accessify-widget 0.3.98 → 0.3.99

package/dist/{text-simplify-W0rPYWpv.js → text-simplify-DUdShGEx.js}

@@ -1,3 +1,50 @@
+function tokenize(s) {
+  const tokens = /* @__PURE__ */ new Set();
+  const parts = s.toLowerCase().replace(/[^\p{L}\p{N}\s]/gu, " ").split(/\s+/);
+  for (const t of parts) {
+    if (t.length >= 3) tokens.add(t);
+  }
+  return tokens;
+}
+function tokenSimilarity(a, b) {
+  if (a.size === 0 || b.size === 0) return 0;
+  let inter = 0;
+  for (const t of a) if (b.has(t)) inter++;
+  const uni = a.size + b.size - inter;
+  return uni === 0 ? 0 : inter / uni;
+}
+const STALE_SIMILARITY_THRESHOLD = 0.4;
+const CONTENT_ANCHOR_DEPTH = 4;
+function isWholePageAncestor(a) {
+  const tag = a.tagName;
+  return tag === "BODY" || tag === "HTML";
+}
+function buildContentAnchors(applied) {
+  const anchors = /* @__PURE__ */ new Set();
+  for (const el of applied) {
+    let a = el;
+    for (let i = 0; i < CONTENT_ANCHOR_DEPTH && a; i++) {
+      if (isWholePageAncestor(a)) break;
+      anchors.add(a);
+      a = a.parentElement;
+    }
+  }
+  return anchors;
+}
+function isNearContent(el, anchors) {
+  let a = el;
+  for (let i = 0; i < CONTENT_ANCHOR_DEPTH && a; i++) {
+    if (isWholePageAncestor(a)) return false;
+    if (anchors.has(a)) return true;
+    a = a.parentElement;
+  }
+  return false;
+}
+function isObviousBoilerplate(el) {
+  return !!el.closest(
+    'nav, [role="navigation"], [role="menu"], [role="menubar"], [role="menuitem"]'
+  );
+}
 function createTextSimplifyModule(aiService, lang = "de", options) {
   let enabled = false;
   const defaultLevel = typeof options === "string" ? options : options?.simplificationLevel || void 0;
@@ -14,6 +61,89 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
   const SIMPLIFIED_ATTR = "data-accessify-simplified";
   const ORIGINAL_ATTR = "data-accessify-original";
   const BLOCK_HASH_ATTR = "data-accessify-block-hash";
+  let diagRun = null;
+  const diagHistory = [];
+  const DIAG = () => !!window.__ACCESSIFY_DIAG;
+  function diagInit(route, siteKey) {
+    diagRun = {
+      ts: (/* @__PURE__ */ new Date()).toISOString(),
+      route,
+      level,
+      siteMode: currentSiteMode,
+      siteKey,
+      manifestLoaded: false,
+      manifestBlocks: 0,
+      elementsFound: 0,
+      phase1Applied: 0,
+      phase1Stale: 0,
+      phase15IdbHits: 0,
+      phase2AICalls: 0,
+      phase2AIErrors: 0,
+      totalReplaced: 0,
+      totalSkipped: 0,
+      blocks: [],
+      restoreLog: []
+    };
+  }
+  function diagBlock(partial) {
+    const entry = {
+      hash: partial.hash,
+      text: partial.text || "",
+      tag: partial.tag || "",
+      manifestMatch: partial.manifestMatch || "none",
+      idbHit: partial.idbHit || false,
+      liveAI: partial.liveAI || false,
+      persistOk: partial.persistOk ?? null,
+      replaced: partial.replaced || false,
+      skipReason: partial.skipReason || null,
+      stale: partial.stale || false,
+      source: partial.source || "skipped"
+    };
+    diagRun?.blocks.push(entry);
+    return entry;
+  }
+  function diagFinish() {
+    if (!diagRun) return;
+    diagRun.totalReplaced = diagRun.blocks.filter((b) => b.replaced).length;
+    diagRun.totalSkipped = diagRun.blocks.filter((b) => !b.replaced).length;
+    diagHistory.push(diagRun);
+    window.__accessifyDiag = diagHistory;
+    if (DIAG()) {
+      console.groupCollapsed(
+        `%c[Accessify DIAG] Run complete: ${diagRun.route}`,
+        "color: #38bdf8; font-weight: bold"
+      );
+      console.table({
+        route: diagRun.route,
+        level: diagRun.level,
+        siteMode: diagRun.siteMode,
+        manifestLoaded: diagRun.manifestLoaded,
+        manifestBlocks: diagRun.manifestBlocks,
+        elementsFound: diagRun.elementsFound,
+        "Phase 1 (manifest)": diagRun.phase1Applied,
+        "Phase 1 stale": diagRun.phase1Stale,
+        "Phase 1.5 (IDB)": diagRun.phase15IdbHits,
+        "Phase 2 (AI)": diagRun.phase2AICalls,
+        "Phase 2 errors": diagRun.phase2AIErrors,
+        totalReplaced: diagRun.totalReplaced,
+        totalSkipped: diagRun.totalSkipped
+      });
+      console.table(diagRun.blocks.map((b) => ({
