'try' is not automatically applied to the 'end' parser.
Added 'try' in calls to 'enclosed' where needed. Slight speed
increase.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@926 788f1e2b-df1e-0410-8736-df70ead52e1b
+ Rearranged parsers in definition of 'inline' so that the most
frequently used would (by and large) be tried first.
+ Removed some unneeded 'try's.
+ Removed tabchar parser, as whitespace handles tabs anyway.
+ All in all, these changes, together with the last two commits,
cut almost in half the time it takes pandoc to parse a large test file.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@924 788f1e2b-df1e-0410-8736-df70ead52e1b
p. 27
at the beginning of a line as an ordered list start, since
it's most likely a page number.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@900 788f1e2b-df1e-0410-8736-df70ead52e1b
is a capital letter followed by a period (including a single-letter
capital roman numeral), then it must be followed by at least two spaces.
The point of this is to avoid accidentally treating people's initials as
list markers: a paragraph may begin:
B. Russell was an English philosopher.
and this shouldn't be treated as a list.
Modified Markdown reader and README documentation.
Added a test case.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@880 788f1e2b-df1e-0410-8736-df70ead52e1b
Markdown test suite under --strict:
+ Removed check for a following setext header in endline.
A full test is too inefficient (doubles benchmark time), and
the substitute we had before is not 100% accurate.
+ Don't use Code elements for autolinks if --strict specified.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@876 788f1e2b-df1e-0410-8736-df70ead52e1b
that anyOrderedListStart did not check for a space following the
ordered list marker. So, 'A.B. 2007' would be parsed as a list item,
then fail because of the lack of space after 'A.' (required by
orderedListStart). Resolves Issue #22.
+ Fixed a similar problem in RST reader.
+ Added regression test.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@861 788f1e2b-df1e-0410-8736-df70ead52e1b
reformatted, etc.) More major changes are documented below:
+ Removed Text.Pandoc.ParserCombinators and moved all its definitions
to Text.Pandoc.Shared.
+ In Text.Pandoc.Shared:
- Removed unneeded 'try' in blanklines.
- Removed endsWith function and rewrote functions to use isSuffixOf instead.
- Added >>~ combinator.
- Rewrote stripTrailingNewlines, removeLeadingSpaces.
+ Moved Text.Pandoc.Entities -> Text.Pandoc.CharacterReferences.
- Removed unneeded functions charToEntity, charToNumericalEntity.
- Renamed functions using proper terminology (character references,
not entities). decodeEntities -> decodeCharacterReferences,
characterEntity -> characterReference.
- Moved escapeStringToXML to Docbook writer, which is the only thing
that uses it.
- Removed old entity parser in HTML and Markdown readers; replaced with
new charRef parser in Text.Pandoc.Shared.
+ Fixed accent bug in Text.Pandoc.Readers.LaTeX: \^{} now correctly
parses as a '^' character.
+ Text.Pandoc.ASCIIMathML is no longer an exported module.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@835 788f1e2b-df1e-0410-8736-df70ead52e1b
+ The changes are documented in README, under Lists.
+ The OrderedList block element now stores information
about list number style, list number delimiter, and
starting number.
+ The readers parse this information, when possible.
+ The writers use this information to style ordered
lists.
+ Test suites have been changed accordingly.
Motivation: It's often useful to start lists with
numbers other than 1, and to have control over the
style of the list.
Added to Text.Pandoc.Shared:
+ camelCaseToHyphenated
+ toRomanNumeral
+ anyOrderedListMarker
+ orderedListMarker
+ orderedListMarkers
Added to Text.Pandoc.ParserCombinators:
+ charsInBalanced'
+ withHorizDisplacement
+ romanNumeral
RST writer:
+ Force blank line before lists, so that sublists will be handled
correctly.
LaTeX reader:
+ Fixed bug in parsing of footnotes containing multiple paragraphs,
introduced by use of charsInBalanced. Fix: use charsInBalanced'
instead.
LaTeX header:
+ use mathletters option in ucs package, so that basic unicode Greek
letters will work properly.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@834 788f1e2b-df1e-0410-8736-df70ead52e1b
+ added try to def of indentSpaces
+ in def of 'reference', check to make sure it's not a note reference
first.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@827 788f1e2b-df1e-0410-8736-df70ead52e1b
Subscript in markdown reader.
+ Also replaced constants like emphStart with literals.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@752 788f1e2b-df1e-0410-8736-df70ead52e1b
unify treatment of embedded brackets in links and inline footnotes.
