Revised 29 October 2026 · v1.0.0 · what changed

Code: github.com/herrrickshaw/fuel-retail-outlets

Watch the walkthrough

Watch on YouTube: youtu.be/_4NBKi2NCGI

Why it is useful

• Who benefits: fuel-retail, CNG and EV-charging planners, consultants and students who need a pump-level list of India with location, company and fuel type.
• What it replaces: collecting listings from each oil company's locator one by one.
• Evidence from the run: counted naively, Delhi appears to have 21,939 pumps; after removing duplicate rows it has 419. The raw file also holds 22,343 duplicate rows (21.3%). If you skip the cleaning step, every state count is wrong.
• After cleaning, the per-capita view shows a clear spread: Bihar 2.7, West Bengal 2.8 and Jharkhand 3.4 pumps per 100,000 people, against Punjab 11.5 and Haryana 11.3.

What it is for

This repo collects Indian petrol, diesel and CNG outlet listings from public sources (the SSRI petrol-pump API, the BPCL dealership locator and CashAtPOS), with two small web apps for the map and a coverage gap-analysis heatmap.

The data is indicative, for planning and visualisation. It is not an official registry. Outlet records are public business listings.

Who it is for

• Energy and mobility analysts sizing pump or charger networks
• Students working on India fuel-retail geography
• Anyone who needs a raw outlet list to join with district or population data

What you need

Get the code and an AI assistant

You do not have to be a programmer. Get the repository, give it to an AI assistant, and ask in plain language. Any of these routes works:

git clone https://github.com/herrrickshaw/fuel-retail-outlets.git
cd fuel-retail-outlets
claude      # or: gemini

Install

1. Clone the repository (needs git)

   git clone https://github.com/herrrickshaw/fuel-retail-outlets.git
   cd fuel-retail-outlets

2. Nothing to install for the sample below. For the web apps, serve a folder with Python

   cd fuel-station-gap-analysis && python3 -m http.server 8001

Quick start: commands and what you should see

Load the committed crawl and deduplicate it the way the repo's notes describe

python3 -I - <<'E'
import gzip,json,collections
p='api-data-integration/outlet_data_ssri_complete/ssri_pumps_raw_20261016.jsonl.gz'
rows=[json.loads(l) for l in gzip.open(p,'rt') if l.strip()]
k={}
for r in rows:
    key=(r['name'],r['address'],r['company'],r['latitude'],r['longitude'])
    if key not in k or r['id']>k[key]['id']: k[key]=r
u=list(k.values())
print(len(rows),len(u),len(rows)-len(u))
print(collections.Counter(r['company'] for r in u).most_common(5))
E

104945 82602 22343, then the company counts: BPCL 25,795, HPCL 23,543, IOCL 19,446, Nayara 9,007, Unknown 2,504.

Open the gap-analysis heatmap

cd fuel-station-gap-analysis && python3 -m http.server 8001

Visit http://localhost:8001 to see the state-level heatmap. The page loads its map tiles from the internet. I did not open it in this run.

Sample outputs

Real output from a run on the date shown in the sources line; trimmed to fit.

Why you must deduplicate: Delhi, raw versus cleaned

raw rows with state == 'Delhi'  : 21,939
  of which district 'New Delhi' : 21,572, all IOCL-heavy
  sitting on only 57 distinct coordinates (3,986 identical rows on each of several points)
after dedup (name+address+company+lat+lon): 419

The repo's own notes (SSRI_REEXTRACTION_20260712.md) document this; the 16 Oct 2026 file is still raw, so you have to apply the step yourself. My dedup gives 82,602; the note's earlier snapshot gave 82,609.

Pumps per 100,000 people by state (cleaned; state populations from the repo's gap-analysis data.js)

state             pumps   per 100k people   per 1000 km2
Uttar Pradesh      8,915        4.5              37.0
Maharashtra        7,317        6.5              23.8
Tamil Nadu         5,812        8.1              44.7
Madhya Pradesh     5,320        7.3              17.3
Karnataka          5,268        8.6              27.5
-- lowest (states with 100+ pumps) --
Delhi                419        2.5             282.5
Bihar              2,779        2.7              29.5
West Bengal        2,522        2.8              28.4
Jharkhand          1,109        3.4              13.9
Assam              1,108        3.6              14.1
-- highest --
Arunachal Pradesh    201       14.6              2.4
Punjab             3,204       11.5              63.6
Haryana            2,855       11.3              64.6
Telangana          3,609       10.3              32.2

Delhi's low per-capita number reflects a dense city counted against its whole population and is not a real gap; read the table as a starting screen, not a verdict.

Data-quality check on state names

state values in the raw file: 53 spellings
unmatched to the 35 states in data.js after dedup: 10,532 of 82,602 outlets
  'unknown' (all Nayara): 9,007
  'Jharkand': 593, 'Jammu And Kashmir': 333, 'Jammu & Kashmir': 230 ...
matched: 72,070

Only the matched outlets feed the table above, so the table understates states with spelling variants (I mapped Jharkand, Uttrakhand and Jammu variants) and cannot place Nayara's 9,007 outlets by state at all.

Ways to use it

Sample prompts for your AI assistant

Copy, paste and edit the parts in [brackets].

• Claude: Read REBUILD_NOTE.md and SSRI_REEXTRACTION_20260712.md, then write a script that deduplicates the crawl and prints pumps per state.
• Gemini: Using the crawl, find the ten districts with the most CNG pumps and explain how you handled spelling variants of state names.
• ChatGPT: Here is a table of pumps per 100,000 people by state. Which states look under-served and which numbers are probably artefacts?
• GitHub (Copilot Chat or Codespaces): Write a function that normalises the 53 state spellings in the crawl to one canonical name per state or UT.
• Claude: Compare the deduplicated total with the PPAC baseline described in PPAC_RETAIL_OUTLETS_MAY2026.md and explain the gap.
• Any assistant: Suggest how to join this outlet list with district population to find gaps for EV charging.

Limits and honest caveats

Sources: Read the committed crawl api-data-integration/outlet_data_ssri_complete/ssri_pumps_raw_20261016.jsonl.gz and fuel-station-gap-analysis/data.js from github.com/herrrickshaw/fuel-retail-outlets (commit 266995b) with the Python standard library on 19 October 2026. Outlet names and addresses were not reproduced; only aggregates are shown. Extractors, web apps and network crawl not run.

About this article: Researched, written and edited by Umashankar Triplicane Dwarakanathan, with AI research assistance; every figure is meant to trace to the primary source cited. See the Editorial Policy for how sourcing, AI use and corrections work.