From cf3352c8a224af219cb062cfe467d7da9289284a Mon Sep 17 00:00:00 2001 From: Chris Sewell Date: Fri, 28 Feb 2020 18:01:24 +1100 Subject: [PATCH] Add performance benchmark (#88) Add CLI and documentation, benchmarking `myst-parser` against other Python based markdown parsers. --- .vscode/settings.json | 7 +- docs/conf.py | 1 + docs/using/benchmark.md | 1000 +++++++++++++++++++++++++ docs/using/index.md | 1 + docs/using/use_api.md | 18 + docstring.fmt.mustache | 20 + myst_parser/__init__.py | 23 + myst_parser/block_tokens.py | 27 +- myst_parser/cli/__init__.py | 0 myst_parser/cli/benchmark.py | 113 +++ myst_parser/docutils_renderer.py | 15 +- setup.py | 4 +- tests/test_cli.py | 11 + tests/test_renderers/test_docutils.py | 13 +- 14 files changed, 1236 insertions(+), 17 deletions(-) create mode 100644 docs/using/benchmark.md create mode 100644 docstring.fmt.mustache create mode 100644 myst_parser/cli/__init__.py create mode 100644 myst_parser/cli/benchmark.py create mode 100644 tests/test_cli.py diff --git a/.vscode/settings.json b/.vscode/settings.json index 2ee89ab6..db5a5166 100644 --- a/.vscode/settings.json +++ b/.vscode/settings.json @@ -8,9 +8,12 @@ "**/__pycache__": true, "**/.pytest_cache": true }, - "editor.rulers": [88], + "editor.rulers": [ + 88 + ], "python.formatting.provider": "black", "python.linting.pylintEnabled": false, "python.linting.flake8Enabled": true, "python.linting.enabled": true, -} + "autoDocstring.customTemplatePath": "docstring.fmt.mustache" +} \ No newline at end of file diff --git a/docs/conf.py b/docs/conf.py index f58a5ea5..3fdc5f7f 100644 --- a/docs/conf.py +++ b/docs/conf.py @@ -117,6 +117,7 @@ def run_apidoc(app): ("py:class", "mistletoe.block_token.Table"), ("py:class", "mistletoe.block_token.Footnote"), ("py:class", "mistletoe.block_token.Paragraph"), + ("py:class", "mistletoe.block_token.ThematicBreak"), ("py:class", "mistletoe.base_renderer.BaseRenderer"), ("py:class", "mistletoe.html_renderer.HTMLRenderer"), ("py:class", "mistletoe.span_token.SpanToken"), diff --git a/docs/using/benchmark.md b/docs/using/benchmark.md new file mode 100644 index 00000000..9cfb8f7c --- /dev/null +++ b/docs/using/benchmark.md @@ -0,0 +1,1000 @@ +Parsing Performance Benchmark +============================= + +The following document describes Markdown syntax, [as written by John Gruber][src]. +It is used to benchmark the parsing speed of the MyST-Parser against +some existing Markdown parsers written in Python: + + [src]: https://daringfireball.net/projects/markdown/syntax + + $ myst-benchmark docs/using/benchmark.md + Test document: benchmark.md + Test iterations: 1000 + Running tests ... + ================= + python-markdown:extra (3.2): 53.78 s + mistune (0.8.4): 15.22 s + commonmark.py (0.9.1): 61.92 s + mistletoe (0.8.0): 43.51 s + myst_parser:html (0.2.0): 52.47 s + myst_parser:docutils (0.2.0): 65.27 s + myst_parser:sphinx (0.2.0): 73.54 s + +As already noted by [mistletoe](https://github.com/miyuchina/mistletoe#performance) +(which this package is built on), although Mistune is the fastest, +this is because it does not strictly follow the CommonMark spec, +which outlines a highly context-sensitive grammar for Markdown. +The simpler approach taken by Mistune means that it cannot handle more +complex parsing cases, such as precedence of different types of tokens, escaping rules, etc. + +The MyST parser is slightly slower than the base mistletoe parser, due to the additional +syntax which it parses. Then the conversion to docutils AST takes some more time, +but is still comparably performant to the core CommonMark.py parser. The sphinx +parse takes some extra time, due to loading the full sphinx environment, +including its roles and directives. + +## Contents + +* [Overview](#overview) + * [Philosophy](#philosophy) + * [Inline HTML](#html) + * [Automatic Escaping for Special Characters](#autoescape) +* [Block Elements](#block) + * [Paragraphs and Line Breaks](#p) + * [Headers](#header) + * [Blockquotes](#blockquote) + * [Lists](#list) + * [Code Blocks](#precode) + * [Horizontal Rules](#hr) +* [Span Elements](#span) + * [Links](#link) + * [Emphasis](#em) + * [Code](#code) + * [Images](#img) +* [Miscellaneous](#misc) + * [Backslash Escapes](#backslash) + * [Automatic Links](#autolink) + +* * * + +

