changelog

texmath (0.12.0.1)

  * OMML writer: Fix overline and accent rendering (#152).
  * OMML reader: Fix dropped arrows (#153). Add tests.

texmath (0.12)

  * Use Text instead of String in data types and functions
    (Christian Despres) [API change].  Note that there are still a few
    places where we unpack Text to String with a view pattern:
    performance could likely be increased with further rewriting.
  * Avoid use of !! with negative index (jgm/pandoc#5853).

texmath (0.11.3)

  * Use error instead of fail to allow building with ghc 8.8.
  * Test output: remove superfluous spaces after control sequences,
    superfluous groups, and unicode VARIATION SELECTOR 1.
  * renderTeX: add space between control sequence and any non-ASCII
    character.  There are differences in behavior of isAlphaNum between
    different ghc versions that would affect test output otherwise.
  * charToLaTeXString: Ignore 65024 VARIATION SELECTOR 1 to avoid putting
    it literally in the output ; it is used in mathml output and occurs
    in many of the test cases.
  * Add cabal.project.
  * Use actions rather than travis for CI.

texmath (0.11.2.3)

  * OMML reader: properly distinguish normal text from math (#136).
    If `m:nor` or `m:lit` is set in `m:rPr`, we interpret the
    contents as literal text and not as math.
  * TeX reader: use different symbol (`_`) for `\underline` (#142).
    This gets the right accent properties on MathML output, so
    that the underline is not lower than it should be.
  * TeX reader: Treat `\bmod` as a relational symbol rather than
    an operator (#143).  This fixes spacing problems in several
    output formats.

texmath (0.11.2.2)

  * OMML writer: use m:nor for normal text (#135).

texmath (0.11.2.1)

  * OMML reader: Don't collapse `fName` to a string (#133).
    This fixes cases where fName has some complexity, e.g.
    a subscript or limit.

texmath (0.11.2)

  * Improved handling of \mathop etc (#126).  We now allow operators like
    `arg\,min`, converting the space into unicode.
  * Support \hspace (#126).
  * Support \hdots as synonym of \ldots (#126).
  * Support \mathds (#126).
  * In parsing array, ignore `|` in column specs (#127).
    We have no way to represent this in EArray, currently.
    Ignoring them seems better than failing altogether.

texmath (0.11.1.2)

  * Eqn writer: properly escape `{` and `}`.
  * Set more accurate bounds (Herbert Valerio Riedel).

texmath (0.11.1.1)

  * Fix building with ghc-8.6.1 by removing need for now missing
    MonadFail instances (Jonas Scholl).
  * TeX reader: Allow operators like `/` to be scaled (#120).
  * TeX reader: Improved efficiency of basicEnclosure.
  * TeX reader: Handle `\bmod` (#115).

texmath (0.11.1)

  * OMML writer: Use m:acc for accents in OMML (#119).  This fixes
    some spacing issues for e.g. the translation of `\dot{m}`.

texmath (0.11.0.1)

  * OMML writer: use zero-width space to avoid dashed box (#118).
    In Word, a dashed box shows up for empty text runs in
    exponents and bases, or empty exponents and bases.  So
    we use a zero-width space in these contexts.

texmath (0.11)

  * Changed treatment of non-ASCII characters.
    Previously we ensured that the output of conversion to tex
    was pure ASCII.  This meant rendering "ä" as "a", for
    example, and it meant that many characters (e.g. Chinese)
    simply got replaced with an empty string, while others
    got replaced with "[?]".

    This was not a particularly helpful behavior.  Including
    the unicode characters verbatim doesn't interfere with
    latex compilation.  They often won't show up in the generated
    math, but that is no worse than what happened before.

    This change passes through unicode characters unchanged
    when they can't be converted to standard LaTeX commands.

    An important reason for including the unicode characters
    is that pandoc uses TeX to represent math in its AST.
    So, for example, if you convert HTML with mathml to docx,
    you'll currently lose all Chinese characters, since they'll
    disappear in the TeX intermediary, even though a direct
    mathml to ooml conversion would have passed them through.
    With this change, these conversions will work better
    (see jgm/pandoc#4642).

