From 03700116261110177f15bddccbbb02401f1c5a4e Mon Sep 17 00:00:00 2001
From: matt cameron <matt@mattcameron.me>
Date: Sun, 10 Nov 2024 19:02:40 +1100
Subject: [PATCH] feat: add Beautiful Soup plugin

#884
---
 .../action/v1/beautifulsoup_action.py         | 113 ++++++++++++++++++
 tracardi/service/setup/setup_plugins.py       |  12 +-
 2 files changed, 124 insertions(+), 1 deletion(-)
 create mode 100644 tracardi/process_engine/action/v1/beautifulsoup_action.py

diff --git a/tracardi/process_engine/action/v1/beautifulsoup_action.py b/tracardi/process_engine/action/v1/beautifulsoup_action.py
new file mode 100644
index 00000000..386fefec
--- /dev/null
+++ b/tracardi/process_engine/action/v1/beautifulsoup_action.py
@@ -0,0 +1,113 @@
+from lxml import etree
+from tracardi.service.plugin.domain.register import Plugin, Spec, MetaData, Documentation, PortDoc, Form, FormGroup, \
+    FormField, FormComponent
+from tracardi.service.plugin.domain.result import Result
+from tracardi.service.plugin.domain.config import PluginConfig
+from tracardi.service.plugin.runner import ActionRunner
+from pydantic import field_validator
+from tracardi.domain.profile import Profile
+from tracardi.service.tracardi_http_client import HttpClient
+
+from bs4 import BeautifulSoup
+
+class Configuration(PluginConfig):
+    html: str
+    method: str = "get_text"
+
+    @field_validator('html')
+    @classmethod
+    def html_must_not_be_empty(cls, value):
+        if value.strip() == "":
+            raise ValueError("HTML must not be empty.")
+        return value
+
+def validate(config: dict):
+    return Configuration(**config)
+
+class BeautifulSoupAction(ActionRunner):
+    
+    config: Configuration
+    
+    async def set_up(self, init):
+        self.config = validate(init)
+
+    async def run(self, payload: dict, in_edge=None) -> Result:
+
+        dot = self._get_dot_accessor(payload)
+        
+        try:
+            
+            html=dot[self.config.html]
+            soup = BeautifulSoup(html, 'html.parser')
+            
+            text = ''
+            
+            if self.config.method == "get_text":
+                text = soup.get_text()
+            else:
+                raise ValueError(f"Unsupported method: {self.config.method}")
+            
+            return Result(port='result', value={'text':text})
+
+        except Exception as e:
+            return Result(value={"message": str(e)}, port="error")  
+
+def register() -> Plugin:
+    return Plugin(
+        start=False,
+        spec=Spec(
+            module=__name__,
+            className=BeautifulSoupAction.__name__,
+            inputs=["payload"],
+            outputs=["result", "error"],
+            version="1.0.4",  
+            init={
+                "html": "",  
+            },
+            form=Form(groups=[
+                FormGroup(
+                    name="Beautiful Soup configuration",
+                    fields=[
+                        FormField(
+                            id="html",
+                            name="HTML",
+                            description="The HTML to be converted.",
+                            component=FormComponent(type="dotPath", props={
+                                "label": "HTML"
+                            }),
+                        ),
+                        FormField(
+                            id="method",
+                            name="Method",
+                            description="The BeautifulSoup method to apply to the HTML",
+                            component=FormComponent(type="select", props={
+                                "label": "Method",
+                                "items": {
+                                    "get_text": "get_text",
+                                }
+                            })
+                        ),
+                    ]),
+            ]),
+            license="MIT",
+            author="Matt Cameron",
+            manual="beautifulsoup",  
+
+        ),
+        metadata=MetaData(
+            name='BeautifulSoup',
+            desc='Converts HTML to text.',
+            icon='BeautifulSoup',
+            group=['Data Processing'],
+            documentation=Documentation(
+                inputs={
+                    "payload": PortDoc(desc="This port takes payload object.")
+                },
+                outputs={
+                    "result": PortDoc(desc="Returns response from Sitemap service."),
+                    "error": PortDoc(desc="Returns error message if plugin fails.")
+                }
+            )
+        )
+    )
+
diff --git a/tracardi/service/setup/setup_plugins.py b/tracardi/service/setup/setup_plugins.py
index 7fa60548..307dcd1f 100644
--- a/tracardi/service/setup/setup_plugins.py
+++ b/tracardi/service/setup/setup_plugins.py
@@ -801,7 +801,17 @@
             }, 
             resource=None),
     ),
-    
+
+    "tracardi.process_engine.action.v1.beautifulsoup_action": PluginMetadata(
+        test=PluginTest(
+            init={
+                'HTML': '',
+                'method': 'get_text',
+                
+            }, 
+            resource=None),
+    ),
+        
 }
 
 if License.has_service(SCHEDULER):