| 1 | n/a | #!/usr/bin/env python3 |
|---|
| 2 | n/a | |
|---|
| 3 | n/a | """ Compare the output of two codecs. |
|---|
| 4 | n/a | |
|---|
| 5 | n/a | (c) Copyright 2005, Marc-Andre Lemburg (mal@lemburg.com). |
|---|
| 6 | n/a | |
|---|
| 7 | n/a | Licensed to PSF under a Contributor Agreement. |
|---|
| 8 | n/a | |
|---|
| 9 | n/a | """ |
|---|
| 10 | n/a | import sys |
|---|
| 11 | n/a | |
|---|
| 12 | n/a | def compare_codecs(encoding1, encoding2): |
|---|
| 13 | n/a | |
|---|
| 14 | n/a | print('Comparing encoding/decoding of %r and %r' % (encoding1, encoding2)) |
|---|
| 15 | n/a | mismatch = 0 |
|---|
| 16 | n/a | # Check encoding |
|---|
| 17 | n/a | for i in range(sys.maxunicode+1): |
|---|
| 18 | n/a | u = chr(i) |
|---|
| 19 | n/a | try: |
|---|
| 20 | n/a | c1 = u.encode(encoding1) |
|---|
| 21 | n/a | except UnicodeError as reason: |
|---|
| 22 | n/a | c1 = '<undefined>' |
|---|
| 23 | n/a | try: |
|---|
| 24 | n/a | c2 = u.encode(encoding2) |
|---|
| 25 | n/a | except UnicodeError as reason: |
|---|
| 26 | n/a | c2 = '<undefined>' |
|---|
| 27 | n/a | if c1 != c2: |
|---|
| 28 | n/a | print(' * encoding mismatch for 0x%04X: %-14r != %r' % \ |
|---|
| 29 | n/a | (i, c1, c2)) |
|---|
| 30 | n/a | mismatch += 1 |
|---|
| 31 | n/a | # Check decoding |
|---|
| 32 | n/a | for i in range(256): |
|---|
| 33 | n/a | c = bytes([i]) |
|---|
| 34 | n/a | try: |
|---|
| 35 | n/a | u1 = c.decode(encoding1) |
|---|
| 36 | n/a | except UnicodeError: |
|---|
| 37 | n/a | u1 = '<undefined>' |
|---|
| 38 | n/a | try: |
|---|
| 39 | n/a | u2 = c.decode(encoding2) |
|---|
| 40 | n/a | except UnicodeError: |
|---|
| 41 | n/a | u2 = '<undefined>' |
|---|
| 42 | n/a | if u1 != u2: |
|---|
| 43 | n/a | print(' * decoding mismatch for 0x%04X: %-14r != %r' % \ |
|---|
| 44 | n/a | (i, u1, u2)) |
|---|
| 45 | n/a | mismatch += 1 |
|---|
| 46 | n/a | if mismatch: |
|---|
| 47 | n/a | print() |
|---|
| 48 | n/a | print('Found %i mismatches' % mismatch) |
|---|
| 49 | n/a | else: |
|---|
| 50 | n/a | print('-> Codecs are identical.') |
|---|
| 51 | n/a | |
|---|
| 52 | n/a | if __name__ == '__main__': |
|---|
| 53 | n/a | compare_codecs(sys.argv[1], sys.argv[2]) |
|---|