Note that the solution adopted here causes one of John Gruber's
markdown tests to fail:
[with_underscore](/url/with_underscore)
Here the whole phrase "underscore](/url/with" is treated as
emphasized. The previous version of the markdown reader handled
this the way Gruber's script handles it, but it ran into trouble
on the following:
[link with verbatim `]`](/url)
where the inner ] was treated as the end of the reference link
label. I don't see any good way to handle both cases in the framework
of pandoc, so I choose to require an escape in the first example:
[with\_underscore](/url/with_underscore)
git-svn-id: https://pandoc.googlecode.com/svn/trunk@729 788f1e2b-df1e-0410-8736-df70ead52e1b
"test^[my [note] contains brackets]" would yield a note with contents
"my [note". Now it yields a note with contents "my [note] contains
brackets". New function: inlinesInBrackets.
Resolves Issue 14.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@665 788f1e2b-df1e-0410-8736-df70ead52e1b
Text.Pandoc.ParserCombinators. This way, all the
Pandoc modules are in one place.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@663 788f1e2b-df1e-0410-8736-df70ead52e1b
being processed. Solution: three-stage parse. First, get
all the reference keys and add information to state. Next,
get all the notes and add information to state. (Reference
keys may be needed at this stage.) Finally, parse everything
else.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@625 788f1e2b-df1e-0410-8736-df70ead52e1b
the parsing code. A colon is now required before every block in a
definition. This fixes a problem with the old syntax, in which the last
block in the following was ambiguous between a regular paragraph in the
definition and a code block following the definition list:
term
: definition
is this code or more definition?
git-svn-id: https://pandoc.googlecode.com/svn/trunk@589 788f1e2b-df1e-0410-8736-df70ead52e1b
end of strings in Main, do it in readMarkdown and readRST.
(Note: the point of this is to ensure that a block at the
end of the file gets treated as if it has blank space after
it, which is generally what is wanted.)
git-svn-id: https://pandoc.googlecode.com/svn/trunk@588 788f1e2b-df1e-0410-8736-df70ead52e1b
structure. Key and Note blocks have been removed. Link and image URLs
are now stored directly in Link and Image inlines, and note blocks
are stored in Note inlines. This requires changes in both parsers
and writers. Markdown and RST parsers need to extract data from key
and note blocks and insert them into the relevant inline elements.
Other parsers can be simplified, since there is no longer any need to
construct separate key and note blocks. Markdown, RST, and HTML writers
need to construct lists of notes; Markdown and RST writers need to
construct lists of link references (when the --reference-links option
is specified); and the RST writer needs to construct a list of image
substitution references. All writers have been rewritten to use the
State monad when state is required. This rewrite yields a small speed
boost and considerably cleaner code.
* Text/Pandoc/Definition.hs:
+ blocks: removed Key and Note
+ inlines: removed NoteRef, added Note
+ modified Target: there is no longer a 'Ref' target; all targets
are explicit URL, title pairs
* Text/Pandoc/Shared.hs:
+ Added 'Reference', 'isNoteBlock', 'isKeyBlock', 'isLineClump',
used in some of the readers.
+ Removed 'generateReference', 'keyTable', 'replaceReferenceLinks',
'replaceRefLinksBlockList', along with some auxiliary functions
used only by them. These are no longer needed, since
reference links are resolved in the Markdown and RST readers.
+ Moved 'inTags', 'selfClosingTag', 'inTagsSimple', and 'inTagsIndented'
to the Docbook writer, since that is now the only module that uses
them.
+ Changed name of 'escapeSGMLString' to 'escapeStringForXML'
+ Added KeyTable and NoteTable types
+ Removed fields from ParserState; 'stateKeyBlocks', 'stateKeysUsed',
'stateNoteBlocks', 'stateNoteIdentifiers', 'stateInlineLinks'.
Added 'stateKeys' and 'stateNotes'.
+ Added clause for Note to 'prettyBlock'.
+ Added 'writerNotes', 'writerReferenceLinks' fields to WriterOptions.
* Text/Pandoc/Entities.hs: Renamed 'escapeSGMLChar' and
'escapeSGMLString' to 'escapeCharForXML' and 'escapeStringForXML'
* Text/ParserCombinators/Pandoc.hs: Added lineClump parser: parses a raw
line block up to and including following blank lines.
* Main.hs: Replaced --inline-links with --reference-links.
* README:
+ Documented --reference-links and removed description of --inline-links.
+ Added note that footnotes may occur anywhere in the document, but must
be at the outer level, not embedded in block elements.
* man/man1/pandoc.1, man/man1/html2markdown.1: Removed --inline-links
option, added --reference-links option
* Markdown and RST readers:
+ Rewrote to fit new Pandoc definition. Since there are no longer
Note or Key blocks, all note and key blocks are parsed on a first pass
through the document. Once tables of notes and keys have been constructed,
the remaining parts of the document are reassembled and parsed.
+ Refactored link parsers.
* LaTeX and HTML readers: Rewrote to fit new Pandoc definition. Since
there are no longer Note or Key blocks, notes and references can be
parsed in a single pass through the document.