Overview

+ +

Philosophy

+ +Markdown is intended to be as easy-to-read and easy-to-write as is feasible. + +Readability, however, is emphasized above all else. A Markdown-formatted +document should be publishable as-is, as plain text, without looking +like it's been marked up with tags or formatting instructions. While +Markdown's syntax has been influenced by several existing text-to-HTML +filters -- including [Setext] [1], [atx] [2], [Textile] [3], [reStructuredText] [4], +[Grutatext] [5], and [EtText] [6] -- the single biggest source of +inspiration for Markdown's syntax is the format of plain text email. + + [1]: http://docutils.sourceforge.net/mirror/setext.html + [2]: http://www.aaronsw.com/2002/atx/ + [3]: http://textism.com/tools/textile/ + [4]: http://docutils.sourceforge.net/rst.html + [5]: http://www.triptico.com/software/grutatxt.html + [6]: http://ettext.taint.org/doc/ + +To this end, Markdown's syntax is comprised entirely of punctuation +characters, which punctuation characters have been carefully chosen so +as to look like what they mean. E.g., asterisks around a word actually +look like \*emphasis\*. Markdown lists look like, well, lists. Even +blockquotes look like quoted passages of text, assuming you've ever +used email. + +

Inline HTML

+ +Markdown's syntax is intended for one purpose: to be used as a +format for *writing* for the web. + +Markdown is not a replacement for HTML, or even close to it. Its +syntax is very small, corresponding only to a very small subset of +HTML tags. The idea is *not* to create a syntax that makes it easier +to insert HTML tags. In my opinion, HTML tags are already easy to +insert. The idea for Markdown is to make it easy to read, write, and +edit prose. HTML is a *publishing* format; Markdown is a *writing* +format. Thus, Markdown's formatting syntax only addresses issues that +can be conveyed in plain text. + +For any markup that is not covered by Markdown's syntax, you simply +use HTML itself. There's no need to preface it or delimit it to +indicate that you're switching from Markdown to HTML; you just use +the tags. + +The only restrictions are that block-level HTML elements -- e.g. `
`, +``, `
`, `
`, etc. -- must be separated from surrounding
+content by blank lines, and the start and end tags of the block should
+not be indented with tabs or spaces. Markdown is smart enough not
+to add extra (unwanted) `
` tags around HTML block-level tags.
+
+For example, to add an HTML table to a Markdown article:
+
+    This is a regular paragraph.
+
+
+ + + +
Foo
+ + This is another regular paragraph. + +Note that Markdown formatting syntax is not processed within block-level +HTML tags. E.g., you can't use Markdown-style `*emphasis*` inside an +HTML block. + +Span-level HTML tags -- e.g. ``, ``, or `` -- can be +used anywhere in a Markdown paragraph, list item, or header. If you +want, you can even use HTML tags instead of Markdown formatting; e.g. if +you'd prefer to use HTML `` or `` tags instead of Markdown's +link or image syntax, go right ahead. + +Unlike block-level HTML tags, Markdown syntax *is* processed within +span-level tags. + + +

Automatic Escaping for Special Characters

+ +In HTML, there are two characters that demand special treatment: `<` +and `&`. Left angle brackets are used to start tags; ampersands are +used to denote HTML entities. If you want to use them as literal +characters, you must escape them as entities, e.g. `<`, and +`&`. + +Ampersands in particular are bedeviling for web writers. If you want to +write about 'AT&T', you need to write '`AT&T`'. You even need to +escape ampersands within URLs. Thus, if you want to link to: + + http://images.google.com/images?num=30&q=larry+bird + +you need to encode the URL as: + + http://images.google.com/images?num=30&q=larry+bird + +in your anchor tag `href` attribute. Needless to say, this is easy to +forget, and is probably the single most common source of HTML validation +errors in otherwise well-marked-up web sites. + +Markdown allows you to use these characters naturally, taking care of +all the necessary escaping for you. If you use an ampersand as part of +an HTML entity, it remains unchanged; otherwise it will be translated +into `&`. + +So, if you want to include a copyright symbol in your article, you can write: + + © + +and Markdown will leave it alone. But if you write: + + AT&T + +Markdown will translate it to: + + AT&T + +Similarly, because Markdown supports [inline HTML](#html), if you use +angle brackets as delimiters for HTML tags, Markdown will treat them as +such. But if you write: + + 4 < 5 + +Markdown will translate it to: + + 4 < 5 + +However, inside Markdown code spans and blocks, angle brackets and +ampersands are *always* encoded automatically. This makes it easy to use +Markdown to write about HTML code. (As opposed to raw HTML, which is a +terrible format for writing about HTML syntax, because every single `<` +and `&` in your example code needs to be escaped.) + + +* * * + + +

Block Elements

+ + +

Paragraphs and Line Breaks

+ +A paragraph is simply one or more consecutive lines of text, separated +by one or more blank lines. (A blank line is any line that looks like a +blank line -- a line containing nothing but spaces or tabs is considered +blank.) Normal paragraphs should not be indented with spaces or tabs. + +The implication of the "one or more consecutive lines of text" rule is +that Markdown supports "hard-wrapped" text paragraphs. This differs +significantly from most other text-to-HTML formatters (including Movable +Type's "Convert Line Breaks" option) which translate every line break +character in a paragraph into a `
` tag. + +When you *do* want to insert a `
` break tag using Markdown, you +end a line with two or more spaces, then type return. + +Yes, this takes a tad more effort to create a `
`, but a simplistic +"every line break is a `
`" rule wouldn't work for Markdown. +Markdown's email-style [blockquoting][bq] and multi-paragraph [list items][l] +work best -- and look better -- when you format them with hard breaks. + + [bq]: #blockquote + [l]: #list + + + + + +Markdown supports two styles of headers, [Setext] [1] and [atx] [2]. + +Setext-style headers are "underlined" using equal signs (for first-level +headers) and dashes (for second-level headers). For example: + + This is an H1 + ============= + + This is an H2 + ------------- + +This is an H1 +============= + +This is an H2 +------------- + +Any number of underlining `=`'s or `-`'s will work. + +Atx-style headers use 1-6 hash characters at the start of the line, +corresponding to header levels 1-6. For example: + + # This is an H1 + + ## This is an H2 + + ###### This is an H6 + +Optionally, you may "close" atx-style headers. This is purely +cosmetic -- you can use this if you think it looks better. The +closing hashes don't even need to match the number of hashes +used to open the header. (The number of opening hashes +determines the header level.) : + + # This is an H1 # + + ## This is an H2 ## + + ### This is an H3 ###### + + +

