Обсуждение: old bug in full text parser

It  looks like there is a very old bug in full text parser (somebody pointed me on it), which appeared after moving tsearch2 into the core.  The problem is in how full text parser process hyphenated words. Our original idea was to report hyphenated word itself as well as its parts and ignore hyphen. That was how tsearch2 works.

This behaviour was changed after moving tsearch2 into the core:
1. hyphen now reported by parser, which is useless.

2.  Hyphenated words with numbers ('4-dot', 'dot-4')  processed differently than ones with plain text words like 'four-dot', no hyphenated word itself reported.

I think we should consider this as a bug and produce fix for all supported versions.

After  investigation we found this commit:

commit 73e6f9d3b61995525785b2f4490b465fe860196b
Author: Tom Lane <tgl@sss.pgh.pa.us>
Date:   Sat Oct 27 19:03:45 2007 +0000

    Change text search parsing rules for hyphenated words so that digit strings
    containing decimal points aren't considered part of a hyphenated word.
    Sync the hyphenated-word lookahead states with the subsequent part-by-part
    reparsing states so that we don't get different answers about how much text
    is part of the hyphenated word.  Per my gripe of a few days ago.

8.2.23

select tok_type, description, token from ts_debug('dot-four');
  tok_type   |          description          |  token
-------------+-------------------------------+----------
 lhword      | Latin hyphenated word         | dot-four
 lpart_hword | Latin part of hyphenated word | dot
 lpart_hword | Latin part of hyphenated word | four
(3 rows)

select tok_type, description, token from ts_debug('dot-4');
  tok_type   |          description          | token
-------------+-------------------------------+-------
 hword       | Hyphenated word               | dot-4
 lpart_hword | Latin part of hyphenated word | dot
 uint        | Unsigned integer              | 4
(3 rows)

select tok_type, description, token from ts_debug('4-dot');
 tok_type |   description    | token
----------+------------------+-------
 uint     | Unsigned integer | 4
 lword    | Latin word       | dot
(2 rows)

8.3.23

select alias, description, token from ts_debug('dot-four');
      alias      |           description           |  token
-----------------+---------------------------------+----------
 asciihword      | Hyphenated word, all ASCII      | dot-four
 hword_asciipart | Hyphenated word part, all ASCII | dot
 blank           | Space symbols                   | -
 hword_asciipart | Hyphenated word part, all ASCII | four
(4 rows)

select alias, description, token from ts_debug('dot-4');
   alias   |   description   | token
-----------+-----------------+-------
 asciiword | Word, all ASCII | dot
 int       | Signed integer  | -4
(2 rows)

select alias, description, token from ts_debug('4-dot');
   alias   |   description    | token
-----------+------------------+-------
 uint      | Unsigned integer | 4
 blank     | Space symbols    | -
 asciiword | Word, all ASCII  | dot
(3 rows)

Regards,

Oleg

On Wed, Feb 10, 2016 at 4:28 AM, Oleg Bartunov <obartunov@gmail.com> wrote:
> It  looks like there is a very old bug in full text parser (somebody pointed
> me on it), which appeared after moving tsearch2 into the core.  The problem
> is in how full text parser process hyphenated words. Our original idea was
> to report hyphenated word itself as well as its parts and ignore hyphen.
> That was how tsearch2 works.
>
> This behaviour was changed after moving tsearch2 into the core:
> 1. hyphen now reported by parser, which is useless.
> 2.  Hyphenated words with numbers ('4-dot', 'dot-4')  processed differently
> than ones with plain text words like 'four-dot', no hyphenated word itself
> reported.
>
> I think we should consider this as a bug and produce fix for all supported
> versions.
>

The Evergreen project has long depended on tsearch2 (both as an
extension and in-core FTS), and one thing we've struggled with is date
range parsing such as birth and death years for authors in the form of
1979-2014, for instance.  Strings like that end up being parsed as two
lexems, "1979" and "-2014".  We work around this by pre-normalizing
strings matching /(\d+)-(\d+)/ into two numbers separated by a space
instead of a hyphen, but if fixing this bug would remove the need for
such a preprocessing step it would be a great help to us.  Would such
strings be parsed "properly" into lexems of the form of "1979" and
"2014" with you proposed change?

