Table of Contents

1. markup-parse
2. Development
3. Main Pipeline types
4. MarkupParse.Patch
5. wiki diff test debug
6. Reference
   1. Prior Art
7. Performance
   1. Profiling

markup-parse

markup-parse parses and prints a subset of common XML & HTML structured data, from and to strict bytestrings

Development

:r
:set prompt "> "
:set -Wno-type-defaults
:set -Wno-name-shadowing
:set -XOverloadedStrings
:set -XTemplateHaskell
:set -XQuasiQuotes
import Control.Monad
import MarkupParse
import MarkupParse.FlatParse
import MarkupParse.Patch
import Data.ByteString qualified as B
import Data.ByteString.Char8 qualified as C
import Data.Map.Strict qualified as Map
import Data.Function
import FlatParse.Basic hiding (take)
import Data.String.Interpolate
import Data.TreeDiff
import Control.Monad
bs <- B.readFile "other/line.svg"
C.length bs

Preprocessing library for markup-parse-0.1.1..
GHCi, version 9.6.2: https://www.haskell.org/ghc/  :? for help
[1 of 3] Compiling MarkupParse.FlatParse ( src/MarkupParse/FlatParse.hs, interpreted )
[2 of 3] Compiling MarkupParse      ( src/MarkupParse.hs, interpreted )
[3 of 3] Compiling MarkupParse.Patch ( src/MarkupParse/Patch.hs, interpreted )
Ok, three modules loaded.
g
Ok, three modules loaded.
7554

Main Pipeline types

:t tokenize Html
:t gather Html
:t normalize
:t degather
:t detokenize Html
:t tokenize Html >=> gather Html >=> (normalize >>> pure) >=> degather Html >=> (fmap (detokenize Html) >>> pure)

tokenize Html :: ByteString -> Warn [Token]
gather Html :: [Token] -> Warn Markup
normalize :: Markup -> Markup
degather :: Standard -> Markup -> Warn [Token]
detokenize Html :: Token -> ByteString
tokenize Html >=> gather Html >=> (normalize >>> pure) >=> degather Html >=> (fmap (detokenize Html) >>> pure)
  :: ByteString -> These [MarkupWarning] [ByteString]

Round trip equality

m = markup_ Xml bs
m == (markup_ Xml $ markdown_ Compact Xml m)

True

MarkupParse.Patch

Obviously, patch doesn’t belong here long-term but has been very useful in testing and development.

showPatch $ patch [1, 2, 3, 5] [0, 1, 2, 4, 6]

[+0, -3, +4, -5, +6]

wiki diff test debug

bs <- B.readFile "other/Parsing - Wikipedia.html"
m = markup_ Html bs
m == (markup_ Html $ markdown_ Compact Html m)

True

Reference

HTML Standard

Are (non-void) self-closing tags valid in HTML5? - Stack Overflow

Extensible Markup Language (XML) 1.0 (Fifth Edition)

Prior Art

html-parse is an attoparsec-based parser for HTML. The HTML parsing here has referenced this parsing logic for HTML elements.

blaze-markup & lucid are HTML DSLs and printers, but not parsers.

xeno is an “event-based” XML parser with a more complicated base markup type.

XMLParser & hexml are parsec-based parsers.

Performance

The perf library has been used to measure performance of the library.

`cabal bench` runs the default benchmarking:

cabal bench

Running 1 benchmarks...
Benchmark markup-parse-speed: RUNNING...
label1          label2          old result      new result      change

gather          time            9.33e4          7.73e4          improvement
html-parse tokenstime            1.21e6          1.17e6
html-parse tree time            6.61e4          7.59e4          slightly-degraded
markdown        time            3.75e5          3.92e5
markup          time            4.72e5          5.20e5          slightly-degraded
normalize       time            2.80e5          3.04e5          slightly-degraded
tokenize        time            8.59e5          8.67e5
Benchmark markup-parse-speed: FINISH

Profiling

Profiling is used to aid library development:

cabal configure --enable-library-profiling --enable-executable-profiling -fprof-auto -fprof --write-ghc-environment-files=always --enable-benchmarks -O2

cabal.project.local

write-ghc-environment-files: always
ignore-project: False
flags: +prof +prof-auto
library-profiling: True
executable-profiling: True

Profiling slow the main functions significantly:

./app/speed -n 1000 --best -c +RTS -s -p -hc -l -RTS
label1              label2              old_result          new_result          status

gather              time                2.08e4              3.01e4              degraded
html-parse tokens   time                4.70e5              1.72e6              degraded
html-parse tree     time                2.30e4              3.85e4              degraded
markdown            time                3.51e5              5.70e5              degraded
markup              time                2.10e5              1.05e6              degraded
normalize           time                8.43e4              1.90e5              degraded
tokenize            time                1.94e5              1.02e6              degraded
   4,520,989,296 bytes allocated in the heap
   2,668,887,592 bytes copied during GC
     287,122,272 bytes maximum residency (21 sample(s))
       1,572,000 bytes maximum slop
             560 MiB total memory in use (0 MiB lost due to fragmentation)

                                     Tot time (elapsed)  Avg pause  Max pause
  Gen  0      1073 colls,     0 par    0.471s   0.479s     0.0004s    0.0024s
  Gen  1        21 colls,     0 par    2.428s   2.575s     0.1226s    0.3303s

  INIT    time    0.007s  (  0.008s elapsed)
  MUT     time    2.142s  (  1.945s elapsed)
  GC      time    1.904s  (  2.071s elapsed)
  RP      time    0.000s  (  0.000s elapsed)
  PROF    time    0.995s  (  0.982s elapsed)
  EXIT    time    0.026s  (  0.000s elapsed)
  Total   time    5.074s  (  5.006s elapsed)

  %GC     time       0.0%  (0.0% elapsed)

  Alloc rate    2,110,654,040 bytes per MUT second

  Productivity  61.8% of total user, 58.5% of total elapsed