    + Removed Text.TeXMath.Unicode.ToASCII (API change).
    + Removed cbits that were needed for that module.
    + Modified Tex.TeXMath.Unicode.ToTeX to pass through
      unicode characters that can't be converted, rather
      than trying to asciify them or remove them.

  * Render degree symbol in tex as `{^\circ}`.

texmath (0.10.1.2)

  * eqn writer:  use uppercase letters in unicode escapes (jgm/pandoc#4597).

texmath (0.10.1.1)

  * Handle multicharacter operators better in Eqn, TeX, OMML (#109).

  * OMML reader: unwrap `<w:...>` tags immediately under `<m:oMath>`
    (#111, Jesse Rosenthal).

texmath (0.10.1)

  * Expose Text.TeXMath.TeX (TeX rendering functions) (#108).
    This is needed in order to use getTeXMath from Text.TeXMath.Unicode.ToTeX.
  * Pandoc writer: don't insert punctuation space before explicit space
    (#107).  E.g. in `2,\!4`.
  * Fix end-line command ('\\') in AMSmath environments (ARATA Mizuki).
    The end-line command in AMSmath environments does not allow spaces
    before its optional argument.
  * Use `\in` for SMALL ELEMENT OF in "base" (Vaclav Haisman).

texmath (0.10)

  * Use `\ni` in base for U+220D (#103).
  * Improved unicode -> tex symbol lookup.  Previously we had many
    cases where the lookup table would map a unicode character to
    the empty string for the base package, and this would print finding
    a good match in another package in the environment.
  * Added support for `\symbf` (#101).
  * Revert "migrating the lookup structures for Unicode/ToTex.hs to
    use C source files to accelerate builds."  This change gave us somewhat
    faster builds (using less memory), but at a huge cost of
    maintainability.
  * Removed AlignDefault from Alignment (API change, #102).
    AlignDefault doesn't make sense for a converter between
    formats that may have different defaults.  We now properly treat
    centering as the default in MathML and OMML input.

texmath (0.9.4.4)

  * Update tests that should have been updated for 0.9.4.3.

texmath (0.9.4.3)

  * MathML writer: put linethickness attribute directly on mfrac
    element. This fixes binomial rendering.

texmath (0.9.4.2)

  * Pandoc writer: better handle accented characters (jgm/pandoc#3922).

texmath (0.9.4.1)

  * Improved handling of accents and upper/lower delimiters (#98).
  * cgi: use cloudflare cdn for mathjax.

texmath (0.9.4)

  * Added writer for GNU eqn format (used with *roff).

    + New module Text.TeXMath.Writers.Eqn, exporting writeEqn.
    + Moved handleDownup to Shared.
    + Add eqn output option in texmath executable.
    + Add eqn writer tests.

texmath (0.9.3)

  * Expose pMacroDefinition in Text.TeXMath.Readers.TeX.Macros,
    with generalized type.

texmath (0.9.2)

  * Support `\newenvironment` and `\renewenvironment`.

texmath (0.9.1.1)

  * Added program to generate cbits/{key,val}ToASCII.c from a data file.
  * Migrate lookup structures for Unicode/ToTex.hs to use C
    source files to accelerate builds (Carter Tazio Schonwald).
  * Allow \boldsymbol + a token without braces.
    And same for other styling commands. Closes #94.
  * Better ghc-prof-options.

texmath (0.9.1)

  * Pandoc writer:  add thin space after math operators.
  * TeX reader: improve parsing of `\mathop` to allow
    multi-character operator names.
  * Minor optimizations.
  * Added Ord instances to Exp and associated types.

texmath (0.9)

  * OMML writer: Properly handle nary inside delimiters (#92).
    Previously under-overs inside delimiters didn't get
    converted to nary the way they did outside of delimiters.
  * TeX reader: Support bm, mathbold, pmd.
  * OMML reader: Handle w:sym (#65).
  * New module, Text.TeXMath.Unicode.Fonts, for MS font code point
    lookup.  Copied from pandoc Text.Pandoc.Readers.Docx.Fonts,
    which will be changed to import this.  [API change]
  * Fixed typo in test program for round-trip tests.
  * Parse math inside text constructions (#62).  E.g.
    `\text{if $y$ is greater than $0$}` Text and math can nest indefinitely.
  * Modify test suite so tests can be marked as "ought to raise error."
    So far this is only used in mml.  If the test is called foo
    and `readers/mml/foo.error` exists, then the test is expected
    to raise a parse error.
  * MathML reader err: Don't print colon after line number.
  * Restore proper error handling to MathML reader.  Note that the tests
    need to be redone, since they assumed the old behavior of just
    returning empty elements on parse errors.

