Replace chompjs with pure Python code

The new implementation is good enough for the current usage (brave)
This commit is contained in:
Alexandre Flament 2023-09-09 10:18:39 +00:00 committed by MatthieuBarbu
parent 33065a619e
commit d4db69b69b
3 changed files with 75 additions and 3 deletions

View File

@ -17,4 +17,3 @@ markdown-it-py==3.0.0
typing_extensions==4.7.1
fasttext-predict==0.9.2.1
pytomlpp==1.0.13
chompjs==1.2.2

View File

@ -104,7 +104,6 @@ from urllib.parse import (
parse_qs,
)
import chompjs
from lxml import html
from searx import locales
@ -112,6 +111,7 @@ from searx.utils import (
extract_text,
eval_xpath_list,
eval_xpath_getindex,
js_variable_to_python,
)
from searx.enginelib.traits import EngineTraits
@ -215,7 +215,7 @@ def response(resp):
datastr = line.replace("const data = ", "").strip()[:-1]
break
json_data = chompjs.parse_js_object(datastr)
json_data = js_variable_to_python(datastr)
json_resp = json_data[1]['data']['body']['response']
if brave_category == 'news':

View File

@ -7,6 +7,7 @@
import re
import importlib
import importlib.util
import json
import types
from typing import Optional, Union, Any, Set, List, Dict, MutableMapping, Tuple, Callable
@ -37,6 +38,9 @@ _BLOCKED_TAGS = ('script', 'style')
_ECMA_UNESCAPE4_RE = re.compile(r'%u([0-9a-fA-F]{4})', re.UNICODE)
_ECMA_UNESCAPE2_RE = re.compile(r'%([0-9a-fA-F]{2})', re.UNICODE)
_JS_QUOTE_KEYS_RE = re.compile(r'([\{\s,])(\w+)(:)')
_JS_VOID_RE = re.compile(r'void\s+[0-9]+|void\s*\([0-9]+\)')
_STORAGE_UNIT_VALUE: Dict[str, int] = {
'TB': 1024 * 1024 * 1024 * 1024,
'GB': 1024 * 1024 * 1024,
@ -645,3 +649,72 @@ def detect_language(text: str, threshold: float = 0.3, only_search_languages: bo
return None
return language
return None
def js_variable_to_python(js_variable):
"""Convert a javascript variable into JSON and then load the value
It does not deal with all cases, but it is good enough for now.
chompjs has a better implementation.
"""
# when in_string is not None, it contains the character that has opened the string
# either simple quote or double quote
in_string = None
# cut the string:
# r"""{ a:"f\"irst", c:'sec"ond'}"""
# becomes
# ['{ a:', '"', 'f\\', '"', 'irst', '"', ', c:', "'", 'sec', '"', 'ond', "'", '}']
parts = re.split(r'(["\'])', js_variable)
# previous part (to check the escape character antislash)
previous_p = ""
for i, p in enumerate(parts):
# parse characters inside a ECMA string
if in_string:
# we are in a JS string: replace the colon by a temporary character
# so quote_keys_regex doesn't have to deal with colon inside the JS strings
parts[i] = parts[i].replace(':', chr(1))
if in_string == "'":
# the JS string is delimited by simple quote.
# This is not supported by JSON.
# simple quote delimited string are converted to double quote delimited string
# here, inside a JS string, we escape the double quote
parts[i] = parts[i].replace('"', r'\"')
# deal with delimieters and escape character
if not in_string and p in ('"', "'"):
# we are not in string
# but p is double or simple quote
# that's the start of a new string
# replace simple quote by double quote
# (JSON doesn't support simple quote)
parts[i] = '"'
in_string = p
continue
if p == in_string:
# we are in a string and the current part MAY close the string
if len(previous_p) > 0 and previous_p[-1] == '\\':
# there is an antislash just before: the ECMA string continue
continue
# the current p close the string
# replace simple quote by double quote
parts[i] = '"'
in_string = None
#
if not in_string:
# replace void 0 by null
# https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Operators/void
# we are sure there is no string in p
parts[i] = _JS_VOID_RE.sub("null", p)
# update previous_p
previous_p = p
# join the string
s = ''.join(parts)
# add quote arround the key
# { a: 12 }
# becomes
# { "a": 12 }
s = _JS_QUOTE_KEYS_RE.sub(r'\1"\2"\3', s)
# replace the surogate character by colon
s = s.replace(chr(1), ':')
# load the JSON and return the result
return json.loads(s)