* RST, Markdown, and HTML writers: Rewrote using state monad new Pandoc
and definition. State is used to hold lists of references footnotes to
and be printed at the end of the document.
* RTF and LaTeX writers: Rewrote using new Pandoc definition. (Because
of the different treatment of footnotes, the "notes" parameter is no
longer needed in the block and inline conversion functions.)
* Docbook writer:
+ Moved the functions 'attributeList', 'inTags', 'selfClosingTag',
'inTagsSimple', 'inTagsIndented' from Text/Pandoc/Shared, since
they are now used only by the Docbook writer.
+ Rewrote using new Pandoc definition. (Because of the different
treatment of footnotes, the "notes" parameter is no longer needed
in the block and inline conversion functions.)
* Updated test suite
* Throughout: old haskell98 module names replaced by hierarchical module
names, e.g. List by Data.List.
* debian/control: Include libghc6-xhtml-dev instead of libghc6-html-dev
in "Build-Depends."
* cabalize:
+ Remove haskell98 from BASE_DEPENDS (since now the new hierarchical
module names are being used throughout)
+ Added mtl to BASE_DEPENDS (needed for state monad)
+ Removed html from GHC66_DEPENDS (not needed since xhtml is now used)
git-svn-id: https://pandoc.googlecode.com/svn/trunk@580 788f1e2b-df1e-0410-8736-df70ead52e1b
+ fixed bug in indentSpaces (which didn't properly handle
cases with mixed spaces and tabs)
+ rewrote definition list code to conform to new syntax
+ include definition lists in list block
+ failIfStrict on definition lists
git-svn-id: https://pandoc.googlecode.com/svn/trunk@572 788f1e2b-df1e-0410-8736-df70ead52e1b
+ definition blocks must be indented throughout (not
just in first line)
+ compact lists can be formed by leaving no blank line
between a definition and the next term
git-svn-id: https://pandoc.googlecode.com/svn/trunk@568 788f1e2b-df1e-0410-8736-df70ead52e1b
syntax:
term 1
Definition 1
Paragraph 2 of definition 1.
term 2
There must be whitespace between entries.
Any kind of block may serve as a definition,
but the first line of each block must be indented.
terms can contain any *inline* elements
If you want to be lazy, you can just
indent the first line of the definition block.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@566 788f1e2b-df1e-0410-8736-df70ead52e1b
+ Lists may begin with lowercase letters only, and only 'a' through
'n'. Otherwise first initials and page references (e.g., p. 400)
are too easily parsed as lists.
+ Numbers beginning list items must end with '.' (not ')', which is
now allowed only after letters).
NOTE: This change may cause documents to be parsed differently.
Users should take care in upgrading.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@561 788f1e2b-df1e-0410-8736-df70ead52e1b
+ remove support for all-caps contractions (too
much potential for conflict with things like
'M. Mitterand')
+ add support for 'm as a contraction
git-svn-id: https://pandoc.googlecode.com/svn/trunk@560 788f1e2b-df1e-0410-8736-df70ead52e1b
treat ' followed by ll, re, ve, then a
non-letter, as a contraction. (e.g.
I've, you're, he'll)
git-svn-id: https://pandoc.googlecode.com/svn/trunk@559 788f1e2b-df1e-0410-8736-df70ead52e1b
reference keys in Markdown parser, instead of parsing
normally, then using setInput to reset input. Slight
performance improvement.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@536 788f1e2b-df1e-0410-8736-df70ead52e1b
especially in "option" contexts.
+ Removed the "try" from the "end" parser in "enclosed"
(Text.Pandoc.Shared). Now "enclosed" behaves like
"option", "manyTill", etc.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@527 788f1e2b-df1e-0410-8736-df70ead52e1b
It is no longer needed now that all entities are processed in the markdown
and HTML readers. All calls to stringToSGML have been replaced by calls
to encodeEntities.
+ Since inTag's attribute handling already encodes entities,
calls to encodeEntities are no longer needed for attribute values, so
they've been removed.
+ The HTML and Markdown readers now call decodeEntities on all raw
strings (e.g. authors, dates, link titles), to ensure that no unprocessed
entities are included in the native representation of the document.
(In the HTML reader, most of this work is done by a change in
extractAttributeName.)
+ The result is a small speed improvement (around 5% on my benchmark)
and cleaner code.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@519 788f1e2b-df1e-0410-8736-df70ead52e1b
Now these are stored as a '"' character, not as '"'.
The function escapeLinkTitle in the Markdown writer is
unnecessary and was removed. Tests modified accordingly.
git-svn-id: https://pandoc.googlecode.com/svn/trunk@517 788f1e2b-df1e-0410-8736-df70ead52e1b
