1 files changed, 59 insertions, 16 deletions

diff --git a/src/lib/db/thaiseed.ts b/src/lib/db/thaiseed.ts
index 687f0f3..5c75345 100644
--- a/src/lib/db/thaiseed.ts
+++ b/src/lib/db/thaiseed.ts
@@ -24,9 +24,9 @@ async function readDump(lang: string) {
   // langrows = langrows.slice(10);
   for (const langrow of langrows) {
     count++;
-    console.log(count);
+    // console.log(count);
     // if (count <= 10000) continue;
-    // if (count > 30) break;
+    // if (count > 100) break;
     const j = JSON.parse(langrow.data);
     const word = j.word.trim();
     if (!word) continue;
@@ -48,7 +48,6 @@ async function handleWord(word: string, j: any) {
   const freq = await getThaiFreq(word);
   const wordId = pdb.addWord(word, "th", freq, null);
   const analyzed = await analyzeTHWord(word);
-  // console.log(analyzed);
   for (let snd of sounds) if ("ipa" in snd) handleIpa(wordId, j, snd, analyzed);
 }
 async function handleIpa(
@@ -66,27 +65,39 @@ async function handleIpa(
   const wikiIpaSplit = wikiIpa.split(".");
   const nlpIpaSplit = nlpIpa.split(".");
   if (wikiIpaSplit.length !== nlpIpaSplit.length) {
-    console.log("ipa mismatch");
-    console.log(wikiIpa);
-    console.log(nlpIpa);
-    // return;
+    // console.log("ipa mismatch");
+    // console.log(wikiIpa);
+    // console.log(nlpIpa);
   }
-  if (analyzed.syllables.length !== wikiIpaSplit.length) {
-    console.log("syllable analysis mismatch", j.word);
-    console.log({ syls: analyzed.syllables, ipa: wikiIpaSplit });
+  if (analyzed.realSyls.length !== wikiIpaSplit.length) {
+    // console.log("syllable analysis mismatch", j.word);
+    // console.log({ syls: analyzed.syllables, ipa: wikiIpaSplit });
     // console.dir(j, { depth: null });
     return;
   }
   pdb.addPronunciation(wordId, ipa, analyzed.syllables.length, tags, null);
+  const writtenSyls = analyzed.syllables;
+  const pronouncedSyls = analyzed.realSyls;
+  let badSyls = false;
+  if (writtenSyls.length !== pronouncedSyls.length) badSyls = true;
 
-  for (let i = 0; i < analyzed.syllables.length; i++) {
-    const spelling = analyzed.syllables[i]!;
+  for (let i = 0; i < pronouncedSyls.length; i++) {
+    const pronounced = pronouncedSyls[i]!.replace(/\u{E3A}/u, "");
+    const written = writtenSyls[i] || "";
+    const syllable = badSyls ? pronounced : written;
     const ipa = wikiIpaSplit[i]!;
+    // TODO insert both??
+    const notes = pronounced === written ? null : `Pronounced ${pronounced}`;
+    if (pronounced !== syllable) {
+      console.log("diff");
+      console.log(pronounced);
+      console.log(written);
+    }
     try {
-      await handleSyllable(spelling, ipa, wordId, i);
+      await handleSyllable(syllable, ipa, wordId, i, notes);
     } catch (e) {
       console.error("syl error", j.word, j.sounds);
-      console.error({ spelling, ipa, wikiIpaSplit });
+      console.error({ analyzed, ipa, wikiIpaSplit });
       console.error(e);
     }
   }
@@ -115,16 +126,48 @@ function parseTone(ipa: string, spelling: string): Tone {
     throw new Error("");
   }
 }
+
 async function handleSyllable(
   spelling: string,
   ipa: string,
   wordId: number | bigint,
   idx: number,
+  notes: string | null,
 ) {
   const sorsyl = await sorSyl(spelling, "th", ipa);
+  const weird = [
+    // "a̯n",
+    // "a̯",
+    // "a̯p",
+    // "a̯w",
+    // "a̯j",
+    // "a̯ŋ",
+    // "a̯k",
+    // "a̯t",
+    // "a̯m",
+    // "a̯ʔ",
+    // "ʔ",
+    "s",
+    "l",
+    "f",
+    "a̯s",
+    "js",
+    "t͡ɕʰ",
+    "ks",
+    "ns",
+    "a̯l",
+    "a̯f",
+    "mk",
+  ];
+  // const weirder = sorsyl.syls.find((s) => weird.includes(s.coda));
+  // if (weirder) {
+  //   console.log("syllable", spelling);
+  //   // console.dir(sorsyl, { depth: null });
+  //   // console.dir(j, { depth: null });
+  // }
   if (sorsyl.syls.length !== 1) throw new Error("wtf sorsyl!");
   const syl = sorsyl.syls[0]!;
-  const tone = syl.tone ? parseTone(syl.tone, spelling) : null;
+  const tone = parseTone(syl.tone, spelling);
   try {
     pdb.addSyllable(
       wordId,
@@ -139,7 +182,7 @@ async function handleSyllable(
       { spelling: syl.coda, ipa: syl.coda },
       { spelling: syl.rhyme, ipa: syl.rhyme },
       tone,
-      null,
+      notes,
     );
   } catch (e) {
     // console.log("well fuck", syl);