Add verse_dictionary SQL function for lexical lookup

Enable clients to query dictionary entries (topic, word, meaning, parse, and forms) for each word in a verse by tokenizing Greek text with greek_to_betacode and parsing XML definitions. Grant web_anon SELECT and EXECUTE rights on dictionary tables and the new function.

Author: placek

sql/12_functions.sql

@@ -631,3 +631,74 @@ BEGIN
   RETURN result;
 END;
 $BODY$;
+
+-- retrieves dictionary entries for words in a given verse
+DROP FUNCTION IF EXISTS public.verse_dictionary(text);
+CREATE OR REPLACE FUNCTION public.verse_dictionary(p_verse_id text)
+  RETURNS TABLE(topic text, word text, meaning text, parse text, forms text[])
+  LANGUAGE 'sql'
+  COST 100
+  STABLE PARALLEL UNSAFE
+  ROWS 1000
+AS $BODY$
+  WITH words AS (
+    SELECT DISTINCT
+           unnest(
+             string_to_array(
+               greek_to_betacode(public.raw_text(v.text)),
+               ' '
+             )
+           ) AS word
+    FROM public._all_verses v
+    WHERE v.id = p_verse_id
+  ),
+  dict AS (
+    SELECT
+        d.topic,
+        xmlparse(content '<root>' || d.definition || '</root>') AS x
+    FROM public._all_dictionary_entries d
+    JOIN words w
+      ON w.word = d.topic
+  ),
+  extracted AS (
+    SELECT
+      d.topic,
+
+      -- dt/a text from first <dl>
+      trim((xpath('/root/dl[1]/dt/a/text()', x))[1]::text) AS word,
+
+      -- meaning
+      trim(
+        array_to_string(
+          xpath('//span[@class="meaning"]//text()', x),
+          ' '
+        )
+      ) AS meaning,
+
+      -- main parse
+      trim(
+        (xpath('/root/dl[1]//span[@class="parse"][1]/text()', x))[1]::text
+      ) AS parse,
+
+      -- raw Greek forms (array of text)
+      xpath(
+        '/root/dl[1]//li/a[starts-with(@href, ''S:'')]/text()',
+        x
+      )::text[] AS forms_raw
+
+    FROM dict d
+  )
+  SELECT
+      topic,
+      word,
+      meaning,
+      parse,
+      COALESCE(
+        (
+          SELECT array_agg(greek_to_betacode(f))
+          FROM unnest(forms_raw) f
+        ),
+        '{}'
+      ) AS forms
+  FROM extracted;
+$BODY$;

sql/13_postgrest.sql

@@ -13,6 +13,7 @@ GRANT SELECT ON public._all_sources TO web_anon;
 GRANT SELECT ON public._all_books TO web_anon;
 GRANT SELECT ON public._all_verses TO web_anon;
 GRANT SELECT ON public._all_commentaries TO web_anon;
+GRANT SELECT ON public._all_dictionary_entries TO web_anon;
 GRANT SELECT ON public._cross_references TO web_anon;
 GRANT SELECT ON public._rendered_stories TO web_anon;
 GRANT EXECUTE ON FUNCTION public.fetch_commentaries(text) TO web_anon;
@@ -22,3 +23,4 @@ GRANT EXECUTE ON FUNCTION public.fetch_verse_with_metadata(text) TO web_anon;
 GRANT EXECUTE ON FUNCTION public.fetch_verses_by_address(text, text, boolean) TO web_anon;
 GRANT EXECUTE ON FUNCTION public.parse_address(text) TO web_anon;
 GRANT EXECUTE ON FUNCTION public.search_verses(text) TO web_anon;
+GRANT EXECUTE ON FUNCTION public.verse_dictionary(text) TO web_anon;