Office & Documents
Office files share a structure: what you see is "how it looks when opened", but what decides behaviour is what is stored inside. A CSV declares no encoding, so the software opening it guesses. An Excel cell has no type, so the content is guessed. PDF page order lives in a page tree, so reordering never touches content.
Once you know that layer, most "inexplicable" behaviour has an explanation: mojibake is a wrong guess about encoding, deformed numbers are a wrong guess about type, and a split PDF grows because shared fonts are duplicated per file.
- CSV Opens as Garbage: BOM, Delimiters, and EncodingWhy a CSV that looks fine in the terminal turns into mojibake in Excel, whether a UTF-8 BOM helps or hurts, how delimiters and quoting break naive parsers, and a five-step diagnostic order.
- Image Compression: How Small Before It Looks BadFile size is driven by pixel count, not by the quality slider. Resize first, then pick a format, then tune quality — with the numbers that actually hold up for web, print and screenshots.
- JSON to Table: Flattening Nested FieldsTurning nested JSON into a table is not a copy-paste problem. It is a decision about what one row means. Here is how to flatten objects, what to do with arrays of arrays, and which output format fits which audience.