Blockquotes

+ +Markdown uses email-style `>` characters for blockquoting. If you're +familiar with quoting passages of text in an email message, then you +know how to create a blockquote in Markdown. It looks best if you hard +wrap the text and put a `>` before every line: + + > This is a blockquote with two paragraphs. Lorem ipsum dolor sit amet, + > consectetuer adipiscing elit. Aliquam hendrerit mi posuere lectus. + > Vestibulum enim wisi, viverra nec, fringilla in, laoreet vitae, risus. + > + > Donec sit amet nisl. Aliquam semper ipsum sit amet velit. Suspendisse + > id sem consectetuer libero luctus adipiscing. + +* * * + +> This is a blockquote with two paragraphs. Lorem ipsum dolor sit amet, +> consectetuer adipiscing elit. Aliquam hendrerit mi posuere lectus. +> Vestibulum enim wisi, viverra nec, fringilla in, laoreet vitae, risus. +> +> Donec sit amet nisl. Aliquam semper ipsum sit amet velit. Suspendisse +> id sem consectetuer libero luctus adipiscing. + +* * * + +Markdown allows you to be lazy and only put the `>` before the first +line of a hard-wrapped paragraph: + + > This is a blockquote with two paragraphs. Lorem ipsum dolor sit amet, + consectetuer adipiscing elit. Aliquam hendrerit mi posuere lectus. + Vestibulum enim wisi, viverra nec, fringilla in, laoreet vitae, risus. + + > Donec sit amet nisl. Aliquam semper ipsum sit amet velit. Suspendisse + id sem consectetuer libero luctus adipiscing. + +* * * + +> This is a blockquote with two paragraphs. Lorem ipsum dolor sit amet, +consectetuer adipiscing elit. Aliquam hendrerit mi posuere lectus. +Vestibulum enim wisi, viverra nec, fringilla in, laoreet vitae, risus. + +> Donec sit amet nisl. Aliquam semper ipsum sit amet velit. Suspendisse +id sem consectetuer libero luctus adipiscing. + +* * * + +Blockquotes can be nested (i.e. a blockquote-in-a-blockquote) by +adding additional levels of `>`: + + > This is the first level of quoting. + > + > > This is nested blockquote. + > + > Back to the first level. + +* * * + +> This is the first level of quoting. +> +> > This is nested blockquote. +> +> Back to the first level. + +* * * + +Blockquotes can contain other Markdown elements, including headers, lists, +and code blocks: + + > ## This is a header. + > + > 1. This is the first list item. + > 2. This is the second list item. + > + > Here's some example code: + > + > return shell_exec("echo $input | $markdown_script"); + +Any decent text editor should make email-style quoting easy. For +example, with BBEdit, you can make a selection and choose Increase +Quote Level from the Text menu. + + +

Lists

+ +Markdown supports ordered (numbered) and unordered (bulleted) lists. + +Unordered lists use asterisks, pluses, and hyphens -- interchangably +-- as list markers: + + * Red + * Green + * Blue + +* * * + +* Red +* Green +* Blue + +* * * + +is equivalent to: + + + Red + + Green + + Blue + +* * * + ++ Red ++ Green ++ Blue + +* * * + +and: + + - Red + - Green + - Blue + +* * * + +- Red +- Green +- Blue + +* * * + +Ordered lists use numbers followed by periods: + + 1. Bird + 2. McHale + 3. Parish + +* * * + +1. Bird +2. McHale +3. Parish + +* * * + +It's important to note that the actual numbers you use to mark the +list have no effect on the HTML output Markdown produces. The HTML +Markdown produces from the above list is: + +
    +
  1. Bird
    2. +
  2. McHale
    3. +
  3. Parish
    4. +
+ +If you instead wrote the list in Markdown like this: + + 1. Bird + 2. McHale + 3. Parish + +or even: + + 3. Bird + 1. McHale + 8. Parish + +* * * + +1. Bird +2. McHale +3. Parish + +* * * + +you'd get the exact same HTML output. The point is, if you want to, +you can use ordinal numbers in your ordered Markdown lists, so that +the numbers in your source match the numbers in your published HTML. +But if you want to be lazy, you don't have to. + +If you do use lazy list numbering, however, you should still start the +list with the number 1. At some point in the future, Markdown may support +starting ordered lists at an arbitrary number. + +List markers typically start at the left margin, but may be indented by +up to three spaces. List markers must be followed by one or more spaces +or a tab. + +To make lists look nice, you can wrap items with hanging indents: + + * Lorem ipsum dolor sit amet, consectetuer adipiscing elit. + Aliquam hendrerit mi posuere lectus. Vestibulum enim wisi, + viverra nec, fringilla in, laoreet vitae, risus. + * Donec sit amet nisl. Aliquam semper ipsum sit amet velit. + Suspendisse id sem consectetuer libero luctus adipiscing. + +But if you want to be lazy, you don't have to: + + * Lorem ipsum dolor sit amet, consectetuer adipiscing elit. + Aliquam hendrerit mi posuere lectus. Vestibulum enim wisi, + viverra nec, fringilla in, laoreet vitae, risus. + * Donec sit amet nisl. Aliquam semper ipsum sit amet velit. + Suspendisse id sem consectetuer libero luctus adipiscing. + +If list items are separated by blank lines, Markdown will wrap the +items in `

