Apertium has moved from SourceForge to GitHub.
If you have any questions, please come and talk to us on #apertium on irc.freenode.net or contact the GitHub migration team.

User:SilentFlame/updatedPipeline

From Apertium
Jump to: navigation, search

For the work done at Progress regarding Automatic_blank_handling

Contents

[edit] Tasks completed

[edit] deformatting prototypes

[edit] pretransfer

[edit] transfer (chunker)

[edit] Interchunk

[edit] Deformatters

[edit] Reformatters

[edit] lttoolbox

[edit] Tasks left to do

[edit] postchunk

[edit] hfst

  • Make hfst-proc correctly disperse inline blanks onto each lexical unit until the next [

[edit] transfer (non-chunking)

[edit] Input and Output at different stages/modes

Input: "<div><i>Hello</i> <b>world</b></div>" 
Testing this input on the entire pipeline.
  • The DIR/DIRECTORY in the below commands refers to the directory address where you have your language pair compiled. Here I have used apertium-en-es language pair.

[edit] deformatter stage

Command: $ echo "<div><i>Hello</i> <b>world</b></div>" |./deshtml

Output: [<div>][{<i>}]Hello[] [{<b>}]world[][][</div>]

[edit] lt-proc(automorph) stage

Command: echo "<div><i>Hello</i> <b>world</b></div>" |./deshtml | lt-proc 'DIRECTORY/apertium-en-es/en-es.automorf.bin'

Output: [<div>][{<i>}]^Hello/Hello<ij>$[] [{<b>}]^world/world<adj>/world<n><sg>$[][][</div>]

[edit] tagger stage

Command: echo "<div><i>Hello</i> <b>world</b></div>" |./deshtml | lt-proc 'DIR/apertium-en-es/en-es.automorf.bin' | apertium-tagger -g $2 'DIR/apertium-en-es/en-es.prob'

Output: [<div>][{<i>}]^Hello<ij>$[] [{<b>}]^world<adj>$[][][</div>]

[edit] pretransfer stage

Command: echo "<div><i>Hello</i> <b>world</b></div>" |./deshtml | lt-proc 'DIR/apertium-en-es/en-es.automorf.bin' | apertium-tagger -g $2 'DIR/apertium-en-es/en-es.prob' \
| apertium-pretransfer

Output:[<div>][{<i>}]^Hello<ij>$[] [{<b>}]^world<adj>$[][][</div>]

[edit] transfer(chunker) stage

Command: echo "<div><i>Hello</i> <b>world</b></div>" |./deshtml | lt-proc 'DIR/apertium-en-es/en-es.automorf.bin' | apertium-tagger -g $2 'DIR/apertium-en-es/en-es.prob' \
| apertium-pretransfer | apertium-transfer -n 'DIR/apertium-en-es/apertium-en-es.en-es.genitive.t1x'  'DIR/apertium-en-es/en-es.genitive.bin'

Output: [<div>][{<i>}]^Hello<ij>$[] [{<b>}]^world<adj>$[][][</div>]

[edit] lt-proc(auto-bilingual) stage

Command: echo "<div><i>Hello</i> <b>world</b></div>" |./deshtml | lt-proc 'DIR/apertium-en-es/en-es.automorf.bin' | apertium-tagger -g $2 'DIR/apertium-en-es/en-es.prob' \
| apertium-pretransfer | apertium-transfer -n 'DIR/apertium-en-es/apertium-en-es.en-es.genitive.t1x'  'DIR/apertium-en-es/en-es.genitive.bin' \
| lt-proc -b 'DIR/apertium-en-es/en-es.autobil.bin'

Output: [<div>][{<i>}]^Hello<ij>/Hola<ij>$[] [{<b>}]^world<adj>/mundial<adj><mf>$[][][</div>]

[edit] lrx-proc(auto-lexical) stage

Command: echo "<div><i>Hello</i> <b>world</b></div>" |./deshtml | lt-proc 'DIR/apertium-en-es/en-es.automorf.bin' | apertium-tagger -g $2 'DIR/apertium-en-es/en-es.prob' \
| apertium-pretransfer | apertium-transfer -n 'DIR/apertium-en-es/apertium-en-es.en-es.genitive.t1x'  'DIR/apertium-en-es/en-es.genitive.bin' \
| lt-proc -b 'DIR/apertium-en-es/en-es.autobil.bin' | lrx-proc -m 'DIR/apertium-en-es/en-es.autolex.bin' 

