@mkterswingman/5mghost-wonder 0.0.16 → 0.0.18

package/dist/wecom/browser-probe.js

@@ -159,6 +159,10 @@ export async function runBrowserNoExportRead(options) {
                 evidence.push("opendoc-response");
             if (extracted?.text)
                 evidence.push("initial-attributed-text");
+            if ((extracted?.tables.length ?? 0) > 0)
+                evidence.push("table-control-characters");
+            if (extracted?.structureHints)
+                evidence.push("structure-hints");
             const images = options.saveDir && extracted?.imageUrls
                 ? await downloadImageResources(extracted.imageUrls, options.saveDir)
                 : (extracted?.imageUrls ?? []).map((url) => ({
@@ -180,6 +184,8 @@ export async function runBrowserNoExportRead(options) {
                 textLength: extracted?.text.length,
                 imageUrls: extracted?.imageUrls,
                 images,
+                tables: extracted?.tables,
+                structureHints: extracted?.structureHints,
                 extraction: extracted?.text ? "opendoc-initial-attributed-text" : "none",
                 evidence,
                 warnings: extracted?.warnings ?? [
@@ -305,16 +311,26 @@ export function extractTextFromOpendoc(body) {
     const decodedChunks = chunks
         .filter((chunk) => typeof chunk === "string")
         .flatMap((chunk) => extractInitialAttributedTextCandidates(chunk));
+    const decodedText = decodedChunks.join("\n");
     const textCandidate = chooseTextCandidate(decodedChunks);
-    const text = normalizeExtractedText(textCandidate ?? decodedChunks.join("\n"));
+    const text = normalizeExtractedText(textCandidate ?? decodedText);
     if (!text)
         return null;
-    const imageUrls = extractImageUrls(decodedChunks.join("\n"));
+    const imageUrls = extractImageUrls(decodedText);
+    const tables = extractTablesFromDecodedText(decodedText);
+    const structureHints = analyzeOpendocStructure({
+        opendocBody: body,
+        htmlData: typeof payload.htmlData === "string" ? payload.htmlData : "",
+        decodedText,
+        decodedCandidateCount: decodedChunks.length,
+    });
     return {
         title: payload.clientVars?.title ?? payload.clientVars?.initialTitle,
         padType: payload.clientVars?.padType ?? payload.padType,
         text,
         imageUrls,
+        tables,
+        structureHints,
         warnings: [
             "No-export text is decoded from opendoc initialAttributedText; rich styles are not reconstructed yet.",
             ...(imageUrls.length > 0
@@ -323,6 +339,52 @@ export function extractTextFromOpendoc(body) {
         ],
     };
 }
+function extractTablesFromDecodedText(decodedText) {
+    const tables = [];
+    const tablePattern = /\x1a([\s\S]*?)\x1b/g;
+    for (const match of decodedText.matchAll(tablePattern)) {
+        const body = match[1] ?? "";
+        const rows = body
+            .split("\x06")
+            .map((row) => row
+            .split("\x07")
+            .map((cell) => normalizeTableCellText(cell))
+            .filter((text) => text.length > 0)
+            .map((text) => ({ text })))
+            .filter((row) => row.length > 0);
+        if (rows.length > 0)
+            tables.push({ rows });
+    }
+    return tables;
+}
+function normalizeTableCellText(value) {
+    return value
+        .replace(/\r/g, "")
+        .replace(/[\u0000-\u0006\u0008-\u001f\u007f-\u009f]+/g, "")
+        .replace(/[^\S\r\n]+/g, " ")
+        .trim();
+}
+function analyzeOpendocStructure(input) {
+    return {
+        opendoc: summarizeStructureSignals(input.opendocBody),
+        htmlData: {
+            ...summarizeStructureSignals(input.htmlData),
+            hasTableMarkup: /<\/?(table|tr|td|th)\b/i.test(input.htmlData),
+        },
+        initialAttributedText: {
+            ...summarizeStructureSignals(input.decodedText),
+            decodedCandidateCount: input.decodedCandidateCount,
+        },
+    };
+}
+function summarizeStructureSignals(value) {
+    return {
+        length: value.length,
+        hasTableSignals: /\b(table|cell|row|column|sheet|workbook)\b/i.test(value),
+        hasMergeSignals: /\b(merge|merged|rowspan|colspan|mergeCell|mergeCells)\b/i.test(value),
+        hasImageSignals: /\b(image|img|pic|picture|media|download_url|qpic)\b/i.test(value),
+    };
+}
 function extractInitialAttributedTextCandidates(chunk) {
     let decoded;
     try {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mkterswingman/5mghost-wonder",
-  "version": "0.0.16",
+  "version": "0.0.18",
   "description": "企微文档读取 CLI — WeCom document reader",
   "type": "module",
   "engines": {

package/skills/setup-5mghost-wonder/SKILL.md

@@ -7,9 +7,9 @@ description: Use this skill when the user wants to install or set up wonder, say
 
