Add DB support for merged_with column

This is meant to provide the foundation for gracefully redirecting requests for
old URLs to the new URLs. For now we can start filtering out search requests,
etc.

Signed-off-by: Dan Scott <[email protected]>

ris2sql

@@ -85,6 +85,7 @@ class RISParser:
             "ALTER TABLE citations ADD COLUMN series TEXT;",
             "ALTER TABLE citations ADD COLUMN pub_title TEXT;",
             "ALTER TABLE citations ADD COLUMN kw_tsv TSVECTOR;",
+            "ALTER TABLE citations ADD COLUMN merged_with TEXT;",
             r"""UPDATE citations SET pub_date = NULL WHERE pub_date !~ '^\d{4}';""",
             "DROP FUNCTION IF EXISTS ris_search(TEXT);",
             "DROP TYPE IF EXISTS search_result CASCADE;",
@@ -146,10 +147,11 @@ class RISParser:
             r"""UPDATE citations SET kw_tsv = kw_tsv || setweight(to_tsvector('english', COALESCE(
                  (SELECT STRING_AGG(author_name, ' ') FROM authored_v WHERE citation = citations.id), '')), 'B');""",
             "CREATE INDEX kw_tsv_index ON citations USING GIN(kw_tsv);",
-            r"""CREATE FUNCTION ris_search(query TEXT) RETURNS SETOF search_result AS $$
+            r"""CREATE OR REPLACE FUNCTION ris_search(query TEXT) RETURNS SETOF search_result AS $$
                  SELECT id, title, substring(pub_date from 1 for 4)::INT AS pub_year, doc_type, ts_rank_cd(kw_tsv, to_tsquery(query))
                  FROM citations
                 WHERE kw_tsv @@ to_tsquery(query)
+                 AND merged_with IS NULL
                 ORDER BY 3 DESC$$ LANGUAGE SQL;""",
 
 

ris2web

@@ -355,7 +355,9 @@ def home_html():
         with conn:
             with conn.cursor() as t:
                 t.execute(r"""SELECT COUNT(*)
-                    FROM citations;"""
+                    FROM citations
+                    WHERE merged_with IS NULL
+		    ;"""
                 )
                 total = t.fetchone()[0]
 
@@ -378,21 +380,27 @@ def home_html():
                 p.execute(r"""SELECT pub_title, COUNT(*)
                     FROM citations
                     WHERE pub_title IS NOT NULL
+                      AND merged_with IS NULL
                     GROUP BY pub_title
                     ORDER BY 2 DESC
                     LIMIT 50;"""
                 )
                 pubs = p.fetchall()
 
             with conn.cursor() as pc:
-                pc.execute(r"""WITH x AS (SELECT DISTINCT pub_title FROM citations)
+                pc.execute(r"""WITH x AS (
+                      SELECT DISTINCT pub_title
+		      FROM citations
+		      WHERE merged_with IS NULL
+                    )
                     SELECT COUNT(*) FROM x"""
                 )
                 pcnt = pc.fetchone()[0]
 
             with conn.cursor() as t:
                 t.execute(r"""SELECT doc_type, COUNT(*)
                     FROM citations
+                    WHERE merged_with IS NULL
                     GROUP BY doc_type
                     ORDER BY 2 DESC"""
                 )
@@ -402,6 +410,7 @@ def home_html():
             with conn.cursor() as c:
                 c.execute(r"""SELECT language, COUNT(*)
                     FROM citations
+                    WHERE merged_with IS NULL
                     GROUP BY language
                     ORDER BY 2 DESC"""
                 )
@@ -411,6 +420,7 @@ def home_html():
                 c.execute(r"""SELECT series, COUNT(*)
                     FROM citations
                     WHERE series IS NOT NULL
+                      AND merged_with IS NULL
                     GROUP BY series 
                     ORDER BY 2 DESC"""
                 )