Curated edition by AFZAL AND ASSOCIATES
Curator: Afzal Hosen Mandal — Lawyer · Legal Researcher · Legal Tech Enthusiast
Practice: Narsingdi Judge Court Area, Narsingdi, Bangladesh
Tagline: Justice Through Knowledge. Advocacy Through Integrity.

A structured research index of Bangladesh legislation, with original titles, source identifiers, bibliographic metadata, provisional classifications, provenance, and explicit verification limitations.

Repository: https://huggingface.co/datasets/afzalhosentipu/bangladesh-code-index
Edition attribution: AFZAL AND ASSOCIATES
Dataset schema version: 3.1
Dataset-declared as-of date: 18 October 2026
Records in the supplied JSON: 1,682

Research and archival material—not an authoritative statement of current Bangladesh law. Uploading these files does not establish legal accuracy, completeness, current applicability, or publication readiness. Consult the relevant official legislation, gazettes, commencement instruments, amendments, and other authoritative sources before relying on any legal proposition.

Release status

The supplied JSON declares:

STRUCTURAL_QA_PASSED_PRODUCTION_RELEASE_BLOCKED

This is a status stored in the dataset, not a certification by Hugging Face or an independent legal review.

Known schema issue: every record includes index_of_bd_code and source_text_available, which are not declared in the embedded record schema despite its prohibition on additional properties. The final enriched export needs schema reconciliation and fresh validation. The stored structural-pass label must not be treated as proof of conformity for this export.

Hugging Face uploads

README.md
full-dataset.json
archives/
  Index-of-the-BD-Code-AFZAL-AND-ASSOCIATES-READY.zip

full-dataset.json: original structured dataset, including a nested records array, embedded schema, taxonomy, and processing summaries.
ZIP archive: uploaded repository capture. This card does not independently enumerate or certify its contents.
README.md: this dataset card, once added to the repository root.

This inventory reflects the owner's reported successful uploads, not a live repository inspection. The expanded structure below describes the owner's supplied repository layout. Folders inside the ZIP are not automatically published as separately browsable Hugging Face folders; they must be uploaded separately if that is desired.

Expanded repository structure

The owner describes the source repository/archive layout as follows. This layout has not been independently checked against the ZIP contents or the current remote repository.

laws/
  by-id/
  by-category/
  by-type/
  by-volume/
indexes/
taxonomy/
research-preview/
source-archive/
  archive-tools/
    BDLawsActs/
      acts/
AFZAL-AND-ASSOCIATES-PROFILE.json

laws/by-id/ — one canonical metadata JSON record per main_id, using filenames such as BD-LAW-000001.json, with a reference to source JSON where available.
laws/by-category/ — compact convenience records grouped by supplied-index subject categories. These classifications remain provisional.
laws/by-type/ — convenience views grouped by the supplied document type.
laws/by-volume/ — convenience views grouped by volume number. Multi-volume records appear in each listed volume without becoming additional unique laws.
indexes/ — master, subject-category, document-type, and volume indexes described as available in JSON, JSONL, and CSV.
taxonomy/ — category labels, document types, volume registry, and the separate approved 26-code legal-service taxonomy.
research-preview/ — HTML research preview and schema 3.1 structured data from the supplied edition, not an authoritative legal publication.
source-archive/ — preserved source records, captured texts, software, and provenance material.
AFZAL-AND-ASSOCIATES-PROFILE.json — supplied professional identity, contact details, website, service links, and other firm information.

Stable identifier

main_id is the canonical cross-file identifier in the described repository layout. For example, BD-LAW-000148 maps to laws/by-id/BD-LAW-000148.json. Upstream numeric record identifiers remain provenance rather than canonical filenames. In the inspected aggregate JSON, main_id and law_id agree across all 1,682 records.

Categories, types, and volumes

The supplied index uses subject category, document type, and volume groupings. Index labels and source URLs remain provisional or unverified. Records retain these attributes under index_of_bd_code, including category_provisional, category_source, volume_numbers, and source_url_verification.

