dp-cli 0.1.0__py3-none-any.whl

dp_cli/commands/misc.py

@@ -0,0 +1,136 @@
+# -*- coding:utf-8 -*-
+"""杂项命令: resize / maximize / state-save / state-load / config-set"""
+import json
+from pathlib import Path
+
+import click
+
+from dp_cli.output import ok, error
+from dp_cli.commands._utils import session_option, _get_page
+
+
+def register(cli):
+
+    @cli.command('resize')
+    @click.argument('width', type=int)
+    @click.argument('height', type=int)
+    @session_option
+    def resize(width, height, session):
+        """调整浏览器窗口大小。
+
+        \b
+        示例:
+          dp resize 1920 1080
+          dp resize 375 812
+        """
+        page = _get_page(session)
+        try:
+            page.set.window.size(width, height)
+            ok({'width': width, 'height': height}, msg='窗口大小已调整')
+        except Exception as e:
+            error(f'调整窗口大小失败', code='RESIZE_FAILED', detail=str(e))
+
+    @cli.command('maximize')
+    @session_option
+    def maximize(session):
+        """最大化浏览器窗口。"""
+        page = _get_page(session)
+        try:
+            page.set.window.max()
+            ok(msg='窗口已最大化')
+        except Exception as e:
+            error(f'最大化失败', code='WINDOW_FAILED', detail=str(e))
+
+    @cli.command('state-save')
+    @click.argument('filename', default='state.json')
+    @session_option
+    def state_save(filename, session):
+        """保存浏览器状态（Cookie + localStorage）到文件。
+
+        \b
+        示例:
+          dp state-save
+          dp state-save auth.json
+        """
+        page = _get_page(session)
+        try:
+            state = {
+                'cookies': page.cookies(all_domains=True).as_dict(),
+                'url': page.url,
+            }
+            try:
+                ls = page.run_js(
+                    'return JSON.stringify(Object.fromEntries(Object.entries(localStorage)))',
+                    as_expr=True)
+                state['localStorage'] = json.loads(ls) if isinstance(ls, str) else ls or {}
+            except Exception:
+                state['localStorage'] = {}
+            Path(filename).write_text(
+                json.dumps(state, ensure_ascii=False, indent=2), encoding='utf-8')
+            ok({'filename': str(Path(filename).absolute()),
+                'cookies_count': len(state['cookies'])}, msg='状态已保存')
+        except Exception as e:
+            error(f'保存状态失败', code='STATE_FAILED', detail=str(e))
+
+    @cli.command('state-load')
+    @click.argument('filename', default='state.json')
+    @session_option
+    def state_load(filename, session):
+        """从文件加载浏览器状态（Cookie + localStorage）。
+
+        \b
+        示例:
+          dp state-load
+          dp state-load auth.json
+        """
+        page = _get_page(session)
+        try:
+            state = json.loads(Path(filename).read_text(encoding='utf-8'))
+            if 'cookies' in state:
+                for name, value in state['cookies'].items():
+                    try:
+                        page.set.cookies({'name': name, 'value': value})
+                    except Exception:
+                        pass
+            if 'localStorage' in state and state['localStorage']:
+                for k, v in state['localStorage'].items():
+                    try:
+                        page.run_js(f'localStorage.setItem({json.dumps(k)}, {json.dumps(v)})',
+                                    as_expr=True)
+                    except Exception:
+                        pass
+            ok({'filename': filename,
+                'cookies_restored': len(state.get('cookies', {}))}, msg='状态已加载')
+        except FileNotFoundError:
+            error(f'状态文件不存在: {filename}', code='FILE_NOT_FOUND')
+        except Exception as e:
+            error(f'加载状态失败', code='STATE_FAILED', detail=str(e))
+
+    @cli.command('config-set')
+    @click.option('-p', '--browser-path', default=None, help='设置浏览器路径')
+    @click.option('-u', '--user-path', default=None, help='设置用户数据路径')
+    @click.option('-c', '--copy-config', is_flag=True, help='复制默认配置文件到当前目录')
+    def config_set(browser_path, user_path, copy_config):
+        """修改 DrissionPage 配置文件。
+
+        \b
+        示例:
+          dp config-set --browser-path /usr/bin/google-chrome
+          dp config-set --user-path /home/user/.chrome-data
+          dp config-set --copy-config
+        """
+        from DrissionPage._configs.chromium_options import ChromiumOptions
+        from DrissionPage._functions.tools import configs_to_here
+
+        if copy_config:
+            configs_to_here()
+            ok(msg='配置文件已复制到当前目录')
+
+        if browser_path or user_path:
+            co = ChromiumOptions()
+            if browser_path:
+                co.set_browser_path(browser_path)
+            if user_path:
+                co.set_user_data_path(user_path)
