Attachment 30318 Details for Bug 9368 – Bug 9368 [ALTERNATE] - specific behavior of yr and acqdate indexes

[patch] Bug 9368 [ALTERNATE] - specific behavior of yr and acqdate indexes

Bug-9368-ALTERNATE---specific-behavior-of-yr-and-a.patch (text/plain), 5.76 KB, created by Fridolin Somers on 2014-07-30 09:18:57 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Fridolin Somers

Created: 2014-07-30 09:18:57 UTC

Size: 5.76 KB

patch

obsolete

>From 87a1eed9e756d4d344f637a55229b408f49f4776 Mon Sep 17 00:00:00 2001
>From: David Cook <dcook@prosentient.com.au>
>Date: Wed, 18 Jun 2014 13:45:31 +1000
>Subject: [PATCH] Bug 9368 [ALTERNATE] - specific behavior of yr and acqdate
> indexes
>
>This patch fixes the "Publication date" and "Acquisition date"
>searches when using non-QueryParser and QueryParser searches.
>
>It adds structure attributes to the template, which is consistent
>with how phrase searching is currently handled.
>
>It removes unnecessary code from Search.pm, adds some necessary
>code which is consistent with existing code, and adds a lot of
>explanatory comments.
>
>_TEST PLAN_
>
>Before applying:
>0) Turn off QueryParser
>1) Try a "Publication date" or "Acquisition date" search from the
>staff client advanced search.
>2) Note that even though the description on the result page makes
>it seem like you're doing an index-specific search, you're actually
>doing a keyword search. You can verify this by checking the 008
>from positions 7 to 10 for "Publication date" or "Date accessioned"
>on items for "Acquisition date".
>
>3) Turn on QueryParser
>4) Try doing the searches from Step 1.
>5) A "Publication date" search should probably produce zero results
>
>After applying patch:
>6) Keep QueryParser on
>7) Try doing the searches from Step 1.
>8) Notice that you're actually getting results consistent with
>your search (ie the 008/7-10 shows the date you searched for,
>and there is a "Date accessioned" in items which matches your
>search)
>
>9) Turn off QueryParser
>10) Note that your results are exactly the same as step 8
>(N.B. this is because QueryParser is falling back to non-QP mode
>instead of producing a bad query.)
>
>Signed-off-by: Fridolin Somers <fridolin.somers@biblibre.com>
>---
> C4/Search.pm                                       | 27 +++++++++++++++++-----
> .../prog/en/includes/search_indexes.inc            | 13 +++++++++--
> 2 files changed, 32 insertions(+), 8 deletions(-)
>
>diff --git a/C4/Search.pm b/C4/Search.pm
>index 219de1e..ae7b850 100644
>--- a/C4/Search.pm
>+++ b/C4/Search.pm
>@@ -1371,17 +1371,28 @@ sub buildQuery {
>                 my $index   = $indexes[$i];
> 
>                 # Add index-specific attributes
>+
>+                #Afaik, this 'yr' condition will only ever be met in the staff client advanced search
>+                #for "Publication date", since typing 'yr:YYYY' into the search box produces a CCL query,
>+                #which is processed higher up in this sub. Other than that, year searches are typically
>+                #handled as limits which are not processed her either.
>+
>                 # Date of Publication
>-                if ( $index eq 'yr' ) {
>-                    $index .= ",st-numeric";
>-                    $indexes_set++;
>+                if ( $index =~ /yr/ ) {
>+                    #weight_fields/relevance search causes errors with date ranges
>+                    #In the case of YYYY-, it will only return records with a 'yr' of YYYY (not the range)
>+                    #In the case of YYYY-YYYY, it will return no results
> 					$stemming = $auto_truncation = $weight_fields = $fuzzy_enabled = $remove_stopwords = 0;
>                 }
> 
>                 # Date of Acquisition
>-                elsif ( $index eq 'acqdate' ) {
>-                    $index .= ",st-date-normalized";
>-                    $indexes_set++;
>+                elsif ( $index =~ /acqdate/ ) {
>+                    #stemming and auto_truncation would have zero impact since it already is YYYY-MM-DD format
>+                    #Weight_fields probably SHOULD be turned OFF, otherwise you'll get records floating to the
>+                      #top of the results just because they have lots of item records matching that date.
>+                    #Fuzzy actually only applies during _build_weighted_query, and is reset there anyway, so
>+                      #irrelevant here
>+                    #remove_stopwords doesn't function anymore so is irrelevant
> 					$stemming = $auto_truncation = $weight_fields = $fuzzy_enabled = $remove_stopwords = 0;
>                 }
>                 # ISBN,ISSN,Standard Number, don't need special treatment
>@@ -1568,9 +1579,13 @@ sub buildQuery {
>     # This is flawed , means we can't search anything with : in it
>     # if user wants to do ccl or cql, start the query with that
> #    $query =~ s/:/=/g;
>+    #NOTE: We use several several different regexps here as you can't have variable length lookback assertions
>     $query =~ s/(?<=(ti|au|pb|su|an|kw|mc|nb|ns)):/=/g;
>     $query =~ s/(?<=(wrdl)):/=/g;
>     $query =~ s/(?<=(trn|phr)):/=/g;
>+    $query =~ s/(?<=(st-numeric)):/=/g;
>+    $query =~ s/(?<=(st-year)):/=/g;
>+    $query =~ s/(?<=(st-date-normalized)):/=/g;
>     $limit =~ s/:/=/g;
>     for ( $query, $query_desc, $limit, $limit_desc ) {
>         s/  +/ /g;    # remove extra spaces
>diff --git a/koha-tmpl/intranet-tmpl/prog/en/includes/search_indexes.inc b/koha-tmpl/intranet-tmpl/prog/en/includes/search_indexes.inc
>index 67725c6..2492ec6 100644
>--- a/koha-tmpl/intranet-tmpl/prog/en/includes/search_indexes.inc
>+++ b/koha-tmpl/intranet-tmpl/prog/en/includes/search_indexes.inc
>@@ -40,6 +40,15 @@
>     [% END %]
>     <option value="pb">Publisher</option>
>     <option value="pl">Publisher location</option>
>-    <option value="yr">Publication date (yyyy)</option>
>-    <option value="acqdate">Acquisition date (yyyy-mm-dd)</option>
>+    [%#
>+        Use non-normalized st-year instead of st-numeric,
>+        since pubdate can include 'u' to signify unkown
>+        dates. See "Legal Characters" at:
>+        http://www.loc.gov/marc/bibliographic/bd008a.html
>+
>+        This search is also for date ranges due to the
>+        special Zebra r=r CCL mapping for 'yr'
>+    %]
>+    <option value="yr,st-year">Publication date (yyyy-yyyy)</option>
>+    <option value="acqdate,st-date-normalized">Acquisition date (yyyy-mm-dd)</option>
> </select>
>-- 
>1.9.1

Actions: View | Diff | Splinter Review

Attachments on bug 9368: 14472 | 15114 | 16268 | 26652 | 28916 | 30318 | 31552