` tags in the HTML output. For example, this input: + + * Bird + * Magic + +will turn into: + +

    +
  • Bird
    • +
  • Magic
    • +
+ +But this: + + * Bird + + * Magic + +will turn into: + +
    +

  • Bird

    • +

  • Magic

    • +
+ +List items may consist of multiple paragraphs. Each subsequent +paragraph in a list item must be indented by either 4 spaces +or one tab: + + 1. This is a list item with two paragraphs. Lorem ipsum dolor + sit amet, consectetuer adipiscing elit. Aliquam hendrerit + mi posuere lectus. + + Vestibulum enim wisi, viverra nec, fringilla in, laoreet + vitae, risus. Donec sit amet nisl. Aliquam semper ipsum + sit amet velit. + + 2. Suspendisse id sem consectetuer libero luctus adipiscing. + +It looks nice if you indent every line of the subsequent +paragraphs, but here again, Markdown will allow you to be +lazy: + + * This is a list item with two paragraphs. + + This is the second paragraph in the list item. You're + only required to indent the first line. Lorem ipsum dolor + sit amet, consectetuer adipiscing elit. + + * Another item in the same list. + +To put a blockquote within a list item, the blockquote's `>` +delimiters need to be indented: + + * A list item with a blockquote: + + > This is a blockquote + > inside a list item. + +* * * + +* A list item with a blockquote: + + > This is a blockquote + > inside a list item. + +* * * + +To put a code block within a list item, the code block needs +to be indented *twice* -- 8 spaces or two tabs: + + * A list item with a code block: + + + + +It's worth noting that it's possible to trigger an ordered list by +accident, by writing something like this: + + 1. What a great season. + +In other words, a *number-period-space* sequence at the beginning of a +line. To avoid this, you can backslash-escape the period: + + 1986\. What a great season. + + + +

Code Blocks

+ +Pre-formatted code blocks are used for writing about programming or +markup source code. Rather than forming normal paragraphs, the lines +of a code block are interpreted literally. Markdown wraps a code block +in both `
` and `` tags.
+
+To produce a code block in Markdown, simply indent every line of the
+block by at least 4 spaces or 1 tab. For example, given this input:
+
+    This is a normal paragraph:
+
+        This is a code block.
+
+Markdown will generate:
+
+    
This is a normal paragraph:

+
+    
This is a code block.
+    

+
+One level of indentation -- 4 spaces or 1 tab -- is removed from each
+line of the code block. For example, this:
+
+    Here is an example of AppleScript:
+
+        tell application "Foo"
+            beep
+        end tell
+
+will turn into:
+
+    
Here is an example of AppleScript:

+
+    
tell application "Foo"
+        beep
+    end tell
+    

+
+A code block continues until it reaches a line that is not indented
+(or the end of the article).
+
+Within a code block, ampersands (`&`) and angle brackets (`<` and `>`)
+are automatically converted into HTML entities. This makes it very
+easy to include example HTML source code using Markdown -- just paste
+it and indent it, and Markdown will handle the hassle of encoding the
+ampersands and angle brackets. For example, this:
+
+        

+            © 2004 Foo Corporation
+        

+
+will turn into:
+
+    
<div class="footer">
+        &copy; 2004 Foo Corporation
+    </div>
+    

+
+Regular Markdown syntax is not processed within code blocks. E.g.,
+asterisks are just literal asterisks within a code block. This means
+it's also easy to use Markdown to write about Markdown's own syntax.
+
+
+
+
Horizontal Rules

+
+You can produce a horizontal rule tag (`
`) by placing three or
+more hyphens, asterisks, or underscores on a line by themselves. If you
+wish, you may use spaces between the hyphens or asterisks. Each of the
+following lines will produce a horizontal rule:
+
+    * * *
+
+    ***
+
+    *****
+
+    - - -
+
+    ---------------------------------------
+
+
+* * *
+
+
Span Elements

+
+
+
+Markdown supports two style of links: *inline* and *reference*.
+
+In both styles, the link text is delimited by [square brackets].
+
+To create an inline link, use a set of regular parentheses immediately
+after the link text's closing square bracket. Inside the parentheses,
+put the URL where you want the link to point, along with an *optional*
+title for the link, surrounded in quotes. For example:
+
+    This is [an example](http://example.com/ "Title") inline link.
+
+    [This link](http://example.net/) has no title attribute.
+
+Will produce:
+
+    
This is 
+    an example inline link.

+
+    
This link has no
+    title attribute.