Output: [<div>][{<i>}]^Hello<ij>/Hola<ij>$[] [{<b>}]^world<adj>/mundial<adj><mf>$[][][</div>]

[edit] transfer stage

Command: echo "<div><i>Hello</i> <b>world</b></div>" |./deshtml | lt-proc 'DIR/apertium-en-es/en-es.automorf.bin' | apertium-tagger -g $2 'DIR/apertium-en-es/en-es.prob' \
| apertium-pretransfer | apertium-transfer -n 'DIR/apertium-en-es/apertium-en-es.en-es.genitive.t1x'  'DIR/apertium-en-es/en-es.genitive.bin' \
| lt-proc -b 'DIR/apertium-en-es/en-es.autobil.bin' | lrx-proc -m 'DIR/apertium-en-es/en-es.autolex.bin' \
| apertium-transfer -b 'DIR/apertium-en-es/apertium-en-es.en-es.t1x'  'DIR/apertium-en-es/en-es.t1x.bin'

Output: [<div>]^default<default>{[{<i>}]^Hola<ij>$}$[] ^Adj<SA><mf><ND>{[{<b>}]^mundial<adj><2><3>$}$[][][</div>]

[edit] interchunk stage

Command: echo "<div><i>Hello</i> <b>world</b></div>" |./deshtml | lt-proc 'DIR/apertium-en-es/en-es.automorf.bin' | apertium-tagger -g $2 'DIR/apertium-en-es/en-es.prob' \
| apertium-pretransfer | apertium-transfer -n 'DIR/apertium-en-es/apertium-en-es.en-es.genitive.t1x'  'DIR/apertium-en-es/en-es.genitive.bin' \
| lt-proc -b 'DIR/apertium-en-es/en-es.autobil.bin' | lrx-proc -m 'DIRApertium/apertium-en-es/en-es.autolex.bin' \
| apertium-transfer -b 'DIR/apertium-en-es/apertium-en-es.en-es.t1x'  'DIR/apertium-en-es/en-es.t1x.bin' \
| apertium-interchunk 'DIR/apertium-en-es/apertium-en-es.en-es.t2x'  'DIR/apertium-en-es/en-es.t2x.bin'

Output: [<div>]^default< default>{[{<i>}]^Hola<ij>$}$[] ^Adj<SA><mf><sg>{[{<b>}]^mundial<adj><2><3>$}$[][][</div>]


[edit] Tasks done

[edit] Pretransfer

All the pretransfer tests pass here.

[edit] Transfer(chunker)

All the tests mentioned in https://github.com/SilentFlame/apertium-1/tree/blank-handling/tests/transfer passes with the updated transfer module.

[edit] Interchunk

  • Here removing "pos=1" from a "<b>" still outputs the right inline blank: This is because If given a "freeblank" which is between chunks and not a wordbound/inline blank so we need to treat it differently. let's say for example we have "^SN<sg>{^cheese<n>$}$🍰^SN<sg>{^sale<n>$}$" as an input. and the rule matches those two chunks and has an action " <out> <chunk pos="1" part="whole"/> <b/> <chunk pos="2" part="whole"/> </out> " so if here we treat "<b/>" as just a space then we'll loose "🍰" which won't give much good feel to our users. So to retain this in the output we handled the freeblanks between chunks.
  • Task: Interchunk was needed to ignore the "pos" argument to b elements, and output each superblank exactly once, preferably where the rule has a b element (if there are not enough b's, output the rest at the end of the rule). Here in this module we didn't deal with wordblanks, since we can't look inside chunks when in interchunk.
  • Category: Code enhancing
  • PR: https://github.com/unhammer/apertium/pull/6
  • Tests: https://github.com/SilentFlame/apertium-1/tree/blank-handling-interchunk/tests/interchunk

All tests mentioned in https://github.com/SilentFlame/apertium-1/blob/blank-handling-interchunk/tests/interchunk/__init__.py passes with the updated interchunk module.

[edit] Deformatters

All the tests run without fail and the run command is $python tests/run_test.py inside the apertium folder.

[edit] Reformatters

All the tests run without fail and the run command is $python tests/run_test.py inside the apertium folder.

[edit] lttoolbox

All the above tests for lt-proc passes with the updated module.

[edit] All commits

https://apertium.projectjj.com/gsoc2017/silentflame.html

Personal tools