From 0585e2ef013aad054b07c594a1af39bd0dabf795 Mon Sep 17 00:00:00 2001
From: Deadbeef <ent3rm4n@gmail.com>
Date: Tue, 24 Oct 2023 23:25:34 +0800
Subject: [PATCH] update pagelinks queries to use linktarget

---
 dbreps2/src/enwiki/boteditcount.rs             |  5 +++--
 dbreps2/src/enwiki/editcount.rs                |  7 ++++---
 dbreps2/src/enwiki/linkedmiscapitalizations.rs |  5 +++--
 dbreps2/src/enwiki/linkedmisspellings.rs       |  5 +++--
 dbreps2/src/enwiki/templatedisambigs.rs        | 13 +++++++------
 dbreps2/src/general/linkedredlinkedcats.rs     |  5 +++--
 dbreps2/src/general/userlinksinarticles.rs     |  3 ++-
 7 files changed, 25 insertions(+), 18 deletions(-)

diff --git a/dbreps2/src/enwiki/boteditcount.rs b/dbreps2/src/enwiki/boteditcount.rs
index c546e250..db721c13 100644
--- a/dbreps2/src/enwiki/boteditcount.rs
+++ b/dbreps2/src/enwiki/boteditcount.rs
@@ -40,11 +40,12 @@ SELECT
 FROM page
 JOIN pagelinks
 ON pl_from = page_id
+JOIN linktarget on pl_target_id = lt_id
 JOIN user
-ON user_name = REPLACE(pl_title, "_", " ")
+ON user_name = REPLACE(lt_title, "_", " ")
 WHERE page_title = ?
 AND page_namespace = 4
-AND pl_namespace IN (2,3);
+AND lt_namespace IN (2,3);
 "#,
             (page,),
             |(user_name, user_editcount)| (user_name, user_editcount),
diff --git a/dbreps2/src/enwiki/editcount.rs b/dbreps2/src/enwiki/editcount.rs
index 4cc6b931..b19e842d 100644
--- a/dbreps2/src/enwiki/editcount.rs
+++ b/dbreps2/src/enwiki/editcount.rs
@@ -32,16 +32,17 @@ async fn get_user_list(conn: &mut Conn, page: &str) -> Result<HashSet<String>> {
             r#"
 /* editcount.rs SLOW_OK */
 SELECT DISTINCT
- pl_title
+ lt_title
 FROM page
 JOIN pagelinks
 ON pl_from = page_id
+JOIN linktarget on pl_target_id = lt_id
 WHERE page_title = ?
 AND page_namespace = 4
-AND pl_namespace IN (2,3);
+AND lt_namespace IN (2,3);
 "#,
             (page,),
-            |(pl_title,)| pl_title,
+            |(lt_title,)| lt_title,
         )
         .await?;
     Ok(rows
diff --git a/dbreps2/src/enwiki/linkedmiscapitalizations.rs b/dbreps2/src/enwiki/linkedmiscapitalizations.rs
index a93ae9b6..4ebd9f79 100644
--- a/dbreps2/src/enwiki/linkedmiscapitalizations.rs
+++ b/dbreps2/src/enwiki/linkedmiscapitalizations.rs
@@ -47,8 +47,9 @@ SELECT
 FROM
   page AS p1
   JOIN categorylinks ON p1.page_id = cl_from
-  JOIN pagelinks ON p1.page_title = pl_title
-  AND pl_namespace = 0
+  JOIN linktarget on p1.page_title = lt_title
+  AND lt_namespace = 0
+  JOIN pagelinks ON pl_target_id = lt_id
   JOIN page AS p2 ON pl_from = p2.page_id
   AND p2.page_namespace = 0
 WHERE
diff --git a/dbreps2/src/enwiki/linkedmisspellings.rs b/dbreps2/src/enwiki/linkedmisspellings.rs
index 2b381709..5f896742 100644
--- a/dbreps2/src/enwiki/linkedmisspellings.rs