The approved 26-code legal-service taxonomy remains separate under categories; it is not silently replaced by the different index labels. Folder groupings are navigation views, not findings about legal scope or legal relationships.

The canonical laws/by-id/ record is the primary metadata record in the described layout. Captured source JSON is described as preserved under source-archive/archive-tools/BDLawsActs/acts/ when available. A source JSON file's presence does not by itself establish complete, correctly decoded, current statutory text. Grouped copies should be treated as convenience views, not independently maintained authoritative records.

Formats

The supplied repository description identifies JSON, JSONL, and CSV master and grouped index manifests, individual JSON law records, and an HTML research preview. Their existence in the archive is owner-described; their upload as separate Hugging Face files is not established by the archive upload alone.

Scope and languages

The supplied JSON contains 1,682 records, unique canonical identifiers, and unique Bangladesh Laws source identifiers. Its recorded index-year range is 1799–2026.

These figures describe the supplied collection. They do not establish that every Bangladesh law is included. A source identifier such as 1,740 is not a record count.

Primary source language	Records
Bengali	905
English	777

This is a mixed-language index, not a complete Bengali–English parallel translation corpus. The supplied edition marks translations as not generated.

Sources and provenance

Records refer to the Bangladesh Laws website and supplied chronological, PDF, and volume indexes. A source URL is a reference, not proof that it currently resolves or that its text has been fully captured.

Provenance fields include source rows, prior-record hashes, URLs, retrieval dates, capture hashes, excerpts, and verification labels. These record the dataset's assertions; they do not independently authenticate source captures or establish legal effect.

Record structure

The top-level JSON object contains metadata and a records array. Principal record groups include:

Fields	Purpose
`law_id`, `main_id`	Canonical and corresponding display identifiers
`act_name`, `titles`	Original titles and language-specific title metadata
`metadata`	Source identity, instrument type, year, number, and derived citations
`legal_status`	Status assertions and verification flags
`lifecycle`	Date observations and commencement information
`governance`, `categories`	Responsibility review and subject classification
`index_of_bd_code`	Supplied-index enrichment and provisional grouping
`lineage`	Candidate or evidenced legislative relationships
`volumes`	Recorded volume membership
`evidence`, `source_provenance`	Evidence references and provenance
`header_extraction`, `source_qa`	Capture scope, encoding, and review state
`content_summary`, `source_text_available`	Content-related fields requiring completeness and usability checks

An index record is not necessarily full statutory text. A landing-page header, preamble, or table of contents is not the complete operative law.

Known limitations

The following counts describe the supplied JSON, not independently verified legal facts:

Measure	Recorded coverage
Identity-matched header captures marked retrieved	1,483
Encoding-unusable captures	198
No-content captures	1
Records with assigned taxonomy categories	325
Records missing assigned taxonomy categories	1,357
Records with header dates	1,439
Records containing evidence entries	7
Verified current legal statuses	0
Verified current ministry allocations	0
Populated assent dates	0
Populated gazette dates	0
Populated commencement dates	1
Records marked release eligible	0

Additional cautions:

Assigned taxonomy categories are title-inferred, not legally validated.
Supplied-index subject groupings remain provisional and are separate from the approved 26-code taxonomy.
Twelve records retain conflicting index and short-title years for review.
Retrieval counters and availability indicators are not fully reconciled. A true availability flag does not guarantee usable or complete text.
Header dates must not become assent, gazette, adoption, or commencement dates without supporting evidence.
Repeal markers do not by themselves establish current status, operative dates, or savings effects.
The single populated commencement date relies on an official-site annotation; the dataset notes that the original SRO gazette was not independently retrieved.
Empty relationship lists do not establish that no amendment or repeal exists.
Identical titles can represent distinct enactments; do not deduplicate by title alone.

Unknown is not false. Preserve the difference between missing evidence and an affirmative negative finding.

Intended uses

Bibliographic browsing and legislation-index development.
Source discovery and human-led legal research triage.
Provenance review, reconciliation, and data-quality assessment.
Clearly labeled research-search prototypes.
Preparation of separately reviewed normalized or translated derivatives.