 ## Skill version
 
-This skill matches **wonder 0.0.16**.
+This skill matches **wonder 0.0.18**.
 
-Once the CLI is installed in Step 1, run `wonder --version`. If the output does not equal `0.0.16`, the CLI on disk has drifted from the skill text loaded in this session. Ask the user to run `/update-5mghost-wonder`, then **start a fresh AI session** (`/exit` and re-enter, or open a new chat) — skill text already loaded into a running session does not refresh after `wonder update`, even though the file on disk has been replaced.
+Once the CLI is installed in Step 1, run `wonder --version`. If the output does not equal `0.0.18`, the CLI on disk has drifted from the skill text loaded in this session. Ask the user to run `/update-5mghost-wonder`, then **start a fresh AI session** (`/exit` and re-enter, or open a new chat) — skill text already loaded into a running session does not refresh after `wonder update`, even though the file on disk has been replaced.
 
 After a successful first install, also remind the user to start a fresh AI session before invoking `/use-5mghost-wonder` for the first time. The skill files were just written to disk; the current session never loaded them.
 

package/skills/update-5mghost-wonder/SKILL.md

@@ -7,7 +7,7 @@ description: Use this skill when the user wants to update or upgrade wonder, say
 
 ## Skill version
 
-This skill matches **wonder 0.0.16**.
+This skill matches **wonder 0.0.18**.
 
 ---
 

package/skills/use-5mghost-wonder/SKILL.md

@@ -10,10 +10,10 @@ the referenced workflow files needed for the current task.
 
 ## Version Gate
 
-This skill matches **wonder 0.0.16**.
+This skill matches **wonder 0.0.18**.
 
 On first use in a session, follow `references/session-init.md`. If the installed
-CLI version differs from `0.0.16`, stop and ask the user to run
+CLI version differs from `0.0.18`, stop and ask the user to run
 `/update-5mghost-wonder`, then start a fresh AI session.
 
 ## Hard Rules

package/skills/use-5mghost-wonder/references/read-no-export-browser.md

@@ -37,8 +37,11 @@ wonder browser read <url> --save /tmp/wonder-browser-read
 ```
 
 This currently attempts text extraction from the browser `opendoc`
-`initialAttributedText`. Treat it as `partial` unless it also proves image and
-table structure.
+`initialAttributedText`, decodes simple table cells from WeCom table control
+characters, downloads detected image resources when `--save` is present, and
+returns safe `structureHints` for `htmlData`, `initialAttributedText`, and the
+full `opendoc` body. Treat it as `partial` unless it also proves merge ranges
+and image anchors.
 
 3. If the read is insufficient, run the evidence probe:
 
@@ -59,6 +62,9 @@ Use `--headed` only when login/debug visibility is needed.
    - WebSocket messages if they expose structured document operations
    - screenshots only as a fallback or visual cross-check
 6. For tables, specifically look for:
+   - `tables[].rows[].text` for simple table cell content
+   - `structureHints.*.hasTableSignals` and `hasMergeSignals`
+   - `structureHints.htmlData.hasTableMarkup`
    - cell coordinates
    - displayed text
    - merge ranges
@@ -82,6 +88,19 @@ Return a concise result:
   "mergeRangesAvailable": true,
   "imagesAvailable": true,
   "imageAnchorsAvailable": true,
+  "tables": [
+    {
+      "rows": [
+        [{ "text": "A1" }, { "text": "B1" }]
+      ]
+    }
+  ],
+  "structureHints": {
+    "htmlData": {
+      "hasTableMarkup": true,
+      "hasMergeSignals": true
+    }
+  },
   "evidence": ["runtime-store", "network-response", "websocket", "screenshot"],
   "missing": []
 }