Another Replace Question

[kamanza]

The answer to Poppeye's question was very helpful for me too, thanks.
I would also be very grateful for help with another problem.
Converting bad PDF files to MOBI I often need to replace garbage text with a space but I have no idea what to put in the replacement text window.
I've tried \s; (\s); [\s], but neither works.
What am I doing wrong?

[DoctorOhh]

Originally Posted by kamanza

Converting bad PDF files to MOBI I often need to replace garbage text with a space but I have no idea what to put in the replacement text window.

I have no experience, but have you tried entering an actual space in the replace window.

[kamanza]

I have. does not work.

[ldolse]

I would think an actual space should have worked, you could try to wrap it in quotes to see if it helps.

An alternate way to specify it would be to use unicode notation, e.g.

\u0020

[Starson17]

Originally Posted by ldolse

I would think an actual space should have worked

So would I. I often suspect (for pdf fixup problems) that the problem is not the failure to insert, it's that the "garbage text" in the pdf isn't getting matched so no insert will ever happen. To make sure, just replace it with some text, which is guaranteed to work, and see if your regex matching is actually working. What works in the wizard, does not always work when it runs, particularly if it's a multiline replace.

[kamanza]

Thanks for all the answers.
The garbage text does get matched- it disappears. But if i try to use an actual space, no space is inserted. In other cases whatever was used for replacement was inserted as the actual text: \s; [\s]; (\s).

[Starson17]

Originally Posted by kamanza

whatever was used for replacement was inserted as the actual text: \s; [\s]; (\s).

Yes, \s is a whitespace regex matching element, not a space. It matches tabs, spaces and other types of whitespace, but doesn't actually define any particular character, so you don't want to use that in the replace field. The replace function wouldn't know which whitespace character to insert even if it was allowed. I've tested a search/replace operation during convert on my system and had no trouble replacing the searched regex with a single space.

[kamanza]

Do you mean, you just press the space bar?
Don't know how it works for you, but i've just tried again & no space is inserted.

[Starson17]

Originally Posted by kamanza

Do you mean, you just press the space bar?

OOPs, sorry, I was too quick. I replaced the word "the" with a space and thought I was seeing multiple spaces, but either I wasn't, or I made some other error. Moreover, calibre won't hold the space as my replace text string if I leave that page or come back after conversion.

Even more interestingly, if I use space to replace "the", it doesn't leave double spaces (the original two that surrounded the original "the"). It only leaves one.

[ldolse]

HTML rendering merges multiple spaces into a single space unless you make them an non-breaking space. If you want the space to be retained when there are multiple consecutive spaces then your replace code should use:

 

Note several Calibre features delete most non-breaking spaces (like Heuristics), so it's still not a guarantee that it will still be there at the end, it depends on the conversion settings.