@braze/sanity-plugin-ai-translation 0.2.66 → 0.2.67

package/dist/index.js

@@ -6,7 +6,6 @@ import defaults from "lodash.defaults";
 import { Card, Stack, Flex, Text, Box, Heading, Button, Spinner, Inline, Checkbox, Grid, Radio, TextInput, Badge, Select, Label, Code, Tooltip } from "@sanity/ui";
 import { toPlainText } from "@portabletext/react";
 import { uuid } from "@sanity/uuid";
-import { XMLParser } from "fast-xml-parser";
 import cloneDeep from "lodash.clonedeep";
 import get from "lodash.get";
 import set from "lodash.set";
@@ -1540,7 +1539,7 @@ const globalTransformRegistry = new TransformRegistry(), callGPT5 = async (apiKe
     cleanedLength: cleanedText.length,
     preview: cleanedText.substring(0, 500)
   });
-  const parsed = parseResponseToNodes(cleanedText, originalDoc);
+  const parsed = parseReviewFormat(cleanedText, originalDoc);
   devLog.debug("Parsed response structure", {
     rootNodeCount: parsed.length,
     rootNodeNames: parsed.map((n) => n.type === "field" && n.name ? n.name : "unknown")
@@ -1562,118 +1561,7 @@ const globalTransformRegistry = new TransformRegistry(), callGPT5 = async (apiKe
     codeBlocksRemoved,
     preview: cleaned.trim().substring(0, 200)
   }), cleaned.trim();
-}, TAG_NAMES = ["field", "block", "span"], looksLikeXmlResponse = (text) => TAG_NAMES.some((tag) => text.includes(`<${tag}`));
-function parseCompactFormatSections(text, originalDoc) {
-  const nodes = [], lines = text.split(`
-`), compactStartRegex = /^---\s+(\w+)\s+---$/;
-  for (let i = 0; i < lines.length; i++) {
-    const match = lines[i].trim().match(compactStartRegex);
-    if (!match) continue;
-    const fieldName = match[1], contentLines = [];
-    for (i++; i < lines.length; ) {
-      const currentLine = lines[i];
-      if (currentLine.trim().match(new RegExp(`^---\\s+/${fieldName}\\s+---$`)))
-        break;
-      contentLines.push(currentLine), i++;
-    }
-    const content = contentLines.join(`
-`);
-    let contentType;
-    try {
-      const fieldValue = get(originalDoc, [fieldName]);
-      if (fieldValue && typeof fieldValue == "object" && fieldValue !== null) {
-        const fieldType = fieldValue._type;
-        typeof fieldType == "string" && fieldType.startsWith("translatable.") && (contentType = fieldType);
-      }
-    } catch {
-    }
-    contentType || (content.includes("<block") || content.includes("<span") ? contentType = "translatable.richtext" : contentType = "translatable.vtt.captions"), nodes.push({
-      type: "field",
-      name: fieldName,
-      text: content,
-      attributes: { type: contentType },
-      children: []
-    });
-  }
-  return nodes;
-}
-const parseResponseToNodes = (text, originalDoc) => {
-  if (looksLikeXmlResponse(text))
-    try {
-      const wrapped = `<root>${text}</root>`, parsed = new XMLParser({
-        ignoreAttributes: !1,
-        attributeNamePrefix: "@_",
-        textNodeName: "#text",
-        parseTagValue: !1,
-        trimValues: !1,
-        // preserve exact whitespace in text (e.g. trailing space in spans)
-        isArray: (tagName) => TAG_NAMES.includes(tagName)
-      }).parse(wrapped);
-      if (parsed && typeof parsed == "object" && parsed.root) {
-        const xmlNodes = convertXmlObjectToParsedNodes(parsed.root), compactNodes = parseCompactFormatSections(text, originalDoc);
-        if (xmlNodes.length > 0 || compactNodes.length > 0) {
-          const merged = [...xmlNodes, ...compactNodes];
-          return devLog.debug("Parsed response with XML parser + compact sections", {
-            xmlNodeCount: xmlNodes.length,
-            compactNodeCount: compactNodes.length,
-            total: merged.length
-          }), merged;
-        }
-      }
-    } catch (err) {
-      devLog.warn("XML parse failed, falling back to regex parser", {
-        error: err instanceof Error ? err.message : String(err)
-      });
-    }
-  return parseReviewFormat(text, originalDoc);
-};
-function normalizeToArray(value) {
