1
0
mirror of https://github.com/kellyjonbrazil/jc.git synced 2025-06-25 00:37:31 +02:00
Files
jc/jc/parsers/proc_vmallocinfo.py
2022-12-27 13:59:10 -08:00

196 lines
4.5 KiB
Python

"""jc - JSON Convert `/proc/vmallocinfo` file parser
This parser will attempt to convert number values to integers. If that is
not desired, please use the `--raw` option (cli) or `raw=True` argument
(module).
Usage (cli):
$ cat /proc/vmallocinfo | jc --proc
or
$ jc /proc/vmallocinfo
or
$ cat /proc/vmallocinfo | jc --proc-vmallocinfo
Usage (module):
import jc
result = jc.parse('proc', proc_vmallocinfo_file)
or
import jc
result = jc.parse('proc_vmallocinfo', proc_vmallocinfo_file)
Schema:
[
{
"start": string,
"end": string,
"size": integer,
"caller": string,
"options": [
string
],
"phys": string
"pages": integer,
"N<id>": integer
}
]
Examples:
$ cat /proc/vmallocinfo | jc --proc -p
[
{
"start": "0xffffb3c1c0000000",
"end": "0xffffb3c1c0005000",
"size": 20480,
"caller": "map_irq_stack+0x93/0xe0",
"options": [
"vmap"
],
"phys": "0x00000000bfefe000"
},
{
"start": "0xffffb3c1c0005000",
"end": "0xffffb3c1c0007000",
"size": 8192,
"caller": "acpi_os_map_iomem+0x1ac/0x1c0",
"options": [
"ioremap"
],
"phys": "0x00000000bfeff000"
},
...
]
$ cat /proc/vmallocinfo | jc --proc-vmallocinfo -p -r
[
{
"start": "0xffffb3c1c0000000",
"end": "0xffffb3c1c0005000",
"size": "20480",
"caller": "map_irq_stack+0x93/0xe0",
"options": [
"vmap"
],
"phys": "0x00000000bfefe000"
},
{
"start": "0xffffb3c1c0005000",
"end": "0xffffb3c1c0007000",
"size": "8192",
"caller": "acpi_os_map_iomem+0x1ac/0x1c0",
"options": [
"ioremap"
],
"phys": "0x00000000bfeff000"
},
...
]
"""
from typing import List, Dict
import jc.utils
class info():
"""Provides parser metadata (version, author, etc.)"""
version = '1.0'
description = '`/proc/vmallocinfo` file parser'
author = 'Kelly Brazil'
author_email = 'kellyjonbrazil@gmail.com'
compatible = ['linux']
tags = ['file']
hidden = True
__version__ = info.version
def _process(proc_data: List[Dict]) -> List[Dict]:
"""
Final processing to conform to the schema.
Parameters:
proc_data: (List of Dictionaries) raw structured data to process
Returns:
List of Dictionaries. Structured to conform to the schema.
"""
for entry in proc_data:
for key in entry:
if isinstance(entry[key], str):
try:
entry[key] = int(entry[key])
except Exception:
pass
return proc_data
def parse(
data: str,
raw: bool = False,
quiet: bool = False
) -> List[Dict]:
"""
Main text parsing function
Parameters:
data: (string) text data to parse
raw: (boolean) unprocessed output if True
quiet: (boolean) suppress warning messages if True
Returns:
List of Dictionaries. Raw or processed structured data.
"""
jc.utils.compatibility(__name__, info.compatible, quiet)
jc.utils.input_type_check(data)
raw_output: List = []
output_line: Dict = {}
if jc.utils.has_data(data):
for line in filter(None, data.splitlines()):
area, size, details = line.split(maxsplit=2)
start, end = area.split('-', maxsplit=1)
detail_split = details.split()
caller = ''
options: List = []
if details == 'unpurged vm_area':
caller = 'unpurged vm_area'
else:
caller = detail_split[0]
for item in detail_split[1:]:
if '=' in item:
key, val = item.split('=')
output_line.update({key: val})
else:
options.append(item)
output_line = {
'start': start,
'end': end,
'size': size,
'caller': caller or None,
'options': options
}
if output_line:
raw_output.append(output_line)
return raw_output if raw else _process(raw_output)