02 / CASE STUDY SDS Chemical Extractor

SDS Chemical
Extractor

Less copying. More useful data. A practical Python tool that turns batches of safety data sheets into a structured Excel workbook.

Open project Explore source

CONTEXT

Client project · Textile industry

MY ROLE

Python tool development

BUILT WITH

Pythonpdfplumberpandasopenpyxl

SDS Chemical Extractor upload interface — Upload SDS PDFs in the browser and download the extracted workbook.

Document workflow: PDF → XLSX
Consistent output columns: 3
Two ways to run it: Web + CLI

01 / THE PROBLEM

The same information, in different layouts.

A corporate textile client needed product names, chemical names, and CAS numbers from SDS PDFs in spreadsheet form. The challenge is that the information can sit in tables, wrap across lines, or appear under different labels.

02 / THE APPROACH

Follow the document’s structure.

The parser reads product identifiers from Section 1 and locates the composition information in Section 3. It tries table extraction first, then falls back to text when tables yield no items. The extracted records become a three-column workbook.

From a batch of documents to one workbook

01SDS PDFsUpload one or more files
02Locate sectionsProduct + ingredients
03Extract recordsTables, then text fallback
04Export workbookProduct · Chemical · CAS

The browser and command-line workflows reuse the same extraction function. The output preserves a product’s identity alongside its chemical records.

03 / KEY DECISIONS

The details that shape the build.

Use structure before pattern matching

Section boundaries narrow the composition search. Product-name patterns handle different labels, while nearby text helps recover chemical names that wrap around CAS numbers.

Make incomplete results visible

Files with no extracted ingredients produce an N/A row. A failed file produces an error row while the batch continues, giving the reviewer a place to investigate.

Deliver something usable in Excel

pandas and openpyxl create the workbook with headers, adjusted column widths, and row heights. The result is ready for review in the client’s existing spreadsheet workflow.

04 / THE RESULT

A repeatable route from PDF to spreadsheet.

The tool accepts multiple PDFs and exports product, chemical, and CAS fields together. It puts extraction and formatting into one workflow, with the original documents available for checking the results.

A constraint worth understanding

Extraction depends on readable PDF text and recognizable layouts. The current pipeline has no OCR stage; scanned documents and unusual formatting need separate handling or manual review.