texmath (0.8.6.7)

  * TeX reader: treat backslash + newline as like backslash + space.
    Previously this case gave an error.  See jgm/pandoc#3189.

texmath (0.8.6.6)

  * Allow pandoc-types 1.17.*.

texmath (0.8.6.5)

  * Fixed transposition of sub/super on operators (jgm/pandoc#3040).

texmath (0.8.6.4)

  * Support 'equation' environment, without the numbering of course.

texmath (0.8.6.3)

  * Use POST instead of GET for texmath-cgi.

texmath (0.8.6.2)

  * Fixed array alignment issues (jgm/pandoc#2864, jgm/pandoc#2310).
  * Use 1 and 0 for _Hide attributes, rather than on and off.
    Officially it seems that either 1/0 or true/false are wanted.
  * Fixed EUnderOver for omml output.  Previously both the under and
    the over part were being placed under (jgm/pandoc#2775).

texmath (0.8.6.1)

  * OMML writer:  Fixed rendering of roots, so that the degree appears
    in the right place.
  * OMML writer:  Don't include empty rPr elements.

texmath (0.8.6)

  * TeX reader: Support hundreds more math symbols (all of those defined in
    Text.TeXMath.Unicode.ToTeX), including `\nwarrow`, `\swarrow`,
    `\nearrow`, `\searrow` (#90).

texmath (0.8.5.1)

  * OMML writer: Fixed order of elements in nary formulas to conform
    to OMML spec (#88, Niko Weh).  `<e>` must follow the `<sup>` and `<sub>`
    parts of `<nary>`. This fixes rendering issues in LibreOffice
    (though Word copes with the incorrect order).
  * Added Paths_texmath to Other-Modules for texmath executable.

texmath (0.8.5)

  * TeX parser: Support limited styling inside \DeclareMathOperator (#85).
  * TeX reader: Correctly parse \mbox.  Its argument is text mode.
  * Updated mathml tests to use mo for operators.
  * TeX reader: support mathopen, mathclose, mathpunct.
  * MathML writer: render EMathOperator as mo, not mi (#86).
  * MathML: handle leading space in EText (#87).
  * Take --version in executable from cabal metadata.
  * Added Paths_texmath to other-modules.

texmath (0.8.4.2)

  * Fixed overbrace, underbrace (#82).  Previously we were using the wrong
    character: U+FE37 instead of U+23DE.  This didn't work in Word.
  * Support \mathop, \mathrel, \mathbin, \mathord
  * MathML - render Symbol Ord as mi, not mo (#83).
  * Handle align environments with > 2 cells per row (#84).

texmath (0.8.4.1)

  * Added stack install instructions.
  * Fixed bold-italic in OMML (#76).  Previously `\mathbfit` didn't work
    properly in OMML output.
  * Ignore `\nonumber` (#79).
  * Allow styling in `\operatorname` e.g. `\operatorname{\mathcal{L}}` (#80).
  * Fixed bug in `supHide` and `subHide` for OMML.  This  led to little
    empty boxes being displayed in integrals with subscripts but no
    superscripts.  See jgm/pandoc#2571.
  * Implemented `\mod` as a math operator (#81).  This doesn't capture all the
    spacing subtleties of the amsmath version, but should be good enough
    for most purposes.
  * Allow pandoc-types < 1.17.

texmath (0.8.4)

  * Improved symbol spacing in Pandoc output (jgm/pandoc#2261).
    This change avoids putting space around binary symbols that
    come at the beginning or end of a group, or appear on their
    own.  It also avoids spacing on a binary symbol that follows
    a Bin, Op, Rel, Open, or Punct atom, in accord with
    TeXBook Appendix G.  We could go farther towards exactly
    matching the TeXBook rules, but this simple change goes some of
    the way.
  * Added stack.yaml.

texmath (0.8.3)

  * Parse uppercase Greek letters as EIdentifier, not ESymbol Op.
    This fixes handling of things like `$Lambda^1$`, particularly in omml.