+
+If you're referring to a local resource on the same server, you can
+use relative paths:
+
+    See my [About](/about/) page for details.
+
+Reference-style links use a second set of square brackets, inside
+which you place a label of your choosing to identify the link:
+
+    This is [an example][id] reference-style link.
+
+You can optionally use a space to separate the sets of brackets:
+
+    This is [an example] [id] reference-style link.
+
+Then, anywhere in the document, you define your link label like this,
+on a line by itself:
+
+    [id]: http://example.com/  "Optional Title Here"
+
+That is:
+
+*   Square brackets containing the link identifier (optionally
+    indented from the left margin using up to three spaces);
+*   followed by a colon;
+*   followed by one or more spaces (or tabs);
+*   followed by the URL for the link;
+*   optionally followed by a title attribute for the link, enclosed
+    in double or single quotes, or enclosed in parentheses.
+
+The following three link definitions are equivalent:
+
+	[foo]: http://example.com/  "Optional Title Here"
+	[foo]: http://example.com/  'Optional Title Here'
+	[foo]: http://example.com/  (Optional Title Here)
+
+**Note:** There is a known bug in Markdown.pl 1.0.1 which prevents
+single quotes from being used to delimit link titles.
+
+The link URL may, optionally, be surrounded by angle brackets:
+
+    [id]:   "Optional Title Here"
+
+You can put the title attribute on the next line and use extra spaces
+or tabs for padding, which tends to look better with longer URLs:
+
+    [id]: http://example.com/longish/path/to/resource/here
+        "Optional Title Here"
+
+Link definitions are only used for creating links during Markdown
+processing, and are stripped from your document in the HTML output.
+
+Link definition names may consist of letters, numbers, spaces, and
+punctuation -- but they are *not* case sensitive. E.g. these two
+links:
+
+	[link text][a]
+	[link text][A]
+
+are equivalent.
+
+The *implicit link name* shortcut allows you to omit the name of the
+link, in which case the link text itself is used as the name.
+Just use an empty set of square brackets -- e.g., to link the word
+"Google" to the google.com web site, you could simply write:
+
+	[Google][]
+
+And then define the link:
+
+	[Google]: http://google.com/
+
+Because link names may contain spaces, this shortcut even works for
+multiple words in the link text:
+
+	Visit [Daring Fireball][] for more information.
+
+And then define the link:
+
+	[Daring Fireball]: http://daringfireball.net/
+
+Link definitions can be placed anywhere in your Markdown document. I
+tend to put them immediately after each paragraph in which they're
+used, but if you want, you can put them all at the end of your
+document, sort of like footnotes.
+
+Here's an example of reference links in action:
+
+    I get 10 times more traffic from [Google] [1] than from
+    [Yahoo] [2] or [MSN] [3].
+
+      [1]: http://google.com/        "Google"
+      [2]: http://search.yahoo.com/  "Yahoo Search"
+      [3]: http://search.msn.com/    "MSN Search"
+
+Using the implicit link name shortcut, you could instead write:
+
+    I get 10 times more traffic from [Google][] than from
+    [Yahoo][] or [MSN][].
+
+      [google]: http://google.com/        "Google"
+      [yahoo]:  http://search.yahoo.com/  "Yahoo Search"
+      [msn]:    http://search.msn.com/    "MSN Search"
+
+Both of the above examples will produce the following HTML output:
+
+    
I get 10 times more traffic from Google than from
+    Yahoo
+    or MSN.

+
+For comparison, here is the same paragraph written using
+Markdown's inline link style:
+
+    I get 10 times more traffic from [Google](http://google.com/ "Google")
+    than from [Yahoo](http://search.yahoo.com/ "Yahoo Search") or
+    [MSN](http://search.msn.com/ "MSN Search").
+
+The point of reference-style links is not that they're easier to
+write. The point is that with reference-style links, your document
+source is vastly more readable. Compare the above examples: using
+reference-style links, the paragraph itself is only 81 characters
+long; with inline-style links, it's 176 characters; and as raw HTML,
+it's 234 characters. In the raw HTML, there's more markup than there
+is text.
+
+With Markdown's reference-style links, a source document much more
+closely resembles the final output, as rendered in a browser. By
+allowing you to move the markup-related metadata out of the paragraph,
+you can add links without interrupting the narrative flow of your
+prose.
+
+
+
Emphasis

+
+Markdown treats asterisks (`*`) and underscores (`_`) as indicators of
+emphasis. Text wrapped with one `*` or `_` will be wrapped with an
+HTML `` tag; double `*`'s or `_`'s will be wrapped with an HTML
+`` tag. E.g., this input:
+
+    *single asterisks*
+
+    _single underscores_
+
+    **double asterisks**
+
+    __double underscores__
+
+will produce:
+
+    single asterisks
+
+    single underscores
+
+    double asterisks
+
+    double underscores
+
+You can use whichever style you prefer; the lone restriction is that
+the same character must be used to open and close an emphasis span.
+
+Emphasis can be used in the middle of a word:
+
+    un*frigging*believable
+
+But if you surround an `*` or `_` with spaces, it'll be treated as a
+literal asterisk or underscore.
+
+To produce a literal asterisk or underscore at a position where it
+would otherwise be used as an emphasis delimiter, you can backslash
+escape it:
+
+    \*this text is surrounded by literal asterisks\*
+
+
+
+
Code

+
+To indicate a span of code, wrap it with backtick quotes (`` ` ``).
+Unlike a pre-formatted code block, a code span indicates code within a
+normal paragraph. For example:
+
+    Use the `printf()` function.
+
+will produce:
+
+    
Use the printf() function.

