Docx reader doesn't read images on Windows.

[jkr]

Not sure why, but it has been reported on the list (Windows 8), and I reproduced the problem using wine. On Linux/OSX, we get the following correct behavior:

jkr@guppy$ pandoc inline_images.docx
<p>This picture <img src="media/image1.jpg" /> is an identicon.</p>
<p>Here is <a href="http://www.google.com">one <img src="media/image2.jpg" /> that</a> links.</p>
jkr@guppy$ pandoc image.docx
<p>An image:</p>
<p><img src="media/image1.jpg" /></p> 

On wine (and Windows 8), we get:

jkr@guppy$ wine pandoc ~/devel/pandoc/tests/docx/inline_images.docx
<p>This picture is an identicon.</p>
<p>Here is <a href="http://www.google.com">one that</a> links.</p>
jkr@guppy$ wine pandoc ~/devel/pandoc/tests/docx/image.docx   
<p>An image:</p>

[jkr]

I'm thinking that, since images are the only place we do path construction, it must have something to do with a disparity between Windows and unix paths. But I don't see why that would produce the problems it does. Unfortunately, I don't have any way to test inside a windows env to be sure.

[jkr]

Since it fails at both inline and block level, I'm going to guess that the culprit is in expandDrawingId. But again, not sure how to test it.

[jgm]

I haven't looked into this, but it's almost certainly a path issue.
I believe that things are being stored in the MediaBag with a "/"
delimiter (even on Windows). And when you do lookup (combine "word"
filepath) e envMedia, on Windows you're combining with "". Try
replacing "combine" with 'intercalate "/"'.

++ Jesse Rosenthal [Sep 02 14 09:46 ]:

Since it fails at both inline and block level, I'm going to guess that
the culprit is in expandDrawingId. But again, not sure how to test it.

—
Reply to this email directly or [1]view it on GitHub.

References

1. Docx reader doesn't read images on Windows. #1597 (comment)

[jkr]

Okay -- would you have an environment for testing?

[jgm]

Yes, I have a Windows VM. Unfortunately, nothing I can give you
access to.

+++ Jesse Rosenthal [Sep 02 14 10:36 ]:

Okay -- would you have an environment for testing?

—
Reply to this email directly or [1]view it on GitHub.

References

1. Docx reader doesn't read images on Windows. #1597 (comment)

[jkr]

I pushed a trivial fix to https://github.com/jkr/pandoc/tree/windowsfix . Would you be able to give it a spin at some point?

[jgm]

Okay, I tried compiling your branch. Ran the test suite and found several failures. (I realize that I never ran the test suite on Windows for this version; probably these failures also turn up if the non-patched pandoc is compiled.)

Test suite test-pandoc: RUNNING...
Old:
  markdown:
    writer:
      basic: [OK]
      tables: [OK]
      lhs to normal: [OK]
      lhs to lhs: [OK]
    reader:
      basic: [OK]
      tables: [OK]
      pipe tables: [OK]
      more: [OK]
      lhs: [OK]
    citations:
      citations: [OK]
  rst:
    writer:
      basic: [OK]
      tables: [OK]
      lhs to normal: [OK]
      lhs to lhs: [OK]
    reader:
      basic: [OK]
      tables: [OK]
      lhs: [OK]
  latex:
    writer:
      basic: [OK]
      tables: [OK]
      lhs to normal: [OK]
      lhs to lhs: [OK]
    reader:
      basic: [OK]
      lhs: [OK]
  html:
    writer:
      basic: [OK]
      tables: [OK]
      lhs to normal: [OK]
      lhs to lhs: [OK]
    reader: [OK]
  s5:
    s5 writer (basic): [OK]
    s5 writer (fancy): [OK]
    html writer (fragment): [OK]
    html writer (inserts): [OK]
  textile:
    writer:
      basic: [OK]
      tables: [OK]
    reader: [OK]
  docbook:
    writer:
      basic: [OK]
      tables: [OK]
    reader: [OK]
  native:
    writer:
      basic: [OK]
      tables: [OK]
    reader: [OK]
  fb2:
    basic: [OK]
    titles: [OK]
    images: [OK]
    images-embedded: [OK]
    math: [OK]
    tables: [OK]
    testsuite: [OK]
  mediawiki:
    writer:
      basic: [OK]
      tables: [OK]
    reader: [OK]
  dokuwiki:
    writer:
      basic: [OK]
      tables: [OK]
    inline_formatting: [OK]
    multiblock table: [OK]
  opml:
    basic: [OK]
    reader: [OK]
  haddock:
    writer:
      basic: [OK]
      tables: [OK]
    reader: [OK]
  txt2tags:
    reader: [OK]
  epub:
    features: [Failed]

