re.search slashes

[pyluke]

I'm parsing LaTeX document and want to find lines with equations blocked
by "\[" and "\]", but not other instances of "\[" like "a & b & c \\[5pt]"

so, in short, I was to match "\[" but not "\\]"

to add to this, I also don't want lines that start with comments.


I've tried:
check_eq = re.compile('(?!\%\s*)\\\\\[')
check_eq.search(line)

this works in finding the "\[" but also the "\\["

so I would think this would work
check_eq = re.compile('(?![\%\s*\\\\])\\\\\[')
check_eq.search(line)

but it doesn't. Any tips?

 

[Scott David Daniels]

I'm parsing LaTeX document and want to find lines with equations blocked
by "\[" and "\]", but not other instances of "\[" like "a & b & c \\[5pt]"
so, in short, I was to match "\[" but not "\\]" .... I've tried:
check_eq = re.compile('(?!\%\s*)\\\\\[')
> check_eq.search(line)
> this works in finding the "\[" but also the "\\["

If you are parsing with regular expressions, you are running a marathon.
If you are doing regular expressions without raw strings, you are running
a marathon barefoot.

Notice: len('(?!\%\s*)\\\\\[') == 13
len(r'(?!\%\s*)\\\\\[') == 15

so I would think this would work
check_eq = re.compile('(?![\%\s*\\\\])\\\\\[')
check_eq.search(line)

but it doesn't. Any tips?

Give us examples that should work and that should not (test cases),
and the proper results of those tests. Don't make people trying to
help you guess about anything you know.

--Scott David Daniels
(e-mail address removed)

 

[Xavier Morel]

I'm parsing LaTeX document and want to find lines with equations blocked
by "\[" and "\]", but not other instances of "\[" like "a & b & c \\[5pt]"
so, in short, I was to match "\[" but not "\\]" .... I've tried:
check_eq = re.compile('(?!\%\s*)\\\\\[')
check_eq.search(line)
this works in finding the "\[" but also the "\\["

If you are parsing with regular expressions, you are running a marathon.
If you are doing regular expressions without raw strings, you are running
a marathon barefoot.

Notice: len('(?!\%\s*)\\\\\[') == 13
len(r'(?!\%\s*)\\\\\[') == 15

so I would think this would work
check_eq = re.compile('(?![\%\s*\\\\])\\\\\[')
check_eq.search(line)

but it doesn't. Any tips?

Give us examples that should work and that should not (test cases),
and the proper results of those tests. Don't make people trying to
help you guess about anything you know.

--Scott David Daniels
(e-mail address removed)

To add to what scott said, two advices:
1. Use Kodos, it's a RE debugger and an extremely fine tool to generate
your regular expressions.
2. Read the module's documentation. Several time. In your case read the
"negative lookbehind assertion" part "(?<! ... )" several time, until
you understand how it may be of use to you.

 

[pyluke]

I'm parsing LaTeX document and want to find lines with equations
blocked by "\[" and "\]", but not other instances of "\[" like "a & b
& c \\[5pt]"
so, in short, I was to match "\[" but not "\\]" .... I've tried:
check_eq = re.compile('(?!\%\s*)\\\\\[')
check_eq.search(line)
this works in finding the "\[" but also the "\\["

If you are parsing with regular expressions, you are running a marathon.
If you are doing regular expressions without raw strings, you are running
a marathon barefoot.

Notice: len('(?!\%\s*)\\\\\[') == 13
len(r'(?!\%\s*)\\\\\[') == 15

so I would think this would work
check_eq = re.compile('(?![\%\s*\\\\])\\\\\[')
check_eq.search(line)

but it doesn't. Any tips?

Give us examples that should work and that should not (test cases),
and the proper results of those tests. Don't make people trying to
help you guess about anything you know.

--Scott David Daniels
(e-mail address removed)

Alright, I'll try to clarify. I'm taking a tex file and modifying some
of the content. I want to be able to identify a block like the following:

\[
\nabla \cdot u = 0
\]


I don't want to find the following

\begin{tabular}{c c}
a & b \\[4pt]
1 & 2 \\[3pt]
\end{tabular}


When I search a line for the first block by looking for "\[", I find it.
The problem is, that this also find the second block due to the "\\[".

I'm not sure what you mean by running a marathon. I do follow your
statement on raw strings, but that doesn't seem to be the problem. The
difference in your length example above is just from the two escaped
slashes... not sure what my point is...

Thanks
Lou

 

[pyluke]

To add to what scott said, two advices:
1. Use Kodos, it's a RE debugger and an extremely fine tool to generate
your regular expressions.

Ok, just found this. Will be helpful.

2. Read the module's documentation. Several time. In your case read the
"negative lookbehind assertion" part "(?<! ... )" several time, until
you understand how it may be of use to you.

Quite a teacher. I'll read it several times...

Thanks anyway.

 

[pyluke]

2. Read the module's documentation. Several time. In your case read the
"negative lookbehind assertion" part "(?<! ... )" several time, until
you understand how it may be of use to you.

OK. lookbehind would be more useful/suitable here...

 

[pyluke]

I'm parsing LaTeX document and want to find lines with equations blocked
by "\[" and "\]", but not other instances of "\[" like "a & b & c \\[5pt]"

so, in short, I was to match "\[" but not "\\]"

to add to this, I also don't want lines that start with comments.


I've tried:
check_eq = re.compile('(?!\%\s*)\\\\\[')
check_eq.search(line)

this works in finding the "\[" but also the "\\["

so I would think this would work
check_eq = re.compile('(?![\%\s*\\\\])\\\\\[')
check_eq.search(line)

but it doesn't. Any tips?

Alright, this seems to work:

re.compile('(?<![(\%\s*)(\\\\)])\\\\\[')

 

[Scott David Daniels]

I... want to find lines with ... "\[" but not instances of "\\["

If you are parsing with regular expressions, you are running a marathon.
If you are doing regular expressions without raw strings, you are running
a marathon barefoot.

I'm not sure what you mean by running a marathon.

I'm referring to this quote from: http://www.jwz.org/hacks/marginal.html
"(Some people, when confronted with a problem, think ``I know, I'll
use regular expressions.'' Now they have two problems.)"

> I do follow your statement on raw strings, but that doesn't seem
> to be the problem.

It is an issue in the readability of your code, not the cause of the
code behavior that you don't like. In your particular case, this is
all made doubly hard to read since your patterns and search targets
include back slashes.

\[
\nabla \cdot u = 0
\]

I don't want to find the following

\begin{tabular}{c c}
a & b \\[4pt]
1 & 2 \\[3pt]
\end{tabular}

how about: r'(^|[^\\])\\\['
Which is:
Find something beginning with either start-of-line or a
non-backslash, followed (in either case) by a backslash
and ending with an open square bracket.

Generally, (for the example) I would have said a good test set
describing your problem was:

re.compile(pattern).search(r'\[ ') is not None
re.compile(pattern).search(r' \[ ') is not None
re.compile(pattern).search(r'\\[ ') is None
re.compile(pattern).search(r' \\[ ') is None

--Scott David Daniels
(e-mail address removed)