Basis: sourced records (URL, sha256 or locator per row), no model.
stored files, documents and chunks, and counts over them
kda.gov.bd: stored files
405
kda.gov.bd: PDF files
405
kda.gov.bd: first upload folder
2024/12
kda.gov.bd: last upload folder
2026/8
kda.gov.bd: files by type
Type
Files
Bytes
pdf
Files: 405
Bytes: 592,298,951
kda.gov.bd: files by upload folder, as printed
Upload folder
Files
2024/12
Files: 377
2026/2
Files: 1
2026/3
Files: 1
kda.gov.bd: text-layer checks
Verdict
PDFs
No text
PDFs: 1
kda.gov.bd: stored files, first 50 of 405
Manifest line
Type
Bytes
Upload folder
37,701
Type: pdf
Bytes: 1,519,592
Upload folder: 2024/12
37,702
Type:
Where these figures come from
The source collections
Source
Publisher
Unit
Caveat
Stored files from government portals
Publisher: Uploading offices of the national portal (stored copies)
Unit: one row per stored file: manifest line, URL, sha256 at fetch, bytes, office, upload folder, file type; source file is relative to the archive disk root
Caveat: upload folder is the object store's <year>/<number> folder exactly as it appears in the URL, not the document's own date; under 2026 the number runs 0 to 8, so it may be a zero-based month, which is not verified. The issuing body is the office-<slug> URL segment; its portal host is the slug with hyphens read as dots. Body and Place links are exact-domain matches only. sha256 verified is set only for re-hashed files (the random sample and the pilot); the rest carry the harvest's own sha256.
Not held
Document and chunk text: Never stored in the lake or served; chunks are located by page range and pinned by the sha256 of their extracted text so a reader re-extracts them.
Documents about a person: CVs, personal data sheets, rosters and beneficiary lists are screened out of the pilot; their printed dates and memo lines would describe individuals.
Phone numbers: Counted per chunk and document at the build (phones_detected), never stored.
A2 oracle_wave spec values: LLM-derived; used only to order pilot candidates, never served.
2026/4
Files: 2
2026/5
Files: 5
2026/6
Files: 3
2026/7
Files: 11
2026/8
Files: 5
pdf
Bytes: 374,203
Upload folder: 2024/12
37,705
Type: pdf
Bytes: 1,175,128
Upload folder: 2024/12
37,706
Type: pdf
Bytes: 297,377
Upload folder: 2024/12
37,707
Type: pdf
Bytes: 1,567,325
Upload folder: 2024/12
37,710
Type: pdf
Bytes: 46,997
Upload folder: 2024/12
37,711
Type: pdf
Bytes: 2,200,170
Upload folder: 2024/12
37,712
Type: pdf
Bytes: 349,346
Upload folder: 2024/12
37,713
Type: pdf
Bytes: 3,024,201
Upload folder: 2024/12
37,714
Type: pdf
Bytes: 8,082,411
Upload folder: 2024/12
37,715
Type: pdf
Bytes: 13,056
Upload folder: 2024/12
37,716
Type: pdf
Bytes: 407,780
Upload folder: 2024/12
37,717
Type: pdf
Bytes: 1,729,667
Upload folder: 2024/12
37,718
Type: pdf
Bytes: 876,340
Upload folder: 2024/12
37,719
Type: pdf
Bytes: 545,431
Upload folder: 2024/12
37,720
Type: pdf
Bytes: 3,111,000
Upload folder: 2024/12
37,721
Type: pdf
Bytes: 229,860
Upload folder: 2024/12
37,722
Type: pdf
Bytes: 624,589
Upload folder: 2024/12
37,723
Type: pdf
Bytes: 320,429
Upload folder: 2024/12
37,724
Type: pdf
Bytes: 179,810
Upload folder: 2024/12
37,725
Type: pdf
Bytes: 2,143,236
Upload folder: 2024/12
37,726
Type: pdf
Bytes: 4,764,153
Upload folder: 2024/12
37,727
Type: pdf
Bytes: 257,771
Upload folder: 2024/12
37,728
Type: pdf
Bytes: 2,245,522
Upload folder: 2024/12
37,729
Type: pdf
Bytes: 3,370,830
Upload folder: 2024/12
37,730
Type: pdf
Bytes: 602,146
Upload folder: 2024/12
37,731
Type: pdf
Bytes: 1,723,610
Upload folder: 2024/12
37,732
Type: pdf
Bytes: 149,564
Upload folder: 2024/12
37,733
Type: pdf
Bytes: 2,857,185
Upload folder: 2024/12
37,734
Type: pdf
Bytes: 3,229,809
Upload folder: 2024/12
37,735
Type: pdf
Bytes: 62,208
Upload folder: 2024/12
37,736
Type: pdf
Bytes: 373,236
Upload folder: 2024/12
37,737
Type: pdf
Bytes: 865,788
Upload folder: 2024/12
37,738
Type: pdf
Bytes: 522,707
Upload folder: 2024/12
37,739
Type: pdf
Bytes: 143,065
Upload folder: 2024/12
37,740
Type: pdf
Bytes: 41,359
Upload folder: 2024/12
37,741
Type: pdf
Bytes: 62,439
Upload folder: 2024/12
37,742
Type: pdf
Bytes: 7,816,134
Upload folder: 2024/12
37,743
Type: pdf
Bytes: 1,690,503
Upload folder: 2024/12
37,744
Type: pdf
Bytes: 236,344
Upload folder: 2024/12
37,745
Type: pdf
Bytes: 574,156
Upload folder: 2024/12
37,746
Type: pdf
Bytes: 294,457
Upload folder: 2024/12
37,747
Type: pdf
Bytes: 3,117,632
Upload folder: 2024/12
37,748
Type: pdf
Bytes: 126,061
Upload folder: 2024/12
37,749
Type: pdf
Bytes: 290,032
Upload folder: 2024/12
37,750
Type: pdf
Bytes: 3,030,211
Upload folder: 2024/12
37,751
Type: pdf
Bytes: 2,183,472
Upload folder: 2024/12
37,752
Type: pdf
Bytes: 182,356
Upload folder: 2024/12
37,753
Type: pdf
Bytes: 2,100,038
Upload folder: 2024/12
37,754
Type: pdf
Bytes: 948,221
Upload folder: 2024/12
Not held: Text layer, Pages, Re-hash verified.
Per-issuing-body manifest datasets
Publisher: Built from the rows above by the inventory pipeline
Unit: one row per uploading office, with its partition file and sha256
Caveat: Each body's inventory partition is the resumable unit of the pipeline; a rerun rewrites only partitions whose rows changed.
200-chunk pilot: documents and chunks with printed fields
Publisher: each document's stored PDF (PyMuPDF text layer)
Unit: one row per pilot document (fields with page and line) and per chunk (page range, counts, text sha256)
Caveat: Pilot documents were chosen in the priority order of the A2 oracle wave specs (LLM extractions, used as pointers only; no spec value is stored or served), one document per body, text-layer PDFs fully chunked in at most 4 chunks. Documents about a person (CVs, data sheets, rosters, beneficiary lists, or two or more mobile numbers on pages 1-2) are screened out before chunking. Kind keywords match only Unicode Bangla in logical order or English; many portal PDFs extract Bangla in visual order or legacy font encodings, so an absent kind is not evidence of none.
Pipeline run log (measured)
Publisher: Measured pipeline runs on a 16-core machine
Unit: one row per stage run: items, bytes, seconds, workers
Caveat: Runs over files a previous run had just read hit the page cache and overstate the rate; the projection uses the cold runs named in its basis.