------------------------------------------------------------------------
--- epub/features.native
+++ C:\Users\jgm\src\pandoc-windowsfix\dist\build\pandoc\pandoc --data-dir ..\data epub/features.epub -r epub -w native
+  39 ,Para [Image [Str "gif",Space,Str "test"] ("img\\check.gif","")]
-  39 ,Para [Image [Str "gif",Space,Str "test"] ("img/check.gif","")]
+  45 ,Para [Image [Str "png",Space,Str "test"] ("img\\check.png","")]
-  45 ,Para [Image [Str "png",Space,Str "test"] ("img/check.png","")]
+  51 ,Para [Image [Str "jpeg",Space,Str "test"] ("img\\check.jpg","")]
-  51 ,Para [Image [Str "jpeg",Space,Str "test"] ("img/check.jpg","")]
+  97 ,Para [Str "The",Space,Str "test",Space,Str "passes",Space,Str "if",Space,Str "the",Space,Str "rendering",Space,Str "looks",Space,Str "like",Space,Image [Str "description",Space,Str "of",Space,Str "imaginary",Space,Str "number:",Space,Str "c",Space,Str "=",Space,Str "a",Space,Str "+bi",Space,Str "with",Space,Str "an",Space,Str "overbrace",Space,Str "reading",Space,Str "'complex",Space,Str "number'",Space,Str "and",Space,Str "underbraces",Space,Str "below",Space,Str "'a'",Space,Str "and",Space,Str "'b",Space,Str "i'",Space,Str "reading",Space,Str "'real'",Space,Str "and",Space,Str "'imaginary'",Space,Str "respectively."] ("img\\complex_number.png",""),Str "."]
-  97 ,Para [Str "The",Space,Str "test",Space,Str "passes",Space,Str "if",Space,Str "the",Space,Str "rendering",Space,Str "looks",Space,Str "like",Space,Image [Str "description",Space,Str "of",Space,Str "imaginary",Space,Str "number:",Space,Str "c",Space,Str "=",Space,Str "a",Space,Str "+bi",Space,Str "with",Space,Str "an",Space,Str "overbrace",Space,Str "reading",Space,Str "'complex",Space,Str "number'",Space,Str "and",Space,Str "underbraces",Space,Str "below",Space,Str "'a'",Space,Str "and",Space,Str "'b",Space,Str "i'",Space,Str "reading",Space,Str "'real'",Space,Str "and",Space,Str "'imaginary'",Space,Str "respectively."] ("img/complex_number.png",""),Str "."]
+ 103 ,Para [Str "The",Space,Str "test",Space,Str "passes",Space,Str "if",Space,Str "the",Space,Str "rendering",Space,Str "looks",Space,Str "like",Space,Link [Str "Cicho\324's",Space,Str "Diagram"] ("Cicho%C5%84's_diagram",""),Str ":",Space,Image [Str "rendering",Space,Str "of",Space,Str "Cicho\324's",Space,Str "diagram."] ("img\\cichons_diagram.png",""),Str "."]
- 103 ,Para [Str "The",Space,Str "test",Space,Str "passes",Space,Str "if",Space,Str "the",Space,Str "rendering",Space,Str "looks",Space,Str "like",Space,Link [Str "Cicho\324's",Space,Str "Diagram"] ("Cicho%C5%84's_diagram",""),Str ":",Space,Image [Str "rendering",Space,Str "of",Space,Str "Cicho\324's",Space,Str "diagram."] ("img/cichons_diagram.png",""),Str "."]
+ 114 ,Para [Str "The",Space,Str "test",Space,Str "passes",Space,Str "if",Space,Str "the",Space,Str "rendering",Space,Str "looks",Space,Str "like",Space,Str "the",Space,Str "following",Space,Str "image:",Space,Image [Str "A",Space,Str "long",Space,Str "division",Space,Str "dividing",Space,Str "1306",Space,Str "by",Space,Str "3,",Space,Str "presented",Space,Str "in",Space,Str "'lefttop'",Space,Str "(US)",Space,Str "notation"] ("img\\ElementaryMathExample.png",""),Str "."]
- 114 ,Para [Str "The",Space,Str "test",Space,Str "passes",Space,Str "if",Space,Str "the",Space,Str "rendering",Space,Str "looks",Space,Str "like",Space,Str "the",Space,Str "following",Space,Str "image:",Space,Image [Str "A",Space,Str "long",Space,Str "division",Space,Str "dividing",Space,Str "1306",Space,Str "by",Space,Str "3,",Space,Str "presented",Space,Str "in",Space,Str "'lefttop'",Space,Str "(US)",Space,Str "notation"] ("img/ElementaryMathExample.png",""),Str "."]
+ 135 ,Para [Image [] ("img\\Maghreb1.png","")]]
- 135 ,Para [Image [] ("img/Maghreb1.png","")]]
------------------------------------------------------------------------
    wasteland: [OK]
    formatting: [OK]
  other writers:
    opendocument:
      basic: [OK]
      tables: [OK]
    context:
      basic: [OK]