+        hash: b.hash,
+        text: b.text,
+        tag: b.tag,
+        source: b.source,
+        manifest: b.manifestMatch,
+        idb: b.idbHit,
+        ai: b.liveAI,
+        persist: b.persistOk,
+        replaced: b.replaced,
+        skip: b.skipReason,
+        stale: b.stale
+      })));
+      console.groupEnd();
+    }
+  }
   const CLIENT_CACHE_DB = "accessify-simplify-cache";
   const CLIENT_CACHE_STORE = "blocks";
   const CLIENT_CACHE_VERSION = 1;
@@ -461,6 +591,7 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
         bySelector.set(block.selector, block);
       }
     }
+    const claimedBlockHashes = /* @__PURE__ */ new Set();
     for (const el of elements) {
       const text = el.dataset.accessifyOriginal || el.textContent?.trim() || "";
       const elHash = hashText(text);
@@ -469,8 +600,11 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
         const success = safeReplace(el, exactMatch.result, elHash);
         if (success) {
           applied.push(el);
+          claimedBlockHashes.add(exactMatch.blockHash);
+          if (DIAG()) diagBlock({ hash: elHash, text: text.slice(0, 60), tag: el.tagName, manifestMatch: "exact", replaced: true, source: "manifest" });
         } else {
           remaining.push(el);
+          if (DIAG()) diagBlock({ hash: elHash, text: text.slice(0, 60), tag: el.tagName, manifestMatch: "exact", replaced: false, skipReason: "layout-unsafe", source: "skipped" });
         }
         continue;
       }
@@ -490,8 +624,11 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
           const success = safeReplace(el, fuzzyMatch.result, elHash);
           if (success) {
             applied.push(el);
+            claimedBlockHashes.add(fuzzyMatch.blockHash);
+            if (DIAG()) diagBlock({ hash: elHash, text: text.slice(0, 60), tag: el.tagName, manifestMatch: "fuzzy", replaced: true, source: "manifest" });
           } else {
             remaining.push(el);
+            if (DIAG()) diagBlock({ hash: elHash, text: text.slice(0, 60), tag: el.tagName, manifestMatch: "fuzzy", replaced: false, skipReason: "layout-unsafe", source: "skipped" });
           }
         } else {
           console.info(
@@ -499,11 +636,67 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
             text.slice(0, 50) + "…"
           );
           stale.push(el);
+          claimedBlockHashes.add(fuzzyMatch.blockHash);
+          if (DIAG()) diagBlock({ hash: elHash, text: text.slice(0, 60), tag: el.tagName, manifestMatch: "fuzzy", stale: true, skipReason: `stale(diff=${diff})`, source: "skipped" });
         }
         continue;
       }
       remaining.push(el);
     }
+    const unclaimedBlocks = manifest.blocks.filter(
+      (b) => b.originalText && b.result && !claimedBlockHashes.has(b.blockHash)
+    );
+    if (unclaimedBlocks.length > 0 && remaining.length > 0) {
+      const blockTokens = unclaimedBlocks.map((b) => ({
+        block: b,
+        tokens: tokenize(b.originalText)
+      }));
+      const candidates = [];
+      for (const el of remaining) {
+        const text = el.dataset.accessifyOriginal || el.textContent?.trim() || "";
+        if (text.length < 20) continue;
+        const elTokens = tokenize(text);
+        if (elTokens.size < 3) continue;
+        let best = null;
+        for (const { block, tokens } of blockTokens) {
+          const sim = tokenSimilarity(elTokens, tokens);
+          if (sim >= STALE_SIMILARITY_THRESHOLD && (!best || sim > best.sim)) {
+            best = { el, block, sim };
+          }
+        }
+        if (best) candidates.push(best);
+      }
+      candidates.sort((a, b) => b.sim - a.sim);
+      const assignedBlocks = /* @__PURE__ */ new Set();
+      const assignedEls = /* @__PURE__ */ new Set();
+      for (const c of candidates) {
+        if (assignedBlocks.has(c.block.blockHash)) continue;
+        if (assignedEls.has(c.el)) continue;
+        assignedBlocks.add(c.block.blockHash);
+        assignedEls.add(c.el);
+        stale.push(c.el);
+        claimedBlockHashes.add(c.block.blockHash);
+        const elText = c.el.dataset.accessifyOriginal || c.el.textContent?.trim() || "";
+        console.info(
