From f78d4c2158439fea62c73b8aa9568f87842b09f7 Mon Sep 17 00:00:00 2001
From: Thomas Hochstein <thh@thh.name>
Date: Thu, 29 May 2025 18:03:44 +0200
Subject: [PATCH] Refactor getting raw headers.

Signed-off-by: Thomas Hochstein <thh@thh.name>
---
 bin/gatherstats.pl | 46 +++++++++++++++++++++++++++++++---------------
 doc/ChangeLog      |  1 +
 2 files changed, 32 insertions(+), 15 deletions(-)

diff --git a/bin/gatherstats.pl b/bin/gatherstats.pl
index ca90cc7..6118a70 100755
--- a/bin/gatherstats.pl
+++ b/bin/gatherstats.pl
@@ -261,21 +261,7 @@ sub HostStats {
 
   my (%Postings,$DBQuery);
 
-  if (!$MID) {
-    # get raw header data from raw table for given month
-    $DBQuery = $DBHandle->prepare(sprintf("SELECT newsgroups,headers FROM %s ".
-                                             "WHERE day LIKE ? AND NOT disregard",
-                                             $DBRaw));
-    $DBQuery->execute($Month.'-%')
-      or &Bleat(2,sprintf("Can't get hosts data for %s from %s: ".
-                          "$DBI::errstr\n",$Month,$DBRaw));
-  } else {
-    $DBQuery = $DBHandle->prepare(sprintf("SELECT newsgroups,headers FROM %s ".
-                                             "WHERE mid = ?", $DBRaw));
-    $DBQuery->execute($MID)
-      or &Bleat(2,sprintf("Can't get hosts data for %s from %s: ".
-                          "$DBI::errstr\n",$MID,$DBRaw));
-  }
+  $DBQuery = GetHeaders($DBHandle,$DBRaw,$Month,$MID);
 
   ### ----------------------------------------------
   print "----- HostStats -----\n" if $Debug;
@@ -372,6 +358,36 @@ sub HostStats {
   };
 };
 
+sub GetHeaders {
+### ----------------------------------------------------------------------------
+### get (newsgroups and) raw headers from database
+### IN : $DBHandle : database handle
+###      $DBRaw    : database table for raw data (to read from)
+###      $Month    : current month to do
+###      $MID      : specific Message-ID to fetch (testing purposes)
+### OUT: DBI statement handle
+  my ($DBHandle,$DBRaw,$Month,$MID) = @_;
+
+  my $DBQuery;
+
+  if (!$MID) {
+    # get raw header data from raw table for given month
+    $DBQuery = $DBHandle->prepare(sprintf("SELECT newsgroups,headers FROM %s ".
+                                             "WHERE day LIKE ? AND NOT disregard",
+                                             $DBRaw));
+    $DBQuery->execute($Month.'-%')
+      or &Bleat(2,sprintf("Can't get header data for %s from %s: ".
+                          "$DBI::errstr\n",$Month,$DBRaw));
+  } else {
+    $DBQuery = $DBHandle->prepare(sprintf("SELECT newsgroups,headers FROM %s ".
+                                             "WHERE mid = ?", $DBRaw));
+    $DBQuery->execute($MID)
+      or &Bleat(2,sprintf("Can't get header data for %s from %s: ".
+                          "$DBI::errstr\n",$MID,$DBRaw));
+  }
+  return $DBQuery;
+}
+
 sub CheckTLH {
 ### ----------------------------------------------------------------------------
 ### count newsgroups from legal TLH(s)
diff --git a/doc/ChangeLog b/doc/ChangeLog
index 9f248cc..3f46f78 100644
--- a/doc/ChangeLog
+++ b/doc/ChangeLog
@@ -1,6 +1,7 @@
 NewsStats 0.4.0 (unreleased)
   * Reformat $Conf{TLH} for GroupStats only.
   * Extract TLH check from HostStats to subroutine, fix no-op check.
+  * Extract getting raw headers from HostStats to subroutine.
 
 NewsStats 0.3.0 (2025-05-18)
   * Extract GroupStats (in gatherstats) to subroutine.