      tables: [OK]
    texinfo:
      basic: [OK]
      tables: [OK]
    icml:
      basic: [OK]
      tables: [OK]
    man:
      basic: [OK]
      tables: [OK]
    plain:
      basic: [OK]
      tables: [OK]
    rtf:
      basic: [OK]
      tables: [OK]
    org:
      basic: [OK]
      tables: [OK]
    asciidoc:
      basic: [OK]
      tables: [OK]
Shared:
  normalize:
    p_normalize_blocks_rt: [OK, passed 100 tests]
    p_normalize_inlines_rt: [OK, passed 100 tests]
    p_normalize_no_trailing_spaces: [OK, passed 100 tests]
  compactify'DL:
    compactify'DL with empty def: [OK]
  collapseFilePath:
    collapse: [OK]
    collapse: [OK]
    collapse: [Failed]
expected: "../foo"
 but got: "..\\foo"
    collapse: [Failed]
expected: "../foo"
 but got: "..\\foo"
    collapse: [OK]
    collapse: [Failed]
expected: "/../baz"
 but got: "/..\\baz"
    collapse: [OK]
    collapse: [OK]
    collapse: [OK]
    collapse: [OK]
    collapse: [OK]
    collapse: [OK]
    collapse: [Failed]
expected: "../.."
 but got: "..\\.."
    collapse: [Failed]
expected: "parent/foo/bar"
 but got: "parent\\foo\\bar"
    collapse: [Failed]
expected: "parent/bar"
 but got: "parent\\bar"
    collapse: [OK]
    collapse: [OK]
    collapse: [Failed]
expected: "/parent/foo"
 but got: "/parent\\foo"
Walk:
  Walk:
    p_walk inlineTrans: [OK, passed 100 tests]
    p_walk blockTrans: [OK, passed 100 tests]
    p_query inlineQuery: [OK, passed 100 tests]
    p_query blockQuery: [OK, passed 100 tests]
Writers:
  Native:
    p_write_rt: [OK, passed 100 tests]
    p_write_blocks_rt: [OK, passed 100 tests]
  ConTeXt:
    inline code:
      with '}': [OK]
      without '}': [OK]
      code property: [OK, passed 100 tests]
    headers:
      level 1: [OK]
    bullet lists:
      nested: [OK]
  LaTeX:
    code blocks:
      in footnotes: [OK]
      identifier: [OK]
      no identifier: [OK]
    definition lists:
      with internal link: [OK]
    math:
      escape |: [OK]
    headers:
      unnumbered header: [OK]
      in list item: [OK]
      in definition list item: [OK]
    inline code:
      struck out and highlighted: [OK]
      struck out and not highlighted: [OK]
      single quotes: [OK]
  HTML:
    inline code:
      basic: [OK]
      haskell: [OK]
      nolanguage: [OK]
    images:
      alt with formatting: [OK]
  Docbook:
    line blocks:
      none: [OK]
      basic: [OK]
      blockquote: [OK]
      footnote: [OK]
    compact lists:
      bullet:
        compact: [OK]
        loose: [OK]
      ordered:
        compact: [OK]
        loose: [OK]
      definition:
        compact: [OK]
        loose: [OK]
  Markdown:
    indented code after list: [OK]
    list with tight sublist: [OK]
  Plain:
    strongly emphasized text to uppercase: [OK]
  AsciiDoc:
    emphasis:
      emph word before: [OK]
      emph word after: [OK]
      emph quoted: [OK]
      strong word before: [OK]
      strong word after: [OK]
      strong quoted: [OK]
    tables:
      empty cells: [OK]
      multiblock cells: [OK]
Readers:
  LaTeX:
    basic:
      simple: [OK]
      space: [OK]
      emphasized: [OK]
    headers:
      level 1: [OK]
      level 2: [OK]
      level 3: [OK]
      emph: [OK]
      link: [OK]
    math:
      escaped $: [OK]
    space and comments:
      blank lines + space at beginning: [OK]
      blank lines + space + comments: [OK]
      comment in paragraph: [OK]
    code blocks:
      identifier: [OK]
      no identifier: [OK]
    citations:
      natbib:
        citet: [OK]
        suffix: [OK]
        suffix long: [OK]
        multiple: [OK]
        group: [OK]
        suffix and locator: [OK]
        suffix only: [OK]
        no author: [OK]
        markup: [OK]
      biblatex:
        textcite: [OK]
        suffix: [OK]
        suffix long: [OK]
        multiple: [OK]
        group: [OK]
        suffix and locator: [OK]
        suffix only: [OK]
        no author: [OK]
        markup: [OK]
        parencite: [OK]
  Markdown:
