carlo said:

I have just become aware that PG now autogenerates HTML for texts that don't have it.

since a number of people are fond of rewriting history here, let's note for the record that i suggested this some time ago. indeed, my recommendation was that the .txt version should be used to autogenerate the .html version for _all_ the books, that hand-crafted .html be abandoned because it is too hard to maintain and to upgrade. i also suggested that conformance to this strategy would enable p.g. to improve the .txt versions... and i predicted that sooner or later, you'd all come around to this workflow. and how you have. so i will say "i told you so."

Unfortunately however sometimes the autogenerated file is garbage (e.g. poetry rewrapped, see 31079).

without even looking at those files, i can guess what's wrong... many of the books that are exclusively poetry are set flush to the left margin, lacking any of the leading spaces that serve as a signal to the conversion program not to wrap the lines... so of course the converter is gonna wrap the lines. this is an error, a major error, in the processing of these books. (and it's so easy to change every linebreak to a linebreak+space.)

Would it be possible to have the autogeneration program to find what is the problem, or at least to preview the autogenerated file and possibly fix either the program or the files?

i've never tried to verify it with a closer analysis, but my impression is that some of the whitewashers use a slightly different converter... and then of course there are a number of different ones over at d.p., including the one in thundercat's app, and another by david garcia... without dedication to making the .txt program correct at the outset, however, it doesn't matter how good the converter might be... -bowerbird