-  return Array.isArray(value) ? value : value != null && typeof value == "object" ? [value] : [];
-}
-function convertXmlObjectToParsedNodes(root) {
-  const result = [];
-  for (const tagName of TAG_NAMES) {
-    const raw = root[tagName];
-    if (raw !== void 0) {
-      for (const item of normalizeToArray(raw))
-        if (item != null && typeof item == "object") {
-          const node = convertXmlElementToParsedNode(item, tagName);
-          node && result.push(node);
-        }
-    }
-  }
-  return result;
-}
-function convertXmlElementToParsedNode(obj, tagName) {
-  const attributes = {};
-  for (const [key, value] of Object.entries(obj))
-    if (key.startsWith("@_")) {
-      const attrName = key.slice(2);
-      attributes[attrName] = value != null ? String(value) : "";
-    }
-  let text;
-  const textVal = obj["#text"];
-  textVal != null && (text = typeof textVal == "string" ? textVal : String(textVal));
-  const children = [];
-  for (const childTag of TAG_NAMES) {
-    const raw = obj[childTag];
-    if (raw !== void 0) {
-      for (const item of normalizeToArray(raw))
-        if (item != null && typeof item == "object") {
-          const child = convertXmlElementToParsedNode(item, childTag);
-          child && children.push(child);
-        }
-    }
-  }
-  return {
-    type: tagName,
-    name: attributes.name,
-    text: text !== void 0 && text !== "" ? unescapeXml(text) : void 0,
-    attributes,
-    children
-  };
-}
-const parseReviewFormat = (text, originalDoc) => {
+}, parseReviewFormat = (text, originalDoc) => {
   const root = [], stack = [], lines = text.split(`
 `);
   devLog.debug("Parsing response text", {
@@ -1761,25 +1649,34 @@ const parseReviewFormat = (text, originalDoc) => {
         attributes,
         children: []
       }, closingTag = `</${tagName}>`;
-      if (content.endsWith(closingTag.replace(/[<>]/g, ""))) {
-        const textContent = content.substring(0, content.lastIndexOf("<"));
-        if (node.text = unescapeXml(textContent), stack.length > 0) {
+      if (content.includes(closingTag)) {
+        const closeIdx = content.indexOf(closingTag);
+        node.text = unescapeXml(content.substring(0, closeIdx));
+        const remainder = content.substring(closeIdx + closingTag.length).trim();
+        if (stack.length > 0) {
           const parent = stack[stack.length - 1];
           parent && parent.children.push(node);
         } else
           root.push(node);
-      } else if (content.includes(closingTag))
-        if (node.text = unescapeXml(content.replace(closingTag, "")), stack.length > 0) {
+        remainder && (lines[i] = remainder, i--);
+      } else if (content.endsWith(closingTag)) {
+        const textContent = content.substring(0, content.lastIndexOf("<"));
+        if (node.text = unescapeXml(textContent), stack.length > 0) {
           const parent = stack[stack.length - 1];
           parent && parent.children.push(node);
         } else
           root.push(node);
-      else if (attributes.type && typeof attributes.type == "string" && attributes.type.startsWith("translatable.")) {
+      } else if (attributes.type && typeof attributes.type == "string" && attributes.type.startsWith("translatable.")) {
         const contentLines = [];
         for (i++; i < lines.length; ) {
           const currentLine = lines[i];
-          if (currentLine.trim().match(/^<\/field>$/))
+          if (currentLine.includes("</field>")) {
+            const idx = currentLine.indexOf("</field>"), before = currentLine.substring(0, idx).trimEnd();
+            before.trim() && contentLines.push(before);
+            const after = currentLine.substring(idx + 8).trim();
+            after && (lines[i] = after, i--);
             break;
+          }
           contentLines.push(currentLine), i++;
         }
         if (node.text = contentLines.join(`
@@ -1788,19 +1685,15 @@ const parseReviewFormat = (text, originalDoc) => {
           parent && parent.children.push(node);
         } else
           root.push(node);