+++ b/dbreps2/src/enwiki/linkedmisspellings.rs
@@ -47,8 +47,9 @@ SELECT
 FROM
   page AS p1
   JOIN categorylinks ON p1.page_id = cl_from
-  JOIN pagelinks ON p1.page_title = pl_title
-  AND pl_namespace = 0
+  JOIN linktarget on p1.page_title = lt_title
+  AND lt_namespace = 0
+  JOIN pagelinks ON pl_target_id = lt_id
   JOIN page AS p2 ON pl_from = p2.page_id
   AND p2.page_namespace = 0
 WHERE
diff --git a/dbreps2/src/enwiki/templatedisambigs.rs b/dbreps2/src/enwiki/templatedisambigs.rs
index 867dfd5b..bea4a027 100644
--- a/dbreps2/src/enwiki/templatedisambigs.rs
+++ b/dbreps2/src/enwiki/templatedisambigs.rs
@@ -44,7 +44,7 @@ impl Report<Row> for TemplateDisambigs {
 /* templatedisambigs.rs SLOW_OK */
 SELECT
   pltmp.page_title AS template_title,
-  pltmp.pl_title AS disambiguation_title,
+  pltmp.lt_title AS disambiguation_title,
   (
     SELECT
       COUNT(*)
@@ -60,21 +60,22 @@ FROM
     SELECT
       page_namespace,
       page_title,
-      pl_namespace,
-      pl_title
+      lt_namespace,
+      lt_title
     FROM
       page
       JOIN pagelinks ON pl_from = page_id
+      JOIN linktarget ON pl_target_id = lt_id
     WHERE
       page_namespace = 10
-      AND pl_namespace = 0
+      AND lt_namespace = 0
     LIMIT
       1000000
   ) AS pltmp
   JOIN page AS pg2
   /* removes red links */
-  ON pltmp.pl_namespace = pg2.page_namespace
-  AND pltmp.pl_title = pg2.page_title
+  ON pltmp.lt_namespace = pg2.page_namespace
+  AND pltmp.lt_title = pg2.page_title
 WHERE
   EXISTS (
     SELECT
diff --git a/dbreps2/src/general/linkedredlinkedcats.rs b/dbreps2/src/general/linkedredlinkedcats.rs
index 9c0dcf84..072fd34d 100644
--- a/dbreps2/src/general/linkedredlinkedcats.rs
+++ b/dbreps2/src/general/linkedredlinkedcats.rs
@@ -45,9 +45,10 @@ SELECT
   cl_to,
   COUNT(*)
 FROM
+  /* FIXME when categorylinks gets normalized as well */
   categorylinks
-  JOIN pagelinks ON pl_title = cl_to
-  AND pl_namespace = 14
+  JOIN linktarget ON lt_title = cl_to AND lt_namespace = 14
+  JOIN pagelinks ON pl_target_id = lt_id
   JOIN page AS p1 ON pl_from = p1.page_id
   AND p1.page_namespace IN (0, 6, 10, 12, 14, 100)
   LEFT JOIN page AS p2 ON cl_to = p2.page_title
diff --git a/dbreps2/src/general/userlinksinarticles.rs b/dbreps2/src/general/userlinksinarticles.rs
index 1af9af3d..f21815fd 100644
--- a/dbreps2/src/general/userlinksinarticles.rs
+++ b/dbreps2/src/general/userlinksinarticles.rs
@@ -45,9 +45,10 @@ SELECT
 FROM
   page
   JOIN pagelinks ON pl_from = page_id
+  JOIN linktarget ON pl_target_id = lt_id
 WHERE
   pl_from_namespace = 0
-  AND pl_namespace IN (2, 3)
+  AND lt_namespace IN (2, 3)
   AND NOT EXISTS (
     SELECT 1 FROM templatelinks
     JOIN linktarget ON tl_target_id = lt_id