+          `[Accessify] Stale block (similarity=${c.sim.toFixed(2)}):`,
+          elText.slice(0, 50) + "…"
+        );
+        if (DIAG()) diagBlock({
+          hash: hashText(elText),
+          text: elText.slice(0, 60),
+          tag: c.el.tagName,
+          manifestMatch: "similarity",
+          stale: true,
+          skipReason: `stale(sim=${c.sim.toFixed(2)})`,
+          source: "skipped"
+        });
+      }
+      if (assignedEls.size > 0) {
+        const stillRemaining = remaining.filter((el) => !assignedEls.has(el));
+        remaining.length = 0;
+        remaining.push(...stillRemaining);
+      }
+    }
     return { applied, remaining, stale };
   }
   function buildSelector(el) {
@@ -549,11 +742,37 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
         if (window.__ACCESSIFY_DEBUG) {
           console.log(`[Accessify] Persisted ${blocks.length} blocks to D1+KV`);
         }
+        if (DIAG() && diagRun) {
+          for (const b of blocks) {
+            const entry = diagRun.blocks.find((d) => d.hash === b.blockHash);
+            if (entry) entry.persistOk = true;
+          }
+          console.log(
+            `%c[Accessify DIAG] Persist OK: ${blocks.length} blocks → ${pageUrl}`,
+            "color: #22c55e"
+          );
+        }
       } else {
         console.warn(`[Accessify] Persist failed: ${res.status}`);
+        if (DIAG() && diagRun) {
+          for (const b of blocks) {
+            const entry = diagRun.blocks.find((d) => d.hash === b.blockHash);
+            if (entry) entry.persistOk = false;
+          }
+          console.warn(
+            `%c[Accessify DIAG] Persist FAILED (${res.status}): ${blocks.length} blocks`,
+            "color: #ef4444"
+          );
+        }
       }
     }).catch((err) => {
       console.warn("[Accessify] Failed to persist live simplification:", err);
+      if (DIAG() && diagRun) {
+        for (const b of blocks) {
+          const entry = diagRun.blocks.find((d) => d.hash === b.blockHash);
+          if (entry) entry.persistOk = false;
+        }
+      }
     });
   }
   function markLoading(el) {
@@ -625,15 +844,24 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
     let fromCache = 0;
     let remaining = elements;
     const staleBlocks = /* @__PURE__ */ new Set();
+    let appliedFromManifest = [];
     const DEBUG = !!window.__ACCESSIFY_DEBUG;
+    if (DIAG()) diagInit(window.location.pathname, siteKey || null);
+    if (diagRun) diagRun.elementsFound = elements.length;
     if (siteKey) {
       showProgress(0, elements.length);
       const manifest = await fetchManifest(siteKey, proxyUrl);
       if (DEBUG) console.log(`[Accessify] Manifest: ${manifest?.blocks?.length ?? 0} blocks for ${elements.length} elements`);
+      if (diagRun) {
+        diagRun.manifestLoaded = !!manifest?.blocks?.length;
+        diagRun.manifestBlocks = manifest?.blocks?.length ?? 0;
+        diagRun.siteMode = currentSiteMode;
+      }
       if (manifest?.blocks?.length) {
         await clearClientCacheForPage();
         const result = applyManifestBlocks(manifest, elements);
         fromCache = result.applied.length;
+        appliedFromManifest = result.applied;
         if (DEBUG) console.log(`[Accessify] Manifest hit: ${fromCache}, stale: ${result.stale.length}, remaining: ${result.remaining.length}`);
         for (const el of result.stale) staleBlocks.add(el);
         remaining = [...result.stale, ...result.remaining];
@@ -642,8 +870,13 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
             setClientCached(clientCacheKey(block.blockHash), block.result);
           }
         }
+        if (diagRun) {
+          diagRun.phase1Applied = result.applied.length;
+          diagRun.phase1Stale = result.stale.length;
+        }
         if (remaining.length === 0) {
           showDone();
+          if (DIAG()) diagFinish();
           return;
         }
       }
@@ -662,8 +895,13 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
           const success = safeReplace(el, cached, blockHash);
           if (success) {
             fromCache++;
+            if (DIAG()) {
