A toolkit for extracting, exploring, and visualizing New Jersey Municipal Court Statistics. It parses the NJ Courts' annual PDF reports directly from content streams (no OCR), stores the results as JSON/CSV, and serves an interactive web dashboard for comparing trends across jurisdictions, metrics, and case types.
Live at municourt.hackjc.org.
go build -o municourt .
# Download all available PDFs from NJ Courts
municourt download -dir data/
# Parse every PDF into JSON + CSV
municourt parse data/
# Start the web dashboard
municourt web -dir data/
# → http://localhost:8080Scrapes njcourts.gov/public/statistics for municipal court PDF links and downloads them.
municourt download [-dir outputDir]
Files are saved as municipal-courts-YYYY-MM.pdf.
Parses one or more PDFs into structured JSON and CSV.
municourt parse <input.pdf|directory> [--json out.json] [--csv out.csv]
When given a directory, every .pdf inside it is parsed. Output files are written alongside the input with the same base name unless overridden.
Includes interactive deduplication: when municipality names change between years (e.g. "TOWNSHIP" vs "TOWN" suffixes), the tool detects candidates that never co-occur in the same time period and prompts you to merge them.
Starts an HTTP server that serves the interactive dashboard and a JSON API.
municourt web [-dir data/] [-port 8080]
All parsed JSON files in the data directory are loaded into memory at startup. There is no database — the server reads *.json files and serves everything from RAM.
Renders charts to the terminal (sparklines) or to a PDF file.
municourt viz [-dir data/] [-metric filings] [-type grand-total] [-pdf output.pdf]
The dashboard is a single-page app embedded in the Go binary. It provides:
- Multi-entity comparison — add any combination of state, county, and municipality-level series to the same chart.
- Multi-metric/type selection — each entity captures the metric and case type set at add-time, so you can plot Filings vs Clearance Rate or DWI vs Parking on the same chart.
- Smart labeling — chip and legend labels only show metric/type suffixes when those dimensions actually vary across the current selection, keeping things clean.
- Dynamic chart title — shows the shared metric/type when uniform, omits dimensions that are mixed.
- Data table modal — view the raw numbers in a sortable table and download as CSV.
- Shareable URLs — the full state (entities, metrics, types) is encoded in the URL for bookmarking and sharing. Old-format URLs with global metric/type params are handled for backwards compatibility.
- Image export — save the chart as a PNG.
Returns the lists of counties, municipalities, metrics, and case types used to populate the UI dropdowns.
{
"counties": ["ATLANTIC", "BERGEN", ...],
"municipalities": {
"ATLANTIC": ["ABSECON", "ATLANTIC CITY", ...],
...
},
"metrics": [
{"value": "filings", "label": "Filings"},
{"value": "resolutions", "label": "Resolutions"},
{"value": "clearance", "label": "Clearance"},
{"value": "clearance-pct", "label": "Clearance %"},
{"value": "backlog", "label": "Backlog"},
{"value": "backlog-per-100", "label": "Backlog/100 Monthly Filings"},
{"value": "backlog-pct", "label": "Backlog %"},
{"value": "active-pending", "label": "Active Pending"}
],
"types": [
{"value": "grand-total", "label": "Grand Total"},
{"value": "indictables", "label": "Indictables"},
{"value": "dp-pdp", "label": "D.P. & P.D.P."},
{"value": "other-criminal", "label": "Other Criminal"},
{"value": "criminal-total", "label": "Criminal Total"},
{"value": "dwi", "label": "D.W.I."},
{"value": "traffic-moving", "label": "Traffic (moving)"},
{"value": "parking", "label": "Parking"},
{"value": "traffic-total", "label": "Traffic Total"}
]
}Returns time-series data for a given metric, case type, and aggregation level.
| Parameter | Values | Default |
|---|---|---|
level |
state, county, municipality |
county |
metric |
Any metric value from metadata | filings |
type |
Any type value from metadata | grand-total |
county |
County name (uppercase) | — |
municipality |
Municipality name (uppercase) | — |
{
"title": "Filings — Grand Total",
"dates": ["2005-06", "2006-06", "2007-06", ...],
"series": [
{
"name": "ATLANTIC",
"values": [1234.0, 1456.0, null, 1678.0, ...]
}
]
}Null values indicate missing data for that time period.
Each municipality produces a record with:
- Header: county, municipality, date range
- 8 sections, each with sub-rows of 9 column values:
| Section | Sub-rows |
|---|---|
| Filings | Prior, Current, % Change |
| Resolutions | Prior, Current, % Change |
| Clearance | Prior, Current |
| Clearance Percent | Prior, Current |
| Backlog | Prior, Current, % Change |
| Backlog/100 Mthly Filings | Prior, Current, % Change |
| Backlog Percent | Prior, Current |
| Active Pending | Prior, Current, % Change |
Columns: Indictables, D.P. & P.D.P., Other Criminal, Criminal Total, D.W.I., Traffic (moving), Parking, Traffic Total, Grand Total.
Values are stored as strings since they may contain commas, %, - -, or negative signs.
- pdf.go — Opens the PDF with pdfcpu, iterates pages, decompresses content streams, and skips non-data pages (cover pages).
- content.go — Tokenizes PDF content streams and extracts text from
TjandTJoperators. WithinTJarrays, kerning values determine whether adjacent strings are concatenated (small spacing) or treated as separate columns (large spacing). Handles hex-encoded strings and ToUnicode CMap decoding. - parser.go — Reads the ordered text items and maps them to
MunicipalityStatsstructs using the known section layout. - main.go — CLI entry point that dispatches to
download,parse,web, orvizsubcommands.
├── main.go CLI entry point and subcommand dispatch
├── cmd/
│ ├── web.go HTTP server, API handlers, data loading
│ ├── web.html Embedded single-page dashboard (HTML/CSS/JS)
│ ├── viz.go Terminal sparkline + shared viz helpers
│ ├── vizpdf.go PDF chart rendering (gonum/plot)
│ ├── parse.go Parse subcommand
│ ├── download.go Download subcommand
│ └── dedupe.go Municipality name deduplication
├── parser/
│ ├── model.go Data structures (MunicipalityStats, RowData, etc.)
│ ├── pdf.go PDF reading and content stream extraction
│ ├── content.go PDF tokenization and text item extraction
│ ├── parser.go Text-to-struct mapping
│ └── cmap.go ToUnicode CMap parsing
├── data/ Parsed JSON/CSV files (not in repo)
├── Dockerfile Multi-stage build for deployment
└── config/deploy.yml Kamal deployment config
The app is deployed via Kamal with a multi-stage Docker build:
docker build -t municourt .
docker run -p 8080:8080 municourtThe Dockerfile copies the data/ directory into the image so the server has all parsed JSON files available at startup.
go build -o municourt .
Requires Go 1.24+.
go test ./...
Test fixtures page.pdf (ATLANTIC/ABSECON) and cover.pdf (cover page) are included in the parser/testdata/ directory.
- pdfcpu — PDF content stream extraction
- gonum/plot — PDF chart rendering
Built by Hack Jersey City.