Join split files
Concatenate parts back into one file, in filename order. Pairs with Split large file. For a recognised .partNNN set, a gap at the start or in the middle is detected and refused (e.g. part001 or part003 missing). A gap at the end (the last part missing entirely) cannot be detected from the parts alone, so verify the count yourself.
Runs on your device. The file is never uploaded.
Join split files concatenates the parts you drop back into one file, sorted numerically so part10 follows part9. When every name matches base.partNNN, the output takes that base and the walk from part 1 upward refuses any hole it finds. A missing final part leaves no trace in the remaining ones, so the count is yours to check.
Questions
What order are the parts joined in?
Filename order, sorted numerically rather than as plain text, so part10 lands after part9 instead of before it. That matters as soon as the part count passes the zero padded width. Anything that is not a recognised part set is still joined in that same sorted order.
Will it notice if a part is missing?
For a recognised .partNNN set, yes, if the gap is at the start or in the middle. The walk is anchored at part 1 and runs up to the highest index present, so a missing first part counts as a gap as much as a hole in the middle, and the run is refused naming the exact file it wants. A missing last part cannot be detected from the parts alone, so check the count yourself.
What counts as a recognised part set?
Every input matching a name of the form base.partNNN with the same base. That is what Split large file produces. When the set is recognised, the base name is used for the output and the gap check runs. Mixed or unrelated names skip both, and are simply concatenated in sorted order.
What is the joined file called?
The shared base name when the parts are a recognised set, so backup.zip.part001 and friends produce backup.zip. Set outputName to override that. If the names are not a part set and you give no name, the result is called joined.bin.
Why did it complain about two parts with the same index?
Because two files claim the same position, for example after mixing parts from two different splits in one folder. The join is refused as ambiguous rather than picking one, since the wrong choice would produce a file that is quietly wrong with no size based tell.
Is anything uploaded?
No. Every part is read into memory and the joined file is built there, all on your device. That does mean the whole reassembled file plus all its parts are in memory at once, so joining is bounded by your browser tab rather than by disk space.