-      } else if (attributes.type === "string" && content && content.includes(closingTag))
-        if (node.text = unescapeXml(content.replace(closingTag, "")), stack.length > 0) {
-          const parent = stack[stack.length - 1];
-          parent && parent.children.push(node);
-        } else
-          root.push(node);
-      else if (attributes.type === "string" && content && !content.includes(closingTag)) {
+      } else if (attributes.type === "string" && content && !content.includes(closingTag)) {
         const contentLines = [content];
         for (i++; i < lines.length; ) {
           const currentLine = lines[i];
           if (currentLine.includes(closingTag)) {
-            const closingIndex = currentLine.indexOf(closingTag), remainingContent = currentLine.substring(0, closingIndex);
+            const closingIndex = currentLine.indexOf(closingTag), remainingContent = currentLine.substring(0, closingIndex).trimEnd();
             remainingContent && contentLines.push(remainingContent);
+            const after = currentLine.substring(closingIndex + closingTag.length).trim();
+            after && (lines[i] = after, i--);
             break;
           }
           contentLines.push(currentLine), i++;
@@ -1815,8 +1708,13 @@ const parseReviewFormat = (text, originalDoc) => {
         const contentLines = [];
         for (i++; i < lines.length; ) {
           const currentLine = lines[i];
-          if (currentLine.trim() === "</field>")
+          if (currentLine.includes("</field>")) {
+            const idx = currentLine.indexOf("</field>"), before = currentLine.substring(0, idx).trimEnd();
+            before.trim() && contentLines.push(before);
+            const after = currentLine.substring(idx + 8).trim();
+            after && (lines[i] = after, i--);
             break;
+          }
           contentLines.push(currentLine), i++;
         }
         if (node.text = contentLines.join(`
@@ -1829,32 +1727,36 @@ const parseReviewFormat = (text, originalDoc) => {
         stack.push(node);
       continue;
     }
-    const closeMatch = trimmed.match(/^<\/(field|block|span)>/);
-    if (closeMatch) {
-      if (stack.length === 0) {
-        devLog.warn("Closing tag found but stack is empty", {
+    if (trimmed.match(/^<\/(field|block|span)>/)) {
+      let afterClosingTag = trimmed;
+      for (; ; ) {
+        const match = afterClosingTag.match(/^<\/(field|block|span)>/);
+        if (!match) break;
+        if (stack.length === 0) {
+          devLog.warn("Closing tag found but stack is empty", {
+            line: i,
+            tagName: match[1],
+            trimmed: afterClosingTag
+          }), afterClosingTag = afterClosingTag.substring(match[0].length).trim(), afterClosingTag = afterClosingTag.replace(/<!--[\s\S]*?-->/g, "").trim();
+          continue;
+        }
+        matchedPatternCount++, devLog.debug("Matched closing tag", {
           line: i,
-          tagName: closeMatch[1],
-          trimmed
+          tagName: match[1],
+          trimmed: afterClosingTag,
+          stackLength: stack.length,
+          stackTopName: stack[stack.length - 1]?.name
         });
-        continue;
+        const node = stack.pop();
+        if (node)
+          if (stack.length > 0) {
+            const parent = stack[stack.length - 1];
+            parent && parent.children.push(node);
+          } else
+            root.push(node);
+        afterClosingTag = afterClosingTag.substring(match[0].length).trim(), afterClosingTag = afterClosingTag.replace(/<!--[\s\S]*?-->/g, "").trim();
       }
-      matchedPatternCount++, devLog.debug("Matched closing tag", {
-        line: i,
-        tagName: closeMatch[1],
-        trimmed,
-        stackLength: stack.length,
-        stackTopName: stack[stack.length - 1]?.name
-      });
-      const node = stack.pop();
-      if (node)
-        if (stack.length > 0) {
-          const parent = stack[stack.length - 1];
-          parent && parent.children.push(node);
-        } else
-          root.push(node);