+            co.save()
+            ok(msg='配置已保存')

dp_cli/commands/network.py

@@ -0,0 +1,169 @@
+# -*- coding:utf-8 -*-
+"""网络命令: listen / listen-stop / http-get / http-post"""
+import json
+from pathlib import Path
+
+import click
+
+from dp_cli.output import ok, error
+from dp_cli.commands._utils import session_option, _get_page
+
+
+def register(cli):
+
+    @cli.command('listen')
+    @session_option
+    @click.option('--filter', 'url_filter', default=None, help='URL 过滤关键字，如 "api/user"')
+    @click.option('--count', default=10, help='最多捕获请求数', show_default=True)
+    @click.option('--timeout', default=30, help='监听超时秒数', show_default=True)
+    @click.option('--method', default=None, help='过滤请求方法，如 POST')
+    def listen(session, url_filter, count, timeout, method):
+        """监听网络请求（抓包）。先 listen，再执行触发操作，最后 listen-stop 读取结果。
+
+        \b
+        这是 DrissionPage 的核心独特能力，可精确捕获 XHR/Fetch/图片等任意请求。
+
+        \b
+        示例:
+          dp listen --filter "api/login"
+          dp listen --count 5 --timeout 10
+        """
+        page = _get_page(session)
+        try:
+            targets = url_filter if url_filter else None
+            page.listen.start(targets=targets, method=method)
+            ok(msg=f'已开始监听，过滤: {url_filter or "全部"}')
+        except Exception as e:
+            error(f'启动监听失败', code='LISTEN_FAILED', detail=str(e))
+
+    @cli.command('listen-stop')
+    @session_option
+    @click.option('--count', default=1, help='等待数据包数量', show_default=True)
+    @click.option('--timeout', default=10, help='等待数据超时秒数', show_default=True)
+    def listen_stop(session, count, timeout):
+        """停止监听并获取捕获的网络请求数据。"""
+        page = _get_page(session)
+        try:
+            packets = page.listen.wait(count=count, timeout=timeout, fit_count=False)
+            results = []
+            if packets:
+                pkts = packets if isinstance(packets, list) else [packets]
+                for pkt in pkts:
+                    try:
+                        item = {
+                            'url': pkt.url,
+                            'method': pkt.method,
+                            'status': pkt.response.status if pkt.response else None,
+                            'type': pkt.resourceType,
+                        }
+                        try:
+                            item['response_body'] = pkt.response.body
+                        except Exception:
+                            pass
+                        results.append(item)
+                    except Exception:
+                        continue
+            page.listen.stop()
+            ok({'packets': results, 'count': len(results)})
+        except Exception as e:
+            error(f'获取监听数据失败', code='LISTEN_FAILED', detail=str(e))
+
+    @cli.command('http-get')
+    @click.argument('url')
+    @click.option('--headers', default=None, help='JSON 格式的 Headers')
+    @click.option('--proxy', default=None, help='代理地址')
+    @click.option('--timeout', default=30, help='超时秒数', show_default=True)
+    @click.option('--output', default=None, help='响应体保存路径')
+    def http_get(url, headers, proxy, timeout, output):
+        """发送 HTTP GET 请求（不启动浏览器，高效爬虫模式）。
+
+        \b
+        示例:
+          dp http-get https://api.example.com/users
+          dp http-get https://example.com --output page.html
+          dp http-get https://api.example.com --headers '{"Authorization":"Bearer xxx"}'
+        """
+        try:
+            from DrissionPage import SessionPage
+            from DrissionPage._configs.session_options import SessionOptions
+
+            so = SessionOptions()
+            if proxy:
+                so.set_proxies(http=proxy, https=proxy)
+            if headers:
+                so.set_headers(json.loads(headers))
+
+            page = SessionPage(session_or_options=so)
+            page.get(url, timeout=timeout)
+
+            result = {
+                'url': page.url,
+                'status_code': page.response.status_code if page.response else None,
+            }
+            try:
+                result['body'] = page.response.json()
+            except Exception:
+                try:
+                    body_text = page.response.text[:3000]
+                except Exception:
+                    body_text = '<binary>'
+                if output:
+                    Path(output).write_text(page.response.text, encoding='utf-8')
+                    result['saved_to'] = output
+                else:
+                    result['body'] = body_text
+
+            ok(result)
+        except Exception as e:
+            error(f'HTTP GET 失败', code='HTTP_FAILED', detail=str(e))
+
+    @cli.command('http-post')
+    @click.argument('url')
+    @click.option('--data', default=None, help='JSON 格式的请求体')