    inline code:
      with attribute: [OK]
      with attribute space: [OK]
    emph and strong:
      two strongs in emph: [OK]
      emph and strong emph alternating: [OK]
      emph with spaced strong: [OK]
      intraword underscore with opening underscore (#1121): [OK]
    raw LaTeX:
      in URL: [OK]
    raw HTML:
      nesting (issue #1330): [OK]
    unbalanced brackets: [OK]
    backslash escapes:
      in URL: [OK]
      in title: [OK]
      in reference link title: [OK]
      in reference link URL: [OK]
    bare URIs:
      http://google.com is a search engine.: [OK]
      <a href="http://foo.bar.baz">http://foo.bar.baz</a>: [OK]
      Try this query: http://google.com?search=fish&time=hour.: [OK]
      HTTPS://GOOGLE.COM,: [OK]
      http://el.wikipedia.org/wiki/Τεχνολογία,: [OK]
      doi:10.1000/182,: [OK]
      git://github.com/foo/bar.git,: [OK]
      file:///Users/joe/joe.txt, and: [OK]
      mailto:someone@somedomain.com.: [OK]
      Use http: this is not a link!: [OK]
      (http://google.com).: [OK]
      http://en.wikipedia.org/wiki/Sprite_(computer_graphics): [OK]
      http://en.wikipedia.org/wiki/Sprite_[computer_graphics]: [OK]
      http://en.wikipedia.org/wiki/Sprite_{computer_graphics}: [OK]
      http://example.com/Notification_Center-GitHub-20101108-140050.jpg: [OK]
      https://github.com/github/hubot/blob/master/scripts/cream.js#L20-20: [OK]
      http://www.rubyonrails.com: [OK]
      http://www.rubyonrails.com:80: [OK]
      http://www.rubyonrails.com/~minam: [OK]
      https://www.rubyonrails.com/~minam: [OK]
      http://www.rubyonrails.com/~minam/url%20with%20spaces: [OK]
      http://www.rubyonrails.com/foo.cgi?something=here: [OK]
      http://www.rubyonrails.com/foo.cgi?something=here&and=here: [OK]
      http://www.rubyonrails.com/contact;new: [OK]
      http://www.rubyonrails.com/contact;new%20with%20spaces: [OK]
      http://www.rubyonrails.com/contact;new?with=query&string=params: [OK]
      http://www.rubyonrails.com/~minam/contact;new?with=query&string=params: [OK]
      http://en.wikipedia.org/wiki/Wikipedia:Today%27s_featured_picture_%28animation%29/January_20%2C_2007: [OK]
      http://www.mail-archive.com/rails@lists.rubyonrails.org/: [OK]
      http://www.amazon.com/Testing-Equal-Sign-In-Path/ref=pd_bbs_sr_1?ie=UTF8&s=books&qid=1198861734&sr=8-1: [OK]
      http://en.wikipedia.org/wiki/Texas_hold%27em: [OK]
      https://www.google.com/doku.php?id=gps:resource:scs:start: [OK]
      http://www.rubyonrails.com: [OK]
      http://manuals.ruby-on-rails.com/read/chapter.need_a-period/103#page281: [OK]
      http://foo.example.com/controller/action?parm=value&p2=v2#anchor123: [OK]
      http://foo.example.com:3000/controller/action: [OK]
      http://foo.example.com:3000/controller/action+pack: [OK]
      http://business.timesonline.co.uk/article/0,,9065-2473189,00.html: [OK]
      http://www.mail-archive.com/ruby-talk@ruby-lang.org/: [OK]
    Headers:
      blank line before header: [OK]
    smart punctuation:
      quote before ellipses: [OK]
      apostrophe before emph: [OK]
      apostrophe in French: [OK]
    footnotes:
      indent followed by newline and flush-left text: [OK]
      indent followed by newline and indented text: [OK]
      recursive note: [OK]
    lhs:
      inverse bird tracks and html: [OK]
    definition lists:
      no blank space: [OK]
      blank space before first def: [OK]
      blank space before second def: [OK]
      laziness: [OK]
      no blank space before first of two paragraphs: [OK]
    +compact_definition_lists:
      basic compact list: [OK]
    lists:
      issue #1154: [OK]
  Org:
    Inlines:
      Plain String: [OK]
      Emphasis: [OK]
      Strong: [OK]
      Strong Emphasis: [OK]
      Strikeout: [OK]
      Verbatim: [OK]
      Code: [OK]
      Math $..$: [OK]
      Math $$..$$: [OK]
      Math \[..\]: [OK]
      Math \(..\): [OK]
      Symbol: [OK]
      Superscript simple expression: [OK]
      Superscript multi char: [OK]