-      let afterClosingTag = trimmed.substring(closeMatch[0].length).trim();
-      if (afterClosingTag = afterClosingTag.replace(/<!--[\s\S]*?-->/g, "").trim(), afterClosingTag) {
+      if (afterClosingTag) {
         const remainingOpenMatch = afterClosingTag.match(/^<(field|block|span)([^>]*)>(.*)$/);
         if (remainingOpenMatch) {
           const [, tagName, attrsStr, content] = remainingOpenMatch, attributes = parseAttributes(attrsStr), remainingNode = {
@@ -1864,13 +1766,17 @@ const parseReviewFormat = (text, originalDoc) => {
             attributes,
             children: []
           }, remainingClosingTag = `</${tagName}>`;
-          if (content.includes(remainingClosingTag))
-            if (remainingNode.text = unescapeXml(content.replace(remainingClosingTag, "")), stack.length > 0) {
+          if (content.includes(remainingClosingTag)) {
+            const closeIdx = content.indexOf(remainingClosingTag);
+            remainingNode.text = unescapeXml(content.substring(0, closeIdx));
+            const remainder = content.substring(closeIdx + remainingClosingTag.length).trim();
+            if (stack.length > 0) {
               const parent = stack[stack.length - 1];
               parent && parent.children.push(remainingNode);
             } else
               root.push(remainingNode);
-          else if (attributes.type === "string" && content)
+            remainder && (lines[i] = remainder, i--);
+          } else if (attributes.type === "string" && content)
             if (remainingNode.text = unescapeXml(content), stack.length > 0) {
               const parent = stack[stack.length - 1];
               parent && parent.children.push(remainingNode);
@@ -1880,8 +1786,13 @@ const parseReviewFormat = (text, originalDoc) => {
             const contentLines = [];
             for (i++; i < lines.length; ) {
               const currentLine = lines[i];
-              if (currentLine.trim() === `</${tagName}>` || currentLine.trim().startsWith(`</${tagName}>`))
+              if (currentLine.includes(`</${tagName}>`)) {
+                const idx = currentLine.indexOf(`</${tagName}>`), before = currentLine.substring(0, idx).trimEnd();
+                before.trim() && contentLines.push(before);
+                const after = currentLine.substring(idx + `</${tagName}>`.length).trim();
+                after && (lines[i] = after, i--);
                 break;
+              }
               contentLines.push(currentLine), i++;
             }
             if (remainingNode.text = contentLines.join(`
@@ -1892,7 +1803,8 @@ const parseReviewFormat = (text, originalDoc) => {
               root.push(remainingNode);
           } else
             stack.push(remainingNode);
-        }
+        } else
+          lines[i] = afterClosingTag, i--;
       }
       continue;
     }
@@ -2009,22 +1921,10 @@ const parseReviewFormat = (text, originalDoc) => {
     });
     return;
   }
-  if (node.text !== void 0 && !node.attributes.readonly) {
-    const existingValue = get(doc, path);
-    if (existingValue === void 0)
-      devLog.debug("Skipping field not present in original document", {
-        path: path.join("."),
-        attemptedValue: node.text
-      });
-    else if (typeof existingValue == "string") {
-      const value = path.length === 1 ? node.text.trim() : node.text;
-      set(doc, path, value);
-    } else
-      devLog.warn("Skipping text apply at path where value is not a string (prevents structure corruption)", {
-        path: path.join("."),
-        actualType: typeof existingValue
-      });
-  }
+  node.text !== void 0 && !node.attributes.readonly && (get(doc, path) !== void 0 ? set(doc, path, node.text) : devLog.debug("Skipping field not present in original document", {
+    path: path.join("."),
+    attemptedValue: node.text
+  }));
 }, applyBlockChanges = (block, blockNode) => {
   if (blockNode.attributes._key && block._key !== blockNode.attributes._key) {
     devLog.warn("Block key mismatch, skipping", {