+
+To include a literal backtick character within a code span, you can use
+multiple backticks as the opening and closing delimiters:
+
+    ``There is a literal backtick (`) here.``
+
+which will produce this:
+
+    
There is a literal backtick (`) here.

+
+The backtick delimiters surrounding a code span may include spaces --
+one after the opening, one before the closing. This allows you to place
+literal backtick characters at the beginning or end of a code span:
+
+	A single backtick in a code span: `` ` ``
+
+	A backtick-delimited string in a code span: `` `foo` ``
+
+will produce:
+
+	
A single backtick in a code span: `

+
+	
A backtick-delimited string in a code span: `foo`

+
+With a code span, ampersands and angle brackets are encoded as HTML
+entities automatically, which makes it easy to include example HTML
+tags. Markdown will turn this:
+
+    Please don't use any `` tags.
+
+into:
+
+    
Please don't use any <blink> tags.

+
+You can write this:
+
+    `—` is the decimal-encoded equivalent of `—`.
+
+to produce:
+
+    
&#8212; is the decimal-encoded
+    equivalent of &mdash;.

+
+
+
+
Images

+
+Admittedly, it's fairly difficult to devise a "natural" syntax for
+placing images into a plain text document format.
+
+Markdown uses an image syntax that is intended to resemble the syntax
+for links, allowing for two styles: *inline* and *reference*.
+
+Inline image syntax looks like this:
+
+    ![Alt text](/path/to/img.jpg)
+
+    ![Alt text](/path/to/img.jpg "Optional title")
+
+That is:
+
+*   An exclamation mark: `!`;
+*   followed by a set of square brackets, containing the `alt`
+    attribute text for the image;
+*   followed by a set of parentheses, containing the URL or path to
+    the image, and an optional `title` attribute enclosed in double
+    or single quotes.
+
+Reference-style image syntax looks like this:
+
+    ![Alt text][id]
+
+Where "id" is the name of a defined image reference. Image references
+are defined using syntax identical to link references:
+
+    [id]: url/to/image  "Optional title attribute"
+
+As of this writing, Markdown has no syntax for specifying the
+dimensions of an image; if this is important to you, you can simply
+use regular HTML `` tags.
+
+
+* * *
+
+
+
Miscellaneous

+
+
+
+Markdown supports a shortcut style for creating "automatic" links for URLs and email addresses: simply surround the URL or email address with angle brackets. What this means is that if you want to show the actual text of a URL or email address, and also have it be a clickable link, you can do this:
+
+    
+
+Markdown will turn this into:
+
+    http://example.com/
+
+Automatic links for email addresses work similarly, except that
+Markdown will also perform a bit of randomized decimal and hex
+entity-encoding to help obscure your address from address-harvesting
+spambots. For example, Markdown will turn this:
+
+    
+
+into something like this:
+
+    address@exa
+    mple.com
+
+which will render in a browser as a clickable link to "address@example.com".
+
+(This sort of entity-encoding trick will indeed fool many, if not
+most, address-harvesting bots, but it definitely won't fool all of
+them. It's better than nothing, but an address published in this way
+will probably eventually start receiving spam.)
+
+
+
+
Backslash Escapes