      Subscript simple expression: [OK]
      Subscript multi char: [OK]
      Linebreak: [OK]
      Inline note: [OK]
      Markup-chars not occuring on word break are symbols: [OK]
      No empty markup: [OK]
      Adherence to Org's rules for markup borders: [OK]
      Inline math must stay within three lines: [OK]
      Single-character math: [OK]
      Markup may not span more than two lines: [OK]
      Sub- and superscript expressions: [OK]
      Image: [OK]
      Explicit link: [OK]
      Self-link: [OK]
      Image link: [OK]
      Plain link: [OK]
      Angle link: [OK]
      Anchor: [OK]
      Inline code block: [OK]
      Inline code block with arguments: [OK]
      Citation: [OK]
      Citation containing text: [OK]
      Inline LaTeX symbol: [OK]
      Inline LaTeX command: [OK]
      Inline LaTeX math symbol: [OK]
      Unknown inline LaTeX command: [OK]
      LaTeX citation: [OK]
    Meta Information:
      Comment: [OK]
      Not a comment: [OK]
      Comment surrounded by Text: [OK]
      Title: [OK]
      Author: [OK]
      Date: [OK]
      Description: [OK]
      Properties drawer: [OK]
      Logbook drawer: [OK]
      Drawer surrounded by text: [OK]
      Drawer start is the only text in first line of a drawer: [OK]
      Drawers with unknown names are just text: [OK]
      Anchor reference: [OK]
      Search links are read as emph: [OK]
      Link to nonexistent anchor: [OK]
      Link abbreviation: [OK]
      Link abbreviation, defined after first use: [OK]
      Link abbreviation, URL encoded arguments: [OK]
      Link abbreviation, append arguments: [OK]
    Basic Blocks:
      Paragraph: [OK]
      First Level Header: [OK]
      Third Level Header: [OK]
      Compact Headers with Paragraph: [OK]
      Separated Headers with Paragraph: [OK]
      Headers not preceded by a blank line: [OK]
      Paragraph starting with an asterisk: [OK]
      Paragraph containing asterisk at beginning of line: [OK]
      Example block: [OK]
      Example block surrounded by text: [OK]
      Horizontal Rule: [OK]
      Not a Horizontal Rule: [OK]
      Comment Block: [OK]
      Figure: [OK]
      Unnamed figure: [OK]
      Figure with `fig:` prefix in name: [OK]
      Footnote: [OK]
      Two footnotes: [OK]
      Footnote followed by header: [OK]
    Lists:
      Simple Bullet Lists: [OK]
      Indented Bullet Lists: [OK]
      Multi-line Bullet Lists: [OK]
      Nested Bullet Lists: [OK]
      Simple Ordered List: [OK]
      Simple Ordered List with Parens: [OK]
      Indented Ordered List: [OK]
      Nested Ordered Lists: [OK]
      Ordered List in Bullet List: [OK]
      Bullet List in Ordered List: [OK]
      Definition List: [OK]
      Compact definition list: [OK]
      Loose bullet list: [OK]
    Tables:
      Single cell table: [OK]
      Multi cell table: [OK]
      Multi line table: [OK]
      Empty table: [OK]
      Glider Table: [OK]
      Table between Paragraphs: [OK]
      Table with Header: [OK]
      Table with final hline: [OK]
      Table in a box: [OK]
      Table with alignment row: [OK]
      Pipe within text doesn't start a table: [OK]
      Missing pipe at end of row: [OK]
      Table with differing row lengths: [OK]
      Table with caption: [OK]
    Blocks and fragments:
      Source block: [OK]
      Source block between paragraphs: [OK]
      Source block with rundoc/babel arguments: [OK]
      Source block with results and :exports both: [OK]
      Source block with results and :exports code: [OK]
      Source block with results and :exports results: [OK]
      Source block with results and :exports none: [OK]
      Example block: [OK]
      HTML block: [OK]
      Quote block: [OK]
      Verse block: [OK]
      LaTeX fragment: [OK]
      Code block with caption: [OK]
      Convert blank lines in blocks to single newlines: [OK]
      Non-letter chars in source block parameters: [OK]
  RST:
    line block with blank line: [OK]
    field list: [OK]
    initial field list: [OK]
