readxmllog

#!/bin/bash

# Read Log which contains some long XML lines and formats those
# by ChatGPT :-)

cat $1| python3 -c 'import sys,re,subprocess
xml_decl=re.compile(r"<\?xml\b")
root_re=re.compile(r"<([A-Za-z_][\w:.-]*)\b[^>]*?>")
end_tmpl="</%s>"
selfclose_tmpl=re.compile(r"<%s\\b[^>]*/>")
buf=[]; inxml=False; root=None
def flush(form=False):
    global buf,inxml,root
    if not buf: return
    s="".join(buf)
    if form:
        p=subprocess.run(["xmllint","--format","-"],input=s.encode(),stdout=subprocess.PIPE,stderr=subprocess.DEVNULL)
        sys.stdout.write(p.stdout.decode() if p.returncode==0 else s)
    else:
        sys.stdout.write(s)
    buf=[]; inxml=False; root=None
for line in sys.stdin:
    if not inxml:
        m=xml_decl.search(line)
        if not m: sys.stdout.write(line); continue
        pre=line[:m.start()]
        if pre: sys.stdout.write(pre)
        buf=[line[m.start():]]; inxml=True; root=None
        continue
    # already buffering XML
    m=xml_decl.search(line)
    if m:
        # new XML starts before previous finished -> flush previous raw, start new
        flush(form=False)
        pre=line[:m.start()]
        if pre: sys.stdout.write(pre)
        buf=[line[m.start():]]; inxml=True; root=None
        continue
    buf.append(line)
    if root is None:
        s="".join(buf)
        # find first element start tag after decl (skip comments/doctypes loosely by just searching)
        rm=root_re.search(s)
        if rm: root=rm.group(1)
    if root:
        s="".join(buf)
        if end_tmpl%root in s or re.search(selfclose_tmpl.pattern%re.escape(root), s):
            flush(form=True)
flush(form=False)'