implement alternate writings (failing tests down to 500)

1 files changed, 20 insertions, 8 deletions

diff --git a/import/jmdict/jmdict.ts b/import/jmdict/jmdict.ts
index 155c423..bf3614e 100644
--- a/import/jmdict/jmdict.ts
+++ b/import/jmdict/jmdict.ts
@@ -1,10 +1,15 @@
 import type { JMdict } from "npm:@scriptin/jmdict-simplified-types";
-import { Tag } from "../../language/tags.ts";
+import { Tag } from "../../search/tags.ts";
+import "../../util/string.ts";
 
 // this script is very messy right now, and doesn't transfer all information
 // present in the dictionary.
 //
 // proceed with caution
+//
+// TODO: separate term and glossary tags
+// TODO: dictionary normalization (numbers/half-width/長音符)
+// TODO: use sql synthesis library instead of garbo format strings
 
 const LANG = "eng";
 
@@ -47,8 +52,8 @@ const tagLookup = {
 console.log(`.param set :dict 'jmdict_${LANG}'`);
 console.log(".param set :lang 'en'");
 
-// TODO: separate term and glossary tags
-console.log("insert into ingest(expression, reading, term_tags, glossary_sort, glossary) values");
+var alts = "insert into alts(expression, reading, normal_expression, normal_reading) values\n";
+var ingest = "insert into ingest(expression, reading, term_tags, glossary_sort, glossary) values\n"
 
 // var max = -100;
 for (let i = 0; i < jmdict.words.length; i++) {
@@ -75,10 +80,17 @@ for (let i = 0; i < jmdict.words.length; i++) {
 	])];
 	var tags = tags.filter(i => i in tagLookup).map(i => tagLookup[i]);
 	// if (writing == "来る") console.log(term);
-	for (let j = 0; j < definitions.length; j++) {
-		var out = `\t('${writing}', '${reading}', '${tags.join(" ")}', ${j+1}, '${definitions[j].replaceAll("'", "''")}')${(last && j == definitions.length-1) ? ';' : ','}`;
-		if (j == 0 && other_writings.length > 0) out += ` -- TODO: alts: ${other_writings.join(", ")}`;
-		console.log(out);
-	}
+	definitions.forEach((definition, j) => {
+		ingest += `\t('${writing}', '${reading}', '${tags.join(" ")}', ${j+1}, '${definition.replaceAll("'", "''")}'),\n`;
+	});
+	other_writings.forEach(alt => {
+		alts += `\t('${alt}', '${reading}', '${writing}', '${reading}'),\n`;
+	});
 }
 
+ingest = ingest.replaceLast(",", ";");
+alts = alts.replaceLast(",", ";");
+
+console.log(ingest);
+console.log(alts);
+