+
+Markdown allows you to use backslash escapes to generate literal
+characters which would otherwise have special meaning in Markdown's
+formatting syntax. For example, if you wanted to surround a word
+with literal asterisks (instead of an HTML `` tag), you can use
+backslashes before the asterisks, like this:
+
+    \*literal asterisks\*
+
+Markdown provides backslash escapes for the following characters:
+
+    \   backslash
+    `   backtick
+    *   asterisk
+    _   underscore
+    {}  curly braces
+    []  square brackets
+    ()  parentheses
+    #   hash mark
+	+	plus sign
+	-	minus sign (hyphen)
+    .   dot
+    !   exclamation mark
diff --git a/docs/using/index.md b/docs/using/index.md
index e393b02b..d8ce3964 100644
--- a/docs/using/index.md
+++ b/docs/using/index.md
@@ -7,5 +7,6 @@ MyST documents.
 install.md
 sphinx.md
 syntax.md
+benchmark.md
 use_api.md
 ```
diff --git a/docs/using/use_api.md b/docs/using/use_api.md
index c05488b3..1c87cb87 100644
--- a/docs/using/use_api.md
+++ b/docs/using/use_api.md
@@ -6,6 +6,24 @@ MyST-Parser may be used as an API *via* the `myst_parser` package.
 The raw text is first parsed to syntax 'tokens',
 then these are converted to other formats using 'renderers'.
 
+The simplest way to parse text is using:
+
+```python
+from myst_parser import parse_text
+parse_text("some *text*", "html")
+```
+
+```html
+'
some text
\n'
+```
+
+The output type can be one of:
+
+- `dict` (a.k.a ast)
+- `html`
+- `docutils`
+- `sphinx`
+
 ## Convert Text to Tokens
 
 To convert some text to tokens:
diff --git a/docstring.fmt.mustache b/docstring.fmt.mustache
new file mode 100644
index 00000000..717a4572
--- /dev/null
+++ b/docstring.fmt.mustache
@@ -0,0 +1,20 @@
+{{! Sphinx Docstring Template }}
+{{summaryPlaceholder}}
+
+{{extendedSummaryPlaceholder}}
+
+{{#args}}
+:param {{var}}: {{descriptionPlaceholder}}
+{{/args}}
+{{#kwargs}}
+:param {{var}}: {{descriptionPlaceholder}}
+{{/kwargs}}
+{{#exceptions}}
+:raises {{type}}: {{descriptionPlaceholder}}
+{{/exceptions}}
+{{#returns}}
+:return: {{descriptionPlaceholder}}
+{{/returns}}
+{{#yields}}
+:yield: {{descriptionPlaceholder}}
+{{/yields}}
diff --git a/myst_parser/__init__.py b/myst_parser/__init__.py
index e26a853d..d5c142b6 100644
--- a/myst_parser/__init__.py
+++ b/myst_parser/__init__.py
@@ -17,6 +17,29 @@ def render_tokens(root_token, renderer, **kwargs):
         return renderer.render(root_token)
 
 
+def parse_text(text: str, output_type: str, **kwargs):
+    """Convert MyST text to another format.
+
+    :param text: the text to convert
+    :param output_type: one of 'dict', 'html', 'docutils', 'sphinx'
+    :param kwargs: parsed to the render initiatiation
+    """
+    if output_type == "dict":
+        from myst_parser.ast_renderer import AstRenderer as renderer_cls
+    elif output_type == "html":
+        from myst_parser.html_renderer import HTMLRenderer as renderer_cls
+    elif output_type == "docutils":
+        from myst_parser.docutils_renderer import DocutilsRenderer as renderer_cls
+    elif output_type == "sphinx":
+        from myst_parser.docutils_renderer import SphinxRenderer as renderer_cls
+    else:
+        raise ValueError("output_type not recognised: {}".format(output_type))
+    from myst_parser.block_tokens import Document
+
+    with renderer_cls(**kwargs) as renderer:
+        return renderer.render(Document(text))
+
+
 def setup(app):
     """Initialize Sphinx extension."""
     from myst_parser.sphinx_parser import MystParser
diff --git a/myst_parser/block_tokens.py b/myst_parser/block_tokens.py
index c417b05b..357b3c36 100644
--- a/myst_parser/block_tokens.py
+++ b/myst_parser/block_tokens.py
@@ -3,13 +3,7 @@
 from mistletoe import block_token, span_token
 import mistletoe.block_tokenizer as tokenizer
 
-from mistletoe.block_token import (  # noqa: F401
-    tokenize,
-    HTMLBlock,
-    ThematicBreak,
-    Footnote,
-    TableRow,
-)
+from mistletoe.block_token import tokenize, HTMLBlock, Footnote, TableRow  # noqa: F401
 
 """
 Tokens to be included in the parsing process, in the order specified.
@@ -147,6 +141,25 @@ def __repr__(self):
         return "MyST.{}(range={})".format(self.__class__.__name__, self.range)
 
 
+class ThematicBreak(block_token.ThematicBreak):
+    """
+    Thematic break token (a.k.a. horizontal rule.)
+    """
+
+    def __init__(self, result):
+        line, lineno = result
+        self.raw = line.splitlines()[0]
+        self.range = (lineno, lineno)
+
+    @classmethod
+    def read(cls, lines):
+        line = next(lines)
+        return line, lines.lineno
+
+    def __repr__(self):
+        return "MyST.{}(range={})".format(self.__class__.__name__, self.range)
+
+
 class BlockBreak(block_token.BlockToken):
     """Block break token ``+++``.
 
diff --git a/myst_parser/cli/__init__.py b/myst_parser/cli/__init__.py
new file mode 100644
index 00000000..e69de29b
diff --git a/myst_parser/cli/benchmark.py b/myst_parser/cli/benchmark.py
new file mode 100644
index 00000000..9603166b
--- /dev/null
+++ b/myst_parser/cli/benchmark.py
@@ -0,0 +1,113 @@
+import argparse
+from importlib import import_module
+import os
+import re
+from time import perf_counter
+
+ALL_PACKAGES = (
+    "python-markdown:extra",
+    "mistune",
+    "commonmark.py",
+    "mistletoe",
+    "myst_parser:html",
+    "myst_parser:docutils",
+    "myst_parser:sphinx",
+)
+
+
+def benchmark(package_name, version=None):
+    def decorator(func):
+        def inner(text, num_parses):
+            try:
+                package = import_module(package_name)
+                print("(" + (version or package.__version__) + ")", end=": ")
+            except ImportError:
+                return "not available."
+            start = perf_counter()
+            for i in range(num_parses):
+                func(package, text)
+            end = perf_counter()
+
+            return end - start