+    @click.option('--form', default=None, help='JSON 格式的表单数据')
+    @click.option('--headers', default=None, help='JSON 格式的 Headers')
+    @click.option('--proxy', default=None, help='代理地址')
+    @click.option('--timeout', default=30, help='超时秒数', show_default=True)
+    def http_post(url, data, form, headers, proxy, timeout):
+        """发送 HTTP POST 请求（不启动浏览器）。
+
+        \b
+        示例:
+          dp http-post https://api.example.com/login --data '{"user":"admin","pass":"123"}'
+          dp http-post https://example.com/form --form '{"field":"value"}'
+        """
+        try:
+            from DrissionPage import SessionPage
+            from DrissionPage._configs.session_options import SessionOptions
+
+            so = SessionOptions()
+            if proxy:
+                so.set_proxies(http=proxy, https=proxy)
+            if headers:
+                so.set_headers(json.loads(headers))
+
+            page = SessionPage(session_or_options=so)
+
+            kwargs = {'timeout': timeout}
+            if data:
+                kwargs['json'] = json.loads(data)
+            elif form:
+                kwargs['data'] = json.loads(form)
+
+            page.post(url, **kwargs)
+
+            result = {
+                'url': page.url,
+                'status_code': page.response.status_code if page.response else None,
+            }
+            try:
+                result['body'] = page.response.json()
+            except Exception:
+                try:
+                    result['body'] = page.response.text[:3000]
+                except Exception:
+                    result['body'] = '<binary>'
+
+            ok(result)
+        except Exception as e:
+            error(f'HTTP POST 失败', code='HTTP_FAILED', detail=str(e))

dp_cli/commands/page.py

@@ -0,0 +1,204 @@
+# -*- coding:utf-8 -*-
+"""页面操作命令: screenshot / pdf / eval / add-init-js / dialog-accept / dialog-dismiss / wait"""
+import json
+from pathlib import Path
+from time import perf_counter, sleep as _sleep
+
+import click
+
+from dp_cli.output import ok, error, format_page_info
+from dp_cli.commands._utils import session_option, _get_page
+
+
+def register(cli):
+
+    @cli.command('screenshot')
+    @session_option
+    @click.option('--locator', default=None, help='截图特定元素')
+    @click.option('--filename', default=None, help='保存路径')
+    @click.option('--full-page', is_flag=True, help='截取完整页面（包括视口外）')
+    @click.option('--format', 'fmt', type=click.Choice(['png', 'jpg', 'jpeg']),
+                  default='png', show_default=True)
+    def cmd_screenshot(session, locator, filename, full_page, fmt):
+        """截图。DrissionPage 支持完整页面截图（含视口外内容）。
+
+        \b
+        示例:
+          dp screenshot
+          dp screenshot --filename page.png
+          dp screenshot --full-page
+          dp screenshot --locator "#header"
+        """
+        page = _get_page(session)
+        try:
+            if not filename:
+                from datetime import datetime
+                ts = datetime.now().strftime('%Y%m%d_%H%M%S')
+                filename = f'screenshot_{ts}.{fmt}'
+            if locator:
+                ele = page.ele(locator)
+                if not ele or ele.__class__.__name__ == 'NoneElement':
+                    error(f'未找到元素: {locator}', code='ELEMENT_NOT_FOUND')
+                    return
+                ele.get_screenshot(path=filename)
+            else:
+                page.get_screenshot(path=filename, full_page=full_page)
+            ok({'filename': str(Path(filename).absolute())}, msg='截图已保存')
+        except Exception as e:
+            error(f'截图失败', code='SCREENSHOT_FAILED', detail=str(e))
+
+    @cli.command('pdf')
+    @session_option
+    @click.option('--filename', default=None, help='保存路径')
+    def cmd_pdf(session, filename):
+        """将当前页面保存为 PDF。
+
+        \b
+        示例:
+          dp pdf
+          dp pdf --filename output.pdf
+        """
+        page = _get_page(session)
+        try:
+            if not filename:
+                from datetime import datetime
+                ts = datetime.now().strftime('%Y%m%d_%H%M%S')
+                filename = f'page_{ts}.pdf'
+            result = page.run_cdp('Page.printToPDF', transferMode='ReturnAsBase64')
+            import base64
+            Path(filename).write_bytes(base64.b64decode(result['data']))
+            ok({'filename': str(Path(filename).absolute())}, msg='PDF 已保存')
+        except Exception as e:
+            error(f'保存 PDF 失败', code='PDF_FAILED', detail=str(e))
+
+    @cli.command('eval')
+    @click.argument('script')
+    @session_option
+    @click.option('--locator', default=None, help='在特定元素上执行（this 指向该元素）')
+    @click.option('--timeout', default=30, help='执行超时秒数', show_default=True)
+    def cmd_eval(script, session, locator, timeout):
