A helper function for extracting a JavaScript variable from a BeautifulSoup object [Example at https://stackoverflow.com/a/76366675/6146136 ]

explanation_jsonload.md

Extract JavaScript variables from BeautifulSoup objects

INPUTS

• inpX: must be a bs4 document/tag/ResultSet or a string or a list of strings
  • ( target variable must be JSON and seaparated from other variables by ; )
• varName: name of the target variable
  • ( only the first variable found with the specified name will be returned )
• selector: a CSS selector for searching the bs4 document/tag for target script
  • ( if inpX is a script-tag/ResultSet/string/list then selector doesn't matter )
• prepFn: should be a univariate function that takes a string and returns a string
  • ( for modifying the script string before searching for and parsing variable )

EXAMPLES:

• jsonload_from_script(BeautifulSoup('<script>y=8</script>'), 'y') --> 8
• jsonload_from_script(['y=8','x=7'], 'x') --> 7
• jsonload_from_script('y=8;x=7', 'x') --> 7
• jsonload_from_script('y=8;x={"a":1};w="lorem";', 'x') --> {'a':1}
• jsonload_from_script('y=8;x={"a":1};w="lorem";', 'w') --> 'lorem'

get_jsScriptVal.py

import json

def get_jsScriptVal(jSoup, valDecl, isJson=True):
    script_finder = lambda s: s and valDecl in s
    for sc in jSoup.find('script', string=script_finder):
        for st in  sc.string.split(';'):
            ls, rs, *_ = [s.strip() for s in (st.split('=', 1) + [''])]
            if ls == valDecl and rs: return json.loads(rs) if isJson else rs

jsonload_from_script.py

## For extracting a [JSON] variable by name from JavaScript string or bs4 script-tag ## 
 
import json
 
def jsonload_from_script(inpX, varName, selector='script', prepFn=lambda x:x):
    try: sList = [inpX] if inpX.name=='script' else inpX.select(selector)
    except: sList = inpX if isinstance(inpX, list) else [inpX]
    for s in sList: 
        s = getattr(s, 'string', s)
        if not (isinstance(s,str) and s.strip()): continue
        sections = prepFn(s.strip()).split(';')
 
        for i, section in enumerate(sections):
            name, val = ['', *section.split('=', 1)][-2:]
            if '=' in section and name.strip()==varName:
                sections,t1 = [val]+sections[i+1:],True 
                try: return json.loads(val)
                except: break
        else: continue
 
        while len(sections) > 1: 
            try: return json.loads(';'.join(sections))
            except: sections = sections[:-1]