browser-pilot-cli 0.3.0-rc.4 → 0.3.0-rc.5

package/README.md

@@ -182,7 +182,7 @@ A project can pin Browser Pilot locally and run the same CLI without a global
 installation:
 
 ```bash
-npm install --save-exact browser-pilot-cli@0.3.0-rc.4
+npm install --save-exact browser-pilot-cli@0.3.0-rc.5
 npx --no-install browser-pilot tabs
 ```
 

package/dist/cli.js

@@ -6,7 +6,7 @@ import { writeFileSync as writeFileSync2, readFileSync as readFileSync4, existsS
 import { resolve as resolvePath } from "path";
 
 // src/version.ts
-var BROWSER_PILOT_VERSION = "0.3.0-rc.4";
+var BROWSER_PILOT_VERSION = "0.3.0-rc.5";
 
 // src/protocol/errors.ts
 var ERROR_CODES = [
@@ -1804,7 +1804,7 @@ var scrollEvidenceSchema = objectSchema({
   afterX: numberSchema({ minimum: 0 }),
   afterY: numberSchema({ minimum: 0 }),
   matchedText: sensitive(stringSchema({ maxLength: 4096 }), "browser_data"),
-  reason: stringSchema({ enum: ["at_boundary", "text_not_found"] })
+  reason: stringSchema({ enum: ["at_boundary", "text_not_found", "text_not_revealed"] })
 }, [
   "action",
   "status",
@@ -4368,9 +4368,6 @@ program.command("close").description("Close current browser tab").option("-a, --
         context: { failedTargetIds: failed }
       });
     }
-    if (remainingTabs.length > 0 && !remainingTabs.some((tab) => tab.active)) {
-      await client.callTool("browser.tabs.switch", { targetId: remainingTabs[0].targetId });
-    }
     emit(
       { ok: true, closed, remaining: remainingTabs.length },
       `\u2713 Closed ${closed} Pilot tab(s)`
@@ -4381,8 +4378,8 @@ program.command("close").description("Close current browser tab").option("-a, --
     const remainingTabs = await client.listTabs("all");
     if (remainingTabs.length > 0) {
       if (!remainingTabs.some((tab) => tab.active)) {
-        const fallback = remainingTabs.find((tab) => tab.origin !== "user_tab") ?? remainingTabs[0];
-        await client.callTool("browser.tabs.switch", { targetId: fallback.targetId });
+        const fallback = remainingTabs.find((tab) => tab.origin !== "user_tab");
+        if (fallback) await client.callTool("browser.tabs.switch", { targetId: fallback.targetId });
       }
       emit({ ok: true, remaining: remainingTabs.length }, "\u2713 Tab closed");
     } else {