The current verification state does not support certifying current law, authoritative automated legal advice, treating title matches as legal relationships, or using these records as verified legal ground truth.

Loading the original JSON

The file is a JSON object containing a nested records array, not JSON Lines. The following example downloads that file directly instead of relying on automatic dataset discovery.

Install huggingface_hub in your Python environment first. For private access, use a token with appropriate read permission. Never commit credentials to the repository.

import getpass
import json
from huggingface_hub import hf_hub_download

# Use a terminal that supports hidden input.
token = getpass.getpass("Hugging Face read token: ")

path = hf_hub_download(
    repo_id="afzalhosentipu/bangladesh-code-index",
    filename="full-dataset.json",
    repo_type="dataset",
    token=token,
    # Add revision="EXACT_COMMIT" for a reproducible snapshot.
)

with open(path, "r", encoding="utf-8-sig") as source:
    dataset = json.load(source)

records = dataset["records"]
assert len(records) == dataset["record_count"]
print("Records:", len(records))
print("First title:", records[0]["act_name"])

This example has not been executed against the private repository as part of preparing this card.

Dataset Viewer and future exports

This card does not configure or guarantee an automatic Dataset Viewer. The nested JSON envelope, archive contents, and account-specific private-dataset viewer availability require separate consideration.

The owner describes JSONL and CSV indexes within the expanded repository. Before configuring a Viewer, verify which of those files are separately uploaded, inspect their actual row structure, and point configuration at the intended index only. If preparing another JSONL or Parquet export, use one law per row, preserve stable identifiers and uncertainty, and document transformations. Preserve the original JSON and ZIP separately.

No training, validation, or test split is claimed by this edition.

Preservation and editorial policy

Preserve original uploads and raw source captures unchanged.
Keep encoding repairs, normalized text, translations, explanations, examples, and commentary as separate, versioned derivatives.
Record source, transformation, review state, and date for derived assertions.
Keep string-match candidates separate from legally verified relationships.
Keep completeness, current-law, amendment, citation, and URL-verification claims source-pending until supported.
Revalidate the final enriched export before declaring a new release ready.

Licensing and third-party rights

No dataset-wide open license is declared in this card. Rights and redistribution conditions for source materials, archived pages, editorial additions, and other components require review.

Public availability does not justify assigning an arbitrary license to the entire collection. Select license metadata only after establishing the applicable rights and scope. This card does not grant additional reuse rights or override source terms or applicable law.

Privacy and security

The intended collection concerns legislation and research metadata, not confidential client matters. Review files before broader sharing and remove credentials, access tokens, client information, and confidential case material if present. This card does not certify a complete privacy or security scan.

AFZAL AND ASSOCIATES

The following professional details are supplied by the owner; this card does not independently verify the contact channels or linked pages.

Website: https://advafzal.blogspot.com
GitHub profile: https://github.com/afzal-hosen-mandal
Email: advafzalhosen@gmail.com · advafzalhosen@outlook.com
Phone: 01726634656
Office: Narsingdi Judge Court Area, Narsingdi, Bangladesh

The supplied repository description identifies AFZAL-AND-ASSOCIATES-PROFILE.json as containing service pages, resources, social profiles, office map links, and other firm information. These professional identity details are separate from legislative source material and do not imply government affiliation or official endorsement.

Suggested acknowledgement

AFZAL AND ASSOCIATES. Index of the BD Code — AFZAL AND ASSOCIATES Edition. Dataset-declared as-of date: 18 October 2026. Hugging Face repository: afzalhosentipu/bangladesh-code-index. Revision: [insert the exact revision used].

This acknowledges the compilation; it does not replace citations to official legal sources or grant reuse rights.

Corrections and maintenance

For a proposed correction, identify the affected record, field, proposed value, supporting source, source date, and verification basis. Distinguish legal corrections from encoding or presentation changes. Recompute quality summaries and validate the final schema before declaring a release ready.

Downloads last month: 173

Total file size:

470 MB