+        """执行 JavaScript 并返回结果。
+
+        \b
+        示例:
+          dp eval "document.title"
+          dp eval "window.innerWidth"
+          dp eval "el => el.textContent" --locator "#header"
+          dp eval "return Array.from(document.links).map(l=>l.href)"
+        """
+        page = _get_page(session)
+        try:
+            if locator:
+                ele = page.ele(locator, timeout=timeout)
+                if not ele or ele.__class__.__name__ == 'NoneElement':
+                    error(f'未找到元素: {locator}', code='ELEMENT_NOT_FOUND')
+                    return
+                if script.strip().startswith(('el =>', 'el=>', 'function')):
+                    result = ele.run_js(f'return ({script})(this)', timeout=timeout)
+                else:
+                    result = ele.run_js(script, as_expr=True, timeout=timeout)
+            else:
+                if script.strip().startswith(('return ', 'function')):
+                    result = page.run_js(script, timeout=timeout)
+                else:
+                    result = page.run_js(script, as_expr=True, timeout=timeout)
+            ok({'result': result})
+        except Exception as e:
+            error(f'JavaScript 执行失败', code='JS_FAILED', detail=str(e))
+
+    @cli.command('add-init-js')
+    @click.argument('script')
+    @session_option
+    def add_init_js(script, session):
+        """添加在每个新页面加载前执行的 JS 脚本（反检测/环境修改等）。
+
+        \b
+        示例:
+          dp add-init-js "Object.defineProperty(navigator,'webdriver',{get:()=>undefined})"
+          dp add-init-js "window.__dp_cli = true"
+        """
+        page = _get_page(session)
+        try:
+            js_id = page.add_init_js(script)
+            ok({'id': js_id, 'script': script[:100]}, msg='初始化脚本已添加')
+        except Exception as e:
+            error(f'添加初始化脚本失败', code='INIT_JS_FAILED', detail=str(e))
+
+    @cli.command('dialog-accept')
+    @click.argument('text', required=False)
+    @session_option
+    def dialog_accept(text, session):
+        """接受（确认）对话框。可选传入 prompt 输入内容。
+
+        \b
+        示例:
+          dp dialog-accept
+          dp dialog-accept "确认文本"
+        """
+        page = _get_page(session)
+        try:
+            page.handle_alert(accept=True, send=text)
+            ok(msg='对话框已接受')
+        except Exception as e:
+            error(f'处理对话框失败', code='DIALOG_FAILED', detail=str(e))
+
+    @cli.command('dialog-dismiss')
+    @session_option
+    def dialog_dismiss(session):
+        """取消对话框。"""
+        page = _get_page(session)
+        try:
+            page.handle_alert(accept=False)
+            ok(msg='对话框已取消')
+        except Exception as e:
+            error(f'处理对话框失败', code='DIALOG_FAILED', detail=str(e))
+
+    @cli.command('wait')
+    @session_option
+    @click.option('--url', default=None, help='等待 URL 变为此值（支持子串匹配）')
+    @click.option('--locator', default=None, help='等待元素出现')
+    @click.option('--locator-gone', default=None, help='等待元素消失')
+    @click.option('--text', default=None, help='等待页面包含指定文本')
+    @click.option('--loaded', is_flag=True, help='等待页面加载完成')
+    @click.option('--timeout', default=30, help='超时秒数', show_default=True)
+    def wait(session, url, locator, locator_gone, text, loaded, timeout):
+        """等待条件满足。
+
+        \b
+        示例:
+          dp wait --loaded
+          dp wait --locator "#result"
+          dp wait --url "success"
+          dp wait --text "操作成功"
+          dp wait --locator-gone "css:.loading"
+        """
+        page = _get_page(session)
+        try:
+            if loaded:
+                page.wait.doc_loaded()
+                ok(format_page_info(page), msg='页面已加载')
+            elif url:
+                page.wait.url_change(url, timeout=timeout)
+                ok(format_page_info(page), msg='URL 已变化')
+            elif locator:
+                ele = page.wait.ele_displayed(locator, timeout=timeout)
+                ok({'locator': locator, 'found': bool(ele)}, msg='元素已出现')
+            elif locator_gone:
+                page.wait.ele_hidden(locator_gone, timeout=timeout)
+                ok({'locator': locator_gone}, msg='元素已消失')
+            elif text:
+                end = perf_counter() + timeout
+                found = False
+                while perf_counter() < end:
+                    if text in page.html:
+                        found = True
+                        break
+                    _sleep(0.3)
+                if found:
+                    ok({'text': text}, msg='文本已出现')
+                else:
+                    error(f'等待超时：文本未出现 "{text}"', code='WAIT_TIMEOUT')
+            else:
+                error('请至少指定一个等待条件', code='INVALID_ARGS')
+        except Exception as e:
+            error(f'等待失败', code='WAIT_FAILED', detail=str(e))