From 002371920d6f9acedaf987f273bb138a95353ffd Mon Sep 17 00:00:00 2001
From: Miquel Sabaté Solà <msabate@suse.com>
Date: Fri, 15 Mar 2024 10:38:05 +0100
Subject: Fixed search for tags with spaces
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

We should not split the body mindlessly, but we have to be more careful
about it.

Moreover, this commit also polishes the order for tags inside of
comments.

Signed-off-by: Miquel Sabaté Solà <msabate@suse.com>
---
 app/models/search.rb | 27 ++++++++++++++++++++++++++-
 1 file changed, 26 insertions(+), 1 deletion(-)

(limited to 'app/models/search.rb')

diff --git a/app/models/search.rb b/app/models/search.rb
index 7823a74..053e641 100644
--- a/app/models/search.rb
+++ b/app/models/search.rb
@@ -31,7 +31,7 @@ class Search < ApplicationRecord
   def parse_body
     res = { tag: [], plain: [] }
 
-    body.split.each do |part|
+    split_body.each do |part|
       if part.include?(':')
         parts = part.split(':', 2)
         parts[1] = clean_clause(part: parts[1])
@@ -47,6 +47,31 @@ class Search < ApplicationRecord
     res
   end
 
+  # Returns the body split into its constituents. Note that we cannot mindlessly
+  # do something like `body.split` because that would ignore quotes that are not
+  # to be split among other things.
+  def split_body
+    cur = ''
+    res = []
+    co  = ''
+
+    body.each_char do |c|
+      if c.match?(/\s/) && co == ''
+        res << cur if cur.present?
+        cur = ''
+      else
+        # If it is a quote character, check whether it's being opened or
+        # closed.
+        co = c == co ? '' : c if c == '"' || c == "'"
+
+        cur += c
+      end
+    end
+
+    res << cur if cur.present? && co.blank?
+    res
+  end
+
   # Returns the argument without any leading/trailing quotes.
   def clean_clause(part:)
     part.gsub(/^("|')+/, '').gsub(/("|')+$/, '')
-- 
cgit v1.2.3