Universal XBRL Parsing#

The Universal Parsing Engine (nse_xbrl_parser.core) is the foundation of the library. It can parse any XBRL 2.1 document filed with NSE or BSE, regardless of whether a specialized typed model exists for that specific filing.

How It Works#

  1. Namespace Normalization: XBRL files often mix namespaces like in-bse-fin, in-capmkt, in-ind-as, or xbrli. The engine indexes tags both by their canonical local name and Clark notation.

  2. Context Resolution: The parser identifies instant snapshots (e.g. Balance Sheet dates) vs. duration periods (e.g. Quarter or Year-To-Date), along with explicit dimensional members (e.g. Standalone vs Consolidated).

  3. Unit Coercion: Automatically recognizes currency symbols (iso4217:INR), share counts, and ratios (pure).

        flowchart LR
    A[XML Stream / File] --> B[AsyncXBRLParser]
    B --> C[XBRLContexts]
    B --> D[XBRLUnits]
    B --> E[XBRLFacts]
    C & D & E --> F[XBRLInstance]

Parsing from Multiple Sources#

From a Local File#

from nse_xbrl_parser import parse_file

instance = await parse_file("filing.xml")

From Raw Bytes or Strings#

from nse_xbrl_parser import parse_xbrl

xml_data = b"<xbrl>...</xbrl>"
instance = await parse_xbrl(xml_data)

From a ZIP Archive#

NSE and BSE frequently bundle filing XML files inside .zip archives. Use parse_archive:

from nse_xbrl_parser import parse_archive

with open("filing_package.zip", "rb") as f:
    zip_bytes = f.read()

instances = await parse_archive(zip_bytes)
for inst in instances:
    print(inst.get_fact_value("NameOfTheCompany"))

Querying Facts#

Direct Tag Lookup#

val = instance.get_fact_value("RevenueFromOperations")

Specific Context Lookup#

When multiple contexts exist (e.g. Current Quarter OneD vs YTD FourD):

val = instance.get_fact_value("RevenueFromOperations", context_ref="OneD")