ext):
    avDoc = Dispatch("AcroExch.AVDoc") # Connect to Adobe Acrobat

    # Open the input file (as a pdf)
    ret = avDoc.Open(f_path, f_path)
    assert(ret) # FIXME: Documentation says "-1 if the file was opened successfully, 0 otherwise", but this is a bool in practise?

    pdDoc = avDoc.GetPDDoc()

    dst = os.path.join(f_path_out, ''.join((f_basename, f_ext)))

    # Adobe documentation says "For that reason, you must rely on the documentation to know what functionality is available through the JSObject interface. For details, see the JavaScript for Acrobat API Reference"
    jsObject = pdDoc.GetJSObject()

    # Here you can save as many other types by using, for instance: "com.adobe.acrobat.xml"
    jsObject.SaveAs(dst, "com.adobe.acrobat.accesstext")

    pdDoc.Close()
    avDoc.Close(True) # We want this to close Acrobat, as otherwise Acrobat is going to refuse processing any further files after a certain threshold of open files are reached (for example 50 PDFs)
    del pdDoc

if __name__ == "__main__":
    assert(5 == len(sys.argv)), sys.argv # <script name>, <script_file_input_path>, <script_file_input_extension>, <script_file_output_path>, <script_file_output_extension>

    #$ python get.txt.from.multiple.pdf.py 'C:\input' '*.pdf' 'C:\output' '.txt'

    ROOT_INPUT_PATH = sys.argv[1]
    INPUT_FILE_EXTENSION = sys.argv[2]
    ROOT_OUTPUT_PATH = sys.argv[3]
    OUTPUT_FILE_EXTENSION = sys.argv[4]

    # tuples are of schema (path_to_file, filename)
    matching_files = ((os.path.join(_root, filename), os.path.splitext(filename)[0]) for _root, _dirs, _files in walk(ROOT_INPUT_PATH) for filename in fnmatch.filter(_files, INPUT_FILE_EXTENSION))

    # patch ERRORS_BAD_CONTEXT as per https://mail.python.org/pipermail/python-win32/2002-March/000265.html
    global ERRORS_BAD_CONTEXT
    ERRORS_BAD_CONTEXT.append(winerror.E_NOTIMPL)

    for filename_with_path, filename_without_extension in matching_files:
        print "Processing '{}'".format(filename_without_extension)
        acrobat_extract_text(filename_with_path, ROOT_OUTPUT_PATH, filename_without_extension, OUTPUT_FILE_EXTENSION)

I have tested this on WinPython x64 2.7.6.3, Acrobat X Pro