    URLs with following punctuation: [OK]
  Docx:
    inlines:
      font formatting: [OK]
      font formatting with character styles: [OK]
      hyperlinks: [OK]
      inline image: [OK]
      inline image in links: [OK]
      handling unicode input: [OK]
      literal tabs: [OK]
      normalizing inlines: [OK]
      normalizing inlines deep inside blocks: [OK]
      move trailing spaces outside of formatting: [OK]
      inline code (with VerbatimChar style): [OK]
    blocks:
      headers: [OK]
      headers already having auto identifiers: [OK]
      lists: [OK]
      definition lists: [OK]
      footnotes and endnotes: [OK]
      blockquotes (parsing indent as blockquote): [OK]
      hanging indents: [OK]
      tables: [OK]
      code block: [OK]
      dropcap paragraphs: [OK]
    track changes:
      insertion (default): [OK]
      insert insertion (accept): [OK]
      remove insertion (reject): [OK]
      deletion (default): [OK]
      remove deletion (accept): [OK]
      insert deletion (reject): [OK]
      keep insertion (all): [OK]
      keep deletion (all): [OK]
    media:
      image extraction: [OK]
    metadata:
      metadata fields: [OK]
      stop recording metadata with normal text: [OK]
  Txt2Tags:
    Inlines:
      Plain String: [OK]
      Emphasis: [OK]
      Strong: [OK]
      Strong Emphasis: [OK]
      Strikeout: [OK]
      Verbatim: [OK]
      Symbol: [OK]
      No empty markup: [OK]
      Inline markup is greedy: [OK]
      Markup must be greedy: [OK]
      Inlines must be glued: [OK]
      Macros: Date: [OK]
      Macros: Mod Time: [OK]
      Macros: Infile: [OK]
      Macros: Outfile: [OK]
      Autolink: [OK]
      Image: [OK]
      Link: [OK]
      Image link: [OK]
      Invalid link: No trailing space: [OK]
    Basic Blocks:
      Paragraph, lines grouped together: [OK]
      Paragraph, ignore leading and trailing spaces: [OK]
      Comment line in paragraph: [OK]
      Paragraph: [OK]
      First Level Header: [OK]
      Third Level Header: [OK]
      Header with label: [OK]
      Invalid header, mismatched delimiters: [OK]
      Invalid header, spaces in label: [OK]
      Invalid header, invalid label character: [OK]
      Headers not preceded by a blank line: [OK]
      Paragraph starting with an equals: [OK]
      Paragraph containing asterisk at beginning of line: [OK]
      Horizontal Rule: [OK]
      Comment Block: [OK]
    Lists:
      Simple Bullet Lists: [OK]
      Indented Bullet Lists: [OK]
      Nested Bullet Lists: [OK]
      Simple Ordered List: [OK]
      Indented Ordered List: [OK]
      Nested Ordered Lists: [OK]
      Ordered List in Bullet List: [OK]
      Bullet List in Ordered List: [OK]
      Definition List: [OK]
      Loose bullet list: [OK]
    Tables:
      Single cell table: [OK]
      Multi cell table: [OK]
      Multi line table: [OK]
      Empty table: [OK]
      Glider Table: [OK]
      Table with Header: [OK]
      Table alignment determined by spacing: [OK]
      Pipe within text doesn't start a table: [OK]
      Table with differing row lengths: [OK]
    Blocks and fragments:
      Source block: [OK]
      tagged block: [OK]
      Quote block: [OK]
  EPUB:
    EPUB Mediabag:
      features bag: [Failed]
"MediaBag did not match:\nExpected: "[("img/ElementaryMathExample.png","image/png",1331),("img/Maghreb1.png","image/png",2520),("img/check.gif","image/gif",1340),("img/check.jpg","image/jpeg",2661),("img/check.png","image/png",2815),("img/cichons_diagram.png","image/png",7045),("img/complex_number.png","image/png",5238),("img/multiscripts_and_greek_alphabet.png","image/png",10060)]
Actual: [("img/multiscripts_and_greek_alphabet.png","image/png",10060),("img\\ElementaryMathExample.png","image/png",1331),("img\\Maghreb1.png","image/png",2520),("img\\check.gif","image/gif",1340),("img\\check.jpg","image/jpeg",2661),("img\\check.png","image/png",2815),("img\\cichons_diagram.png","image/png",7045),("img\\complex_number.png","image/png",5238)]

