libretto 0.6.23 → 0.6.24

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -88,16 +88,23 @@ function readPngDimensions(buffer) {
88
88
  height: buffer.readUInt32BE(20)
89
89
  };
90
90
  }
91
- async function takeViewportScreenshot(page) {
92
- const viewport = page.viewportSize();
93
- if (!viewport) {
94
- throw new Error("Viewport size not found");
95
- }
96
- const screenshot = await page.screenshot({
97
- fullPage: false,
98
- scale: "css",
99
- timeout: 1e4
100
- });
91
+ function toPositiveNumber(value) {
92
+ return typeof value === "number" && Number.isFinite(value) && value > 0 ? value : void 0;
93
+ }
94
+ async function getViewportFromPage(page) {
95
+ const metrics = await page.evaluate(() => ({
96
+ visualViewportWidth: window.visualViewport?.width,
97
+ visualViewportHeight: window.visualViewport?.height,
98
+ innerWidth: window.innerWidth,
99
+ innerHeight: window.innerHeight,
100
+ documentElementClientWidth: document.documentElement?.clientWidth,
101
+ documentElementClientHeight: document.documentElement?.clientHeight
102
+ }));
103
+ const width = toPositiveNumber(metrics.visualViewportWidth) ?? toPositiveNumber(metrics.innerWidth) ?? toPositiveNumber(metrics.documentElementClientWidth);
104
+ const height = toPositiveNumber(metrics.visualViewportHeight) ?? toPositiveNumber(metrics.innerHeight) ?? toPositiveNumber(metrics.documentElementClientHeight);
105
+ return width && height ? { width, height } : null;
106
+ }
107
+ function screenshotState(screenshot, viewport) {
101
108
  const dimensions = readPngDimensions(screenshot);
102
109
  return {
103
110
  screenshot,
@@ -110,6 +117,18 @@ async function takeViewportScreenshot(page) {
110
117
  }
111
118
  };
112
119
  }
120
+ async function takeViewportScreenshot(page) {
121
+ const viewport = page.viewportSize() ?? await getViewportFromPage(page).catch(() => null);
122
+ if (!viewport) {
123
+ throw new Error("Viewport size not found");
124
+ }
125
+ const screenshot = await page.screenshot({
126
+ fullPage: false,
127
+ scale: "css",
128
+ timeout: 1e4
129
+ });
130
+ return screenshotState(screenshot, viewport);
131
+ }
113
132
  async function executeBrowserAction(page, action, logger = defaultLogger) {
114
133
  switch (action.type) {
115
134
  case "click": {
@@ -247,9 +266,9 @@ async function executeRecoveryAgent(page, instruction, logger, model, maxSteps =
247
266
  }
248
267
  const log = logger ?? defaultLogger;
249
268
  log.info("Executing vision-based recovery agent", { instruction });
250
- let screenshotState;
269
+ let screenshotState2;
251
270
  try {
252
- screenshotState = await takeViewportScreenshot(page);
271
+ screenshotState2 = await takeViewportScreenshot(page);
253
272
  } catch (screenshotError) {
254
273
  log.warn("Failed to take screenshot for recovery agent, skipping", {
255
274
  screenshotError: screenshotError instanceof Error ? screenshotError.message : String(screenshotError)
@@ -258,7 +277,7 @@ async function executeRecoveryAgent(page, instruction, logger, model, maxSteps =
258
277
  }
259
278
  const steps = [];
260
279
  for (let step = 1; step <= maxSteps; step++) {
261
- const { screenshot, dimensions, scale } = screenshotState;
280
+ const { screenshot, dimensions, scale } = screenshotState2;
262
281
  const { object: result } = await generateObject({
263
282
  model,
264
283
  schema: recoveryActionSchema,
@@ -305,7 +324,7 @@ Analyze the screenshot and decide what action to take. If the task is complete o
305
324
  await executeBrowserAction(page, action, log);
306
325
  await delay(2e3);
307
326
  if (step < maxSteps) {
308
- screenshotState = await takeViewportScreenshot(page);
327
+ screenshotState2 = await takeViewportScreenshot(page);
309
328
  }
310
329
  }
311
330
  log.info("Recovery agent execution completed");
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "libretto",
3
- "version": "0.6.23",
3
+ "version": "0.6.24",
4
4
  "description": "AI-powered browser automation library and CLI built on Playwright",
5
5
  "license": "MIT",
6
6
  "homepage": "https://libretto.sh",
@@ -4,7 +4,7 @@ description: "Browser automation CLI for building, maintaining, and running brow
4
4
  license: MIT
5
5
  metadata:
6
6
  author: saffron-health
7
- version: "0.6.23"
7
+ version: "0.6.24"
8
8
  ---
9
9
 