+              diagBlock({ hash: blockHash, text: text.slice(0, 60), tag: el.tagName, idbHit: true, replaced: true, source: "idb" });
+              if (diagRun) diagRun.phase15IdbHits++;
+            }
           } else {
             stillRemaining.push(el);
+            if (DIAG()) diagBlock({ hash: blockHash, text: text.slice(0, 60), tag: el.tagName, idbHit: true, replaced: false, skipReason: "layout-unsafe", source: "skipped" });
           }
         } else {
           stillRemaining.push(el);
@@ -672,22 +910,48 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
       remaining = stillRemaining;
       if (remaining.length === 0) {
         showDone();
+        if (DIAG()) diagFinish();
         return;
       }
     }
     const manifestWasLoaded = fromCache > 0 || siteKey && staleBlocks.size > 0;
     if (manifestWasLoaded && remaining.length > 0) {
-      const nonStaleRemaining = remaining.filter((el) => !staleBlocks.has(el));
-      if (nonStaleRemaining.length > 0 && DEBUG) {
+      const contentAnchors = buildContentAnchors(appliedFromManifest);
+      const skipAsBoilerplate = [];
+      const keepForAI = [];
+      for (const el of remaining) {
+        if (staleBlocks.has(el)) {
+          keepForAI.push(el);
+          continue;
+        }
+        if (isObviousBoilerplate(el)) {
+          skipAsBoilerplate.push(el);
+          continue;
+        }
+        if (currentSiteMode === "auto" && isNearContent(el, contentAnchors)) {
+          keepForAI.push(el);
+        } else {
+          skipAsBoilerplate.push(el);
+        }
+      }
+      if (skipAsBoilerplate.length > 0 && DEBUG) {
         console.info(
-          `[Accessify] Manifest loaded — skipping ${nonStaleRemaining.length} non-manifest block(s) (nav/footer/sidebar). Only ${staleBlocks.size} stale block(s) will use live AI.`
+          `[Accessify] Safety Gate: ${skipAsBoilerplate.length} boilerplate skipped, ${keepForAI.length} queued for AI (stale + near-content).`
         );
       }
-      for (const el of nonStaleRemaining) clearLoading(el);
-      remaining = remaining.filter((el) => staleBlocks.has(el));
+      if (DIAG()) {
+        for (const el of skipAsBoilerplate) {
+          const t = el.dataset.accessifyOriginal || el.textContent?.trim() || "";
+          const reason = isObviousBoilerplate(el) ? "structural-boilerplate(nav/menu)" : currentSiteMode === "auto" ? "no-content-anchor" : `manual-mode(siteMode=${currentSiteMode})`;
+          diagBlock({ hash: hashText(t), text: t.slice(0, 60), tag: el.tagName, skipReason: reason, source: "skipped" });
+        }
+      }
+      for (const el of skipAsBoilerplate) clearLoading(el);
+      remaining = keepForAI;
       if (remaining.length === 0) {
         showDone();
         showDisclaimer();
+        if (DIAG()) diagFinish();
         return;
       }
     } else if (currentSiteMode !== "auto" && remaining.length > 0) {
@@ -696,6 +960,12 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
           `[Accessify] siteMode=${currentSiteMode} — skipping live AI for ${remaining.length} uncached block(s). Site owner: trigger a crawl from the dashboard or switch to 'auto' mode.`
         );
       }
+      if (DIAG()) {
+        for (const el of remaining) {
+          const t = el.dataset.accessifyOriginal || el.textContent?.trim() || "";
+          diagBlock({ hash: hashText(t), text: t.slice(0, 60), tag: el.tagName, skipReason: `manual-mode(siteMode=${currentSiteMode})`, source: "skipped" });
+        }
+      }
       if (staleBlocks.size > 0) skippedBlocks += staleBlocks.size;
       for (const el of remaining) clearLoading(el);
       if (fromCache > 0) {
@@ -704,6 +974,7 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
       } else {
         removeProgress();
       }
+      if (DIAG()) diagFinish();
       return;
     }
     if (!aiService && remaining.length > 0) {
@@ -737,6 +1008,11 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
         continue;
       }
       const blockHash = hashText(text);
+      let diagEntry = null;
+      if (DIAG()) {
+        diagEntry = diagBlock({ hash: blockHash, text: text.slice(0, 60), tag: el.tagName, liveAI: true, stale: isStale(el), source: "ai" });
+        if (diagRun) diagRun.phase2AICalls++;