texmath (0.8.2.2)

  * Handle `.` after number with no following digits.

texmath (0.8.2.1)

  * Handle bare hyphen in `\text{...}`.  Closes jgm/pandoc#2274.
  * Support `\ltimes` and `\rtimes` in the TeX reader (Arata Mizuki).
  * Slightly more efficient number parser.

texmath (0.8.2)

  * Better handling of decimal points. Decimal points are now parsed
    as parts of numbers, not as separate symbols.  E.g. in MathML they
    now appear in `<mn>` elements.  Closes #74.

texmath (0.8.1)

  * OMML: Don't force everything into Roman font by default.
    This change ensures that variables will be italic by
    default in Word.  See jgm/pandoc#2075.

texmath (0.8.0.2)

  * Fixed typo in `defaultEnv` to include `amssymb` (#68).
  * Moved some lookup tables to C, and disabled aggressive
    profiling defaults, to avoid excessive memory usage in
    compiling with clang (#70).
  * Support `\newcommand*` in `parseMacroDefinition` (jgm/pandoc#2005).
  * Fixed order bug for over/under in OMML reader (#66).
  * Support `\boldsymbol` (#67).

texmath (0.8.0.1)

  * Added network-uri flag. This addresses the split of network
    and network-uri packages.
  * OMML reader: change default accent (Jesse Rosenthal).
    The default had previously been set as accute (possibly as a
    placeholder). It appears to be circumflex/hat instead.

texmath (0.8)

  * Added OMML reader (Jesse Rosenthal).
  * Support latex \substack (#57).
  * Added EBoxed and implemented in readers and writers (#58).
  * Handle latex \genfrac.  Use \genfrac for \brace, \brack,
    etc. when amsmath is available.
  * Improvements in handling of space characters.
  * Use ESpace rather than EText when a mathml mtext just contains
    a space.
  * Use \mspace when needed to get latex spaces with odd sizes, rather
    than finding the closest simple command.
  * Use Rational instead of Double in ESpaced, EScaled.
  * Shared: Export getSpaceWidth, getSpaceChars.
  * Shared: Export fixTree, isEmpty, empty (formerly in MathML reader).

texmath (0.7.0.2)

  * TeX reader:  further improvements in error reporting.
    Instead of reporting line and column, a snippet is printed
    with a caret indicating the position of the error.  Also
    fixed bad position information when control sequences are
    followed by a letter.

texmath (0.7.0.1)

  * TeX reader:

    + Improved error reporting.
    + Optimized parser.
    + Treat `\ ` as ESpaced rather than ESymbol.
    + Internal improvements, including using the parsec3 interface
      instead of the older parsec2 compatibility interface.

  * Added tests for phantom.

texmath (0.7)

  * Changes in Exp type:

    + Removed EUp, EDown, EDownup, EUnary, EBinary.
    + Added EFraction (and FractionType), ESqrt, Eroot, EPhantom.
    + Added boolean "convertible" parameter to EUnder, EOver, EUnderover.
    + Changed parameter of EScaled from String to Double.
    + Changed parameter of ESpace from String to Double.
    + Removed EStretchy.
    + Added EStyled, corresponding to mstyle in mathml, and \mathrm,
      \mathcal, etc. in TeX (which can contain arbitrary math content,
      not just text).
    + Changed the type of EDelimited.  The contents of an EDelimited are
      now either Right Exp or Left String (the latter case represents a
      fence in middle position, e.g. \mid| in LaTeX).

  * Module reorganisation:  the exposed interface has been completely
    changed, and modules for reading MathML and writing TeX math
    have been added:

    + All writers now reside in Text.TeXMath.Writers.
        - Text.TeXMath.MathML -> Text.TeXMath.Writers.MathML.
          toMathML and showExp are removed, writeMathML added.
        - Text.TeXMath.OMML -> Text.TeXMath.Writers.OMML.
          toOMML and showExp removed, writeOMML added.
        - Text.TeXMath.Pandoc -> Text.TeXMath.Writers.Pandoc.
          toPandoc removed, writePandoc added.
        - New module Text.TeXMath.Writers.TeX, exporting writeTeX,
          writeTeXWith, addLaTeXEnvironment (the latter giving control
          over which packages are assumed to be available).

