Mailing-List: contact dev-help@madlib.incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@madlib.incubator.apache.org
From: mktal <git@git.apache.org>
To: dev@madlib.incubator.apache.org
Reply-To: dev@madlib.incubator.apache.org
References: <git-pr-29-incubator-madlib@git.apache.org>
In-Reply-To: <git-pr-29-incubator-madlib@git.apache.org>
Subject: [GitHub] incubator-madlib pull request: Path: Return results for each
 match
Content-Type: text/plain
Message-Id: <20160324183511.3C231DFD43@git1-us-west.apache.org>
Date: Thu, 24 Mar 2016 18:35:11 +0000 (UTC)

Github user mktal commented on a diff in the pull request:

    https://github.com/apache/incubator-madlib/pull/29#discussion_r57366247
  
    --- Diff: src/ports/postgres/modules/utilities/path.py_in ---
    @@ -118,140 +120,175 @@ def path(schema_madlib, source_table, output_table, partition_expr,
             #   string produced by concatenating the symbols. The exact rows that
             #   produce the match are identified by correlating the matched string
             #   indices with another array containing row ids.
    -        #
    -        #   matched_partitions: For each partition (group), concatenate all symbols
    -        #       into a single string (sym_str). Keep corresponding ids in an array in the
    -        #       same order as the symbols. This is performed only for partitions
    -        #       that contain a match.
    -        #   build_multiple_matched_rows:
    -        #       q1: Split sym_str into an array containing the lengths of the
    -        #           strings between the matches.
    -        #       q2: Store lengths of matches into an array
    -        #       q3: Merge q1 and q2 and unnest the arrays (ensuring same length).
    -        #           Also right shift the matches array.
    -        #       q4: Compute the cumulative sum of the arrays.
    +
    +        # matched_partitions: For each partition, concatenate all symbols
    +        #       into a single string (sym_str). Keep corresponding ids in an
    +        #       array (match_to_row_id) in the same order as the symbols.
    +        #       This is performed only for partitions that contain a match.
    +        match_id_name = "__madlib_path_match_id__" if "match_id" in all_input_cols else "match_id"
    +        symbol_name = "__madlib_path_symbol__" if "symbol" in all_input_cols else "symbol"
             plpy.execute("""
                 CREATE TEMP TABLE {matched_partitions} AS
                     SELECT
                         {p_col_name_str},
    -                    array_to_string(array_agg({symbol_name_str} ORDER BY {order_expr}), '') as sym_str,
    -                    array_agg({id_col_name} ORDER BY {order_expr}) as matched_ids
    +                    array_to_string(array_agg({short_sym_name_str} ORDER BY {order_expr}), '') as sym_str,
    +                    array_agg({id_col_name} ORDER BY {order_expr}) as {match_to_row_id}
                     FROM {input_with_id}
    +                WHERE {short_sym_name_str} is NOT NULL
                     GROUP BY {p_col_name_str}
    -                HAVING array_to_string(array_agg({symbol_name_str} ORDER BY {order_expr}), '') ~* '{pattern_expr}'
    +                HAVING array_to_string(array_agg({short_sym_name_str} ORDER BY {order_expr}), '')
    +                                ~* '{new_pattern_expr}'
                 """.format(**locals()))
    +
    +        # length_of_matches: For each partition in matched_partitions:
    +        #       - find all matches and compute the lengths of each match.
    +        #       - output these lengths in an array (matches), along with
    +        #         an array of corresponding rank of each match (match_indices).
    +        length_of_matches = unique_string("match_length_view")
    +        plpy.execute("""
    +            CREATE VIEW {length_of_matches} AS
    +            SELECT
    +                {p_col_name_str},
    +                {match_to_row_id},
    +                array_agg(length(matches) ORDER BY match_index) AS matches,
    +                array_agg(match_index ORDER BY match_index) AS match_indices
    +            FROM (
    +                SELECT
    +                    {p_col_name_str},
    +                    {match_to_row_id},
    --- End diff --
    
    Is it necessary to have {match_to_row_id} in both {length_of_matches} and {length_of_between_matches}?


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---