Thanks!

--
Mike Rylander

> After  investigation we found this commit:
>
> commit 73e6f9d3b61995525785b2f4490b465fe860196b
> Author: Tom Lane <tgl@sss.pgh.pa.us>
> Date:   Sat Oct 27 19:03:45 2007 +0000
>
>     Change text search parsing rules for hyphenated words so that digit
> strings
>     containing decimal points aren't considered part of a hyphenated word.
>     Sync the hyphenated-word lookahead states with the subsequent
> part-by-part
>     reparsing states so that we don't get different answers about how much
> text
>     is part of the hyphenated word.  Per my gripe of a few days ago.
>
>
> 8.2.23
>
> select tok_type, description, token from ts_debug('dot-four');
>   tok_type   |          description          |  token
> -------------+-------------------------------+----------
>  lhword      | Latin hyphenated word         | dot-four
>  lpart_hword | Latin part of hyphenated word | dot
>  lpart_hword | Latin part of hyphenated word | four
> (3 rows)
>
> select tok_type, description, token from ts_debug('dot-4');
>   tok_type   |          description          | token
> -------------+-------------------------------+-------
>  hword       | Hyphenated word               | dot-4
>  lpart_hword | Latin part of hyphenated word | dot
>  uint        | Unsigned integer              | 4
> (3 rows)
>
> select tok_type, description, token from ts_debug('4-dot');
>  tok_type |   description    | token
> ----------+------------------+-------
>  uint     | Unsigned integer | 4
>  lword    | Latin word       | dot
> (2 rows)
>
> 8.3.23
>
> select alias, description, token from ts_debug('dot-four');
>       alias      |           description           |  token
> -----------------+---------------------------------+----------
>  asciihword      | Hyphenated word, all ASCII      | dot-four
>  hword_asciipart | Hyphenated word part, all ASCII | dot
>  blank           | Space symbols                   | -
>  hword_asciipart | Hyphenated word part, all ASCII | four
> (4 rows)
>
> select alias, description, token from ts_debug('dot-4');
>    alias   |   description   | token
> -----------+-----------------+-------
>  asciiword | Word, all ASCII | dot
>  int       | Signed integer  | -4
> (2 rows)
>
> select alias, description, token from ts_debug('4-dot');
>    alias   |   description    | token
> -----------+------------------+-------
>  uint      | Unsigned integer | 4
>  blank     | Space symbols    | -
>  asciiword | Word, all ASCII  | dot
> (3 rows)
>
>
> Regards,
> Oleg

Oleg Bartunov <obartunov@gmail.com> writes:
> It  looks like there is a very old bug in full text parser (somebody
> pointed me on it), which appeared after moving tsearch2 into the core.  The
> problem is in how full text parser process hyphenated words. Our original
> idea was to report hyphenated word itself as well as its parts and ignore
> hyphen. That was how tsearch2 works.

> This behaviour was changed after moving tsearch2 into the core:
> 1. hyphen now reported by parser, which is useless.
> 2.  Hyphenated words with numbers ('4-dot', 'dot-4')  processed differently
> than ones with plain text words like 'four-dot', no hyphenated word itself
> reported.

> I think we should consider this as a bug and produce fix for all supported
> versions.

I don't see anything here that looks like a bug, more like a definition
disagreement.  As such, I'd be pretty dubious about back-patching a
change.  But it's hard to debate the merits when you haven't said exactly
what you'd do instead.

I believe the commit you mention was intended to fix this inconsistency:

http://www.postgresql.org/message-id/6269.1193184058@sss.pgh.pa.us

so I would be against simply reverting it.  In any case, the examples
given there make it look like there was already inconsistency about mixed
words and numbers.  Do we really think that "4-dot" should be considered
a hyphenated word?  I'm not sure.

                        regards, tom lane