jReads does not exist

[npx]

Hey there,

I compiled and installed all dependencies and now wanna run some of the examples presented here.

>>> from jNlp.jConvert import *
Traceback (most recent call last):
  File "<input>", line 1, in <module>
  File "/Users/xxx/VirtualEnvs/venv-python2.7-django/lib/python2.7/site-packages/jProcessing-0.1-py2.7.egg/j
Nlp/jConvert.py", line 4, in <module>
    from jNlp.jTokenize import jTokenize, jReads
ImportError: cannot import name jReads

I tried replacing the jReads with the jTokenize method but I didn't expect that to work :)

I found and old implementation that I took and changed to using cabocha().

def jReads(target_sent):
    sentence = etree.fromstring(cabocha(target_sent).encode('utf-8'))
    jReadsToks = []
    for chunk in sentence:
        for tok in chunk.findall('tok'):
            if tok.get("read"): jReadsToks.append(tok.get("read"))
    return jReadsToks

However, I don't seem to be getting a valid XML:

気象庁が２１日午前４時４８分、発表した天気概況によると、
tokenizedRomaji(input_sentence)

iconv_open is not supported
                           Traceback (most recent call last):
  File "<input>", line 1, in <module>
Nlp/jConvert.py", line 42, in tokenizedRomaji2.7-django/lib/python2.7/site-packages/jProcessing-0.1-py2.7.egg/j
    for kataChunk in jReads(jSent):
  File "/Users/xxx/VirtualEnvs/venv-python2.7-django/lib/python2.7/site-packages/jProcessing-0.1-py2.7.egg/j
Nlp/jTokenize.py", line 29, in jReads
    sentence = etree.fromstring(cabocha(target_sent).encode('utf-8'))
  File "<string>", line 124, in XML
ParseError: not well-formed (invalid token): line 1, column 4

I compiled and installed iconv but is this related to the problem?

Also, I verified my installation of mecab and cabocha and both seem to work fine.

But jReads really does not exist xP

[kevincobain2000]

Thanks man, infact the method doesn't exist. Will fix soon

[timmahrt]

Any update on this? I'm working on my own code to convert kanji to katakana/hiragana, based on the unidic dictionary from ninjal.

[kevincobain2000]

tentative fix, I ll try to do the fix over this weekend by added the previous code.

def jReads(target_sent):
    sentence = etree.fromstring(cabocha(target_sent).encode('utf-8'))
    jReadsToks = []
    for chunk in sentence:
        for tok in chunk.findall('tok'):
            if tok.get("read"): jReadsToks.append(tok.get("read"))
    return jReadsToks

Although i never had that error iconv_open is not supported in the past, but let me confirm.

[timmahrt]

When this function runs on my machine: jNlp.jCabocha.cabocha

it doesn't return xml. However, it works fine if I change the line command = ['cabocha', '-f', '3 <', temp.name] to command = ['cabocha', '-f', '3', temp.name]

Also, you're looking for the 'read' label in the xml? But the xml I get back is pretty sparse (still useable for my purposes though):

荒く

I'm not sure if maybe I configured cabocha or one of its dependencies incorrectly?

[anyong]

Did you ever get the 'read' label issue fixed?

[kevincobain2000]

@anyong https://github.com/kevincobain2000/jProcessing/commit/d59425fb8b63365340458ec683037f4d98e7e255 fixes it. Just didn't close the issue.