package/dist/daemon.js

@@ -747,7 +747,7 @@ var scrollEvidenceSchema = objectSchema({
   afterX: numberSchema({ minimum: 0 }),
   afterY: numberSchema({ minimum: 0 }),
   matchedText: sensitive(stringSchema({ maxLength: 4096 }), "browser_data"),
-  reason: stringSchema({ enum: ["at_boundary", "text_not_found"] })
+  reason: stringSchema({ enum: ["at_boundary", "text_not_found", "text_not_revealed"] })
 }, [
   "action",
   "status",
@@ -9012,6 +9012,10 @@ var ScrollService = class {
       const exact=${exact};
       const normalize=value=>String(value||'').replace(/\\s+/g,' ').trim();
       const needle=normalize(query).toLocaleLowerCase();
+      const slash=String.fromCharCode(92);
+      const regexSpecials='^$.*+?()[]{}|'+slash;
+      const escapeRegex=value=>Array.from(value,char=>regexSpecials.includes(char)?slash+char:char).join('');
+      const queryPattern=exact?null:new RegExp(normalize(query).split(' ').map(escapeRegex).join(slash+'s+'),'iu');
       const roots=[document];
       let visited=0;
       while(roots.length&&visited<20000){
@@ -9029,21 +9033,50 @@ var ScrollService = class {
           if(!element)continue;
           const tag=String(element.tagName||'').toLowerCase();
           if(['script','style','noscript','template'].includes(tag))continue;
-          const content=normalize(node.nodeValue);
+          const raw=String(node.nodeValue||'');
+          const content=normalize(raw);
           const normalized=content.toLocaleLowerCase();
           if(!(exact?normalized===needle:normalized.includes(needle)))continue;
+          let matchStart=0;
+          let matchEnd=raw.length;
+          if(exact){
+            matchStart=Math.max(0,raw.search(/\\S/u));
+            matchEnd=Math.max(matchStart,raw.length-((raw.match(/\\s*$/u)||[''])[0].length));
+          }else{
+            const match=queryPattern.exec(raw);
+            if(!match)continue;
+            matchStart=match.index;
+            matchEnd=match.index+match[0].length;
+          }
           const style=getComputedStyle(element);
           if(style.display==='none'||style.visibility==='hidden'||style.visibility==='collapse'||element.getClientRects().length===0)continue;
+          const viewportWidth=Math.max(0,window.innerWidth||document.documentElement.clientWidth||0);
+          const viewportHeight=Math.max(0,window.innerHeight||document.documentElement.clientHeight||0);
+          const range=document.createRange();
+          range.setStart(node,matchStart);
+          range.setEnd(node,matchEnd);
+          const rangeRect=()=>range.getBoundingClientRect();
+          const intersectsViewport=rect=>rect.width>0&&rect.height>0&&rect.bottom>0&&rect.right>0&&rect.top<viewportHeight&&rect.left<viewportWidth;
           const beforeX=Math.max(0,window.scrollX||document.documentElement.scrollLeft||0);
           const beforeY=Math.max(0,window.scrollY||document.documentElement.scrollTop||0);
           element.scrollIntoView({block:'center',inline:'nearest',behavior:'instant'});
-          const afterX=Math.max(0,window.scrollX||document.documentElement.scrollLeft||0);
-          const afterY=Math.max(0,window.scrollY||document.documentElement.scrollTop||0);
+          let afterX=Math.max(0,window.scrollX||document.documentElement.scrollLeft||0);
+          let afterY=Math.max(0,window.scrollY||document.documentElement.scrollTop||0);
+          let rect=rangeRect();
+          if(!intersectsViewport(rect)){
+            const left=Math.max(0,afterX+rect.left-Math.max(0,(viewportWidth-rect.width)/2));
+            const top=Math.max(0,afterY+rect.top-Math.max(0,(viewportHeight-rect.height)/2));
+            window.scrollTo({left,top,behavior:'instant'});
+            afterX=Math.max(0,window.scrollX||document.documentElement.scrollLeft||0);
+            afterY=Math.max(0,window.scrollY||document.documentElement.scrollTop||0);
+            rect=rangeRect();
+          }
           const round=value=>Math.round(Number(value)||0);
           const moved=Math.abs(afterX-beforeX)>=1||Math.abs(afterY-beforeY)>=1;
-          return{ok:true,action:'scroll',status:'verified',mode:'text',target:'text',moved,
+          const revealed=intersectsViewport(rect);
+          return{ok:true,action:'scroll',status:revealed?'verified':'mismatch',mode:'text',target:'text',moved,
             deltaX:round(afterX-beforeX),deltaY:round(afterY-beforeY),beforeX:round(beforeX),beforeY:round(beforeY),
-            afterX:round(afterX),afterY:round(afterY),matchedText:content.slice(0,4096)};
+            afterX:round(afterX),afterY:round(afterY),matchedText:content.slice(0,4096),...(revealed?{}:{reason:'text_not_revealed'})};
         }
       }
       return{ok:true,action:'scroll',status:'mismatch',mode:'text',target:'text',moved:false,
@@ -9066,11 +9099,13 @@ var ScrollService = class {
 var MAX_ANNOTATIONS = 200;
 function annotationScript(data, annotations, viewport) {
   return `(async() => {
-    const source=${JSON.stringify(`data:image/png;base64,${data}`)};
+    const base64=${JSON.stringify(data)};
     const annotations=${JSON.stringify(annotations)};
     const viewport=${JSON.stringify(viewport)};
-    const response=await fetch(source);
-    const bitmap=await createImageBitmap(await response.blob());
+    const binary=atob(base64);
+    const bytes=new Uint8Array(binary.length);
+    for(let index=0;index<binary.length;index+=1)bytes[index]=binary.charCodeAt(index);
+    const bitmap=await createImageBitmap(new Blob([bytes],{type:'image/png'}));
     const canvas=document.createElement('canvas');
     canvas.width=bitmap.width;canvas.height=bitmap.height;
     const context=canvas.getContext('2d');
@@ -13979,7 +14014,7 @@ var ManagedTargetJanitorClient = class {
 };
 