+      }
       try {
         const simplified = await aiService.simplifyText(text, level, lang);
         if (abortController?.signal.aborted) return;
@@ -748,9 +1024,25 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
             if (siteKey) {
               persistToManifest(siteKey, blockHash, text, simplified, buildSelector(el), proxyUrl);
             }
+            if (diagEntry) diagEntry.replaced = true;
+          } else {
+            if (diagEntry) {
+              diagEntry.replaced = false;
+              diagEntry.skipReason = "layout-unsafe";
+            }
+          }
+        } else {
+          if (diagEntry) {
+            diagEntry.replaced = false;
+            diagEntry.skipReason = simplified ? "ai-returned-identical" : "ai-returned-empty";
           }
         }
       } catch (err) {
+        if (diagEntry) {
+          diagEntry.skipReason = `error:${err?.message?.slice(0, 50) || "unknown"}`;
+          diagEntry.replaced = false;
+        }
+        if (diagRun) diagRun.phase2AIErrors++;
         if (err?.message?.includes("401") || err?.message?.includes("403")) {
           console.warn("[Accessify] AI auth failed, stopping live simplification");
           for (const r of remaining) {
@@ -772,6 +1064,10 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
                 if (siteKey) {
                   persistToManifest(siteKey, blockHash, text, retry, buildSelector(el), proxyUrl);
                 }
+                if (diagEntry) {
+                  diagEntry.replaced = true;
+                  diagEntry.skipReason = null;
+                }
               }
             }
           } catch {
@@ -804,10 +1100,15 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
         showDisclaimer();
       }
     }
+    if (DIAG()) diagFinish();
   }
   function restorePage() {
     disconnectObservers();
+    const restoreReason = abortController?.signal.aborted ? "abort(navigation/toggle)" : "toggle-off";
     for (const { el, originalHtml, savedTextNodes, ancestorPatches } of savedParagraphs) {
+      const blockHash = el.getAttribute(BLOCK_HASH_ATTR) || "unknown";
+      let textNodeRestore = false;
+      let innerHtmlFallback = false;
       if (savedTextNodes.length > 0) {
         let allRestored = true;
         for (const sn of savedTextNodes) {
@@ -819,9 +1120,23 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
         }
         if (!allRestored) {
           el.innerHTML = originalHtml;
+          innerHtmlFallback = true;
+        } else {
+          textNodeRestore = true;
         }
       } else {
         el.innerHTML = originalHtml;
+        innerHtmlFallback = true;
+      }
+      if (DIAG() && diagRun) {
+        diagRun.restoreLog.push({
+          ts: (/* @__PURE__ */ new Date()).toISOString(),
+          hash: blockHash,
+          text: (el.textContent?.trim() || "").slice(0, 60),
+          textNodeRestore,
+          innerHtmlFallback,
+          reason: innerHtmlFallback ? `${restoreReason}+innerHTML-fallback(text-nodes-detached)` : restoreReason
+        });
       }
       el.removeAttribute(ORIGINAL_ATTR);
       el.removeAttribute(SIMPLIFIED_ATTR);
@@ -834,6 +1149,17 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
       }
       revertPatches(ancestorPatches);
     }
+    if (DIAG() && diagRun) {
+      const rl = diagRun.restoreLog;
+      console.log(
+        `%c[Accessify DIAG] Restore: ${rl.length} blocks, ${rl.filter((r) => r.textNodeRestore).length} text-node, ${rl.filter((r) => r.innerHtmlFallback).length} innerHTML-fallback`,
+        "color: #f59e0b; font-weight: bold"
+      );
+      if (rl.some((r) => r.innerHtmlFallback)) {
+        console.warn("[Accessify DIAG] innerHTML fallbacks detected — framework may have detached text nodes:");
+        console.table(rl.filter((r) => r.innerHtmlFallback));
+      }
+    }
     savedParagraphs = [];
     skippedBlocks = 0;
   }
@@ -927,4 +1253,4 @@ function createTextSimplifyModule(aiService, lang = "de", options) {
 export {
   createTextSimplifyModule as default
 };
-//# sourceMappingURL=text-simplify-W0rPYWpv.js.map
+//# sourceMappingURL=text-simplify-DUdShGEx.js.map