Search regex problem

[ColMac]

My wife has a number of books that have masses of "Extra" carriage returns inserted in them making reading difficult.

A sample is shown below

<p class="calibre1">And  Alandra  looking  at  him  as  he  came  into  the  room,  found  that </p>
<p class="calibre1">although  she  had  not  wanted  to  allow  her  grandfather  one  courtesy, </p>
<p class="calibre1">she was getting to her feet. </p>
<p class="calibre1">Silently,  she  watched  and  waited  as  he  came  closer.  And  when  he </p>
<p class="calibre1">stopped and for long seconds stared at her, she saw deep frown lines </p>
<p class="calibre1">groove  on  his  forehead.  But  she  had  no  word  to  say  to  him,  and  he </p>
<p class="calibre1">none for her as he turned to the man who, keeping his eyes steady on </p>
<p class="calibre1">the two of them, had now moved from his position by the door, and </p>
<p class="calibre1">was coming in their direction. </p>
<p class="calibre1">And  it  was  left  to  Matt  Carstairs  to  introduce  the  two—the  elderly </p>
<p class="calibre1">man  who  still  had  the  gait  of  a  man  years  younger,  and  the  young </p>
<p class="calibre1">woman whose solemn face was giving nothing away of the very low </p>
<p class="calibre1">regard in which she held the other. </p>
<p class="calibre1">'This,'  said  Matt  Carstairs,  pausing  only  marginally  as  if  to  assess </p>
<p class="calibre1">how  the  older  man  would  take  it,  'this  woman  claims  to  be  your </p>
<p class="calibre1">granddaughter, sir—she says she is Edward's child.' </p>

I managed to find a saved search from Zajora that partly solves the problem

(?<![".!?>*”“…~’])</(?P<tag>\w+)>\s*<(?P=tag) [^/>]+>

Replaced with "Null"

However, it also removes the genuine end of paragraph returns (In the example above, the ones with full stops and single quote).

Is there any way to amend this search to exclude the more obvious "Genuine" carriage returns. I know that there are other ways to end a sentence other than full stops, and that any such search will not catch everything, and will get some wrong. But it would be better than what she currently has.

Thanks

[cybmole]

Search instead for lines that begin with a lower case character and strip the opening tag.

[cybmole]

And also the previous closing tag. I can't face typing the code via my tablet but there will be examples in older threads

[cybmole]

ok at my PC now here is how I'd do it with Sigil
1. highlight a relevant fragement e.g. this from your example
</p>
<p class="calibre1">h
2. paste that into the find part of the regex. (pasting it as-is will take care of the space between lines issue. )
3. now replace that closing h with ([a-z]) so that it matches any lower case
the replace string for the regex is (blank space)\1

what that all does is removes the closing tag + the next opening tab & 1st letter, and then puts back the initial opening letter, preceded by a blank space.
test it carefully and make a backup before you "replace all" !

outside of poetry and titles, there's no valid reason for a paragraph to begin with lower case, so that will fix most issues.

you can still get awkward cases like
then he said,
"this"

but go can extrapolate code as needed for those

[theducks]

</p>\s+<p
will take care of any Indention or CR (that may vary in the document due to other tags like Blockquotes and Div)

[ColMac]

Originally Posted by cybmole

ok at my PC now here is how I'd do it with Sigil
1. highlight a relevant fragement e.g. this from your example
</p>
<p class="calibre1">h
2. paste that into the find part of the regex. (pasting it as-is will take care of the space between lines issue. )
3. now replace that closing h with ([a-z]) so that it matches any lower case
the replace string for the regex is (blank space)\1

That works perfectly.

First one I tried found almost 4,000 occurrences. I'm pretty sure that there may be an error in that 4,000, but it is a massive improvement on what I had before.

Thanks for the help.

[ColMac]

Originally Posted by theducks

</p>\s+<p
will take care of any Indention or CR

Not sure how this one is meant. Is this an addition to the code from cybmole, or instead of.

I tried using it as is, but wasn't sure it was giving me valid results

Colin

[cybmole]

Originally Posted by ColMac

Not sure how this one is meant. Is this an addition to the code from cybmole, or instead of.

I tried using it as is, but wasn't sure it was giving me valid results

Colin

that last one is the "proper" way to detect & remove the blank stuff between end of one line and start of next. my lazy way is to copy paste the whole shebang into the find field & it usually works!



I used to have a file with several of these all tested and vetted but I can't find it. and I could not face re-reading all of the regex examples in the sigil forum stickies

I'm sure there are detailed old threads but writing the appropirate search expression for the forum search engine has me beat

[theducks]

\s is whitespace of any kind (space, return, tab) the + is one or more (in a row of the condition)

[phossler]

Sorry

OBE

[gbm]

This is one I think is from the Sigil forums, use to find and join split paras:

Search

</p>\s*<p[^>]+>([a-z])

replace

\1

Have it as a saved search in ebook editor.
Using the OP example this is the result.

Spoiler:

<p class="calibre1">And Alandra looking at him as he came into the room, found that although she had not wanted to allow her grandfather one courtesy, she was getting to her feet. </p>
<p class="calibre1">Silently, she watched and waited as he came closer. And when he stopped and for long seconds stared at her, she saw deep frown lines groove on his forehead. But she had no word to say to him, and he none for her as he turned to the man who, keeping his eyes steady on the two of them, had now moved from his position by the door, and was coming in their direction. </p>
<p class="calibre1">And it was left to Matt Carstairs to introduce the two—the elderly man who still had the gait of a man years younger, and the young woman whose solemn face was giving nothing away of the very low regard in which she held the other. </p>
<p class="calibre1">'This,' said Matt Carstairs, pausing only marginally as if to assess how the older man would take it, 'this woman claims to be your granddaughter, sir—she says she is Edward's child.' </p>

bernie

Originally Posted by cybmole

that last one is the "proper" way to detect & remove the blank stuff between end of one line and start of next. my lazy way is to copy paste the whole shebang into the find field & it usually works!



I used to have a file with several of these all tested and vetted but I can't find it. and I could not face re-reading all of the regex examples in the sigil forum stickies

I'm sure there are detailed old threads but writing the appropirate search expression for the forum search engine has me beat

[eschwartz]

And if there is no class?

[gbm]

Originally Posted by eschwartz

And if there is no class?

</p>\s*<p>+([a-z])

bernie

[eschwartz]

Au contraire.

</p>\s*<p(?: [^>]+)?>([a-z])

I was hoping you'd notice the previous regex didn't cover all cases... and fix that, rather than create another problem.
The whole idea behind regex is to, you know, create one pattern to rule them all.

[PeterT]

Personally I'd head back to whoever supplied me with those books and ask for clean versions...