    + All readers now reside in Text.TeXMath.Readers.
        - Text.TeXMath.MathMLParser -> Text.TeXMath.Readers.MathML,
	  exporting readMathML.
        - Text.TeXMath.Readers.TeX nows exports readTeX rather than
          parseFormula.

    + New modules for unicode conversion: Text.TeXMath.Unicode.ToASCII,
      Text.TeXMath.Unicode.ToTeX, Text.TeXMath.Unicode.ToUnicode.

    + Two MathML specific modules: Text.TeXMath.Readers.MathML.EntityMap,
      Text.TeXMath.Readers.MathML.MMLDict.

    + In Text.TeXMath, all the XtoY functions have been removed
      in favour of rexporting raw reader and writer functions. The
      data type Exp is now also exported.

  * Bug fixes and improvements.

    + TeX writer: Properly handle accents inside \text{}.
      Use real text environments for EText (\textrm, not \mathrm).
      Improved handling of scalers (\Big etc.).  Use amsmath matrix
      environments when appropriate.  Fixed \varepsilon.
    + MathML writer:  Omit superfluous outer mrows.  Add position
      information to fences.
    + OMML writer:  Handle \phantom.
    + Pandoc writer:  Use unicode characters to support Fraktur and
      other text styles.
    + TeX reader:  Use EUnder/Over for \stackrel, \overset, \underset.
      Improved handling of primes.  Fixed \notin.  Avoid superfluous
      grouping of single elements.  Improved handling of scalers (\Big etc.).
      Handle \choose, \brace, \brack, \bangle (#21).
    + Macros:  Don't raise an error if applying a macro fails to
      resolve to a fixed point; instead, just return the original string.

  * Rewrote test suite as a proper cabal test suite.  Added
    --regenerate-tests and --round-trip options.

  * Updated texmath online demo for bidirectional conversion.

  * Removed cgi and test flags.  Added executable flag to build texmath.

  * Modified texmath so it works like a cgi script when run as
    texmath-cgi (through symlink or renaming).  Removed dependency on
    the cgi package.

texmath (0.6.7)

  * New Module: Text.TeXMath.Unidecode, a module for approximating
    unicode characters in ASCII.
  * New Module: Text.TeXMath.Shared, a module for shared lookup
    tables between the various readers and writers
  * New Module: Text.TeXMath.MathMLParser, exporting readMathML.
  * New Module: Text.TeXMath.EntityMap, exporting getUnicode,
    a conversion table from MathML entities to unicode characters.
  * New Module: Text.TeXMath.UnicodeToLaTeX, exporting getLaTeX,
    converting a string of unicode characters to a string of equivalent LaTeX
    commands.
  * New Module: Text.TeXMath.LaTeX, replacing Text.TeXMath.Parser,
    exporting toTeXMath.
  * New Module: Text.TeXMath.MMLDict, implements a lookup table from
    operators to their default values as defined by the MML dictionary,
    exporting getOperator.
  * New Module: Text.TeXMath.Compat, maintaining compatibility with
    mtl < 2.2.1.
  * Modified Text.TeXMath to export the primitive readers, as well as
    mathMLTo{Writer} for all writers
  * Modified: Text.TeXMath.Types: added additional record types for
    Text.TeXMath.MMLDict and Text.TeXMath.UnicodeToLaTeX.
    New Exports: Operator(..), Record(..).
  * Modified test suite:  use cabal test, added significant number of tests.
  * Added recognition of the LaTeX command \phantom

texmath (0.6.6.3)

  * Use combining tilde accent for \tilde.  Closes pandoc #1324.

texmath (0.6.6.2)

  * Allow \left to be used with ), ], etc.  Ditto with \right.
    Previously only (, [, etc. were allowed with \left.  Closes pandoc #1319.

texmath (0.6.6.1)

  * Support \multline (previously it was mispelled "multiline")
  * Changed data-files to extra-source-files.

texmath (0.6.6)

  * Insert braces around macro expansions to prevent breakage (#7).
  * Support \operatorname and \DeclareMathOperator (rekka) (#17).
  * Support \providecommand (#15).
  * Fixed spacing bugs in pandoc rendering (#24).
  * Ignore \hline at end of array row instead of failing (#19).