document-generation

Create a DocumentTemplate of Type MicrosoftWord with TokenMappingMethodType OmniDataTransform, referencing one Extract and one Transform ODT by exact name. Build the .docx with mustache tokens, wire the ODTs while the template is in Draft status, then reactivate and test with a generation run.

Place a plain-text {{IMG_name}} token in the template and have the Transform output a nested object with src, width, and height. The src must be a ContentDocument ID (069 prefix), all three fields are mandatory, and the Integration User needs Content Library access.

Blank output usually means the Extract is failing: check for duplicate object query items, null OutputObjectName values, or dot notation in InputFieldName paths. Validate items with docgen_odt_validate.py and execute the Extract standalone to isolate the failing stage.

No. Salesforce documents OmniDataTransform records as internal use only, so create/edit/delete via SObject REST is appropriate only for scratch-org experimentation and debugging. For committed or production work, author Metadata API XML and deploy it.

Phantom rows occur when field mappings for the same output array read from mixed hierarchy depths and a deeper join is a child-of relationship. Fix it with the redundant join pattern so all mappings read from uniform child depth, and verify with docgen_odt_inspect_hierarchy.py.

Supported functions include IF, ISBLANK, CONCAT, SUBSTRING, LIST, FUNCTION, arithmetic, comparisons, AND/OR/NOT, and ABS/ROUND/FLOOR/CEILING/MAX/MIN. Unsupported functions like CASE, LEN, UPPER, TEXT, and MOD save silently with a null FormulaConverted and produce no output.