Changelog for scholdoc-citeproc-0.6

pandoc-citeproc (0.6) * The CSL parser has been replaced with a new, xml-conduit based module, which does not rely on C libraries and should be easier to repair and extend. The module has been checked against the old module with all CSL files in the citeproc repository. The only differences are with attribute values beginning or ending with a nonbreaking space (e.g. with French guillemets), and the new parser's behavior (preserving the spaces) is clearly correct. Parsing was measured as about twice as fast in a benchmark. * The `hexpat` cabal flag has been removed. * The old `Text.CSL.Pickle` and `Text.CSL.Pickle.*` modules have been removed. * Plural locator labels (e.g. 'pp.') are now used for page ranges containing en-dashes (#84). * `Elements` has been removed from `Element`. It was entirely unnecessary, used only once (and dispensibly) in `Text.CSL.Eval`. This change results in cleaner code. * Improved man page generation. * Updated `chicago-author-date.csl` default style. pandoc-citeproc (0.5) * Revised locator parsing: + parseLocator now looks for the "short" forms of terms in the style's locale(s). So, in English you'd use "p." or "ch."; in German, "S." or "Kap." + Note that the locator label must match what is in the locale file, including the period. Before this change, you could omit the period: "p 12" or "p. 12" would both give you a "page." Also, previously, the locators were case-insensitive; now they must be in the same case as in the locale. (English "p.", German "S.".) + "no." no longer gets parsed as a "note": closes #74. + Text.CSL.Reference no longer exports parseLocator. parseLocator is now used only locally, in Text.CSL.Pandoc. * Data.getLocale: Try 2-letter locale lookup if longer locale not found. This should fix an error that occured with less common locales (jgm/pandoc#1548). * Added parseNames field to Agent. Set parse-names=true when writing CSL JSON and collapsing suffix or particles into first or last names. * Properly handle agents with 'parse-names' set to 'true' (#77). Note that parse-names defaults to true if not set, as in Zotero. * Reference: added explicit exports. * Styles: Added explicit export list. * Eval: Fixed isNumericString to recognize en dash (#74). * Allow bibtex double quotes to be escaped inside {} (jgm/pandoc#1568). * Titlecase transform improvements (#76). * Added two csl files missing from tests/. * Performance improvements: Avoid unnecessary Output groupings. Eliminated some unnecessary generic traversals. (#71) * Makefile: Use cpphs. Text.CSL.Data seems to break on OSX 10.9 without it. * Added prof target to Makefile * Test suite: don't exit with success if there were errors! pandoc-citeproc (0.4.0.1) * Interpret date literals with underscores (e.g. "2004_2006") as ranges. This covers a common workaround for a deficiency in Zotero (#65). * Correctly handle "literal" in CSL date (#65). * Fixed erroneous capitalization of the first ibid. in a citation, even in the middle of a sentence (#68). Removed "ibid-c", "ibid-locator-c" classifications. * Pandoc: Capitalize first word of citation note, unless span=nocase. Previously we only capitalized citations. This captures ibid and the like. * Text.CSL.Util.toCapital: respect `<span class="nocase">`. * Turn small caps into SmallCaps element in reading CSL JSON (#67). Either the <span style="font-variant:small-caps"> or the <sc> variant may be used. * Added "ca", "c", "et" to list of short words in titlecase transforms (#66). * Fixed some errors in the no-author test case. * test-pandoc-citeproc: Avoid running tests twice! * Check for en dash in checking for page ranges (#70). * Removed unused texmath build-depend. pandoc-citeproc (0.4) * Fixed #58 - previously not all substitute alternatives were tried. * Made first parameter of OName an Agent, not a String (API change). Also moved Agent from Reference to Style. * test-pandoc-citeproc: run tests on all matching files in the directory. This removes the need to recompile when new tests are added. * Style: Fixed appendWithPunct (mappend for Formatted). One definition was wrong, causing commas to be lost in certain places (#57). * Support \nocite{*}. If the `nocite` metadata filed contains a citation with identifier `*`, all entries in the bibliography are included in the references (#64). * Bump upper bound for aeson. * Removed span nodecor around 'ser.' (#60). * Print series as "3rd ser." (using locale's ordinal) (#60). * Added scaffolding for 'ordinalize' and infrastructure for accessing locale information to Bibtex. * Export new parseLocale from Text.CSL.Parser. * Move punctuation before fn in in-text citation before punctuation (#59). * Updated locales and chicago-author-date.csl. * Use collection-title for series, even for articles (#60). * Fixed locator parsing so capital roman numerals are legal numbers (#61). * Map bookinbook -> chapter (#62). * Better error message from parseString. * Remove space before superscript citation (pandoc #1327). * Fixed spurious "et al" problem (#27). * Match biblatex's method for separating von from last name, for biblatex. For bibtex, we match the documented bibtex algorithm as before (#50). Test