diff options
| author | J08nY | 2022-07-09 12:36:08 +0200 |
|---|---|---|
| committer | J08nY | 2022-07-09 12:39:42 +0200 |
| commit | 4b3692cb14282ef6eca1735b78062d2d5f0a5684 (patch) | |
| tree | 7dbf4addc2fa8c8beacd819ef8ad1bec5d480f4f | |
| parent | f11765221cf4474e46dbcfdd3b8a8289ed416d09 (diff) | |
| download | sec-certs-4b3692cb14282ef6eca1735b78062d2d5f0a5684.tar.gz sec-certs-4b3692cb14282ef6eca1735b78062d2d5f0a5684.tar.zst sec-certs-4b3692cb14282ef6eca1735b78062d2d5f0a5684.zip | |
Rename thresholds to more meaningful names.
| -rw-r--r-- | sec_certs/constants.py | 11 | ||||
| -rw-r--r-- | sec_certs/utils/pdf.py | 20 |
2 files changed, 16 insertions, 15 deletions
diff --git a/sec_certs/constants.py b/sec_certs/constants.py index fcac29d4..784ea18a 100644 --- a/sec_certs/constants.py +++ b/sec_certs/constants.py @@ -61,8 +61,9 @@ STOP_ON_UNEXPECTED_NUMS = False APPEND_DETAILED_MATCH_MATCHES = False MAX_ALLOWED_MATCH_LENGTH = 300 LINE_SEPARATOR = " " -LINES_THRESHOLD = 30 -SIZE_THRESHOLD = 1000 -AVG_LLEN_THRESHOLD = 10 -EVERY_SECOND_CHAR_THRESHOLD = 15 -ALPHA_CHARS_THRESHOLD = 0.5 + +GARBAGE_LINES_THRESHOLD = 30 +GARBAGE_SIZE_THRESHOLD = 1000 +GARBAGE_AVG_LLEN_THRESHOLD = 10 +GARBAGE_EVERY_SECOND_CHAR_THRESHOLD = 15 +GARBAGE_ALPHA_CHARS_THRESHOLD = 0.5 diff --git a/sec_certs/utils/pdf.py b/sec_certs/utils/pdf.py index cf2fb574..11d09947 100644 --- a/sec_certs/utils/pdf.py +++ b/sec_certs/utils/pdf.py @@ -14,11 +14,11 @@ from PyPDF2.generic import BooleanObject, FloatObject, IndirectObject, NumberObj from sec_certs import constants as constants from sec_certs.constants import ( - ALPHA_CHARS_THRESHOLD, - AVG_LLEN_THRESHOLD, - EVERY_SECOND_CHAR_THRESHOLD, - LINES_THRESHOLD, - SIZE_THRESHOLD, + GARBAGE_ALPHA_CHARS_THRESHOLD, + GARBAGE_AVG_LLEN_THRESHOLD, + GARBAGE_EVERY_SECOND_CHAR_THRESHOLD, + GARBAGE_LINES_THRESHOLD, + GARBAGE_SIZE_THRESHOLD, ) logger = logging.getLogger(__name__) @@ -235,19 +235,19 @@ def text_is_garbage(text: str) -> bool: alpha = 0 # If number of lines is small, this is garbage. - if lines < LINES_THRESHOLD: + if lines < GARBAGE_LINES_THRESHOLD: return True # If the file size is small, this is garbage. - if size < SIZE_THRESHOLD: + if size < GARBAGE_SIZE_THRESHOLD: return True # If the average length of a line is small, this is garbage. - if avg_line_len < AVG_LLEN_THRESHOLD: + if avg_line_len < GARBAGE_AVG_LLEN_THRESHOLD: return True # If there a small amount of lines that have more than one character at every second character, this is garbage. # This detects the ANSSI spacing issues. - if every_second < EVERY_SECOND_CHAR_THRESHOLD: + if every_second < GARBAGE_EVERY_SECOND_CHAR_THRESHOLD: return True # If there is a small ratio of alphanumeric chars to all chars, this is garbage. - if alpha < ALPHA_CHARS_THRESHOLD: + if alpha < GARBAGE_ALPHA_CHARS_THRESHOLD: return True return False |
