Note that the test cases are really awkward so this is unlikely to
actually happen normally, but implementation is much easier if it
doesn't have to worry about these cases.
[SVN r86704]
This is based on Rene's implementation, but I used my own glob function,
and adjusted a few things since it's now always using ascii. It would be
nice to support unicode, but that would require at the very least a
normalization library, and perhaps more than that.
[SVN r86699]
Actually a subtle change, as includes and anchors could previously
contain spaces. Possibly should still alow that, although I expect it's
never used.
[SVN r86688]
Fix bug where a code block at the start of a file was not recognised.
I was actually adding this test to check that the `source_mode_unset`
mechanism in `doc_info_grammar.cpp` was working, but it ended exposing
this bug.
[SVN r85590]
There's a minor bug with escaped templates in 1.6 and I don't have time to
fix it, this will help limit its effect. As far as I'm aware no one uses
escaped templates with punctuation templates. Escaped templates are an
undocumented feature anyway.
[SVN r85285]
Doesn't work for: templates in lists or explicit list block elements
(because it detects when it's in a list based on the parser), but it is
usually at least as good as the older version.
This kind of thing would be much easier if I represented the lists in
data rather than translating them as they are parsed. Not sure if I'll
do that soon though.
[SVN r85247]
Since this is often indented, does not treat indented blocks as code
blocks. This is inconsistent for lists, so maybe indented blocks in
lists should always be code, or never be code. Or maybe it should be a
bit smarter about nested indented blocks.
[SVN r85126]
It's about to break because of a 1.6 change, but that doesn't matter because
it's an obsolete test. I've already removed it from my 1.7 development.
[SVN r84935]
If the whitespace to be removed when unindenting a block is not
consistent, then replace all the indentation in that block with spaces
(i.e. expand tabs). This fixes some odd results that could happen, e.g.
....Line 1
--->Line 2
(dots = spaces, arrow = tab) would become:
...Line 1
Line 2
Because the first character was removed from each line. This now works
okay.
But I don't expand tabs when the whitespace that isn't removed is
inconsistent, which can have odd results, for example:
..Line 1
..->Line 2
......Line 3
Will just remove the first two spaces, resulting in:
Line 1
--->Line 2
....Line 3
This should be rare, as tabs after spaces are a bit odd. But maybe I
should convert to spaces more often. Perhaps when indentation is
inconsistent, or perhaps whenever a mixture of tabs and spaces are used,
or even always just do it.
[SVN r84150]
I think that the dependencies should just be files that were loaded, so
I changed it to do that, but added an extra (hidden for now) option to
list all paths that were checked, staring with a `+` for files that are
present, and `-` for files that are absent.
Also fixed a bug with the path resolver.
[SVN r77973]
- Include relevant locations on path where file wasn't found.
- Improve `include_search` while I'm at it.
- Normalize paths for non-existent files.
- Tests, not integrated with Boost.Build yet. Needs to check that
python is available then run the script with the location of the
quickbook executable.
[SVN r77598]
Again, less gratuitous changes from older versions (although it's still
different). Also more consistent with generated markup from indented code
blocks.
[SVN r75732]
The macro syntax has been tightened up in 1.6, so that fewer characters
are available for macro identifiers. But macros can still be defined
using the older quickbook versions When such macros are defined, prevent
them from expanding in a quickbook 1.6 document.
[SVN r75630]
`value` doesn't track what a string is encoded as, and sometimes it's
plain text not boostbook. It would be nice to track the encoding and
automatically do the right thing, but I'm not sure if it's worth the
effort just now.
[SVN r75619]
Because code blocks are added in sections, a nested section could be
unindented too far. Code is unindented in the main quickbook actions, so
there's actually no need to do it when extracting code snippets.
But this lead to left over whitespace that would have been stripped. So
change the `start_snippet` and `end_snippet` grammars to swallow up the
line they're on.
I also added a newline before the end of the snippet, this cleaned up
the occasional syntax error.
But due to these changes there is more leading and trailing whitespace
in the generated code blocks, so I just changed the code block grammar
to strip them out. This means there are no newlines at the beginning or
end of code blocks, but that should be fine.
[SVN r75613]
- Fixes extension check.
- Looks for images relative to quickbook file instead of the current
working directory, so that it will work when run from another
directory.
- Allow user to specify the image location.
- Use `bbk_value` for synthesized attributes rather than `qbk_value`.
It's the right thing to do, but doesn't make any practical difference.
Still not happy with image path handling. I feel that they should be
relative to the quickbook file and be rewritten - in a similar manner to
`xinclude`.
[SVN r75612]
In quickbook 1.6, nested documents default to C++, regardless of the
parent's source mode so that they are interpreted as if they are
standalone documents.
[SVN r75528]
In the old parser, the `eol` never matched anything because the `space`
would gobble up any newlines, so I'm pretty sure the intent was to use
`blank` to remove any leading spaces. So do that in 1.6.
[SVN r75521]
Quite a big change. I moved the list logic into the grammar so that it's
easier to tell how to parse different blocks. Also reworked some of the
block vs. phrase stuff - it's a lot cleaner now which helped implement
this. It generates terrible markup at the moment, but at least the
parser is in place.
[SVN r75446]