         Properties  Test Cases  Total 
 Passed  10          447         457   
 Failed  0           9           9     
 Total   10          456         466   
Test suite test-pandoc: FAIL
Test suite logged to: dist\test\pandoc.windows.log

[jkr]

Looks like the failures are all in the epub reader and the implementation of collapsePath there (i.e. we don't want os-sensitive slash-direction for an archive path). I'll put mine up as a PR and ping @mpickering to see if he'll be able to fix those.

[mpickering]

I think a better solution than cryptically interspersing "/" would either be to make the insertItem function do so in the MediaBag module or use the split path as a key ([String]]) rather than by convention saying we use "/" as a separator as it has to be changed to be platform agnostic anyway when we write the contents.

@jkr thoughts?

[jkr]

Sure -- I'd be okay with that too. Not sure I see the need, though. We
know that the path separator inside an archive is a forward-slash,
regardless of os, so it doesn't seem particularly cryptic to me. And
right now, writeMedia takes care of the conversion.

Witht the simple fix, we'd have:

(string + "/"-sep'd string) ->
("/"-sep'd string) ->
(os-specific string)

You seem to be suggesting

(string + "/"-sep'd string) ->
(string + [string]) ->
([string]) ->
os-specific string

It seems fine, and I'm happy to switch to it. But I don't see any
particular gain in generality. Especially because the splitting into a
list and rejoining is currently done inside writeMedia. (I don't see
any loss in generality either, though.)

[mpickering]

The problem I foresee with this convention is that someone unfamiliar may assume that it is fine to make the key for MediaBag with the functions from System.FilePath (as you should construct any path) and the compiler wouldn't complain but we would get the wrong result.

Approach 1: The programmer can construct the path with the functions as expected and the insertion function deals with this os-specific quirk.

Approach 2: we make a distinction between a path and key for the mediabag so it's clear to the programmer that they are different things.

#1 is cleaner I think but the fact that FilePath is just a synonym for String is just annoying. It would be better to use a filepath library with a stronger idea of types to make the os distinction invisible until trying to print paths.

[mpickering]

There were actually three problems here, I will describe them for posterity.

1. collapseFilePath assumed unix separators.
2. FilePaths with different seperators were being inserted into the MediaBag
3. Images were written out with different path separators (which is the correct behaviour) but this caused the test to fail on windows.