 // src/version.ts
-var BROWSER_PILOT_VERSION = "0.3.0-rc.4";
+var BROWSER_PILOT_VERSION = "0.3.0-rc.5";
 
 // src/daemon.ts
 var CLI_EXECUTABLE_PATH = publicExecutablePath(import.meta.url);

package/docs/plans/v0.3.0-stabilization.md

@@ -1,7 +1,7 @@
 # Browser Pilot v0.3.0 Stabilization Plan
 
 Status: **In progress**
-Candidate baseline: `v0.3.0-rc.3`; correction target: `v0.3.0-rc.4`, protocol `1.2`
+Candidate baseline: `v0.3.0-rc.4`; correction target: `v0.3.0-rc.5`, protocol `1.2`
 
 ## Goal
 
@@ -35,8 +35,10 @@ products depend on. No Agent-specific behavior enters Browser Pilot.
   integration checks against the installed standalone release executable.
 - [x] Run one controlled user-Chrome host acceptance with explicit Profile
   choices and a single authorization attempt.
-- [ ] Publish `v0.3.0-rc.4` with Broker-backed CLI browser status and verify
+- [x] Publish `v0.3.0-rc.4` with Broker-backed CLI browser status and verify
   that passive discovery reports the live authorized connection.
+- [ ] Publish `v0.3.0-rc.5` with the real-site soak corrections and rerun the
+  same public information, form, capture, and managed-cleanup tasks.
 - [ ] Soak several real Agent tasks through the installed candidate.
 - [ ] Tag `v0.3.0`, publish npm `latest`, and verify upgrade/rollback metadata.
 

package/docs/releases/v0.3.0-rc.5.md

@@ -0,0 +1,26 @@
+# Browser Pilot 0.3.0-rc.5
+
+This release candidate fixes three edge cases found by running RC.4 through
+real Agent tasks in two live Chrome Profiles.
+
+- Text-targeted scrolling now verifies that the matched text actually entered
+  the viewport. If a site's `scrollIntoView` path is ineffective, Browser Pilot
+  falls back to the matched text Range's page coordinates; an unrevealed match
+  returns an explicit mismatch instead of a false success.
+- Screenshot annotations no longer decode captured PNGs through
+  `fetch(data:)`. They use in-memory base64, Blob, and ImageBitmap conversion in
+  the isolated world, so restrictive site `connect-src` policies such as
+  GitHub's cannot break annotation.
+- Tab cleanup no longer switches control to a user tab when Chrome has not yet
+  reported its next active target. `bp close --all` still closes only managed
+  tabs.
+
+The soak otherwise completed public GitHub reading and targeted search, a
+disposable HTTP form with verified input and control state, submission result
+verification, multi-Profile routing, ordinary screenshots, annotated form
+screenshots, PDF export, connection-status checks, and preservation of every
+user tab and exported file.
+
+RC.5 keeps protocol 1.2 and the Agent-neutral stdio contract unchanged.
+Candidate gates passed with 276 Node tests, 109 isolated Playwright tests, and
+TypeScript type checking.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "browser-pilot-cli",
-  "version": "0.3.0-rc.4",
+  "version": "0.3.0-rc.5",
   "description": "CLI tool to control your browser via Chrome DevTools Protocol",
   "repository": "https://github.com/relixiaobo/browser-pilot",
   "type": "module",
@@ -48,6 +48,7 @@
     "docs/releases/v0.3.0-rc.2.md",
     "docs/releases/v0.3.0-rc.3.md",
     "docs/releases/v0.3.0-rc.4.md",
+    "docs/releases/v0.3.0-rc.5.md",
     "README.md",
     "LICENSE"
   ],