Convert a PDF Parts Book to a Structured Catalog

A typical OEM parts book runs to hundreds of pages: dozens of assemblies, each with an exploded-view diagram and a parts-list table. The information is valuable — it is the definitive record of what goes where — but as a PDF it is frozen at the time of publication. This playbook turns that frozen document into live catalog data.

Prerequisites

• The PDF parts book (preferably with selectable text, not scanned images)
• A reference to the equipment model, serial-number range and publication date the book covers
• Access to the current catalog or product-identity system for part-number resolution
• A target data model that separates the four layers (see the parts-book data model guide)

The workflow

1. Inventory the book

   Before extracting anything, survey the document:

   • How many assemblies does it cover?
   • How many pages total? How many contain exploded views vs. parts lists vs. front matter?
   • What is the publication date?
   • What equipment model and serial range does it cover?
   • Are the parts-list tables consistent in format, or does the layout vary by section?

   The inventory tells you the scope of the conversion and surfaces format inconsistencies early.

2. Extract the visual layer

   Isolate each exploded-view diagram as a separate image file. For each:

   • Assign an assembly identifier (from the book’s table of contents or section numbering)
   • Record the page number for traceability
   • Note the image quality — if the diagram is too low-resolution for the callout numbers to be legible, flag it

   The diagram images become referenced assets in the structured catalog. They do not need to be redrawn — they just need to be separable from the table data.

3. Extract the parts-list tables

   Parse each parts-list table into structured rows. For each row, extract:

   • Position / callout number
   • Part number (as printed in the book)
   • Description
   • Quantity per assembly
   • Any notes (material, color, variant)

   Common extraction challenges: merged cells, multi-line descriptions, footnotes that apply to specific rows, tables that span multiple pages.

4. Parse applicability headers

   Each assembly section typically has a header identifying which equipment it applies to. Extract:

   • Equipment make and model
   • Serial-number range (if specified)
   • Variant or option package (if specified)
   • Effective date or revision number

   Structure this as a typed applicability record, not a free-text note.

5. Resolve part numbers against the current catalog

   For each part number extracted from the book:

   • Normalize the number (strip spaces, fix formatting)
   • Look it up in the current catalog or product-identity system
   • If found and active: link the position to this product identity
   • If found but discontinued: follow the supersession chain to the current successor
   • If not found: flag for manual review — the number may be a typo, a format variant, or a part not yet in the catalog

   Record the resolution outcome for every row: matched (active), matched (superseded → successor), or unresolved.

6. Link positions to product identities

   Connect each position in the extracted data to the resolved product identity (not the static part number from the PDF). This is the step that makes the parts book live: when a part is superseded in the future, the position’s reference resolves to the new part automatically.

7. Run quality checks

   Before publishing the structured parts book:

   • Are all positions accounted for? (Compare the count of extracted positions to the callout numbers visible in the diagram)
   • Are there duplicate position numbers within the same assembly?
   • Are there part numbers that appeared in the table but were not resolved?
   • Do the quantities look reasonable? (A quantity of 999 is likely a parsing error)
   • Does the applicability data match the book’s cover page?
8. Load and verify

   Load the structured data into the catalog system. Verify:

   • Each assembly displays the correct diagram
   • Each position maps to the correct (current) part number
   • Superseded parts show the successor, not the original number
   • Applicability filters work: selecting a serial number returns the correct assembly version

What the output looks like

For one assembly from the conversion:

Common pitfalls

• Treating the PDF part number as the final answer — it may be years out of date.
• Skipping the applicability extraction — without it, the structured parts book cannot filter by serial number.
• Not verifying position counts — a missing position means a part the technician cannot find.
• Storing the extracted data as a new static table instead of linking to live product identities — this recreates the original problem.
• Ignoring multi-page tables — a table that spans two pages often loses a row at the page break.

Where Claro fits

Claro provides the identity-resolution layer for the conversion. When a PDF parts book is processed, Claro normalizes the extracted part numbers, resolves them against the product identity graph, follows supersession chains, and flags unresolved numbers for the catalog team. The result is a structured parts book where every position points to a live, sourced product identity.

Related

FAQ