case: Drummond de Andrade, Carlos. bibtex makes "Drummond de" a dropping particle. biblatex takes "Drummond de Andrade" to be the last name. * Leave off bibliography if `suppress-bibliography` has a true value in the document metadata (#40, with thanks to Jesse Rosenthal). * Try empty (null) 'csl' as if no 'csl' was specified. (The default chicago-author-date.csl is then used.) * Fixed #51: empty braces with date range. * Set pageFirst automatically from page range (#31). * Bibtex input: set page-first automatically. * Fixed capitalization of first word in sentence (#25). E.g. "foo bar baz: bazbaz foo" should turn into "Foo Bar Baz: Bazbaz Foo" in titlecase, but was turning into "foo Bar Baz: bazbaz Foo." * pandoc-citeproc: Allow conversion of mods bibliographies (#28). * readCSLFile now supports dependent styles (#30). * Removed network flag. We now use pandoc's fetchItem. * Changed type of findFile, so it returns a Maybe. * Documented fact that you can specify a URL under 'csl:'. * Added locale parameter to readCSLFile (API change). readCSLFile also now looks in ~/.csl for the file, if it isn't found locally. * Moved findFile to Util. * Bibtex/biblatex: map "unpublished" to "speech" if the record contains "eventtitle", "eventdate", or "venue". Otherwise map it to "manuscript" as before (#44). * Look for chicago-author-date.csl in ~/.csl if no csl specified. If not found there, use the chicago-author-date.csl that comes with the distribution (#48). * Improved case transform behavior with punctuation. Previously capitalization might change after a word-internal period or apostrophe. Now internal punctuation only signals a word boundary if it is a dash or single quote character. pandoc-citeproc (0.3.1) * Marked final heading (if present) as "unnumbered" (pandoc #1277). * Treat empty 'references' (empty string) as empty list. * Fixed titlecase transformation of words containing '. * Fixed punctuation in notes. * Improved test suite. * FromJSON for Formatted: handle block lists as well as inline. * Require yaml >= 0.8.8.2. * Fixed spacing problem with citations in notes in footnote styles (pandoc #1036). * Updated chicago-author-date.csl and locales (#34). * Move punctuation inside quotes, depending on style. This is activated by the `punctuation-in-quote` locale option, which has a default per locale but can be overridden in the style. This gives more flexibility. (Thanks to Jesse Rosenthal.) * Moved `isPunctuationInQuote` to `Text.CSL.Style` and rewrote. Previously it was in `Text.CSL.Output.Pandoc`. It doesn't need to use generics and no longer does. * Fixed recognition of "byzantine" names. Names with curly apostrophes were being counted as non-byzantine, and printed given-first with no space (#39). * Fixed representation of reference types to conform to CSL (#24): `motion_picture`, `legal_case`, `musical_score`, `personal_communication`. pandoc-citeproc (0.3.0.1) * Require ghc-prim if ghc < 7.6 (needed for generics). pandoc-citeproc (0.3) * Removed biblio2yaml. Instead, use pandoc-citeproc with the --bib2yaml or --bib2json option. (#20) * pandoc-citeproc --bib2json will convert bibtex/biblatex to CSL compatible JSON bibliographies. * Updated locale files for correct textual date ranges in e.g. the da-DK locale. Use form="ordinal" instead of a period as suffix. (#16, #18) * Support new langid and langidopts biblatex fields. (#11) * Made test-pandoc-citeproc work in all locales. (#19) It now tests the executable rather than the library function. * Fixed disambiguation with multivolume works (#14). * Improved titlecase/unTitlecase to be sensitive to colons, question marks, and exclamation marks, after which the next character sohuld be capitalized even in sentence case (pandoc #1102). * The "locale" filed in metadata will now override a style's default locale. This can also be set by the command line: --metadata locale=fr-FR (#10). * Use CSL-compatible date-parts for dates in YAML/JSON bibliographies. * Made FromJSON for Reference more forgiving. Suitable strings are coerced into integer values. (pandoc #1094) * Fixed extra punctuation at end of footnote (#13). * Reference: Always use array in JSON for dates, agents. * Bibtex parser: Don't fail on entry with no title. * Bibliography output: 'given' is now a string, not an array. The string will be split on spaces into a list of given names. Note that an array can also be used. But we write a plain string, to match the citeproc json schema. * Fixed spacing bug in date ranges (#7). * Names: Fixed formatLabel so it works for editortranslator. This fixes a bug for cases where editor = translator (#9). * Text.CSL.Eval.Date: Fixed bugs with date ranges. Ranges within the same year were raising an error "Prelude.init: empty list" (#7). * Util: Export tr' and init'. * Text.CSL.Proc.Disamb: Correct definition of allTheSame. * Improved disambiguation. Now we correctly handle the case with no author but title. * Consider names too when determining date disambiguations. Previously if you had two distinct names needing disambiguation for the same date, the letters would not start over with 'a' for the second name. * Added Generic instances for the types in Style and Reference.