+
+        return inner
+
+    return decorator
+
+
+@benchmark("markdown")
+def run_python_markdown_extra(package, text):
+    return package.markdown(text, extensions=["extra"])
+
+
+@benchmark("mistune")
+def run_mistune(package, text):
+    return package.markdown(text)
+
+
+@benchmark("commonmark", "0.9.1")
+def run_commonmark_py(package, text):
+    return package.commonmark(text)
+
+
+@benchmark("mistletoe")
+def run_mistletoe(package, text):
+    return package.markdown(text)
+
+
+@benchmark("myst_parser")
+def run_myst_parser_html(package, text):
+    package.parse_text(text, "html")
+
+
+@benchmark("myst_parser")
+def run_myst_parser_docutils(package, text):
+    package.parse_text(text, "docutils", config={"ignore_missing_refs": True})
+
+
+@benchmark("myst_parser")
+def run_myst_parser_sphinx(package, text):
+    package.parse_text(text, "sphinx", load_sphinx_env=True)
+
+
+def run_all(package_names, text, num_parses):
+    prompt = "Running {} test(s) ...".format(len(package_names))
+    print(prompt)
+    print("=" * len(prompt))
+    for package_name in package_names:
+        print(package_name, end=" ")
+        func_name = re.sub(r"[\.\-\:]", "_", package_name.lower())
+        print(
+            "{:.2f} s".format(globals()["run_{}".format(func_name)](text, num_parses))
+        )
+    return True
+
+
+def main(args=None):
+    parser = argparse.ArgumentParser(description="Run benchmark test.")
+    parser.add_argument("path", type=str, help="the path to the file to parse")
+    parser.add_argument(
+        "-n",
+        "--num-parses",
+        metavar="NPARSES",
+        default=1000,
+        type=int,
+        help="The number of parse iterations (default: 1000)",
+    )
+    parser.add_argument(
+        "-p",
+        "--package",
+        action="append",
+        default=[],
+        help="The package(s) to run (use -p multiple times).",
+        choices=ALL_PACKAGES,
+        metavar="PACKAGE_NAME",
+    )
+    args = parser.parse_args(args)
+
+    assert os.path.exists(args.path), "path does not exist"
+    print("Test document: {}".format(os.path.basename(args.path)))
+    print("Test iterations: {}".format(args.num_parses))
+    with open(args.path, "r") as handle:
+        text = handle.read()
+    return run_all(args.package or ALL_PACKAGES, text, args.num_parses)
diff --git a/myst_parser/docutils_renderer.py b/myst_parser/docutils_renderer.py
index c015a237..190e1bc8 100644
--- a/myst_parser/docutils_renderer.py
+++ b/myst_parser/docutils_renderer.py
@@ -186,7 +186,9 @@ def render_strikethrough(self, token):
         raise NotImplementedError
 
     def render_thematic_break(self, token):
-        self.current_node.append(nodes.transition())
+        node = nodes.transition()
+        self.add_line_and_source_path(node, token)
+        self.current_node.append(node)
 
     def render_block_break(self, token):
         block_break = nodes.comment(token.content, token.content)
@@ -284,11 +286,14 @@ def render_heading(self, token):
         self.document.note_implicit_target(section, section)
         self.current_node = section
 
-    def handle_cross_reference(self, token, destination, ref_node):
+    def handle_cross_reference(self, token, destination):
         # TODO use the docutils error reporting mechanisms, rather than raising
-        raise NotImplementedError(
-            "reference not found in current document: {}".format(destination)
-        )
+        if not self.config.get("ignore_missing_refs", False):
+            raise NotImplementedError(
+                "reference not found in current document: {}\n{}".format(
+                    destination, token
+                )
+            )
 
     def render_link(self, token):
         ref_node = nodes.reference()
diff --git a/setup.py b/setup.py
index 8e527bb4..4e412994 100644
--- a/setup.py
+++ b/setup.py
@@ -16,7 +16,9 @@
     author_email="chrisj_sewell@hotmail.com",
     license="MIT",
     packages=find_packages(),
-    entry_points={"console_scripts": []},
+    entry_points={
+        "console_scripts": ["myst-benchmark = myst_parser.cli.benchmark:main"]
+    },
     classifiers=[
         "Development Status :: 3 - Alpha",
         "Intended Audience :: Developers",
diff --git a/tests/test_cli.py b/tests/test_cli.py
new file mode 100644
index 00000000..4b069ca9
--- /dev/null
+++ b/tests/test_cli.py
@@ -0,0 +1,11 @@
+import pathlib
+import tempfile
+
+from myst_parser.cli import benchmark
+
+
+def test_benchmark():
+    with tempfile.TemporaryDirectory() as tempdir:
+        path = pathlib.Path(tempdir).joinpath("test.md")
+        path.write_text("a b c")
+        assert benchmark.main(["-n", "1", "-p", "myst_parser:html", str(path)])
diff --git a/tests/test_renderers/test_docutils.py b/tests/test_renderers/test_docutils.py
index 31963b00..60698363 100644
--- a/tests/test_renderers/test_docutils.py
+++ b/tests/test_renderers/test_docutils.py
@@ -4,7 +4,7 @@
 from mistletoe.block_token import tokenize
 from mistletoe.span_token import tokenize_inner
 
-from myst_parser import text_to_tokens, render_tokens
+from myst_parser import text_to_tokens, render_tokens, parse_text
 from myst_parser.block_tokens import Document
 from myst_parser.docutils_renderer import SphinxRenderer
 
@@ -21,7 +21,7 @@ def render_token(
     render_func(mock_token)
 
 
-def test_render_tokens():
+def test_text_to_tokens():
     root = text_to_tokens("abc")
     document = render_tokens(
         root,
@@ -34,6 +34,15 @@ def test_render_tokens():
     )
 
 
+def test_parse_text():
+    document = parse_text(
+        "abc", "sphinx", load_sphinx_env=True, sphinx_conf={"project": "MyST Parser"}
+    )
+    assert document.pformat() == (
+        '\n    \n        abc\n'
+    )
+
+
 def test_strong(renderer_mock):
     render_token(renderer_mock, "Strong")
     assert renderer_mock.document.pformat() == dedent(