Boost C++ Libraries: Ticket #2860: misdetected unexpected character

jgresula@… — Sun, 15 Mar 2009 21:59:22 GMT

fixed formatting:

[python]
``
print "\xfapln\xeck"
``

warning: in column:9, unexpected character: "
warning: in column:22, unexpected character: "

owner, status, component changed

Daniel James — Mon, 16 Mar 2009 19:06:31 GMT

owner changed from Joel de Guzman to Daniel James
status new → assigned
component quickbook → spirit

I think this is a bug in spirit's c_escape_ch_p. I'll look into it.

component changed

Daniel James — Mon, 16 Mar 2009 19:16:22 GMT

component spirit → quickbook

Sorry, my mistake. It's a bug in quickbook, it's using a C character escape parser for python and the syntax is different. Although I think it should be fairly liberal about how it parses either language, so I think I might change both to accept a string like that.

Daniel James — Tue, 24 Mar 2009 08:34:14 GMT

(In [51949]) Use our own escape character parser.

We're currently using the spirit escaped character parser. This is a strict C parser and causes parse errors for code that we probably should accept, such as any hexadecimal encoded characters with a value greater than the maximum value of the current platform's 'char'. This is \x7F on some platforms so it rejects '\x80' upwards. As well as rejecting characters that might be valid, it also means that quickbook will act differently on different platforms.

In python strings, '\xaaa' is equivalent to '\xaa' + 'a', but the spirit parser interprets this as a character with value '0xaaa'. So we probably should accept these.

I also think we should be liberal about what we accept. IMO it isn't our job to enforce correct C++/python, just to create a reasonable rendering of our input. So rather than write a parser which understands the various types of escapes, I just wrote one that ignores any character following a backslash.

Refs #2860.

status changed; resolution set

Daniel James — Mon, 30 Mar 2009 17:49:18 GMT

status assigned → closed
resolution → fixed

(In [52064]) Fix bug handling python and C++ strings.

Fixes #2860.

Merged revisions 51949 via svnmerge from https://svn.boost.org/svn/boost/trunk

........

r51949 | danieljames | 2009-03-24 08:34:09 +0000 (Tue, 24 Mar 2009) | 21 lines

Use our own escape character parser.

We're currently using the spirit escaped character parser. This is a strict C parser and causes parse errors for code that we probably should accept, such as any hexadecimal encoded characters with a value greater than the maximum value of the current platform's 'char'. This is \x7F on some platforms so it rejects '\x80' upwards. As well as rejecting characters that might be valid, it also means that quickbook will act differently on different platforms.

In python strings, '\xaaa' is equivalent to '\xaa' + 'a', but the spirit parser interprets this as a character with value '0xaaa'. So we probably should accept these.

I also think we should be liberal about what we accept. IMO it isn't our job to enforce correct C++/python, just to create a reasonable rendering of our input. So rather than write a parser which understands the various types of escapes, I just wrote one that ignores any character following a backslash.

Refs #2860.

........