10
10
  ## How Libretto Works
@@ -4,7 +4,7 @@ description: "Read-only Libretto workflow for diagnosing live browser state with
4
4
  license: MIT
5
5
  metadata:
6
6
  author: saffron-health
7
- version: "0.6.23"
7
+ version: "0.6.24"
8
8
  ---
9
9
 
10
10
  ## How Libretto Read-Only Works
@@ -153,21 +153,41 @@ function readPngDimensions(buffer: Buffer): ImageDimensions {
153
153
  };
154
154
  }
155
155
 
156
- async function takeViewportScreenshot(page: Page): Promise<{
156
+ function toPositiveNumber(value: unknown): number | undefined {
157
+ return typeof value === "number" && Number.isFinite(value) && value > 0
158
+ ? value
159
+ : undefined;
160
+ }
161
+
162
+ async function getViewportFromPage(page: Page): Promise<ImageDimensions | null> {
163
+ const metrics = await page.evaluate(() => ({
164
+ visualViewportWidth: window.visualViewport?.width,
165
+ visualViewportHeight: window.visualViewport?.height,
166
+ innerWidth: window.innerWidth,
167
+ innerHeight: window.innerHeight,
168
+ documentElementClientWidth: document.documentElement?.clientWidth,
169
+ documentElementClientHeight: document.documentElement?.clientHeight,
170
+ }));
171
+ const width =
172
+ toPositiveNumber(metrics.visualViewportWidth) ??
173
+ toPositiveNumber(metrics.innerWidth) ??
174
+ toPositiveNumber(metrics.documentElementClientWidth);
175
+ const height =
176
+ toPositiveNumber(metrics.visualViewportHeight) ??
177
+ toPositiveNumber(metrics.innerHeight) ??
178
+ toPositiveNumber(metrics.documentElementClientHeight);
179
+
180
+ return width && height ? { width, height } : null;
181
+ }
182
+
183
+ function screenshotState(
184
+ screenshot: Buffer,
185
+ viewport: ImageDimensions,
186
+ ): {
157
187
  screenshot: Buffer;
158
188
  dimensions: ImageDimensions;
159
189
  scale: CoordinateScale;
160
- }> {
161
- const viewport = page.viewportSize();
162
- if (!viewport) {
163
- throw new Error("Viewport size not found");
164
- }
165
-
166
- const screenshot = await page.screenshot({
167
- fullPage: false,
168
- scale: "css",
169
- timeout: 10000,
170
- });
190
+ } {
171
191
  const dimensions = readPngDimensions(screenshot);
172
192
  return {
173
193
  screenshot,
@@ -181,6 +201,25 @@ async function takeViewportScreenshot(page: Page): Promise<{
181
201
  };
182
202
  }
183
203
 
204
+ async function takeViewportScreenshot(page: Page): Promise<{
205
+ screenshot: Buffer;
206
+ dimensions: ImageDimensions;
207
+ scale: CoordinateScale;
208
+ }> {
209
+ const viewport =
210
+ page.viewportSize() ?? (await getViewportFromPage(page).catch(() => null));
211
+ if (!viewport) {
212
+ throw new Error("Viewport size not found");
213
+ }
214
+
215
+ const screenshot = await page.screenshot({
216
+ fullPage: false,
217
+ scale: "css",
218
+ timeout: 10000,
219
+ });
220
+ return screenshotState(screenshot, viewport);
221
+ }
222
+
184
223
  async function executeBrowserAction(
185
224
  page: Page,
186
225
  action: BrowserAction,